The Routledge Handbook of Second Language Acquisition and Psycholinguistics

THE ROUTLEDGE HANDBOOK OF
SECOND LANGUAGE ACQUISITION AND

PSYCHOLINGUISTICS
The Routledge Handbook of Second Language Acquisition and Psycholinguistics provides a comprehensive
survey of the latest research at the intersection of linguistics, cognitive psychology, and applied
linguistics, for those seeking to understand the mental architecture and processes that shape the
acquisition of additional languages. The handbook represents the full complexity of second language
acquisition across the lifespan, spanning childhood bilinguals and adult L2 learners, and is inclusive of
heritage languages, early bilingualism and multilingualism, and language attrition.
An authoritative selection of diverse, global, leading psycholinguists synthesize the latest research
to provide a thorough overview in a single volume and set the agenda for the future. The volume is
organized into fve key parts for ease of use: psycholinguistics across the lifespan; methods; theoretical
perspectives; the psycholinguistics of learning; and transdisciplinary perspectives.
This handbook will be an indispensable resource for scholars and students of psycholinguistics,
second language acquisition, applied linguistics, bilingualism, cognitive science, psychology, and
research methodology.
Aline Godfroid is an Associate Professor of Second Language Studies and TESOL at Michigan
State University, USA. She is the winner of the 2019 TESOL Award for Distinguished Research.
Holger Hopp is a Professor of English Linguistics at Technische Universität Braunschweig (Institute

of Technology), Germany. He is an Executive Editor of the journal Linguistic Approaches to Bilingualism.
ROUTLEDGE HANDBOOKS IN SECOND
LANGUAGE ACQUISITION
Susan M. Gass and Alison Mackey, Series Editors
Kimberly L. Geeslin, Associate Editor
The Routledge Handbooks in Second Language Acquisition are a comprehensive, must-have

survey of this core sub-discipline of applied linguistics. With a truly global reach and featuring
diverse contributing voices, each handbook provides an overview of both the fundamentals and new
directions for each topic.
The Routledge Handbook of Second Language Acquisition and Pragmatics

Edited by Naoko Taguchi
The Routledge Handbook of Second Language Acquisition and Corpora

Edited by Nicole Tracy-Ventura and Magali Paquot
The Routledge Handbook of Second Language Acquisition and Language Testing

Edited by Paula Winke and Tineke Brunfaut
The Routledge Handbook of Second Language Acquisition and Technology

Edited by Nicole Ziegler and Marta González-Lloret
The Routledge Handbook of Second Language Acquisition and Writing

Edited by Rosa M. Manchón and Charlene Polio
The Routledge Handbook of Second Language Acquisition and Speaking

Edited by Tracey M. Derwing, Murray J. Munro, and Ron I. Thomson
The Routledge Handbook of Second Language Acquisition and Sociolinguistics

Edited by Kimberly Geeslin
The Routledge Handbook of Second Language Acquisition and Individual

Differences
Edited by Shaofeng Li, Phil Hiver, and Mostafa Papi
The Routledge Handbook of Second Language Acquisition and Psycholinguistics

Edited by Aline Godfroid and Holger Hopp
For more information about this series, please visit: www.routledge.com/The-Routledge-Handbooks-

in-Second-Language-Acquisition/book-series/RHSLA
THE ROUTLEDGE HANDBOOK
OF SECOND LANGUAGE
ACQUISITION AND
PSYCHOLINGUISTICS
Edited by Aline Godfroid and Holger Hopp

Cover image: Getty Images
First published 2023
by Routledge
605 Third Avenue, New York, NY 10158
and by Routledge
4 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN
Routledge is an imprint of the Taylor & Francis Group, an informa business
© 2023 selection and editorial matter, Aline Godfroid and Holger Hopp;
individual chapters, the contributors
The right of Aline Godfroid and Holger Hopp to be identifed as the
authors of the editorial material, and of the authors for their individual
chapters, has been asserted in accordance with sections 77 and 78 of the
Copyright, Designs and Patents Act 1988.
All rights reserved. No part of this book may be reprinted or reproduced or
utilised in any form or by any electronic, mechanical, or other means, now
known or hereafter invented, including photocopying and recording, or in
any information storage or retrieval system, without permission in writing
from the publishers.
Trademark notice: Product or corporate names may be trademarks or registered
trademarks, and are used only for identifcation and explanation without
intent to infringe.
Library of Congress Cataloging-in-Publication Data
Names: Godfroid, Aline, editor. | Hopp, Holger, editor.
Title: The Routledge handbook of second language acquisition and
psycholinguistics/edited by Aline Godfroid and Holger Hopp.
Description: New York, NY: Routledge, 2023. | Series: The Routledge
Handbooks in Second Language Acquisition | Includes bibliographical
references and index.
Identifers: LCCN 2022022966 (print) | LCCN 2022022967 (ebook)
Subjects: LCSH: Second language acquisition—Psychological aspects. |
Psycholinguistics. | LCGFT: Essays.
Classifcation: LCC P118.2 .R68524 2023 (print) | LCC P118.2 (ebook) |
DDC 401/.93019—dc23/eng/20220606
LC record available at https://lccn.loc.gov/2022022966
LC ebook record available at https://lccn.loc.gov/2022022967
ISBN: 978-0-367-89376-7 (hbk)
ISBN: 978-1-032-37293-8 (pbk)
ISBN: 978-1-003-01887-2 (ebk)
DOI: 10.4324/9781003018872
Typeset in Bembo
by Apex CoVantage, LLC
CONTENTS
List of fgures ix
List of tables x
List of contributors xi
1 Introduction: second language acquisition and psycholinguistics 1

Holger Hopp and Aline Godfroid
SECTION I
Psycholinguistics across the lifespan 11
2 The psycholinguistics of early bilingualism 13

Christine E. Potter and Casey Lew-Williams
3 The psycholinguistics of child second language acquisition 26

Sarah Schimke
4 The psycholinguistics of adult second language acquisition 38

Carrie N. Jackson
5 The psycholinguistics of L3/Ln acquisition 50

Jorge González Alonso and Jason Rothman
6 The psycholinguistics of frst language attrition 61

Monika S. Schmid, Concepción Soto, and Julia Heimann
7 The psycholinguistics of heritage language acquisition 73

Silvina Montrul
v
Contents
8 Learning a new language in older age: psycholinguistic and cross-

disciplinary approaches 85
Jessica G. Cox and Cristina Sanz
9 Synthesis: psycholinguistics across the lifespan 97

Theres Grüter
SECTION II
Research methods in second language psycholinguistics 109
10 Studying second language comprehension 111

Irina Elgort and Paul Warren
11 Studying second language production 124

Sarah Bernolet
12 Studying multimodal language processing: the integration of speech

and gestures 137
Marianne Gullberg
13 Conducting reaction time research in second language psycholinguistics 150

Phillip Hamrick
14 Using corpora in research on second language psycholinguistics 164

Sandra C. Deshors and Stefan Th. Gries
15 Synthesis: research methods in second language psycholinguistics 178

Nan Jiang
SECTION III
Theoretical perspectives on second language psycholinguistics 189
16 Word and multi-word processing: memory-based and linguistic approaches 191

Kira Gor
17 Word and multiword processing: cognitive approaches 203

Kathy Conklin and Rüdiger Thul
18 Sentence processing in a second language: linguistic approaches 216

Holger Hopp
19 Sentence processing: cognitive approaches to second language

morphosyntactic and morphological processing 229
Nuria Sagarra
vi
Contents
20 Discourse processing 242

Saveria Colonna
21 Code-switching 255
Janet G. van Hell
22 Synthesis: transdisciplinary approaches to understanding variability in

bilingual processing and use 268
Paola E. Dussias
SECTION IV
The psycholinguistics of learning 279
23 Implicit learning and second language acquisition: a cognitive

psychology perspective 281
John N. Williams and Patrick Rebuschat
24 Hypotheses about the interface between explicit and implicit

knowledge in second language acquisition 294
Aline Godfroid
25 Automatization and practice 308

Yuichi Suzuki
26 Declarative and procedural memory in second language learning:

psycholinguistic considerations 322
Kara Morgan-Short and Michael T. Ullman
27 The psycholinguistics of second language interaction 335

Michael H. Long
28 Working memory and second language learning: a critical and

synthetic review 348
Shaofeng Li
29 Synthesis: the psycholinguistics of learning 361

Brian MacWhinney
SECTION V
Transdisciplinary perspectives: contexts and future directions 371
30 The infuence of learning contexts on the psycholinguistics of second

language learning 373
Ronald P. Leow
vii
Contents
31 Psycholinguistics for the language classroom 387

Rebecca Sachs, Melissa Baralt, and Laura Gurzynski-Weiss
32 The psycholinguistics of second language writing 400

Rosa M. Manchón
33 The psycholinguistics of second language assessment 413

Rob Schoonen
34 Cognitive efects of bilingualism 426

Gregory J. Poarch
35 Synthesis: transdisciplinary perspectives on second language

acquisition: exploration versus explanation 439
Jan Hulstijn
Index 450
viii
FIGURES
11.1 Examples of possible stimuli in (a) regular picture naming experiments,

(b) picture-word interference experiments, and (c) picture-picture interference
experiments (pictures taken from the MultiPic database [Duñabeitia et al., 2018]). 126
11.2 Target picture eliciting active and passive transitive sentences (used in Bernolet
et al., 2009, and reprinted with permission from Elsevier). 132
13.1 Illustration of timing errors that normally occur (bottom) versus the ideal timing
that most experimenters have in mind (top), inspired by Plant and Quinlan (2013). 152
14.1 Concordance lines for particles used in phrasal verb constructions (in the International
Corpus of Learner English). 168
17.1 Panel A. Efect of additional occurrences on low frequency ”gnat,” higher
frequency “rat,” and even higher frequency “cat,” demonstrating that exposure
to a lexical item raises its RLA in proportion to its distance from the activation
threshold/asymptote. Panel B. Efect of additional occurrences on monolingual
high- and low-frequency words and bilingual L1 and L2, high- and low-frequency
words, demonstrating that increased exposure in bilinguals has a greater
efect on RLA. 206
17.2 Relationship between RLA and RT distribution. Each solid straight line
represents the time evolution of the activation level for a word starting from a
specifc RLA (open upright triangle) until the activation threshold is reached
(open upside-down triangle). 210
24.1 A three-stage representation of implicit, explicit, and automatized explicit
knowledge (Cleeremans, 2008) with representational redescription (Karmilof-
Smith, 1992) and analysis (e.g., Bialystok, 2001) marking the transition between
diferent stages. Cleeremans (2008, Figure 1) reprinted and modifed with
permission from the author and from Elsevier. 297
24.2 Main areas of research on the interface and the reverse interface hypothesis. 298
25.1 Power law curve: proceduralization and automatization. 310
25.2 Performance changes (losses and gains) over four practice sessions under
shorter-spaced and longer-spaced conditions. 314
34.1 Flanker task (panel A) and Simon task (panel B) congruent and incongruent
conditions. Stimulus displayed on screen requires indicated button press. 430
ix
TABLES
13.1 A sample of benchmarked timing latencies for typical response devices used in
psycholinguistic research. 154
13.2 Response device options for L2 psycholinguistics research and their
chronometric-relevant properties. 155
13.3 Software options for L2 psycholinguistics research and their chronometric-
relevant properties. 156
13.4 Suggested details to report on technology used in chronometric research in L2
psycholinguistics. 160
25.1 Key topics and research questions in instructed L2 research on practice and
automatization. 311
33.1 An example Q-matrix for a seven-item test and three presumed subskills
(after Lee & Sawaki, 2009b). 420
35.1 Timeline and schools (paradigms) in linguistics and psychology. 441
x
CONTRIBUTORS
Melissa Baralt is an applied psycholinguist and an associate professor of linguistics at Florida Inter-
national University (USA). She specializes in second language acquisition, bilingualism, bilingual
language development in children, and language teacher training.
Sarah Bernolet is an associate professor in Dutch linguistics at the University of Antwerp (Bel-
gium). She investigates sentence production patterns in order to determine which memory rep-
resentations, procedures, and memory systems are involved in the production and comprehension
of language. She is interested in sentence production by experienced bi-/multilinguals, but also by
speakers and writers who have not fully mastered a language.
Saveria Colonna is a professor at the University of Paris 8 and member of the Structures Formelles du
Langage laboratory (CNRS & Paris 8, France). A psycholinguist with an interest in discourse compre-
hension, her research focuses mainly on the acquisition of referential cohesion markers in a frst language
and in a second language, using experimental methods such as the visual world eye-tracking paradigm.
Kathy Conklin is a full professor of psycholinguistics at the Centre for Applied Linguistics in the
School of English at the University of Nottingham (United Kingdom). A major focus of her research
is on the application of psycholinguistic methods, in particular eye-tracking, to the exploration of
lexical and formulaic language processing in a frst and second language. She is co-author, along
with Ana Pellicer-Sánchez and Gareth Carrol, of the seminal book Eye-Tracking: A Guide for Applied
Linguistics Research (CUP, 2018).
Jessica G. Cox is an associate professor of Spanish and linguistics at the Franklin and Marshall Col-
lege (USA). She researches relationships between language and cognition in multilingual individu-
als, including older adult language learners, aptitude for learning new languages, and how bilingual
experiences afect other areas of cognition. Her publications can be found in journals such as Studies
of Second Language Acquisition, Bilingualism: Language and Cognition, and Language Teaching.
Sandra C. Deshors is an associate professor in the Department of Linguistics, Languages, and Cul-
tures at Michigan State University (USA) where she is a core faculty in the Second Language Studies
Ph.D. program and the Master of Arts in TESOL program. Her research specializes in quantitative
corpus-based approaches to learner language.
xi
Contributors
Paola E. Dussias is a professor of Spanish, linguistics, and psychology at Pennsylvania State Univer-
sity (USA), where she and her students examine the variables that modulate how bilingual speakers
process sentences in each of their languages. Her work, using behavioral methods and the recoding
of brain activity, has been supported by grants from the National Science Foundation (NSF) and the
National Institutes of Health. She and Judith F. Kroll are the principal investigators on an NSF PIRE
grant to translate fndings from cognitive and brain science to language learning environments.
Irina Elgort is an associate professor in higher education at Victoria University of Wellington (New
Zealand). Her research interests include explicit and implicit vocabulary learning and processing; lexi-
cal and semantic representations and their development; and reading. She uses research paradigms from
applied linguistics, cognitive psychology, and education to better understand, predict, and infuence
learning. Her research has been published in Applied Linguistics, Bilingualism: Language and Cognition,
Language, Cognition and Neuroscience, Language Learning, and Studies in Second Language Acquisition.
Aline Godfroid is an associate professor in Second Language Studies and TESOL at Michigan State
University (USA), where she teaches courses in psycholinguistics, vocabulary learning and teaching,
quantitative research methods, and eye-tracking methodology. She is the co-director of the Second
Language Studies Eye-Tracking Lab and the recipient of the 2019 TESOL Award for Distinguished
Research and the co-recipient of the Best of the Modern Language Journal 2019. Her book Eye
Tracking in Second Language Acquisition and Bilingualism: A Research Synthesis and Methodological Guide
was published by Routledge in 2020.
Jorge González Alonso is a senior researcher at the Nebrija Research Center in Cognition
(CINC), Universidad Nebrija (Spain), and the AcqVA Aurora Center at UiT, the Arctic University
of Norway. His research focuses on the acquisition and processing of languages throughout the lifes-
pan, and the involvement of core cognitive processes (attention, memory, perception, and emotion)
in language learning and processing.
Kira Gor is a professor of second language acquisition at the University of Maryland (USA), where
she teaches courses in the graduate program in second language acquisition. Her research focuses on
the phonological and morphological aspects of nonnative lexical processing and the structure of the
nonnative mental lexicon. She has worked on cross-linguistic phonetic perception, the phonology-
orthography interface, and phonological and morphological processing in heritage and late learners.
Stefan Th. Gries obtained his PhD at the University of Hamburg (Germany) in 2000. After a few
years at the University of Southern Denmark and the Max Planck Institute for Evolutionary Anthro-
pology, he has been working at the University of California at Santa Barbara (USA) since 2005 and
at the University of Giessen (Germany) since 2018.
Theres Grüter is a professor of second language studies at the University of Hawai’i at Mānoa
(USA). Her research investigates how language users of all kinds—adults and children, monolingual
and bilingual—process and learn structural aspects of language. She is particularly interested in how
we integrate information from linguistic and non-linguistic sources, how we do so as quickly and
successfully as we typically do in everyday communication, and how this ability develops across the
lifespan.
Marianne Gullberg is a professor of psycholinguistics at Lund University (Sweden). She studies acquisi-
tion and processing in second language and multilingual speakers targeting semantics, discourse, and gesture
xii
Contributors
production and comprehension. She is co-founder of the Nijmegen Gesture Centre, the frst of its kind,
and was vice president of the European Second Language Association (EuroSLA) from 2000 to 2007.
Laura Gurzynski-Weiss (PhD, Georgetown University, USA) is an associate professor and director
of undergraduate studies in the Department of Spanish and Portuguese at Indiana University (USA),
where she teaches courses in second language acquisition, task-based language teaching, research and
teaching methods, individual diferences, and Hispanic linguistics.
Phillip Hamrick (PhD, Georgetown University, USA) is an associate professor and principal inves-
tigator of the Language and Cognition Research Laboratory at Kent State University (USA). His
research focuses on the roles of general learning and memory phenomena in language and meta-
research aimed at improving the rigor and validity of language acquisition research.
Julia Heimann is a PhD researcher of frst language attrition and psycholinguistics at the University
of Essex (United Kingdom).
Holger Hopp is a professor of English linguistics at the Technische Universität Braunschweig (Ger-
many). In his research, he investigates child and adult second and third language acquisition and
processing as well as heritage language acquisition and frst language attrition. He uses several psy-
cholinguistic methods, e.g., eye-tracking and priming, to determine the directionality, scope, and
degree of cross-linguistic infuence in bi- and multilingual speakers of diferent ages as well as the
limits on multilingual learning.
Jan Hulstijn is a professor emeritus of second language acquisition at the Amsterdam Center for Lan-
guage and Communication (ACLC) of the University of Amsterdam (The Netherlands). He received
the 2018 Distinguished Scholar Award from the European Second Language Association (EuroSLA).
Carrie N. Jackson is a professor of German and linguistics at Pennsylvania State University (USA).
Her research focuses on how second language speakers process lexical and grammatical information
during language comprehension and production to advance our understanding of the linguistic and
cognitive mechanisms that drive second language acquisition and use.
Nan Jiang is a professor of second language acquisition at the University of Maryland (USA). His
research interests include lexical representation and development in a second language, morphosyn-
tactic processing, semantic development, and bilingual language processing. He has published numer-
ous articles in second language acquisition, applied linguistics, and psycholinguistics journals. He is
also the author of Conducting Reaction Time Research in Second Language Studies (Routledge, 2012) and
Second Language Processing: An Introduction (Routledge, 2018).
Ronald P. Leow is a professor of applied linguistics and director of Spanish language instruction
at Georgetown University (USA). His areas of expertise include language curriculum development,
teacher education, instructed language learning, psycholinguistics, cognitive processes in language
learning, research methodology, CALL, and writing/written corrective feedback. He has published
extensively in prestigious journals and has initiated several strands of research in the feld of instructed
second language acquisition.
Casey Lew-Williams is a professor in the Department of Psychology at Princeton University

(USA). He works with postdocs, graduate students, and undergraduates in the Princeton Baby Lab
xiii
Contributors
to study how young children learn, and his research focuses in particular on language learning and
communication. He is a chief editor of Frontiers for Young Minds and a co-founder of ManyBabies.
Shaofeng Li is an associate professor and program leader of Foreign and Second Language Acquisi-
tion at Florida State University (USA). His main research interests include language aptitude, work-
ing memory, task-based language teaching and learning, and research methods. He is the editor in
chief of Research Methods in Applied Linguistics, a journal devoted exclusively to the research methods
of language-related studies.
Michael H. Long was a professor of second language acquisition at the University of Maryland
(USA). His many publications included the Handbook of Language Teaching (Blackwell, 2009), Second
Language Acquisition and Task-Based Language Teaching (Wiley-Blackwell, 2015), and The Cambridge
Handbook of Task-Based Language Teaching (Cambridge, 2021). He passed away in 2021, when the
current handbook was in progress.
Brian MacWhinney is the Teresa Heinz Professor of Psychology at Carnegie Mellon University
(USA). He has formulated a model of normal and disordered language processing and learning called
the Competition Model. He has also developed the open TalkBank database system and related pro-
grams for the study of child language, aphasia, and second language learning.
Rosa M. Manchón is a professor of applied linguistics in the Department of English at the Univer-
sity of Murcia (Spain). Her research explores second language writing processes and strategies. She is
a past editor of the Journal of Second Language Writing (2008–2014), and is the chief editor of the new
book series Research Methods in Applied Linguistics (John Benjamins).
Silvina M. Montrul is a professor in the Department of Spanish and Portuguese and the Department of
Linguistics at the University of Illinois at Urbana-Champaign (USA), and she directs the Second Language
Acquisition and Bilingualism Lab. She co-edits the journal Second Language Research. Her research focuses
on linguistic and psycholinguistic approaches to adult second language acquisition and bilingualism, in
particular syntax, semantics, and morphology, and on language loss and retention in heritage speakers.
Kara Morgan-Short holds a PhD from Georgetown University (USA) and is a professor in the
Department of Hispanic and Italian Studies and the Department of Psychology at the University of
Illinois at Chicago (USA) where she directs the Cognition of Second Language Acquisition Labora-
tory. Informed by the felds of linguistics, cognitive psychology, and neuroscience, her research aims
to elucidate the neurocognitive processes underlying adult-learned language acquisition and use. Her
research has been published in journals such as Language Learning, Studies in Second Language Acquisi-
tion, Bilingualism: Language and Cognition, and the Journal of Cognitive Neuroscience.
Gregory J. Poarch is an assistant professor of English linguistics and English as a Second Language
(ESL) at the University of Groningen (The Netherlands). He received his PhD in psycholinguistics
from Radboud University Nijmegen (The Netherlands) and has held various positions as researcher/
lecturer in Frankfurt (Germany), Toronto (Canada), Tübingen, and Münster (Germany). His research
covers psycholinguistic, cognitive, and social aspects of multilingualism, such as cross-linguistic inter-
action and cognitive control in multilinguals, the multilingual mental lexicon, and the implications of
societal multilingualism for language education.
Christine E. Potter is an assistant professor of psychology at the University of Texas at El Paso

(USA). Her research explores the role of experience in language learning, including examining
learning among bilingual infants and children.
xiv
Contributors
Patrick Rebuschat is a professor of linguistics and cognitive science at Lancaster University (United
Kingdom), and director of the Heritage Language 2 Consortium (HL2C), a strategic partnership that
brings together six European universities and the Portuguese Ministry of Foreign Afairs. He also
co-directs the Lancaster Language Learning Lab (4L), a cross-disciplinary lab dedicated to the study
of implicit-statistical learning and additional language learning in children, adolescents, and adults.
Jason Rothman is a professor of psycholinguistics at UiT, the Arctic University of Norway, direc-
tor of the Psycholinguistics of Language Representation (PoLaR) lab, and a senior researcher at the
Centro de Investigación Nebrija en Cognición (CINC) at Universidad Nebrija (Spain). He primar-
ily works on language acquisition and processing focusing on the mutually benefcial, bi-directional
relationship between linguistic theory and experimental methods and evidence from psycho- and
neurolinguistics. His work also investigates language-induced or associated links to neurocognition
in various bilingual and multilingual populations.
Rebecca Sachs is currently teaching high school linguistics, English for academic purposes, and writing
at Sandy Spring Friends School in Maryland (USA). Previously, she was an associate professor and chair
of the master’s programs in applied linguistics and TESOL at Virginia International University (USA).
Her research interests include individual diferences in language learning, feedback, task-based language
teaching, attention and awareness, depth of processing, teacher cognition, and research methods.
Nuria Sagarra (PhD, University of Illinois at Urbana-Champaign, USA) is an associate professor

of Spanish at Rutgers University (USA). Her overarching aim is to understand the cognitive mecha-
nisms underlying adult second language acquisition and how language processing relates to other
cognitive skills. She investigates linguistic and cognitive factors modulating morphosyntactic and
syntactic processing, and lexical and morphosyntactic prediction in a second language.
Cristina Sanz is a professor of Spanish and linguistics at Georgetown University (USA) and an expert
on multilingualism. She is interested in the interaction between learning context, including study
abroad, and individual diferences such as aging, experience, motivation, and aptitude. She has pub-
lished several volumes and over 100 chapters, articles, and reviews in venues such as the Journal of Cog-
nitive Neuroscience, Bilingualism: Language and Cognition, Language Learning, and Applied Psycholinguistics.
Sarah Schimke is a professor of German linguistics at Ludwig-Maximilians-Universität in Munich (Ger-

many). After completing a PhD at Radboud University Nijmegen (The Netherlands), she worked as a
post-doctoral researcher, lecturer, junior professor, substitute and full professor at the Universities of Paris
8 (France), Osnabrück, Münster, Bielefeld, and Dortmund (Germany). Her research centers on the frst
and second language acquisition and processing of morphosyntactic and discourse-related phenomena.
Monika S. Schmid obtained her PhD in English linguistics in 2000 from the Heinrich-Heine Univer-
sity of Düsseldorf (Germany) with a thesis on frst language attrition, and she has since held positions at
the Vrije Universiteit Amsterdam, the Rijksuniversiteit Groningen (The Netherlands), and the Univer-
sity of Essex (United Kingdom). She is a professor of linguistics and head of department of Language and
Linguistic Science at the University of York (United Kingdom). Her work has focused on various aspects
of frst language attrition, and she has published two monographs and edited several collected volumes
and special issues of journals on this topic, most recently the Oxford Handbook of Language Attrition (2019).
Rob Schoonen is a professor of applied linguistics at the Department of Language and Commu-
nication and the Centre for Language Studies of the Radboud University Nijmegen (The Neth-
erlands). His research concerns (psycholinguistic) aspects of language profciency, both language
comprehension and production, and also the assessment of language profciency.
xv
Contributors
Concepción Soto is a PhD candidate in applied linguistics at the University of Essex (United King-
dom). She is working on bilingual lexical processing at the word level via an eye-tracking paradigm. Her
main areas of research interest include language processing within the bilingual development spectrum
(language attrition, heritage languages, bilingualism), eye-tracking technology, and code-switching.
Yuichi Suzuki is an associate professor at Kanagawa University (Japan). His research focuses on
explicit and implicit learning and knowledge, skill acquisition theory, automatization, and optimiza-
tion of second language practice. He received the Valdman’s Award from Studies in Second Language
Acquisition (2017) and the IRIS Replication Award (2018).
Rüdiger Thul is an associate professor of applied mathematics in the School of Mathematical Sci-
ences at the University of Nottingham (United Kingdom). Prior to that, he was an assistant professor
and held a Leverhulme Trust Early Career fellowship at the University of Nottingham.
Michael T. Ullman is a professor in the Department of Neuroscience at Georgetown University

(USA), with secondary appointments in the departments of Neurology and Psychology. He is the
director of the Brain and Language Laboratory and the Georgetown EEG/ERP Lab, as well as direc-
tor of medical neuroscience at Georgetown University Medical School. His research examines the
neurocognition of frst and second language, math, reading, and memory; how these domains are
afected in various disorders (e.g., autism, dyslexia, developmental language disorder, Alzheimer’s,
Parkinson’s, and Huntington’s diseases); and how they may be modulated by factors such as genetic
variability, sex, handedness, and aging.
Janet G. van Hell is a professor of psychology and linguistics at the Pennsylvania State University
(USA) and also serves as co-director of the Center for Language Science. Her research focuses on
the neural and cognitive processes related to second language learning and bilinguals’ use of two
languages. She combines neuropsychological (ERPs) and behavioral techniques to study word and
sentence processing, patterns of cross-language interaction, in child and adult second language learn-
ers and bilinguals at diferent levels of profciency as well as the neural and cognitive mechanisms
involved in code-switching and the comprehension of foreign accented speech.
Paul G. Warren is a professor of linguistics in the School of Linguistics and Applied Language Studies
at Victoria University of Wellington (New Zealand). His main research interests are in psycholinguistics,
experimental phonetics, including intonation, sociophonetics, and New Zealand English phonetics and
phonology. His book Uptalk: The Phenomenon of Rising Intonation (2016, Cambridge University Press),
explores the use of high rising terminal intonation across varieties of English and in other languages.
John N. Williams graduated in psychology from the University of Durham (United Kingdom),
and then went on to do a PhD in psychology at the Cognition and Brain Sciences Unit, Cambridge
(United Kingdom). He is currently a lecturer and researcher in the Department of Theoretical and
Applied Linguistics, University of Cambridge (United Kingdom). He has published research on
second language lexical and syntactic processing and the cognitive mechanisms of second language
learning, with a special emphasis on implicit learning.
xvi
1
INTRODUCTION
Second language acquisition and
psycholinguistics
The Routledge Handbook of Second Language Acquisition and Psycholinguistics presents a state-of-the-art
overview of psycholinguistic research on second language acquisition (SLA). From a quick glance
at the title of the handbook, Second Language Acquisition and Psycholinguistics, a reader could get the
impression that SLA and psycholinguistics are separate areas of study or that psycholinguistics refers
to a particular feld in SLA or Applied Linguistics, as is true for other handbooks in this series, e.g.,
Second Language Acquisition and Writing (Manchón & Polio, 2021) or Second Language Acquisition and
Pragmatics (Taguchi, 2019).
We believe it is more appropriate to think of the psycholinguistic study of SLA as an approach to
SLA. Psycholinguists study the cognitive mechanisms and processes of language production, compre-
hension, acquisition, and use. They share the belief that unravelling the architecture and processes in
the minds of learners is central to understanding SLA. Of course, one can approach SLA from difer-
ent, say, sociological or educational viewpoints. Adopting a psycholinguistic approach by no means
precludes studying the social, in line with the central tenet of embedded cognition that the mind is
not just “in the head” (Ellis, 2019). Depending on how wide a notion of the mind we entertain, the
psycholinguistic study of SLA may thus encompass a large share of SLA.
At the same time, we can think of SLA as an area within psycholinguistics. Psycholinguists study
the acquisition and processing of languages, and second language (L2) acquisition falls squarely
within psycholinguistics in that it refers to the study of a particular type of language user. Seeing that
bilinguals increasingly outnumber monolingual language users across the world, the study of SLA and
bilingualism is moving center stage in psycholinguistics (Kroll et al., 2014; Prior & van Hell, 2021).
As its name suggests, the psycholinguistic study of SLA itself is informed by various scientifc
disciplines in which the researchers have been trained. Many psycholinguists have a background in
linguistics and investigate the nature of learners’ developing knowledge of more than one language
and how it is deployed in real-time production and comprehension. Conversely, researchers coming
from cognitive psychology typically focus on L2 development and how patterns of L2 knowledge
emerge shaped by general cognitive learning mechanisms. Lastly, researchers with backgrounds in
foreign language teaching conduct psycholinguistic research motivated by questions in the context of
foreign language instruction and aim to understand and improve L2 teaching and learning.
This handbook brings together linguistic, cognitive-psychological, and applied views on SLA and
thus highlights that studying how people acquire and use an L2 is a joint enterprise and that it takes
a village to understand SLA. The complementary perspectives represented in this volume point to
potential synergies between research traditions and invite dialogue and collaboration that transcend
DOI: 10.4324/9781003018872-1 1
traditional disciplinary boundaries. In this introductory chapter, we intend to map central and shift-
ing themes in the psycholinguistic study of SLA, present the major topics in the feld, and give an
overview of the chapters assembled in this handbook.
Shifting themes and perspectives within psycholinguistic research on SLA

Over the years, psycholinguistic studies have addressed a large range of questions about how second
languages are learned and processed, using a wide variety of methods and tasks. For decades, arguably,
the central question was how bilingual language acquisition and processing compare to its monolin-
gual counterparts. In no small part, the ways in which this question have been studied were shaped by
the types of bilinguals under investigation. When the psycholinguistic study of SLA frst took hold in
the 1970s, the majority of people described in research became bilingual through foreign language
learning in schools or through migration. As a consequence, instructed language learners in second-
ary schooling or adult immigrants constituted the typical populations under investigation.1 These
learners start learning an L2 late, typically around puberty, and they often show variable and non-
native learning outcomes and persistent infuences of the frst language, manifesting, for instance,
in foreign accents or grammatical mistakes refecting properties of the frst language. As a result,
psycholinguistic research studied how these learners developed or why they failed to develop skills in
the foreign language for speaking, reading, listening, and writing.
With these learners in mind, psycholinguists sought to test whether diferences in outcomes across
diferent language skills between native monolingual and late bilingual language acquisition can be
traced to age-induced constraints on learning. In the context of the Critical Period Hypothesis (Len-
neberg, 1967), many studies examined whether adult L2 learners can achieve similar commands of
the L2 as native monolingual speakers of that language or whether their developmental paths and
learning rates difer from monolingual children acquiring their L1.
Potential constraints on learning could afect either the types of linguistic knowledge that late
learners can readily recruit or the types of learning mechanisms available to them. With respect to
linguistic knowledge, researchers have addressed whether late L2 learners are able to use grammati-
cal knowledge to the same degree as, for instance, semantic information (for discussion, see Hopp,
this volume) or whether they rely on diferent cues in the input (Sagarra, this volume). In terms of
learning mechanisms, late language learning may force learners to primarily rely on explicit rather
than implicit knowledge (for discussion, see Godfroid; Williams & Rebuschat; Ullmann & Morgan-
Short, this volume) or it may lead to language processing being less automatic (see Suzuki, this
volume), especially in speaking and listening. These questions continue to inform current research
(e.g., van der Slik et al., 2022; Hartshorne et al., 2018), and they have become more nuanced, seeing
that constraints on learning an L2 may be afected by many factors other than age, such as learning
contexts, language learning experience, education, and L1, among others (Pfenninger & Singleton,
2021). Not least, these questions still abound in research because of their implications for educational
policy about the best starting ages of foreign language instruction; yet, they also reinforce the social
perception of L2 speakers as “defcient” compared to monolinguals (e.g., Long, 2007).
Next to age of learning, another factor that sets apart L2 from L1 acquisition is the presence
of another language which can afect L2 acquisition and processing by virtue of cross-linguistic
infuence (CLI). CLI surfaces at diferent language levels. Although CLI is most pronounced
among learners who begin to learn an L2 in their foreign accents, it has become clear that CLI is
pervasive across all levels of language processing, also extending to the lexicon and morphosyntax,
as well as semantics and pragmatics (see Section 3, this volume). Moreover, CLI surfaces across
all types of bilinguals, whether they learn an L2 early or late, and it persists to high profciency
levels (see Section 1, this volume). Therefore, CLI must be a central component in modeling the
bilingual mind.
2
Introduction
To this day, these research themes continue to dominate the feld; yet, since bilingualism is
becoming more diverse, these themes are also becoming increasingly nuanced. For instance,
research on age efects has expanded its scope beyond the traditional comparisons between L2
learners and monolingual natives by comparing diferent types of bilinguals who difer in their
ages of acquisition, such as early and child L2 learners and late L2 learners (e.g., Unsworth, 2016).
In a similar vein, studies of CLI now often contrast forward CLI from the L1 to the L2 in L2
acquisition and backward CLI from the L2 to the L1 in heritage languages (e.g., Montrul, 2016)
and L1 attrition (e.g., Schmid & Köpke, 2019). In addition, psycholinguists study whether CLI is
similar in L2 and L3 learners (see González Alonso & Rothman, this volume). Common to these
changing trends is a shift from comparing monolinguals to bilinguals to studying variation across
bilingual populations.
Just as populations of bilinguals vary, the bilingual experience varies substantially across individu-
als: People become bilingual at diferent ages, in diferent languages, and in diferent contexts. As a
consequence, another quickly growing trend is the study of individual diferences among bilinguals
to map out the wider spectrum of bilingual experiences. Currently, psycholinguistic research aims to
identify and model the contributions of environmental (external) and learner-internal factors to the
process of SLA as continuous variables. For instance, studies weigh and delineate the relative impacts
of cognitive factors, such as working memory (Li, this volume), attention or executive function
(Poarch, this volume), as well as the impact of external variables, such as the number of siblings, inter-
actions with native speakers, or media exposure in the L2 (see Section 5, this volume). As a result,
psycholinguists have abandoned the traditional focus on specifc skills in language production or
comprehension, and research is addressing broader questions about the interconnected architecture
of the bilingual mind and how bilinguals learn and interact in diferent environments and social con-
texts. For instance, it is interesting to know whether learners recruit the same learning mechanisms
and processes in naturalistic versus instructed settings (e.g., Leow, this volume) or in classroom-based
versus digital and multimodal environments (e.g., Li & Lan, 2021).
All these trends suggest that the traditional view of psycholinguistics as narrowly lab-focused
research with low ecological validity focusing on a carefully selected group of L2 learners is giving
way to a research feld that embraces learner diferences, integrates social contexts of learning (Leow,
this volume), and also considers the cognitive consequences of becoming bilingual (see Poarch, this
volume). This way, psycholinguistic research is getting closer to the multifaceted and complex reali-
ties of being bilingual. As a corollary, the increasing attention research pays to diferences among
bilinguals breaks with the idea that bilinguals are a population best studied in dichotomous contrast
to monolinguals. In this respect, approaches and models are beginning to be proposed that model
bilingualism in its own right instead of viewing it through the lens of models and theories developed
for monolingual acquisition and processing (Grüter, this volume). For instance, in current models of
the bilingual mental lexicon, language is not represented by virtue of separate processing paths for
each language or the annotation of lexical items for language membership. Instead, these models hold
that access to the mental lexicon or sentence processing is language-non-selective in that informa-
tion across languages is stored and integrated in the same system and continuously co-activated (e.g.,
Shook & Marian, 2013; Dijkstra et al., 2019).
Not least, severing models and approaches to the psycholinguistics of bilingualism from
monolingual models implicates a diferent view of bilingualism. In the last decades, the focus
of research has shifted from investigating how and why L2 learners can, or often fail to, master
an L2 towards understanding how they successfully “juggle” two or more languages in their
minds (Kroll et al., 2012). As a result, the traditional view of bilingualism as a defcient form of
monolingualism has given way to an appreciation of bilingualism and multilingualism as a more
complex mastery. The chapters in this handbook give an overview of the psycholinguistic facets
of this mastery.
3
Topics and themes of the handbook

To cover the diverse and multifaceted domains of inquiry in psycholinguistic research on SLA, we
divided the handbook into fve sections:
I. Psycholinguistics Across the Lifespan

II. Research Methods in L2 Psycholinguistics
III. Theoretical Perspectives on L2 Psycholinguistics
IV. The Psycholinguistics of Learning
V. Transdisciplinary Perspectives: Contexts and Future Directions
The frst section is organized by population, the second by method, and the third through the lens
of diferent theories. The fourth section addresses the cognitive architecture and mechanisms of
(second) language learning. The fnal section explores psycholinguistic dimensions of other topics
in SLA, applied linguistics, and bilingualism. Invariably, there is some overlap between chapters and
sections; yet, each addresses diferent themes that are central to the feld. Each section concludes
with a synthesis, written by an eminent scholar in the feld. The synthesis chapters ofer authoritative
summaries and point out major avenues of future research. In the following, we outline the chapters
with respect to the major themes of the sections.
The frst section, “Psycholinguistics Across the Lifespan,” presents research fndings for diferent
types of bilingual and L2 groups. This section represents the growing interest in populations other
than classroom adult L2 learners. For one thing, widening the scope of psycholinguistic research marks
a shift of the feld towards a more inclusive discipline and allows researchers to study these diferent
populations in their own right. For another, it allows the feld to systematically compare and contrast
diferent types of bilinguals and L2 learners. We can investigate which developmental trajectories or
which difculties with particular aspects of the L2, e.g., grammar or discourse, are general features of
bilingualism and which ones are specifc to a particular type of bilinguals. For instance, systematic com-
parisons between bilingual children (Lew-Williams and Potters, Chapter 2), child L2 learners (Schimke,
Chapter 3), and adult L2 learners (Jackson, Chapter 4), on the one hand, and long-term emigrants
who potentially undergo L1 attrition (Schmid and colleagues, Chapter 6), and heritage speakers of the
L1 (Montrul, Chapter 7), on the other, address the role of age efects on learning and processing (see
Hopp & Schmid, 2013). Further, juxtaposing these bilingual groups sheds light on how diferences in
the timing and type of the bilingual experience impact the ability to perform well in time-sensitive tasks
tapping into implicit knowledge and automatic processing compared to untimed tasks allowing for the
use of explicit knowledge and controlled processing (see Chapter 24 by Godfroid for discussion). Inves-
tigating multilinguals, e.g., L3 learners, can uncover diferent sources of cross-linguistic infuence and
transfer (Rothman and González Alonso, Chapter 5). Moreover, it shows whether fndings from the
psycholinguistic study of bilingualism scale up to multilingualism, which is increasingly common across
the globe. Finally, comparisons of subgroups within adult L2 learners that difer in experience elucidate
in how far long-term experience with the L1 and changes in cognitive and social circumstances that
come with ageing among elderly learners afect L2 learning (Cox and Sanz, Chapter 8).
On the one hand, the chapters in the section are testimony to the fact that bilingualism breaks
down into diferent types of bilinguals and L2 learners. Each chapter charts the progress that has been
made in studying and mapping the particular psycholinguistic profles of each population. On the
other hand, each chapter includes implicit or explicit between-population comparisons, either to set
apart the populations in characteristic features of language processing or to establish commonalities
between diferent types of bilinguals. These comparisons allow for dialogue between the subfelds
and also raise the possibility of conceptualizing diferent types of the bilingual experience accord-
ing to age of acquisition, language use, and contexts of language learning as continuous, rather than
4
Introduction
between-group, variables. In the synthesis chapter of this section (Chapter 9), Grüter discusses how
transcending the boundaries of subfelds and bilingual populations opens the road for the feld to
address broader questions about language learning and processing that characterize our species.
In the second section of the handbook, “Research Methods in L2 Psycholinguistics,” the chapters
present overviews of the central methods used to study comprehension (Elgort and Warren, Chap-
ter 10) and production (Bernolet, Chapter 11). In Chapter 12, Gullberg outlines methods for study-
ing the specifcs of multimodal L2 production and comprehension involving paralinguistic features
such as gestures. While psycholinguistic research is typically conducted in laboratory settings using
stationary equipment, the feld is beginning to adopt methods for data collection outside the lab.
Hamrick (Chapter 13) discusses the opportunities, challenges, and pitfalls that come with collecting
time-sensitive data in the lab and via the internet. Beyond data from experimental studies, psycho-
linguists increasingly turn to corpus data in order to supplement experimental data with ecologically
more valid data from natural conversation, unprompted production, or actual writing. In Chapter 14,
Gries and Deshors outline how corpus studies operationalize core psycholinguistic concepts and how
corpora can be used to investigate psycholinguistic processes such as priming. In conjunction, the
chapters in this section attest to the methodological advances the feld has been making in the last
decade. Beyond ofering L2 psycholinguists a choice to select from a widening array of methods,
they highlight the need for researchers to have a “theory of method” (Foss, 1998) and be aware of the
assumptions and scope of a particular method in relation to other methods, as argued by Jiang in his
synthesis chapter of this section (Chapter 15). Finally, the greater diversity of available methods ofers
chances to triangulate and cross-validate methods in psycholinguistic studies.
Section 3, “Theoretical Perspectives on L2 Psycholinguistics,” is organized around the major
language domains studied in L2 psycholinguistics. For lexical and sentence-level comprehension,
the chapters approach the same topic from a linguistic or a cognitive-psychological perspective,
respectively. Prioritizing learner-internal factors, the linguistic view is predominantly concerned
with how language knowledge of the L1 or the L2 guides and constrains processing routines and
outcomes. Capitalizing on learner-external factors, the cognitive-psychological view is primarily
interested in how experience, e.g., frequency and cues in the input, interacts with learner-internal
cognitive factors, such as attention and working memory, to shape language processing. The chapters
in this section serve to illustrate the diferent research questions and traditions across major language
domains. For word and multi-word comprehension, Gor (Chapter 16) exemplifes how variability in
the internal processing at diferent levels of language comprehension, namely sublexical phonological
and orthographic processing, may lead to fuzzy lexical representations and slower and error-prone
grammatical processing. Focusing on external variation in the language input to the learner, Conklin
and Thul (Chapter 17) charts how diferences in the lexical frequency of words in an L2 or between
the L1 and the L2 occasion diferent speeds of lexical recognition in bilinguals. For sentence process-
ing, Hopp discusses in Chapter 18 whether and how diferences in how the parser uses and integrates
grammatical information may account for diferences in sentence comprehension between L2 learn-
ers and monolingual native speakers. In Chapter 19, Sagarra explores how external features of the
input, like perceptual salience and the availability and reliability of cues, together with the learner
experience of attending to salient features of the L1 (learned attention), shape sentence processing
patterns among adult L2 learners.
For each domain, the pairs of chapters illustrate how variability and difculties characteristic of
the performance among adult L2 learners can be explained either with recourse to features of the
language processing system that may difer between monolinguals and L2 learners or by appealing to
experiential diferences between L1 and L2 speakers in the amount and type of input. Chapters 20
and 21 incorporate each perspective in their reviews of discourse processing and code-switching, two
phenomena in which external and internal factors meet. In Chapter 20, Colonna addresses the ques-
tion of how L2 learners establish discourse coherence with pronouns or connectives and how sensitive
5
they are to information structure in an L2. What the rapidly growing body of psycholinguistic studies
suggests is that L2 learners can acquire discourse markers and their uses in the L2, but that they may
keep experiencing challenges in integrating discourse information with other types of, e.g., syntactic
information, in real time. In Chapter 21, van Hell reviews linguistic triggers of code-switching, such
as cognate words and word order overlap, as well as external triggers of code-switching, e.g., the (per-
ceived) language identity of the interlocutor. Moreover, she outlines how cognitive factors, such as
executive control, modulate the ability and speed with which bilinguals code-switch. Probably more
than any of the other chapters, these two chapters illustrate how the joint investigation of linguistic
and cognitive factors as well as their interactions leads to a deeper understanding of typical features of
L2 comprehension and production. In this vein, the synthesis chapter by Dussias (Chapter 22) makes a
persuasive case that individual diferences in cognitive processing, experience, and diferent contexts of
L2 acquisition and bilingual language use may lead to distinct psycholinguistic profles within bilingual
speakers and between bilingual and monolingual language users.
Section 4, “The Psycholinguistics of Learning,” addresses key psycholinguistic constructs and
mechanisms of language learning. Set within the broad domain of cognitive psychology, these chap-
ters discuss which general cognitive mechanisms L2 learners bring to the task of L2 acquisition and
use, and how experience or language profciency shapes the process of L2 development. Williams
and Rebuschat (Chapter 23) review classic research paradigms from cognitive psychology used to
study implicit learning and statistical learning. Although a lot of this research is conducted with arti-
fcial languages, the studies Williams and Rebuschat review ofer useful and generalizable insights for
SLA. Godfroid (Chapter 24) provides a complementary perspective from SLA, reviewing research
on implicit and explicit instruction, learning, and use conducted with artifcial and natural languages
in the lab and the classroom. A central theme in these two chapters is the interface issue, or whether
and how implicit and explicit learning and knowledge interact. Suzuki (Chapter 25) provides a
modern conceptualization of practice that emphasizes the quality as well as the quantity of practice
on the road to high language profciency. Suzuki argues that practice activities allow for the joint
contribution of explicit and implicit learning, with the ultimate goal being to enable automatic use
of L2 knowledge and skills. Suzuki’s chapter and the next chapter by Morgan-Short and Ullman
(Chapter 26) share a focus on declarative and procedural memory. Following a historical review of
diferent models of declarative and procedural memory, Morgan-Short and Ullman address how the
declarative/procedural model (Ullman, 2001), which is at heart a neurolinguistic model, can beneft
from psycholinguistic research approaches to test its key predictions. Long (Chapter 27) focuses on
spoken interaction as a powerful site for language learning thanks to the interactional modifcations
that arise when interlocutors work to establish mutual comprehension. Through a detailed histori-
cal recount of landmark studies, Long sketches current topics and future directions for research in
the cognitive-interactionist framework. Lastly, Li (Chapter 28) ofers a critical and synthetic review
of working memory as a domain-general resource that underlies most of the diferent models and
theories reviewed in this section. Despite the importance of working memory in L2 learning, Li
notes that its role could be theorized in more detail, and he provides some contingent theoretical
discussion interlaced with an empirical review. In conclusion, MacWhinney (Chapter 29) provides
a comprehensive synthesis of the themes that the aforementioned chapters cover within a unifed
approach to SLA (e.g., MacWhinney, 2017). He makes an argument for focusing more directly on
learning (as opposed to processing), which can take place across varying timescales, including those
beyond the ones that are typically studied in the lab. The chapter makes some specifc suggestions on
how to investigate the many paths of L2 learning across these diferent timescales, using web-based
formats of data collection. Collectively, the chapters in this section advance our understanding of
psycholinguistic processes and learning mechanisms that shape L2 acquisition.
A handbook of second language acquisition and psycholinguistics would not be complete without
a section that addresses the place of psycholinguistics within the wider feld of SLA. In Section 5,
6
Introduction
“Transdisciplinary Perspectives: Contexts and Future Directions,” six experts write about the inter-
section between psycholinguistics and core areas in SLA and the study of bilingualism. For instructed
second language acquisition, Leow (Chapter 30) sets the stage with a process-oriented perspective
to SLA (Leow, 2015), which foregrounds the mental or cognitive processes that classroom L2 learn-
ers bring to the task of L2 learning. Such a process-oriented perspective, Leow argues, is valuable
(albeit underrepresented) in research on learning contexts because of the unique insights that it can
provide into how local/global contexts may infuence learner-internal factors. Also espousing a
process-oriented perspective, Sachs, Baralt, and Gurzynski-Weiss (Chapter 31) review the many ways
in which psycholinguistic research is relevant to the language classroom. Their review highlights the
complex and multifaceted nature of language instruction, as seen in aptitude-treatment interactions,
task complexity research, and work on learner and teacher cognition. Methodologies such as think-
alouds and eye-tracking can capture moment-to-moment learning processes and are thus particularly
well suited to capturing the dynamic nature of L2 learning. In the chapter on the psycholinguistics
of L2 writing, Manchón (Chapter 32) similarly adopts a process-based approach to understanding
L2 writing. Drawing on psycholinguistic and cognitive learning mechanisms, she makes the case for
writing as a driver of L2 learning and illustrates how “writing to learn” may be studied at diferent
levels and using diferent tasks. For L2 assessment, Schoonen (Chapter 33) discusses how work in
psycholinguistics has informed the understanding of two main constructs in L2 assessment—language
profciency and language learning potential, or language aptitude. Schoonen highlights the role of
psycholinguistic models of language learning and use in research on language assessment, while also
pinpointing areas where the two felds converge less in spite of their common goals of obtaining valid
measurement of language profciency and processing. For Schoonen, the questions of granularity of
measurement and individual diferences, both important in diagnostic language assessment, stand
out as shared priorities for future research. Poarch (Chapter 34) reviews psycholinguistic and neu-
rolinguistic evidence regarding the efects of bilingualism on cognitive development, in particular,
domain-general executive function and its subcomponent, inhibitory control, across diferent age
ranges (children, young adults, and older adults). Amongst other things, this rich body of work has
revealed a need to move away from a monolingual-bilingual dichotomy in order to do more justice
to the full spectrum of bilingual experiences, including those traditionally researched in SLA. Finally,
Hulstijn (Chapter 35) synthesizes the diferent chapters in this section from the perspective of scien-
tifc inquiry as a cyclical process of exploration and explanation. Adopting this perspective, Hulstijn
appraises the preceding chapters as well as the Douglas Fir Group (2016), which called for broadening
the scope of SLA by proposing “a transdisciplinary framework for SLA in a multilingual world.” On
balance, Hulstijn concludes that it is not only through transdisciplinary approaches and collabora-
tion, but also through rigorous cycles of exploration, explanation, and hypothesis testing that our
understanding of SLA can increase. By critically looking back and ahead, this closing chapter paves
the way to an integrated study of SLA.
Notes for readers and instructors

This handbook is primarily intended as a reference and overview compendium for researchers,
graduate students, and upper-level undergraduate students interested in SLA, bilingualism, language
learning, or the psycholinguistics of language users other than monolinguals. While not designed as
a textbook for teaching as such, we strove to make the handbook suitable for use in teaching. We
structured each chapter in a comparable way and included helpful textboxes intended for students.
1. Introduction with defnitions

2. Historical perspectives
3. Critical issues and topics
7
4. Theoretical perspectives and approaches

5. Current contributions and research
6. Current trends and future directions
7. Further reading
The chapters in Section 2 (Methods) follow a slightly diferent, though comparable, structure. The
textboxes in each chapter ofer handy overviews of the key terms and they present open questions
and future issues that may inspire the reader’s own research. Further reading is annotated and lists a
few central and accessible texts suitable for students looking for more detail and ideas for research
projects. We also added cross-references to other chapters in all chapters of the handbook, so that
readers can more easily establish links between the topics and chapters. We suggest that instructors
may use the handbook chapters as overview reading in various courses in (applied) linguistics that
have a focus on diferent populations (Section 1), psycholinguistic research methods (Section 2),
theoretical perspectives on diferent linguistic phenomena (best illustrated by Section 3), or key topics
in the psychology of language learning and applied linguistics (Sections 4 and 5).
Acknowledgements
First and foremost, we are deeply grateful to the contributors of this handbook, who wrote and submit-
ted excellent chapters during an extraordinarily challenging time. We appreciate their unfailing devotion
to the handbook, their willingness to write chapters in line with the spirit and the template of the hand-
book, and their stamina in revising the chapters after internal and external reviews. We also owe a debt
of gratitude to the series editors Susan Gass, Alison Mackey, and Kimberly Geeslin for entrusting this
ambitious project to us. We express our thanks to the former series editors at Routledge, Ze’ev Sudry
and Helena Parkinson, as well as the current editors for their enthusiastic support along the way. Paula
Winke and Tineke Brunfaut most generously shared the online submission system they created for the
Routledge Handbook of Second Language Acquisition and Assessment with us, which proved immensely useful
to streamline the practical aspects of collecting manuscripts, revisions, and administrative forms from our
contributors. Lastly, our sincere thanks go to our research assistants, Brittany Finch, Joanne Koh, Wenyue
(Melody) Ma, and Emilia Peinecke, for their careful editorial work behind the scenes. We edited this
handbook with the future generation of psycholinguists in mind, of which these students are prime
examples. We thank all our students, present and past, for inspiring us to do this project and hope it will
be helpful for them as they help build the future of second language acquisition and psycholinguistics.
Note
1 An exception was a series of studies on children acquiring second languages in the context of the Morpheme
Order studies. Inspired by the discovery of a natural order in the acquisition of infection in child L1 acquisi-
tion (Brown, 1973), many researchers investigated whether children learning an L2 would follow the same
order (e.g., Dulay & Burt, 1974). After this initial surge of studies on child L2 acquisition, the focus shifted
to adult SLA, and research on how children acquire a second language only garnered interest in the 2000s
(Chondrogianni, 2018; Schimke, this volume).
References
Brown, R. (1973). Development of the frst language in the human species. American Psychologist, 28(2), 97.
Chondrogianni, V. (2018). Child L2 acquisition. In D. Miller, F. Bayram, J. Rothman, & L. Serratrice (Eds.),
Bilingual cognition and language: the state of the science across its subfelds (Vol. 54, pp. 103–126). John Benjamins.
Dijkstra, T., Wahl, A., Buytenhuijs, F., Van Halem, N., Al-Jibouri, Z., De Korte, M., & Rekké, S. (2019). Mul-
tilink: A computational model for bilingual word recognition and word translation. Bilingualism: Language
and Cognition, 22(4), 657–679.
8
Introduction
Douglas Fir Group. (2016). A transdisciplinary framework for SLA in a multilingual world. The Modern Language
Journal, 100(S1), 19–47.
Dulay, H. C., & Burt, M. K. (1974). Natural sequences in child second language acquisition. Language Learning,
24(1), 37–53.
Ellis, N. C. (2019). Essentials of a theory of language cognition. The Modern Language Journal, 103, 39–60.
Foss, D. J. (1998). Two strands of scholarship on language comprehension: Phoneme monitoring and discourse
context. Journal of Psycholinguistic Research, 27(2), 191–201.
Gass, S. M., & Mackey, A. (2013). Introduction in S. M. Gass & A. Mackey (Eds.), The Routledge handbook of
second language acquisition (pp. 1–4). Routledge.
Hartshorne, J. K., Tenenbaum, J. B., & Pinker, S. (2018). A critical period for second language acquisition:
Evidence from 2/3 million English speakers. Cognition, 177, 263–277.
Hopp, H., & Schmid, M. S. (2013). Perceived foreign accent in frst language attrition and second language
acquisition: The impact of age of acquisition and bilingualism. Applied Psycholinguistics, 34(2), 361–394.
Kroll, J. F., Bobb, S. C., & Hoshino, N. (2014). Two languages in mind: Bilingualism as a tool to investigate
language, cognition, and the brain. Current Directions in Psychological Science, 23(3), 159–163.
Kroll, J. F., Dussias, P. E., Bogulski, C. A., & Krof, J. R. V. (2012). Juggling two languages in one mind: What
bilinguals tell us about language processing and its consequences for cognition. In Psychology of learning and
motivation (Vol. 56, pp. 229–262). Academic Press.
Lenneberg, E. H. (1967). Biological foundations of language. Wiley.
Leow, R. P. (2015). Explicit learning in the L2 classroom: A student-centered approach. Routledge.
Li, P., & Lan, Y. J. (2021). Digital language learning (DLL): Insights from behavior, cognition, and the brain.
Bilingualism: Language and Cognition, 1–18.
Long, M. H. (2007). Problems in SLA. Routledge.
MacWhinney, B. (2017). A unifed model of frst and second language learning. In M. Hickmann, E. Vene-
ziano, & H. Jisa (Eds.), Sources of variation in frst language acquisition: Languages, contexts, and learners (pp. 287–
310). John Benjamins.
Manchón, R. M., & Polio, C. (Eds.). (2021). The Routledge handbook of second language acquisition and writing.
Routledge.
Montrul, S. (2016). The acquisition of heritage languages. Cambridge University Press.
Pfenninger, S. E., & Singleton, D. (2021). Is there a best age for second language learning? Evidence from across
the lifespan. Debates in Second Language Education, 52–65.
Prior, A., & van Hell, J. G. (2021). Including bilingual participants in cognitive research. In Bilingualism across the
lifespan (pp. 47–65). Routledge.
Schmid, M. E., Köpke, B. E., Cherciov, M. C., Karayayla, T. C., Keijzer, M. C., De Leeuw, E. C., Mehotcheva,
T. H., Montrul, S., & Polinsky, M. C. (2019). The Oxford handbook of language attrition. Oxford University Press.
Shook, A., & Marian, V. (2013). The bilingual language interaction network for comprehension of speech.
Bilingualism: Language and Cognition, 16(2), 304–324.
Taguchi, N. (Ed.). (2019). The Routledge handbook of second language acquisition and pragmatics. Routledge.
Ullman, M. T. (2001). The neural basis of lexicon and grammar in frst and second language: The declarative/
procedural model. Bilingualism: Language and Cognition, 4(2), 105–122.
Unsworth, S. (2016). Quantity and quality of language input in bilingual language development. In E. Nicoladis
& S. Montanari (Eds.), Bilingualism across the lifespan: factors moderating language profciency (pp. 103–122). De
Gruyter Mouton.
van der Slick, F., Schepens, J., Bongaerts, T., & van Hout, R. (2022). Critical period claim revisited: Reanalysis
of Hartshorne, Tenenbaum, and Pinker (2018) suggests steady decline and learner-type diferences. Language
Learning, 72(1), 87–112.
9
SECTION I
Psycholinguistics across
the lifespan
2
THE PSYCHOLINGUISTICS
OF EARLY BILINGUALISM
Introduction
By defnition, bilingual infants face a challenging learning problem. Language input contains a mixture of
sounds, words, sentences, constructions, and conventions, and every language follows its own patterns at
each level. Bilingual infants must master the systems for multiple languages simultaneously, despite con-
siderably less experience with each language than monolinguals. Furthermore, bilingual infants do not
merely receive less exposure to each language individually; they are also typically surrounded by speakers
who vary in profciency, mix languages, and have unequal balance in their use of each language (Byers-
Heinlein & Fennell, 2014). Remarkably, even though they must absorb up to twice the unique informa-
tion, bilingual infants’ language skills do not develop twice as slowly as monolinguals’ (Oller et al., 2007).
A growing body of research has begun to systematically examine the cognitive and psycholinguistic
processes by which infants learn two languages concurrently. We summarize ten fndings from contem-
porary research on early bilingualism, focusing on how children learn two languages simultaneously.
As is typical of the literature, we defne “bilingual infants” as children under 36 months who regularly
hear two languages in their household or community environment, with each language representing at
least 20% of their total input. We attempt to cover phenomena from a range of important domains: how
infants become familiar with the sounds of their languages, how they learn the meanings of words, how
they process diferent types of sentences, and how bilingualism may afect cognition beyond language.
Textbox 2.1 Key terms and concepts
Perceptual narrowing: The process by which infants become less sensitive over time to perceptual contrasts
that are absent or rare in the environment (e.g., sounds not present in bilinguals’ ambient languages)
Translation equivalents: Words in diferent languages for the same concept (e.g., dog in English and perro
in Spanish refer to the same animal)
Total conceptual vocabulary: A measure of bilinguals’ vocabulary that counts the number of concepts for
which the child has a unique word (e.g., if a child knows both dog and perro, they receive credit only once)
Mutual exclusivity: The inference that a novel label is likely to refer to a novel object, rather than a
known object
Code switching: The use of two languages within a sentence or conversation (see van Hell, 2023 [this volume])
DOI: 10.4324/9781003018872-3 13
Critical issues and topics
Sounds
Learning language begins with the perception of sound, at least among hearing populations. To make
sense of language, learners must attend to meaningful distinctions and ignore those that do not carry
meaning. For bilingual infants, this requires learning to categorize their two languages as separate
systems and adjusting their reliance on diferent cues when hearing each language. For instance, a
Spanish-English bilingual must learn that the distinction between /b/ and /v/ can diferentiate words
such as bat and vat in English, but those sounds are part of a single category in Spanish. In this section,
we discuss some of the frst challenges that bilingual infants must resolve.
Bilingual infants can successfully discriminate between their two languages

A fundamental question about the nature of early bilingualism is whether bilingual infants
“know” they are bilingual. That is, do infants recognize that they are hearing two languages, or
do they interpret their input as one system with high inconsistency (Leopold, 1954)? Through-
out their frst year, bilingual infants detect changes from one of their languages to the other, even
when hearing related languages (e.g., Spanish and Catalan: Bosch & Sebastián-Gallés, 2001),
suggesting that they can distinguish languages perceptually. Thus, from a young age, children
appear to recognize that there are diferent languages in their input (Byers-Heinlein, 2014).
Bilinguals’ ability to separate their languages is consistent with evidence from monolinguals that
infants attend closely to sounds. From birth, monolingual infants distinguish languages with
diferent rhythmic properties and show increased interest in their primary caregiver’s native
language (Mehler et al., 1988). Within the frst six months, monolinguals develop even greater
sensitivity to their native language (Nazzi et al., 2000). Therefore, it is unsurprising that bilin-
gual infants develop similar expertise for each language that they hear regularly. In fact, young
bilinguals may be more sensitive to diferences between languages than monolinguals (Molnar
et al., 2014; Kuipers & Thierry, 2012). Thus, bilingual infants’ capacity to diferentiate their
languages seems to emerge from the combination of infants’ auditory processing skills and sen-
sitivity to their experiences.
Bilingual infants are slower than monolinguals to specialize in

the sounds of their native languages
During the latter part of the frst year, monolinguals lose their ability to discriminate non-native
contrasts and become more attuned to contrasts that are meaningful in their native language. This
specialization process (“perceptual narrowing”; see Textbox 2.1) is positively correlated with gaining
profciency in understanding the native language (Kuhl et al., 2006). In contrast, bilingual infants
retain the ability to perceive non-native contrasts longer than monolinguals. For instance, around ten
months, bilinguals continue to diferentiate non-native consonant contrasts that monolinguals do not
(Singh et al., 2017) and perceive categorical diferences in lexical tone, even when neither of their
languages uses tones contrastively (Petitto et al., 2012), suggesting that exposure to diverse sounds
changes how infants attend to sounds more generally.
Bilingual infants learn more easily from familiar-sounding speakers

In addition to variability across languages, bilingual infants must contend with the variability found
within languages, such as diferences between speakers or accents. Accented speech presents a particular
14
The psycholinguistics of early bilingualism
challenge because subtle pronunciation diferences carry social meaning yet must be disregarded to
recognize speakers’ intended referents. Monolingual infants learn new words more easily when they
are produced by monolinguals (Mattock et al., 2010) and display social biases for native-accented
speakers (Kinzler et al., 2007), which could be because these speakers sound familiar or because the
speech that they produce is more consistent. Interestingly, bilingual infants learn new words more
easily when they are produced by bilinguals (Fennell & Byers-Heinlein, 2014; Mattock et al., 2010),
suggesting that they too beneft from familiar-sounding speech. Moreover, even though they tend
to encounter non-native speakers, bilingual children, like monolinguals, prefer to befriend individu-
als with native (vs. foreign) accents (Souza et al., 2013). Thus, both monolinguals and bilinguals
appear to process information more efectively when they hear speakers who sound typical of their
environments.
Words
While children’s knowledge of language was once evaluated mainly by what they produced, newer
evidence shows that infants understand words well before they produce them (Bergelson & Swing-
ley, 2012), highlighting the importance of considering both comprehension and production to
fully understand language development (Thordardottir, 2011). At the same time, children’s recep-
tive and expressive vocabularies tend to be highly correlated (Marchman et al., 2010). Here,
we describe the trajectory of early bilingual vocabulary growth and discuss how bilingual chil-
dren’s understanding of familiar words and strategies for acquiring new words compare to those
of monolingual peers.
Bilingual children build their vocabularies at a similar

rate as monolingual children
It is often assumed that bilingual children will show delays in vocabulary growth. However,
evidence from diary studies, standardized questionnaires, and lab-based experiments all suggests
that bilinguals demonstrate comprehension and produce their frst words at similar ages as mono-
linguals (for review, see Fennell & Lew-Williams, 2018). Moreover, just as there are individual
diferences in the size and growth trajectory of vocabulary among monolingual children (Fenson
et al., 1994), there is considerable heterogeneity among bilinguals (Conboy & Thal, 2006), and
variability in children’s early vocabulary predicts later language skills for both monolinguals and
bilinguals (Duf et al., 2015; Hurtado et al., 2014). The supposition that bilingual children’s
vocabulary lags behind monolinguals’ likely stems from comparing children’s knowledge in a
single language; bilinguals typically know fewer words than monolinguals of the same age when
their knowledge is assessed in just one language (Oller et al., 2007; Pearson et al., 1997). But
comparisons that account for both languages often show that bilingual children know more total
words than monolinguals (de Houwer et al., 2014). A more conservative approach is to measure
“total conceptual vocabulary,” which approximates the number of unique concepts for which a
child has a label (Textbox 2.1). It has been suggested that parents overreport children’s knowledge
of cross-language synonyms (also called “translation equivalents,” Textbox 2.1), infating estimates
of vocabulary (Legacy et al., 2016). Based on total conceptual measures, monolinguals and bilin-
guals are often reported to know a similar number of words (Pearson et al., 1993; Thordardottir,
2011), although bilinguals’ conceptual vocabularies may initially lag behind monolinguals’ until
around the age of three (Gonzalez-Barrero et al., 2020). Therefore, basic mechanisms and growth
patterns in early vocabulary development appear to be shared among monolingual and bilingual
children.
15
Bilingual children’s vocabulary in each language is related to

the input they receive within, but not across, languages
A key question for early bilingual research is whether children’s knowledge of their languages
develops together (Marchman et al., 2010). Generally, the size of a child’s vocabulary in one lan-
guage is not a good predictor of their vocabulary in the other (Conboy & Thal, 2006; Hurtado
et al., 2014). Instead, children’s vocabulary in a particular language depends on their history of
experience with that language (Hurtado et al., 2014; Marchman et al., 2017; Place & Hof, 2011;
Thordardottir, 2011). Children know more words in their more frequently heard language and
show deeper processing of words in that language (Legacy et al., 2016; Oller et al., 2007). From
studies with monolinguals, it is widely accepted that children’s knowledge of language is related
to the quantity/quality of the input they receive (e.g., Hart & Risley, 1995). Interestingly, studies
with monolingual children have inherent limitations in that they cannot separate child-specifc
factors (e.g., cognitive abilities) or general environmental factors (e.g., family socioeconomic sta-
tus, parent investment, community norms) from the infuence of language experience. However,
studies with bilingual children can better dissociate these infuences, since the same child receives
diferent input in diferent languages (Marchman et al., 2010; Pearson et al., 1997). Hurtado
and colleagues (2014) reported that bilingual toddlers’ relative exposure to Spanish vs. English
predicted the ratio of Spanish:English words that they produced. This fnding underscores that a
child’s learning in one language is driven by experience with that language, not by overall experi-
ence across both languages.
Strategies for learning new words can differ between

bilingual and monolingual children
Language experience not only changes the number of words children know, but also how they
learn new words. For instance, bilingual infants may be more fexible in the cues they deem rel-
evant for meaning. At 18 months, monolingual infants do not map candidate words that difer
only by non-native features (e.g., pitch contours) onto diferent objects (Hay et al., 2015), but
bilinguals can use these cues to distinguish reference (Graf Estes & Hay, 2015). As their knowledge
of their native language(s) increases, both monolingual and bilingual infants become less inclined
to interpret tones or clicks as referentially meaningful when those features are not contrastive in
their language (Graf Estes & Hay, 2015; May & Werker, 2014), providing additional evidence that
experience changes the cues that infants consider for meaning. In addition to shaping the inter-
pretation of sounds, bilingual experience seems to lead children to rely less on the assumption that
there will be one-to-one mappings between labels and objects. Monolingual children tend to infer
that unknown labels refer to novel objects (Markman & Wachtel, 1988), a phenomenon known
as “mutual exclusivity” (see Textbox 2.1). To test how language experience infuences children’s
use of this strategy, Byers-Heinlein and Werker (2009) presented monolingual, bilingual, and tri-
lingual 17-month-olds with familiar and unfamiliar objects. While monolinguals reliably looked
to the unfamiliar object upon hearing a novel label, bilinguals were only marginally more likely
to do so, and trilinguals were at chance. Subsequent studies showed that bilingual infants and even
adults are more willing than monolinguals to accept multiple labels for the same object (Benitez
et al., 2016; Byers-Heinlein et al., 2014; Kandhadai et al., 2017). Importantly, bilinguals are not
permanently incapable of using the heuristic that novel labels likely refer to novel objects; by three
years, monolingual and bilingual children use mutual exclusivity equally (Frank & Poulin-Dubois,
2002). Thus, it appears that bilingual children develop this assumption more slowly and perhaps
apply it more conservatively.
16
Sentences
Beyond learning individual words, infants must learn to combine words and understand them fu-
ently in diferent contexts. Examining bilingual children’s comprehension is important because it
allows us to explore global efects of bilingual experience on children’s language processing, to inves-
tigate the efects of experience with a specifc language, and to observe how children separate and
integrate two languages.
Bilingual infants’ processing of each of their languages is largely independent

Just as children’s vocabulary knowledge can vary both within and across languages, so too can their
comprehension. For monolingual children, vocabulary development is tightly coupled with under-
standing spoken language (Fernald & Marchman, 2012). Bilingual children’s processing efciency
is related to their vocabulary knowledge within a language, but not across languages (Hof et al.,
2018; Hurtado et al., 2014; Marchman et al., 2010). In addition, predictably, children show better
comprehension of their more frequent language (Hurtado et al., 2014). While many studies show no
signifcant relation between children’s understanding of their two languages (Hurtado et al., 2014;
Marchman et al., 2010; Potter et al., 2019), there have been reports of negative relations between
children’s skills across languages. For instance, for Spanish-English bilingual preschoolers in the U.S.,
the development of grammatical skills in English can be a negative predictor of Spanish profciency
(Hof et al., 2018). Thus, children’s specifc experience with each language, rather than the total
amount of input they experience, appears to drive their understanding.
Language switching can disrupt bilingual infants’ processing, but only sometimes
Bilingual infants have to develop efcient processing skills in their two languages independently, and
they also must be able to integrate the two languages in real time, including understanding sentences
that contain words in both languages (e.g., Look at the perro!). Sentences that include such “code
switching” (see van Hell, 2023 [this volume]) tend to slow listeners’ processing, although not sizably.
Byers-Heinlein and colleagues (2017) reported that 20-month-olds learning French and English were
less accurate to recognize familiar nouns that appeared in mixed-language sentences compared to
single-language sentences. Complementary studies showed that mixed-language sentences can disrupt
comprehension for Spanish-English bilinguals of similar ages (Morini & Newman, 2019; Potter et al.,
2019). A careful look at these results shows that not all switches between languages afect processing.
For instance, switches spanning sentence boundaries (e.g., That one looks fun! The dog!) do not disrupt
comprehension (Byers-Heinlein et al., 2017). Furthermore, there are important diferences between
children’s dominant vs. non-dominant languages. Across diferent sentence types, children show better
comprehension of nouns in their dominant language (Potter et al., 2019). Relatedly, children experi-
ence greater processing costs when hearing a switch from sentence frames in their dominant language
to nouns in their non-dominant language, compared to vice versa (Byers-Heinlein et al., 2017; Potter
et al., 2019). Given that infants can efciently process the nouns in single-language sentences, this sug-
gests subtle diferences in how lexical knowledge interacts with language switching.
Consequences of bilingualism
Language processing recruits domain-general cognitive processes such as memory and attention (Saf-
fran, 2003) and is intrinsically linked to social learning (Kinzler et al., 2007). Because dual-language
environments contain complex interactions between diferent people speaking diferent languages,
the added challenges of managing multilingual input can infuence children’s processing across a
17
variety of domains (Akhtar & Menjivar, 2012). Here, we review fndings and debates about the
efects of bilingualism on children’s non-linguistic learning and behavior and consider mutual infu-
ences of language and cognition.
Bilingualism may or may not confer cognitive benefts

A widely-discussed topic in contemporary bilingual research is the idea that bilinguals often outperform
monolinguals on tasks that require cognitive fexibility and control (Bialystok et al., 2012; Kovács &
Mehler, 2009; Poulin-Dubois et al., 2011). Termed the “bilingual advantage” (see Poarch, 2023 [this
volume]), diferences between monolingual and bilingual participants have often been interpreted as
evidence that the constant need to suppress distraction from another language supports the develop-
ment of attentional control (Bialystok et al., 2012). Enhancements in bilinguals’ ability to shift attention
have even been observed in preverbal infants (Kovács & Mehler, 2009). This has led to speculation
that monolingual-bilingual diferences could arise from bilinguals’ experience monitoring the use of
language (Byers-Heinlein et al., 2017; Sebastián-Gallés et al., 2012). However, recently, researchers
have questioned the reliability and size of these efects (Dick et al., 2019; Duñabeitia et al., 2014; Nich-
ols et al., 2020; Paap & Greenberg, 2013), and it has been argued diferences between monolingual
and bilingual populations may be explained by other factors, such as socioeconomic status diferences
(Morton & Harper, 2007). Nevertheless, given that bilingual infants may show diferences in their
development of cognitive skills, we suggest these efects may be similar to benefts of other enriching
experiences. For instance, exposure to rich linguistic input in monolingual environments is related
to growth in cognitive skills including memory and attention (Rowe et al., 2017). Improvements on
cognitive tasks have also been found after children receive training with music (Moreno et al., 2011) or
engage in exercise programs (Davis et al., 2011). Therefore, cognitive skills appear to develop through
diverse experiences, and exposure to multiple languages may be just one contributor.
Bilingual experience may heighten children’s use of social cues

Using language is not a purely cognitive task, and bilingual environments present unique social chal-
lenges. To follow the conventions of their language communities and determine which people speak
each of their languages, children must use a variety of information. For example, language input
is often accompanied by visual cues, such as the presence of a talking face. Attending to speakers’
mouths provides cues that support speech perception (Munhall & Vatikiotis-Bateson, 1998), while
eyes can ofer information about speakers’ intent (Emery, 2000). Early in development, infants may
need additional information to decode the speech signal, and younger infants tend to look more at
mouths and shift their gaze more to eyes as they become increasingly skilled at processing speech
(Lewkowicz & Hansen-Tift, 2012). Ayneto and Sebastián-Gallés (2017) showed that at eight months,
bilingual infants tend to continue to look at mouths, while monolinguals have shifted to the eyes.
Perhaps as a result of these diferent patterns of attention, bilingual, but not monolingual, infants of
this age can diferentiate languages using dynamic visual cues alone (Sebastián-Gallés et al., 2012),
further supporting the view that bilingual exposure may change infants’ social experience. However,
some recent studies fnd no reliable diferences in how monolingual and bilingual infants scan faces
or use gaze-related cues (Byers-Heinlein et al., 2021; Morin-Lessard et al., 2019). The specifc lan-
guages that children hear could help explain diferences across studies. For instance, bilingual infants
learning similar languages (Spanish and Catalan) show increased interest in the mouth compared to
bilingual infants learning distinct pairs of languages (e.g., Spanish and German; Birulés et al., 2019).
Infants hearing two similar-sounding languages may need more disambiguating information, leading
them to attend to speakers’ mouths. Alternatively, it may be that language experience exerts only
minor infuence on infants’ attention, because infants in all environments must seek out information
18
that best supports their perception and learning (Potter & Lew-Williams, 2019). As children get
older, their social environments become more complex, and it has been suggested that the need to
communicate with people using diferent languages supports children’s understanding of the social
world (Byers-Heinlein et al., 2014; Yow & Markman, 2011). Young children often struggle to real-
ize that other people’s knowledge is diferent from theirs (Gopnik & Astington, 1988), and there is
evidence that bilingual children outperform monolinguals in understanding other people’s mental
states, beliefs, and intentions. For example, bilingual children may be better able to understand that
an informant cannot always see objects that are visible to the child (Fan et al., 2015), or may be more
sensitive to social cues that help resolve ambiguity, such as gaze direction (Yow & Markman, 2011).
Theoretical perspectives and approaches

Across these ten key domains of research in early bilingualism, we have described how perceptual,
cognitive, and social abilities shape children’s learning of two languages. Although bilingualism is
often described as a unique context for learning, we now seek to ofer theoretical explanations that
unite research on monolingual and bilingual learning. We argue that bilingual language development
is best understood by considering how general learning mechanisms are deployed in complex audi-
tory and social environments.
Sounds
Interestingly, diferences between monolingual and bilingual infants’ learning of sounds can be con-
strued either as refecting delays in bilinguals’ language development, or as a perceptual advantage. The
slower development of native-language biases could suggest that bilingual experience leads to defcits
in early language skills, because monolinguals’ ability to perceive non-native contrasts is negatively cor-
related with knowledge of their native language (Kuhl et al., 2005). Because bilingual infants have less
experience with each language individually, they have fewer opportunities to learn which distinctions
are relevant. Furthermore, bilingual environments contain more variability, including input from both
native and non-native speakers (Byers-Heinlein & Fennell, 2014). The Processing Rich Information
from Multidimensional Interactive Representations (PRIMIR) model (Curtin et al., 2011) posits that
infants use distributional patterns of their input to infer the presence of diferent phonetic categories,
and exposure to less consistent input could slow their convergence onto a particular set of contrasts,
leading bilinguals to develop native language biases more slowly than monolinguals.
However, it can also be argued that the ability to perceive more sound contrasts refects percep-
tual enhancement. Petitto and colleagues (2012) put forward the “perceptual wedge” hypothesis and
propose that bilinguals’ longer-lasting sensitivities refect fexibility in cognition as well as perception
that could both support infants’ learning of the languages to which they are currently exposed and
facilitate learning of new languages in the future. Regardless of whether bilingual infants’ speech
perception is viewed as delayed, enhanced, or merely diferent from monolinguals’, the divergent
developmental trajectory of monolingual and bilingual infants highlights the fact that infants’ sensi-
tivities are tuned by their experience with language.
Words
In monolingual environments, there are likely to be highly reliable co-occurrences between an
object and its label, and theories of cross-situational word learning suggest that children learn new
words by uncovering these associations (Smith & Yu, 2008). In contrast, in bilingual environments,
the same object is called by diferent names, and learners must track multiple labels for the same
object. Regular experience with multiple languages appears to change learners’ assumptions about
19
the statistical relations in the environment. Benitez and colleagues (2016) presented adults with
objects that consistently co-occurred with a single label and other objects that co-occurred equally
often with two diferent labels, creating competition. Adults with multilingual experience appeared
to have less difculty resolving competition between labels. Thus, being regularly presented with
cases in which multiple names are used with roughly equal probability to refer to the same object
reduces the bias toward mutual exclusivity. The emergence of diferent learning strategies highlights
how examining diferences between children growing up in diferent types of language environments
can provide insight into the fundamental mechanisms of early learning.
Sentences
We have argued that the efects of language switching can be understood by considering how young
learners combine information in processing, notably their learning of statistical regularities of their
environment and their emerging knowledge of individual words (Potter et al., 2019). Hearing words
following a familiar sentence frame (e.g., Look at the . . .) facilitates comprehension for both mono-
linguals and bilinguals (Fernald & Hurtado, 2006; Morini & Newman, 2019). This is consistent with
statistical learning accounts that emphasize that infants are attuned to predictable co-occurrences
(Safran, 2003). Because language switches are less frequent than single-language utterances (Byers-
Heinlein, 2014), they interfere with comprehension. Likewise, because children encounter words in
their dominant language more often, they likely have more robust knowledge of these words (Good-
man et al., 2008), which allows them to overcome the difculty of a switch from their non-dominant
to dominant language more easily than in the reverse direction. Therefore, it appears that diferences
in how bilingual infants process single- and mixed-language sentences in their two languages can be
attributed to general principles of statistical learning and sensitivity to word frequency.
Consequences of bilingualism
If, as some studies show, bilingual experience can shape infants’ cognition and use of social cues, then
the question becomes why. We consider three potential explanations, but they are neither mutually
exclusive, nor likely to be complete. One possibility is that processing the dynamics of bilingual input
is demanding, and therefore could be enriching. For instance, Byers-Heinlein and colleagues (2017)
showed that comprehending certain types of bilingual input, such as code switches, requires increased
efort, which could be a “desirable difculty” that underlies later-observed diferences in cognitive
fexibility and learning (see also Bogulski et al., 2019). Another possibility is that bilingual experience
heightens children’s attention to nonverbal cues, such as eye gaze and gesture, which may be useful
across language contexts for engaging in successful communication. This dynamic use of reliable cues
may support children’s learning (Potter & Lew-Williams, 2019), with benefts for social cognition. A
third possibility is that experience interacting with people with variable language knowledge helps
children understand that diferent people hold diferent knowledge states. Bilingual infants recognize
that monolingual speakers do not understand words in a foreign language (Byers-Heinlein et al.,
2014), and bilingual children readily adjust their language use when interacting with monolingual or
bilingual adults (Genesee et al., 1996). This adaptation refects sensitivity to other people’s knowl-
edge, and diferences in children’s early social experience may have downstream consequences for
understanding the social world, as well as cross-domain efects on their cognition.
Current trends and future directions

There has a been a dramatic increase in research on early bilingualism over the last ten years (Fen-
nell & Lew-Williams, 2018), which aims to capture the linguistic realities of many infants’ experience
20
and take advantage of the “natural experiment” of bilingual environments (Pearson et al., 1997). We
have highlighted important progress that this feld has made in deepening our understanding of both
the unique psycholinguistic challenges that bilingual infants face and the general cognitive processes
involved in early language learning. We aimed to increase links between bilingualism research and
the cognitive sciences by explaining how ten key fndings about early bilingualism can be understood
through the same domain-general cognitive mechanisms that support monolingual development.
Signifcant gaps in knowledge still remain that need to be addressed to provide a clearer picture of
the nature of bilingual experience and learning (see Textbox 2.2). Here, we propose a few areas that
we hope will be addressed in future research.
Commonalities and differences across different

languages and populations
We have mostly described research fndings as though they apply to all groups of bilinguals, yet
there may be very diferent challenges associated with learning diferent pairs of languages (or more
than two languages) as well across diferent cultural contexts. For instance, children learning related
languages, such as Spanish and Catalan, may rely on diferent cues to segregate their languages,
compared to children learning typologically distinct languages, such as English and Mandarin. Since
related languages share more cognates, this overlap could infuence how children learn words across
languages. Moreover, some communities are more bilingual than others, and languages have a more
or less prestigious status (political or otherwise) in the larger society. Children’s learning of the major-
ity and minority languages often difer (Wright et al., 2000), further highlighting the importance of
exploring similarities and diferences across bilingual populations.
Scrutinizing children’s daily experience

Every bilingual home is diferent, and parents often struggle to report how much their child hears of
one language over another. There have been recent endeavors to describe the speech that children
hear in diferent cultures (e.g., Bergelson et al., 2019; Casillas et al., 2020; Cristià et al., 2019), but a
relative paucity of data exists from bilingual homes (but see Hurtado et al., 2014; Marchman et al.,
2017; Orena et al., 2019). Observations from bilingual homes could ofer the opportunity to explore
“how much” input children typically need in each language to become profcient and to test how
diferences in parents’ strategies (e.g., mixing vs. separating languages; having diferent people interact
with the child in diferent languages) infuence children’s language outcomes. In addition, there is
little information available about the language(s) that bilingual children produce in diferent settings,
and it could be informative to understand how children’s skills and preferences infuence their learn-
ing environments.
Assessing and supporting children at risk for

language diffculties
Finally, it is important to remember that many bilingual families are concerned about whether the
presence of multiple languages will impede vs. support their child’s development and success in
school. In particular, many want to know how to best determine when bilingual children have lan-
guage delays and how to accurately assess their knowledge in multiple languages at once. A deeper
understanding of how to evaluate childhood bilingualism and second language learning will be
critical for building theories of cognition and language, as well as informing policies, therapies, and
educational eforts focused on bilingual populations.
21
Textbox 2.2 Open questions and issues
How similar or diferent is children’s language learning across diferent bilingual communities?
How do characteristics of language input that bilingual children hear change over time?
How much experience do bilingual children need with a language to become profcient?
What are the cognitive and linguistic benefts of early bilingualism?
What are the learning-related vulnerabilities, if any, of growing up in a bilingual environment?
Further reading
Grosjean, F., & Byers-Heinlein, K. (2018). The listening bilingual: Speech perception, comprehension, and bilingualism.
John Wiley & Sons.
This book discusses early learning of sounds.
Grüter, T., & Paradis, J. (Eds.). (2014). Input and experience in bilingual development (Vol. 13). John Benjamins.
This book reviews research on language input in bilingual environments.
Marchman, V. A., Fernald, A., & Hurtado, N. (2010). How vocabulary size in two languages relates to efciency
in spoken word recognition by young Spanish-English bilinguals. Journal of Child Language, 37(4), 817–840.
This article describes research on vocabulary knowledge and language processing.
References
Akhtar, N., & Menjivar, J. A. (2012). Cognitive and linguistic correlates of early exposure to more than one
language. In J. B. Benson (Ed.), Advances in child development and behavior (Vol. 42, pp. 41–78). Academic Press.
Ayneto, A., & Sebastián-Gallés, N. (2017). The infuence of bilingualism on the preference for the mouth region
of dynamic faces. Developmental Science, 20(1), e12446.
Benitez, V. L., Yurovsky, D., & Smith, L. B. (2016). Competition between multiple words for a referent in cross-
situational word learning. Journal of Memory and Language, 90, 31–48.
Bergelson, E., Casillas, M., Soderstrom, M., Seidl, A., Warlaumont, A. S., & Amatuni, A. (2019). What do
north American babies hear? A large-scale cross-corpus analysis. Developmental Science, 22(1), e12724.
Bergelson, E., & Swingley, D. (2012). At 6–9 months, human infants know the meanings of many common
nouns. Proceedings of the National Academy of Sciences, 109(9), 3253–3258.
Bialystok, E., Craik, F. I., & Luk, G. (2012). Bilingualism: Consequences for mind and brain. Trends in Cognitive
Sciences, 16(4), 240–250.
Birulés, J., Bosch, L., Brieke, R., Pons, F., & Lewkowicz, D. J. (2019). Inside bilingualism: Language background
modulates selective attention to a talker’s mouth. Developmental Science, 22(3), e12755.
Bogulski, C. A., Bice, K., & Kroll, J. F. (2019). Bilingualism as a desirable difculty: Advantages in word learn-
ing depend on regulation of the dominant language. Bilingualism: Language and Cognition, 22(5), 1052–1067.
Bosch, L., & Sebastián-Gallés, N. (2001). Evidence of early language discrimination abilities in infants from
bilingual environments. Infancy, 2(1), 29–49.
Byers-Heinlein, K. (2014). Languages as categories: Reframing the “one language or two” question in early
bilingual development. Language Learning, 64(s2), 184–201.
Byers-Heinlein, K., Chen, K. H., & Xu, F. (2014). Surmounting the tower of babel: Monolingual and bilingual
2-year-olds’ understanding of the nature of foreign language words. Journal of Experimental Child Psychology,
119, 87–100.
Byers-Heinlein, K., & Fennell, C. T. (2014). Perceptual narrowing in the context of increased variation: Insights
from bilingual infants. Developmental Psychobiology, 56(2), 274–291.
Byers-Heinlein, K., & The ManyBabies Consortium (2021). The development of gaze following in monolin-
gual and bilingual infants: A multi-lab study. Infancy, 26, 4–38.
Byers-Heinlein, K., Morin-Lessard, E., & Lew-Williams, C. (2017). Bilingual infants control their languages as
they listen. Proceedings of the National Academy of Sciences, 114(34), 9032–9037.
22
Byers-Heinlein, K., & Werker, J. F. (2009). Monolingual, bilingual, trilingual: Infants’ language experience
infuences the development of a word-learning heuristic. Developmental Science, 12(5), 815–823.
Casillas, M., Brown, P., & Levinson, S. C. (2020). Early language experience in a Tseltal Mayan village. Child
Development, 91(5), 1819–1835.
Conboy, B. T., & Thal, D. J. (2006). Ties between the lexicon and grammar: Cross-sectional and longitudinal
studies of bilingual toddlers. Child Development, 77(3), 712–735.
Cristià, A., Dupoux, E., Gurven, M., & Stieglitz, J. (2019). Child-directed speech is infrequent in a forager-
farmer population: A time allocation study. Child Development, 90(3), 759–773.
Curtin, S., Byers-Heinlein, K., & Werker, J. F. (2011). Bilingual beginnings as a lens for theory development:
PRIMIR in focus. Journal of Phonetics, 39(4), 492–504.
Davis, C. L., Tomporowski, P. D., McDowell, J. E., Austin, B. P., Miller, P. H., Yanasak, N. E., Allison, J. D., &
Naglieri, J. A. (2011). Exercise improves executive function and achievement and alters brain activation in
overweight children: A randomized, controlled trial. Health Psychology, 30(1), 91.
De Houwer, A., Bornstein, M. H., & Putnick, D. L. (2014). A bilingual—monolingual comparison of young
children’s vocabulary size: Evidence from comprehension and production. Applied Psycholinguistics, 35(6),
1189–1211.
Dick, A. S., Garcia, N. L., Pruden, S. M., Thompson, W. K., Hawes, S. W., Sutherland, M. T., Reidel, M. C.,
Laird, A. R., & Gonzalez, R. (2019). No evidence for a bilingual executive function advantage in the nation-
ally representative ABCD study. Nature Human Behaviour, 3(7), 692–701.
Duf, F. J., Reen, G., Plunkett, K., & Nation, K. (2015). Do infant vocabulary skills predict school-age language
and literacy outcomes? Journal of Child Psychology and Psychiatry, 56(8), 848–856.
Duñabeitia, J. A., Hernández, J. A., Antón, E., Macizo, P., Estévez, A., Fuentes, L. J., & Carreiras, M. (2014).
The inhibitory advantage in bilingual children revisited: Myth or reality? Experimental Psychology, 61(3),
234–251.
Emery, N. J. (2000). The eyes have it: The neuroethology, function and evolution of social gaze. Neuroscience &
Biobehavioral Reviews, 24(6), 581–604.
Fan, S. P., Liberman, Z., Keysar, B., & Kinzler, K. D. (2015). The exposure advantage: Early exposure to a mul-
tilingual environment promotes efective communication. Psychological Science, 26(7), 1090–1097.
Fennell, C. T., & Byers-Heinlein, K. (2014). You sound like Mommy: Bilingual and monolingual infants learn
words best from speakers typical of their language environments. International Journal of Behavioral Develop-
ment, 38(4), 309–316.
Fennell, C. T., & Lew-Williams, C. (2018). Early bilingual word learning. In G. Westermann & N. Mani (Eds.),
Early word learning (pp. 110–122). Routledge.
Fenson, L., Dale, P. S., Reznick, J. S., Bates, E., Thal, D. J., Pethick, S. J., Tomasello, M., Mervis, C. B., &
Stiles, J. (1994). Variability in early communicative development. Monographs of the Society for Research in Child
Development, 59(5), i.
Fernald, A., & Hurtado, N. (2006). Names in frames: Infants interpret words in sentence frames faster than
words in isolation. Developmental Science, 9(3), F33–F40.
Fernald, A., & Marchman, V. A. (2012). Individual diferences in lexical processing at 18 months predict vocabu-
lary growth in typically developing and late-talking toddlers. Child Development, 83(1), 203–222.
Frank, I., & Poulin-Dubois, D. (2002). Young monolingual and bilingual children’s responses to violation of the
mutual exclusivity principle. International Journal of Bilingualism, 6(2), 125–146.
Genesee, F., Boivin, I., & Nicoladis, E. (1996). Talking with strangers: A study of bilingual children’s commu-
nicative competence. Applied Psycholinguistics, 17(4), 427–442.
Gonzalez-Barrero, A. M., Schott, E., & Byers-Heinlein, K. (2020). Bilingual adjusted vocabulary: A develop-
mentally-informed bilingual vocabulary measure. PsyArXiv. https://doi.org/10.31234/osf.io/x7s4u
Goodman, J. C., Dale, P. S., & Li, P. (2008). Does frequency count? Parental input and the acquisition of vocabu-
lary. Journal of Child Language, 35(3), 515–531.
Gopnik, A., & Astington, J. W. (1988). Children’s understanding of representational change and its relation to
the understanding of false belief and the appearance-reality distinction. Child Development, 26–37.
Graf Estes, K., & Hay, J. F. (2015). Flexibility in bilingual infants’ word learning. Child Development, 86(5),
1371–1385.
Hart, B., & Risley, T. R. (1995). Meaningful diferences in the everyday experience of young American children. Paul H
Brookes Publishing.
Hay, J. F., Graf Estes, K., Wang, T., & Safran, J. R. (2015). From fexibility to constraint: The contrastive use of
lexical tone in early word learning. Child Development, 86(1), 10–22.
Hof, E., Quinn, J. M., & Giguere, D. (2018). What explains the correlation between growth in vocabulary and
grammar? New evidence from latent change score analyses of simultaneous bilingual development. Develop-
mental Science, 21(2), e12536.
23
Hurtado, N., Grüter, T., Marchman, V. A., & Fernald, A. (2014). Relative language exposure, processing ef-
ciency and vocabulary in Spanish—English bilingual toddlers. Bilingualism: Language and Cognition, 17(1),
189–202.
Kandhadai, P., Hall, D. G., & Werker, J. F. (2017). Second label learning in bilingual and monolingual infants.
Developmental Science, 20(1), e12429.
Kinzler, K. D., Dupoux, E., & Spelke, E. S. (2007). The native language of social cognition. Proceedings of the
National Academy of Sciences, 104(30), 12577–12580.
Kovács, A. M., & Mehler, J. (2009). Cognitive gains in 7-month-old bilingual infants. Proceedings of the National
Academy of Sciences of the United States of America, 106(16), 6556–6560.
Kuhl, P. K., Conboy, B., Padden, D., Nelson, T., & Pruitt, J. (2005). Early speech perception and later language
development: Implications for the “critical period”. Language Learning and Development, 1(3&4), 237–264.
Kuhl, P. K., Stevens, E., Hayashi, A., Deguchi, T., Kiritani, S., & Iverson, P. (2006). Infants show a facilita-
tion efect for native language phonetic perception between 6 and 12 months. Developmental Science, 9(2),
F13–F21.
Kuipers, J. R., & Thierry, G. (2012). Event-related potential correlates of language change detection in bilingual
toddlers. Developmental Cognitive Neuroscience, 2, 97–102.
Legacy, J., Zesiger, P., Friend, M., & Poulin-Dubois, D. (2016). Vocabulary size, translation equivalents, and
efciency in word recognition in very young bilinguals. Journal of Child Language, 43, 760–783.
Leopold, W. (1954). A child’s learning of two languages. In H. J. Mueller (Ed.), Georgetown university round table
on languages and linguistics (Vol. 7, pp. 19–30). Georgetown University Press.
Lewkowicz, D. J., & Hansen-Tift, A. M. (2012). Infants deploy selective attention to the mouth of a talking face
when learning speech. Proceedings of the National Academy of Sciences, 109(5), 1431–1436.
Marchman, V. A., Martínez, L. Z., Hurtado, N., Grüter, T., & Fernald, A. (2017). Caregiver talk to young
Spanish-English bilinguals: Comparing direct observation and parent-report measures of dual-language
exposure. Developmental Science, 20(1), e12425.
Markman, E. M., & Wachtel, G. F. (1988). Children’s use of mutual exclusivity to constrain the meanings of
words. Cognitive Psychology, 20(2), 121–157.
Mattock, K., Polka, L., Rvachew, S., & Krehm, M. (2010). The frst steps in word learning are easier when the
shoes ft: Comparing monolingual and bilingual infants. Developmental Science, 13(1), 229–243.
May, L., & Werker, J. F. (2014). Can a click be a word?: Infants’ learning of non-native words. Infancy, 19(3),
281–300.
Mehler, J., Jusczyk, P., Lambertz, G., Halsted, N., Bertocini, J., & Amiel-Tison, C. (1988). A precursor of lan-
guage acquisition in young infants. Cognition, 29, 143–178.
Molnar, M., Gervain, J., & Carreiras, M. (2014). Within-rhythm class native language discrimination abilities of
basque-Spanish monolingual and bilingual infants at 3.5 months of age. Infancy, 19(3), 326–337.
Moreno, S., Bialystok, E., Barac, R., Schellenberg, E. G., Cepeda, N. J., & Chau, T. (2011). Short-term music
training enhances verbal intelligence and executive function. Psychological Science, 22(11), 1425–1433.
Morini, G., & Newman, R. S. (2019). Dónde está la ball? Examining the efect of code switching on bilingual
children’s word recognition. Journal of Child Language, 46(6), 1238–1248.
Morin-Lessard, E., Poulin-Dubois, D., Segalowitz, N., & Byers-Heinlein, K. (2019). Selective attention to the
mouth of talking faces in monolinguals and bilinguals aged 5 months to 5 years. Developmental Psychology,
55(8), 1640–1655.
Morton, J. B., & Harper, S. N. (2007). What did Simon say? Revisiting the bilingual advantage. Developmental
Science, 10(6), 719–726.
Munhall, K. G., & Vatikiotis-Bateson, E. (1998). The moving face during speech communication. Hearing by
Eye II: Advances in the Psychology of Speechreading and Auditory-Visual Speech, 123–139.
Nazzi, T., Jusczyk, P. W., & Johnson, E. K. (2000). Language discrimination by English-learning 5-month-olds:
Efects of rhythm and familiarity. Journal of Memory and Language, 19, 1–19.
Nichols, E. S., Wild, C. J., Stojanoski, B., Battista, M. E., & Owen, A. M. (2020). Bilingualism afords no gen-
eral cognitive advantages: A population study of executive function in 11,000 people. Psychological Science,
31(5), 548–567.
Oller, D. K., Pearson, B. Z., & Cobo-Lewis, A. B. (2007). Profle efects in early bilingual language and literacy.
Applied Psycholinguistics, 28(2), 191–230.
Orena, A. J., Byers-Heinlein, K., & Polka, L. (2019). Reliability of the language environment analysis record-
ing system in analyzing French—English bilingual speech. Journal of Speech, Language, and Hearing Research,
62(7), 2491–2500.
24
Paap, K. R., & Greenberg, Z. I. (2013). There is no coherent evidence for a bilingual advantage in executive
processing. Cognitive Psychology, 66(2), 232–258.
Pearson, B. Z., Fernandez, S. C., Lewedeg, V., & Oller, D. K. (1997). The relation of input factors to lexical
learning by bilingual infants. Applied Psycholinguistics, 18, 41–58.
Pearson, B. Z., Fernández, S. C., & Oller, D. K. (1993). Lexical development in bilingual infants and toddlers:
Comparison to monolingual norms. Language Learning, 43(1), 93–120.
Petitto, L. A., Berens, M. S., Kovelman, I., Dubins, M. H., Jasinska, K., & Shalinsky, M. (2012). The “perceptual
wedge” hypothesis as the basis for bilingual babies’ phonetic processing advantage: New insights from fNIRS
brain imaging. Brain and Language, 121(2), 130–143.
Place, S., & Hof, E. (2011). Properties of dual language exposure that infuence two-year-olds’ bilingual prof-
ciency. Child Development, 82(6), 1834–1849.
Poarch, G. J. (2023). Cognitive efects of bilingualism. In A. Godfroid & H. Hopp (Eds.). The Routledge handbook
of second language acquisition and psycholinguistics (pp. 426–438). Routledge.
Potter, C. E., Fourakis, E., Morin-Lessard, E., Byers-Heinlein, K., & Lew-Williams, C. (2019). Bilingual tod-
dlers’ comprehension of mixed sentences is asymmetrical across their two languages. Developmental Science,
22(4), e12794.
Potter, C. E., & Lew-Williams, C. (2019). Infants’ selective use of reliable cues in multidimensional language
input. Developmental Psychology, 55(1), 1–8.
Poulin-Dubois, D., Blaye, A., Coutya, J., & Bialystok, E. (2011). The efects of bilingualism on toddlers’ execu-
tive functioning. Journal of Experimental Child Psychology, 108(3), 567–579.
Rowe, M. L., Leech, K. A., & Cabrera, N. (2017). Going beyond input quantity: Wh-questions matter for tod-
dlers’ language and cognitive development. Cognitive Science, 41, 162–179.
Safran, J. R. (2003). Statistical language learning: Mechanisms and constraints. Current Directions in Psychological
Science, 12(4), 110–114.
Sebastián-Gallés, N., Albareda-Castellot, B., Weikum, W. M., & Werker, J. F. (2012). A bilingual advantage in
visual language discrimination in infancy. Psychological Science, 23(9), 994–999.
Singh, L., Loh, D., & Xiao, N. G. (2017). Bilingual infants demonstrate perceptual fexibility in phoneme dis-
crimination but perceptual constraint in face discrimination. Frontiers in Psychology, 8, 1563.
Smith, L., & Yu, C. (2008). Infants rapidly learn word-referent mappings via cross-situational statistics. Cognition,
106(3), 1558–1568.
Souza, A. L., Byers-Heinlein, K., & Poulin-Dubois, D. (2013). Bilingual and monolingual children prefer
native-accented speakers. Frontiers in Psychology, 4, 953.
Thordardottir, E. (2011). The relationship between bilingual exposure and vocabulary development. International
Journal of Bilingualism, 15(4), 426–445.
van Hell, J. G. (2023). Code-switching. In A. Godfroid & H. Hopp (Eds.). The Routledge handbook of second
language acquisition and psycholinguistics (pp. 255–267). Routledge.
Wright, S. C., Taylor, D. M., & Macarthur, J. (2000). Subtractive bilingualism and the survival of the Inuit lan-
guage: Heritage-versus second-language education. Journal of Educational Psychology, 92(1), 63.
Yow, W. Q., & Markman, E. M. (2011). Young bilingual children’s heightened sensitivity to referential cues
young bilingual children’s heightened sensitivity to referential cues. Journal of Cognition and Development,
12(1), 12–31.
25
3
OF CHILD SECOND LANGUAGE
ACQUISITION
Sarah Schimke
Introduction
This chapter provides a selective review of the psycholinguistics of child second language (L2) acquisi-
tion. Child L2 acquisition refers to language acquisition in learners that have previous knowledge of
a frst language (L1), and who are children. This defnition separates child L2 acquisition from other
types of multilingualism, in particular, simultaneous bilingual L1 acquisition, in which the onset of both
languages occurs simultaneously, and adult L2 acquisition, in which the learners are not children. The
review will include studies on learners with an age of onset of the L2 between approximately one year
of age and puberty, and in diferent learning situations, e.g., with difering amounts and types of input
in the L1 and the L2. I will focus on situations in which the L2 is the majority language of the society,
and in which the age of onset of the L2 is between four and eight years of age. Many studies have
studied child L2 acquisition in this context for practical reasons, since it frequently occurs, but also for
theoretical reasons, because it is often assumed that the acquisition of the L1 reaches an important mile-
stone around age four, with core aspects of the L1 in place, and that language learning processes may
start to change signifcantly from age eight onwards (Unsworth, 2005), resembling adult L2 learning.
In my review, I will privilege insights provided by studies that look at language processing in real time.
Critical period/sensitive period: The critical period hypothesis (Lenneberg, 1967) assumes that neuro-
biological prerequisites for language learning are available during a specifc time window only, which
is assumed to close at puberty at the latest. Later approaches have assumed a sensitive rather than a
critical period, meaning that the diferences between learning processes at diferent ages are gradual
rather than categorical. There may be diferent sensitive periods for diferent linguistic domains
(Meisel, 2013).
Child-internal and child-external variables: Child-external factors refer to all variables that directly or
indirectly infuence the amount and type of input, and child-internal factors refer to age, L1, or other
learner properties.
Explicit and implicit knowledge: Explicit knowledge is conscious and typically declarative, while implicit
knowledge is not consciously accessible and typically procedural. The weight of the two types of
knowledge may change with age of onset (Ellis, 2017, see also Godfroid, 2023 [this volume]).
26 DOI: 10.4324/9781003018872-4
The psycholinguistics of child second language acquisition
Historical and theoretical perspectives

On top of giving insights into the course of child L2 acquisition, one motivation to study child L2
acquisition has been to arrive at a better understanding of other types of acquisition. In particular,
child L2 acquisition was studied to understand diferences between (monolingual or bilingual) child
L1 acquisition, on the one hand, and adult L2 acquisition, on the other. Making such a comparison is
critical for diferent theoretical approaches: From a generative perspective, it is informative to address
the question whether the innate and domain-specifc mechanisms that are assumed to determine
L1 acquisition are still in place in L2 acquisition (Schwartz, 2004). From a functional perspective, a
central interest is to determine whether the weight of language-specifc and language-independent
principles of form-function associations is diferent in L1 compared to L2 acquisition (Dimroth,
2018). Finally, usage-based approaches to language acquisition often ask how general and domain-
independent principles of input processing interact with the previously acquired knowledge in L2
acquisition as a function of experience relative to L1 acquisition (Wulf & Ellis, 2018).
However, any direct comparison between L1 children and L2 adults confounds two variables: age
of onset of acquisition and the infuence of the previously acquired L1. Studying child L2 acquisi-
tion can help dissociating these variables for each of the questions raised by the diferent approaches
mentioned earlier, because child L2 learners start acquiring the L2 at a relatively early age, yet they
already have knowledge of an L1 in place. This means that they are more similar to L1 children than
to L2 adults concerning their age of onset, and more similar to L2 adults than to L1 children regard-
ing the presence of the L1. Comparing them to each of these two groups thus may help in evaluating
the relative impact of age and L1 on L2 acquisition.
Early studies that have looked at child L2 acquisition from this comparative perspective often came
to the conclusion that child L2 learners are more similar to child L1 learners than to L2 adults. Early
research predominantly studied outcome diferences between acquisition types. While child L1 learn-
ers arrive at the same linguistic system as used by their adult interlocutors relatively homogenously,
the outcomes of adult L2 acquisition are much more variable. Johnson and Newport (1989) studied
child L2 learners of English as adults, and found a generally higher performance in grammaticality
judgment tasks for learners with a low compared to a higher age of onset. They suggest that age of
onset, not L1 infuence, explains diferences between native and non-native speakers. In contrast, a
study by Dulay and Burt (1974) looked at the process of acquisition, namely by studying the sequence
of acquisition of diferent grammatical morphemes in child L2 learners of English from diferent L1s.
They observed that the sequence was similar across diferent L1s and similar to that observed in L1
acquisition, and they also argued for child L2 and child L1 acquisition to be similar.
Both lines of research—those looking at outcomes and those looking at the process of child L2
acquisition—continued after these early studies. Research questions became more nuanced, asking in
which respects and which linguistic domains child L2 learners pattern with one or the other learner
group and in which respects their linguistic profle can only be fully accounted for by treating them
as a group in their own right. This question continues to provide the overarching perspective of the
feld. In the last ten years, an additional dimension has been added by studies that include online
processing data.
Age of onset
While it is uncontested that, overall, a lower age of onset of an L2 typically leads to an outcome that is
closer to that of native speakers, there is no agreement that this relation must be due to the existence
of a critical period. Some assume that child L1 and early child L2 acquisition are guided by innate
language processing principles that may become unavailable from a certain age on (Bley-Vroman,
27
Sarah Schimke
1990). Others argue that language acquisition mechanisms may change gradually with age of onset
due to factors typically or necessarily coming with age, such as increases in the overall length and
amount of exposure, the entrenchment of the frst language, general cognitive capacities, and chang-
ing socio-afective factors such as motivation (Birdsong, 2006; Bialystok & Hakuta, 1994; Pallier,
2007). Child L2 acquisition studies provide important data points to this debate in two respects: On
the one hand, many outcome-oriented studies have retrospectively related performance in adulthood
to the age of onset of child L2 learners. On the other hand, process-oriented studies directly compare
child L2 and child L1 learners and relate diferences in developmental trajectories to age of onset
efects (Paradis, 2007; Roesch & Chondrogianni, 2016).
The role of the L1 and the relation between the two languages
For all types of multilingual speakers, it has been asked whether lexical and grammatical knowledge
of diferent languages forms separate cognitive representations, whether access to these representa-
tions is selective or not, and whether they infuence each other during processing (Conklin & Thul,
2023 [this volume]; Hartsuiker & Bernolet, 2017). The answers to these questions may be diferent
for diferent types of multilingualism. In particular, L1 efects are clearly more prevalent in adult L2
learners than in bilingual L1 learners, especially for grammatical representations, where cross-linguis-
tic infuence is often subtle and confned to specifc domains in bilingual children (see Serratrice,
2013, for review). The fnding that L1 infuence is stronger when the onset of the L2 is later can
plausibly be related to several factors, in particular the degree of entrenchment of the L1 and the rela-
tive amount of exposure and profciency in each of the two languages (see Flege et al., 2006; Navarra
et al., 2005; Paradis & Jia, 2017). Studying L2 children makes it possible to ask whether there is a
gradual increase of L1 infuence with an increase of age of onset of the L2, or whether the onset of L1
efects is more categorical. Yet, as reviewed in the following, the amount of cross-linguistic infuence,
its directionality, and its relation to age of onset may be diferent for diferent linguistic domains.
Child-internal and child-external variables

The large number of factors infuencing the process of child L2 acquisition have often been classi-
fed in terms of “child-internal” and “child-external” variables (Paradis, 2011). Next to age of onset,
central child-internal variables are typological or structural properties of the L1 and its distance from
the L2, as well as cognitive variables such as working memory. Child-external variables are those
that are likely to infuence the amount and type of input that the child will receive in a language,
for instance, the length of exposure, the contexts of use of each language (whether a language is
spoken at home, in educational settings, and/or in society as a whole), the relative amount of native
and non-native input, and the socio-economic background of a child’s interlocutors in a given lan-
guage (see Grüter & Paradis, 2014, for overview). With respect to all of these variables, studying L2
children may help determine their respective infuence on language acquisition, since they are often
confounded with age when studied in child L1 or adult L2 acquisition.
Differences between linguistic domains

Another aspect that may impact the outcome and process of acquisition is the linguistic domain.
Many approaches make a principled diference between the lexicon and grammar (phonology, mor-
phology, and syntax). These two types of knowledge difer in that lexical knowledge is declarative in
nature, while grammatical knowledge is (at least partially) procedural (Morgan-Short & Ullman, 2023
[this volume]). Further, lexical knowledge consists of open-class items and may constantly expand
throughout a lifetime, while grammatical units form closed-class groups. It has been claimed that
28
lexical knowledge will more likely be afected by child-external factors, in particular the amount of
input, the learning situation, and related diferences in language dominance, whereas grammatical
knowledge is governed more by child-internal factors such as age of onset (Veríssimo et al., 2018;
Paradis, 2011). Others made more fne-grained diferences within each domain. For instance, it has
been suggested that age of onset efects likely difer between phonology, morphology, and syntax
(Meisel, 2013). Within the domain of morphosyntax, it has further been claimed that those phenom-
ena that are acquired late in L1 acquisition are less prone to age of onset efects in child L2 acquisition
than phenomena that are acquired early (Schulz & Grimm, 2019; Tsimpli, 2014). Importantly, the
same phenomenon may be acquired at a diferent point in time in diferent languages, pointing to
the fact that diferences between domains in order of acquisition may not hold cross-linguistically
(Vasić et al., 2012).
Differences between explicit and implicit knowledge

A related distinction is that between explicit and implicit knowledge. It has often been assumed that,
relative to adults, children are more successful at acquiring implicit and less successful at acquiring
explicit knowledge (see Nemeth et al., 2013; yet see Vinter & Perruchet, 2000). In child L1 acquisi-
tion, at least during the frst two years of life, almost all linguistic knowledge is implicit, and young
children are often not able to perform any meta-linguistic task. As a consequence, researchers often
use online methods that are well suited for insights into implicit knowledge to assess L1 children’s
knowledge. This research has provided insights into knowledge that would be impossible to docu-
ment with explicit measures (see for instance Safran et al., 1999; see also Höhle et al., 2020, for
an overview of linguistic development in the frst two years of life as tested with implicit methods).
Conversely, adult L2 learners have more explicit meta-linguistic knowledge, and more developed
cognitive capacities in general. This facilitates performance in ofine tasks, such as making gram-
maticality judgments. As a consequence, adult L2 speakers have often been found to reveal less
native-like linguistic knowledge during online processing than in ofine tasks (see Clahsen & Felser,
2006), which is the reverse pattern compared to L1 children. It is thus possible that the relative
importance of both types of knowledge changes with age of onset, and that child L2 learners may
form the middle ground between L1 children and L2 adults in this respect. It is also possible that
being multilingual in itself afects the relative importance, since multilingualism has been claimed to
contribute to meta-linguistic awareness (Friesen & Bialystok, 2012) and general cognitive develop-
ment (Bialystok, 2017; for discussion, see Poarch, 2023 [this volume]).
Current contributions and research

Against the backdrop of these issues, I summarize current research across diferent linguistic domains.
For each domain, I address the question whether L2 children pattern with either child L1 or adult
L2 learners, or whether they show a distinct profle, both with respect to outcomes of acquisition and
to the evolving knowledge during the process of acquisition.
Phonological production and perception

Knowledge about the phoneme categories of the native language develops rapidly during the frst
year of age (Bosch & Sebastián-Gallés, 2003). When adult L2 learners are exposed to the sounds of
the L2, their L1 phonemic categories act as a flter, in that there is a strong tendency among non-
native speakers to assimilate sounds of the L2 to the categories of their L1 (see Flege, 2018). This
makes it hard for non-native speakers to use phonetic cues particular to the L2 to reliably identify and
produce sounds of the L2, in particular for those categories in which two L2 phonemes extend across
29
Sarah Schimke
a single L1 phoneme. For L2 children, this raises the question whether there is similarly strong infu-
ence of the L1 on the L2, or whether the relation between the two languages is more symmetrical
in child compared to adult L2 learners.
Concerning outcomes, the production and perception of sound categories of the L2 is among the
phenomena that are not always in native-like ranges even for L2 children with an early age of onset.
For instance, in the study by Abrahamsson and Hyltenstam (2009), only around 75% of Spanish L2
learners of Swedish with an age of onset below 11 years performed native-like in the production
and perception of Swedish consonants. Around 40% of L2 learners with an age of onset above 11
years also performed in this range, which speaks against a principled impossibility to acquire pho-
netic properties of L2 phonemes (see also Flege et al., 2006). Furthermore, non-native-like out-
comes among child L2 learners have also been documented in more implicit tasks. For instance,
Navarra et al. (2005) found diferences in between native and non-native participants reaction times
in an unrelated task while they listened to expected or unexpected phonemes. Such efects have
been interpreted as indicating a particularly early sensitive period in the acquisition of phonology.
Others have pointed to the fact that L1 phonological knowledge is acquired particularly early, and
therefore may exert a strong infuence from early on (Navarra et al., 2005; Pallier, 2007), or have
proposed that length and amount of exposure is an especially important factor for phonological
processing (Flege, 2018).
Studies that look at phonological knowledge in L2 children who are in the process of acquiring the
L2 have often documented rapid development in this area of knowledge. For instance, in a longitudi-
nal study, McCarthy et al. (2014) observed that, initially, four-year-old children with Sylheti as their
L1 and English as their L2 showed L1 infuence on voice onset time of plosives in production and
perception, yet these efects were no longer present after increased exposure to English one year later.
While such efects are in seeming contradiction to studies focusing on outcomes, it is possible that
this is due to process-oriented studies investigating diferent phonemic contrasts and/or L1–L2 pairs
than outcome studies. Moreover, while outcome studies have used both ofine and online measures,
I am not aware of online phonological processing studies in L2 children.
Lexical and morphological processing

Current models of the bilingual mental lexicon assume that multilingual speakers do not store or
access words separately in each language, but instead have shared representations. L2 speakers typi-
cally have a smaller lexicon in their L2 than native speakers, and lexical knowledge of a non-native
and weaker language is also typically accessed more slowly than lexical knowledge of a dominant
and native language (see van Hell & Tanner, 2012, for an overview). In addition, many studies have
found a stronger infuence of the L1 on the L2 than vice versa, as can be seen in stronger cognate
facilitation efects in L2 processing, and in more competition from L1 word form candidates during
L2 lexical access (Weber & Cutler, 2004). Finally, concerning morphology, it has been claimed that
infectional morphology in particular is not processed in a native-like way by adult L2 learners, with
less online decomposition of complex word forms in non-native compared to native speakers (e.g.,
Neubauer & Clahsen, 2009; see Gor, 2023 [this volume]).
Concerning outcomes, very few studies have directly compared the lexicon or lexical processing in
native speakers to that of child L2 acquirers as adults. One of the few studies that did, Abrahamsson
and Hyltenstam (2009), found diferences in some learners with respect to expressive knowledge
of formulaic language, with less knowledge about proverbs and idioms in L2 acquirers com-
pared to native speakers. A study by Veríssimo et al. (2018) looked at the processing of morphology in
adult native speakers of German and adult L2 speakers of German with an age of onset in childhood,
using a priming paradigm. While priming for derivational morphology was identical in the two
groups, priming efects for infectional morphology were much stronger for native speakers and for
30
L2 speakers with an age of onset below six years of age compared to L2 speakers with a higher age
of onset, suggesting that the processing of morphology may be an area in which age of onset plays
an important role.
As for lexical and morphological knowledge during the process of child L2 acquisition, some stud-
ies have asked how long L2 children take to “catch up” in this domain with their monolingual peers.
Paradis and Jia (2017) found that after four to six years of L2 input, there was no longer a diference
in standardized tasks of vocabulary knowledge for L2 compared to L1 children. The study found,
moreover, that diferences between L1 and L2 children were more persistent in production tasks than
in perception tasks. In line with this, Pham and Kohnert (2013) observed steeper developmental
curves for perception than for production, in particular for an online processing task that measured
speed of picture recognition.
Regarding the interaction between the two languages, Poarch and van Hell (2012) found a cog-
nate facilitation efect in L2 English in diferent groups of child L2 learners, but they reported that
cognate facilitation in the L1 was restricted to more profcient L2 speakers. Similarly, von Holzen
et al. (2019) found that there was a facilitating efect of phonological overlap with the L1 German in
L2 English word recognition in toddlers acquiring English as a second language, while the recogni-
tion of the L1 remained unaltered. These studies thus suggest L1 infuence on L2 lexical knowledge.
Interestingly, von Holzen and Mani (2012) also observed infuence of the L2 on the L1 in a prim-
ing study. In sum, these studies show that the child L2 lexicon seems to be very similar in structure
and processing to the adult L2 lexicon in that there can be cross-language activation, with a higher
likelihood of the (dominant) L1 infuencing the L2 than vice versa. Contrary to what is the case for
grammatical knowledge, there seems to be consensus that such asymmetric efects are due to difer-
ences in language dominance and length and amount of exposure, and may not necessarily be related
to the order of acquisition.
Sentence level and discourse-level processing

Regarding sentence-level and discourse-level knowledge, adult L2 speakers often difer in sentence
processing from native speakers, even in cases in which such diferences cannot be detected in ofine
data. In particular, it has been suggested that due to critical period efects, during sentence process-
ing, adult L2 learners would be unable to use infectional markers and to build up syntactic structures
in the same way as native speakers do (Clahsen & Felser, 2006), at least when the relevant morpho-
syntactic categories do not exist in the L1 (Jiang, 2004). Other researchers claim that such diferences
in processing profles are not necessarily due to a critical period for morphology or syntax, but that
they may refect diferences between native and non-native speakers that are due to other factors, for
instance, diferences in underlying lexical knowledge and processing (Hopp, 2018).
Against this background, diferent predictions seem plausible for sentence and discourse pro-
cessing in L2 children. In the Domain-by-Age Model, Schwartz (2004) suggested that, in child L2
acquisition, infectional morphology may be acquired (and, presumably, processed) in ways similar to
L1 acquisition, but that there may be early efects of the L1 syntax on the acquisition process. This
proposal receives support from studies that have observed developmental stages in the acquisition of
syntax by L2 children that are similar to L2 adults, but diferent from L1 children (Song & Schwartz,
2009, but see Tracy & Thoma, 2009). It does not exclude that native-like processing and outcomes
can be reached eventually. By contrast, proposals according to which there are sensitive periods for
infectional morphology and syntax (Meisel, 2013) would predict that, for child L2 learners with an
age of onset later than four years, the representations and processes used by L2 children should be and
remain diferent from those of L1 children. This assumption would be compatible with L2 children
showing processing profles that are diferent from L1 children both during the course of acquisition
and when they have reached adulthood.
31
Sarah Schimke
A number of ofine studies have used grammaticality judgments, similar to or as direct replica-
tions of the study by Johnson and Newport (1989), discussed previously (for review, see Birdsong,
2006). These studies confrmed the fnding that non-native adults typically perform indistinguish-
ably from native speakers for grammaticality judgments when their age of onset is below eight years.
There is more variability with ages of onset above eight years, even though native-like outcomes in
grammaticality judgments remain frequent even in child learners with a higher age of onset.
Turning to studies that assess sentence and discourse-level knowledge during the process of acquisi-
tion, very few studies have directly compared the processing of the same phenomenon in child L2 and
adult L2 learners. Yet, a number of studies have looked at the processing of sentences and discourses
in L2 children, and compared their profles to those of age- or profciency-matched L1 children, or,
in some cases, to those of monolingual children with developmental language disorders (DLD). All of
these studies investigated children acquiring the target language in an immersion setting. Chondro-
gianni and Marinis (2012) examined the production and online processing of tense morphology in
six- to nine-year-old children with English as L1, or English as L2 and Turkish as L1. For production,
the results confrmed previous studies according to which L2 children did not reliably produce tense
morphemes in obligatory contexts even after three years of exposure. In an online word-spotting
task, however, all children showed longer processing times for sentences in which these morphemes
were missing in obligatory contexts. Chondrogianni and Marinis (2012) conclude that L2 children,
similarly to typically developing L1 children but diferently from DLD children, have intact underly-
ing knowledge about tense and agreement, and that they apply this knowledge during sentence pro-
cessing. What seems to be more difcult for L2 than for L1 children, however, is to actively retrieve
the correct morphemes during production. This lends support to approaches like the missing surface
infection hypothesis according to which producing infectional morphology is particularly challeng-
ing in adult and child L2 acquisition, even if the underlying knowledge is in place (Haznedar, 2001).
A similar pattern was observed in a number of studies that examined the production and processing
of grammatical gender marking in L2 Dutch and L2 Greek by children with Turkish as their L1
(Blom & Vasic, 2011; Vasić et al., 2012). In these studies, L2 children uniformly behaved more simi-
larly to L1 children during self-paced listening than during production. This pattern was modulated
by the transparency of the target language, with overall more native-like behavior in L2 children
learning Greek than in those learning Dutch. Overall, this suggests that native-like morphosyntactic
processing is possible even for categories that cannot be transferred from the L1, such as, in this case,
gender. Converging evidence comes from studies on the production and processing of indefnite
articles, defnite articles, and clitic pronouns (see Chondrogianni, 2018, for an overview), which all
documented more knowledge during processing than during production, irrespective of the L1.
Lemmerth and Hopp (2019) studied potential L1 efects during child L2 syntactic processing,
using the visual world paradigm. They assessed how quickly children with Russian as L1 and German
as L2 looked to a gender-matching object on a visual display following the onset of a gender-marked
article or adjective. The study was able to dissociate potential L1 infuence on the lexical level from
L1 infuence on the syntactic level. For lexical efects, the nouns used either had the same gender in
the L1 as in the L2 or not, and, for syntactic efects, the use of gender marking on adjectives, which
exists in both languages, was compared to that of gender marking on articles, which exists in Ger-
man, but not in Russian. The study found that L2 children patterned with L1 children for lexically
congruent nouns, but showed reduced predictive processing for lexically incongruent nouns having
diferent genders. There was no efect of syntactic diferences. Lemmerth and Hopp (2019) conclude
that there is a strong infuence of the L1 lexicon on the L2 lexicon, yet—unlike in adult L2 learners
(Hopp & Lemmerth, 2018)—no evidence for L1 efects in syntax, as syntactic processing patterns
were similar to those of monolingual children.
Other studies explored the processing of complex sentences in L2 compared to L1 children (Marinis,
2007; Cristante & Schimke, 2018). They asked whether L2 children are able to use morphosyntactic
32
cues during online processing to distinguish non-canonical (passive and object-initial) sentences from
canonical active sentences. L2 children processed German and English passive sentences in a way
similar to L1 children (Marinis, 2007; Cristante & Schimke, 2018). As for object-initial sentences,
L2 children were less successful in reaching target-like interpretations than L1 children (Cristante &
Schimke, 2018). This pattern was present both in ofine judgments and during online processing. It
thus appears that, once children have acquired a morphosyntactic phenomenon, there is no evidence
for particular difculties in applying this knowledge during online processing, contrary to what has
been suggested for L2 adults. As for L1 infuence on complex syntax, a study on the production of
complex sentences in children from diferent L1s in L2 English suggests that the amount of complex
syntactic structures produced was infuenced by input and cognitive factors, but not by the L1 (Para-
dis et al., 2017). Interestingly, this study also showed an advantage for L2 children over L1 children
concerning the length of time needed to acquire complex syntax (see also Rothman et al., 2016, for
advantages in older compared to younger L2 children in the comprehension of passives).
Finally, there is a small number of studies on the processing of discourse coherence markers such as
pronouns and connectives. These studies confrm that L2 children may deviate in their production of
such markers, but typically show more native-like patterns during online processing of anaphoric personal
pronouns (Klages & Gerwien, 2018) as well as connectives (Schimke, 2015), similarly to L1 children.
To sum up, sentence- and discourse-level processing are domains wherein diferences have been
documented between adult L2 learners and adult native speakers. L2 children seem to pattern dif-
ferently from both groups. On the one hand, L2 children have difculties in the use of infectional
morphology in sentence contexts, in particular in production tasks, but also in comprehension (see
Chondrogianni & Marinis, 2012, for verb morphology in English; Vasić et al., 2012, for gender in
Dutch; and Cristante & Schimke, 2018, for case in German). In this respect, they difer from mono-
lingual speakers, at least during the process of acquisition. On the other hand, the available sentence
processing studies in L2 children do not report the non-target processing patterns that have been
claimed to be typical of adult L2 learners (Clahsen & Felser, 2006), with no evidence for shallow
processing or L1-based processing in the data available so far.
Taken together, L2 children thus seem to pattern more closely with native speakers in online
processing tasks, but seem to deviate from native patterns more often in their ofine production,
making them more similar to L2 adults in this domain. Together, this leads to a pattern which seems
to be unique to this group, in which efects of task (online vs. ofine) and modality (production vs.
comprehension) seem to be particularly strong when compared to such task efects in other groups of
language users. However, direct comparisons between L2 children and L2 adults are largely missing,
and a number of the phenomena in which L2 adults difer from native speakers during processing
have not yet been tested in L2 children.
Profle effects
While there are important diferences between the three diferent domains that I have reviewed,
there are also common fndings. Firstly, regarding outcomes, studies that have scrutinized L2 children’s
linguistic behavior as adults typically do fnd phenomena in which they deviate from native speakers,
yet these deviations are very restricted in scope, and there is individual variability in which domains
are afected (Abrahamsson & Hyltenstam, 2009). L2 children are thus often almost indistinguishable
from monolingual speakers in their linguistic behavior as adults across diferent domains. It may thus
be a fair generalization to conclude that L2 children are typically more similar to L1 children than to
L2 adults as far as outcomes are concerned.
This relatively homogeneous picture contrasts with evidence from the process of L2 acquisition.
L2 children often pattern with L1 children, in particular when evidence from online measures is
taken into account. However, they also show a unique profle, where in many studies, knowledge
33
Sarah Schimke
seems to be more easily and consistently detectable during online processing for comprehension
than during production. This clearly is the case for morphosyntactic knowledge, but also for lexical
knowledge. The observation that knowledge seems to be particularly hard to be put to use during
production tasks applies to other groups of language users as well, in particular heritage speakers and
attritted speakers (see Montrul, 2023 [this volume], Schmid et al., 2023 [this volume]).

Although some aspects of language processing and use seem to be typical of child L2 language
users as a group, there is also a large variability among child L2 learners. This is not surprising given
the very heterogeneous circumstances under which child L2 acquisition may take place. Diferent
domains of study have often concentrated on particular groups, such that comparisons across domains
are difcult. For instance, studies on morphosyntactic processing have often looked at children in
migration, i.e., second-language, contexts, while studies on lexical processing have often looked at
children acquiring a foreign language in immersion school or kindergarten programs. Finally, studies
on outcomes have typically included participants who were self-selected, while L2 children studied
as children are usually recruited in a less selective way. All these diferences mean that variability and
individual diferences in child L2 acquisition are important topics for future research, in particular for
those areas in which the available results suggest a relatively homogeneous picture.
A second important direction for future research is the relations between language processing and the
stages and outcome of language acquisition (for discussion, see Jackson, 2023 [this volume]). There can
be no acquisition without some processing of the input, but input can only be processed based on
some form of knowledge. Given this close relationship, any dissociation between online and ofine
behavior is puzzling. A promising approach towards better understanding these relations are longitu-
dinal studies that combine online and ofine measures, such that changes in processing profles can be
tracked over time and be related to changes in ofine performance data. Ideally, such studies should
be conducted with diferent learner groups in order to understand which factors modulate the rela-
tion between knowledge visible in ofine tasks and knowledge visible in online tasks.
Finally, important avenues for future research concern applied questions, in particular, whether and
how child L2 acquisition can be fostered, or how to support multilingual language development in
children in general. A further step would be to conduct intervention studies to fnd out whether
what we know about language processing and natural acquisition sequences in child L2 acquisition
can be used to design intervention methods, for instance by modeling natural acquisition sequences
in the classroom input.
Variability within child L2 acquisition: How variable is child L2 acquisition, in particular concern-
ing online processing patterns and ultimate outcomes? How pronounced are individual diferences
between L2 children, and how much of this variance can be explained by taking into account the
moderating variables we know of?
Relations between processing and acquisition: What is the relation between language processing and lan-
guage acquisition sequences and outcomes in child L2 learners? Is there a uniform relation between
language processing and language acquisition in diferent learner groups?
Generalizability and applied issues: How generalizable are results from child L2 acquisition research to
specifc (educational) contexts? Which interventions have an infuence on language acquisition and
use in L2 children?
34
Further reading
Chondrogianni, V. (2018). Child L2 acquisition. In F. Bayram, D. Miller, J. Rothman, & L. Serratrice (Eds.),
Bilingual cognition and language: The state of the science across its subfelds (pp. 103–126). John Benjamins.
This chapter provides an overview of recent child L2 acquisition studies from a generative perspective, with
a focus on studies looking at online processing.
Unsworth, S., Hulk, A., & Marinis, T. (2011). Internal and external factors in child second language acquisition.
Linguistic Approaches to Bilingualism, 1(3).
This special issue contains an introduction and fve empirical articles that together provide an overview of the
diferent role of child-internal and child-external factors in child L2 acquisition.
Paradis, J. (2007). Second language acquisition in childhood. In E. Hof & M. Shatz (Eds.), Blackwell handbook of
language development (pp. 387–405). Blackwell Publishing.
This chapter provides an overview of child L2 acquisition with a focus on acquisition patterns and rates, in
particular compared to monolingual L1 acquisition.
References
Abrahamsson, N., & Hyltenstam, K. (2009). Age of onset and nativelikeness in a second language: Listener
perception versus linguistic scrutiny. Language Learning, 59(2), 249–306.
Bialystok, E. (2017). The bilingual adaptation: How minds accommodate experience. Psychological Bulletin,
143(3), 233–262.
Bialystok, E., & Hakuta, K. (1994). In other words: The psychology and science of second language acquisition. Basic Books.
Birdsong, D. (2006). Age and second language acquisition and processing: A selective overview. Language Learn-
ing, 56, 9–49.
Bley-Vroman, R. (1990). The logical problem of foreign language learning. Linguistic Analysis, 20, 3–39.
Blom, E., & Vasic, N. (2011). The production and processing of determiner-noun agreement in L2 Dutch.
Linguistic Approaches to Bilingualism, 1(3), 265–290.
Bosch, L., & Sebastián-Gallés, N. (2003). Simultaneous bilingualism and the perception of a language-specifc
vowel contrast in the frst year of life. Language and Speech, 46(2–3), 217–243.
Chondrogianni, V., & Marinis, T. (2012). Production and processing asymmetries in the acquisition of tense
morphology by sequential bilingual children. Bilingualism-Language and Cognition, 15(1), 5–21.
Clahsen, H., & Felser, C. (2006). Grammatical processing in language learners. Applied Psycholinguistics, 27(1), 3–42.
Conklin, K. & Thul, R. (2023). Word and multiword processing: cognitive approaches. In A. Godfroid &
H. Hopp (Eds.), The Routledge handbook of second language acquisition and psycholinguistics (pp. 203–215).
Routledge.
Cristante, V., & Schimke, S. (2018). Die Verarbeitung von Passiv- und OVS-Sätzen im kindlichen Zweitsprach-
erwerb. In S. Schimke & H. Hopp (Eds.), Sprachverarbeitung im Zweitspracherwerb (pp. 169–194). De
Gruyter.
Dimroth, C. (2018). Beyond statistical learning: Communication principles and language internal factors shape
grammar in child and adult beginners learning Polish through controlled exposure. Language Learning, 68(4),
863–905.
Dulay, H., & Burt, M. (1974). Natural sequences in child second language acquisition. Language Learning, 24(1),
37–53.
Ellis, N. C. (2017). Implicit and explicit knowledge about language. In J. Cenoz & D. Gorter (Eds.), Language
awareness and multilingualism. Encyclopedia of language and education (pp. 113–124). Springer International Pub-
lishing Switzerland.
Flege, J. E. (2018). It’s input that matters most, not age. Bilingualism: Language and Cognition, 21(5), 919–920.
Flege, J. E., Birdsong, D., Bialystok, E., Mack, M., Sung, M., & Tsukada, M. (2006). Degree of foreign accent
in English sentences produced by Korean children and adults. Journal of Phonetics, 34(2), 153–157.
Friesen, D. C., & Bialystok, E. (2012). Metalinguistic ability in bilingual children: The role of executive control.
Rivista di Psicolinguistica Applicata, 12(3), 47–56.
Godfroid, A. (2023). Hypotheses about the interface between explicit and implicit knowledge in second lan-
guage acquisition. In A. Godfroid & H. Hopp (Eds.), The Routledge handbook of second language acquisition and
psycholinguistics (pp. 294–307). Routledge.
35
Sarah Schimke
Gor, K. (2023). Word and multiword processing: memory-based and linguistic approaches. In A. Godfroid
& H. Hopp (Eds.), The Routledge handbook of second language acquisition and psycholinguistics (pp. 191 -202).
Routledge.
Grüter, T., & Paradis, J. (Eds.). (2014). Input and experience in bilingual development. John Benjamins.
Hartsuiker, R. J., & Bernolet, S. (2017). The development of shared syntax in second language learning. Bilin-
gualism: Language and Cognition, 20, 219–234.
Haznedar, B. (2001). The acquisition of the IP system in child L2 acquisition. Studies in Second Language Acquisi-
tion, 23, 1–39.
Höhle, B., Bijeljac-Babic, R., & Nazzi, T. (2020). Variability and stability in early language acquisition: Com-
paring monolingual and bilingual infants’ speech perception and word recognition. Bilingualism: Language and
Cognition, 23(1), 56–71.
Hopp, H. (2018). The bilingual mental lexicon in L2 sentence processing. Second Language, 17, 5–27.
Hopp, H., & Lemmerth, N. (2018). Lexical and syntactic congruency in L2 predictive gender processing. Studies
in Second Language Acquisition, 40(1), 171–199.
Jackson, C. N. (2023). The psycholinguistics of adult second language acquisition. In A. Godfroid & H. Hopp
(Eds.), The Routledge handbook of second language acquisition and psycholinguistics (pp. 38–49). Routledge.
Jiang, N. (2004). Morphological insensitivity in second language processing. Applied Psycholinguistics, 25(4),
603–634.
Johnson, J., & Newport, E. (1989). Critical period efects in second language learning: The infuence of matu-
rational state on the acquisition of English as a second language. Cognitive Psychology, 21, 60–99.
Klages, H., & Gerwien, J. (2018). Die Interpretation anaphorischer Pronomina bei Kindern im L1- und frühen
L2-Erwerb. In S. Schimke & H. Hopp (Eds.), Sprachverarbeitung im Zweitspracherwerb (pp. 247–274). De Gruyter.
Lemmerth, N., & Hopp, H. (2019). Gender processing in simultaneous and successive bilingual children: Cross-
linguistic lexical and syntactic infuences. Language Acquisition, 26(1), 21–45.
Lenneberg, E. (1967). Biological foundations of language. Wiley.
Marinis, T. (2007). Online processing of passives in L1 and L2 children. In A. Belikova, L. Meroni, & M. Umeda
(Eds.), Proceedings of the 2nd conference on generative approaches to language acquisition (pp. 265–276). Cascadilla
Proceedings Project.
McCarthy, K., Mahon, M., Rosen, S., & Evans, B. (2014). Speech perception and production by sequential
bilingual children: A longitudinal study of voice onset time acquisition. Child Development, 85(5), 1965–1980.
Meisel, J. M. (2013). Sensitive phases in successive language acquisition: The Critical Period Hypothesis revis-
ited. In C. Boeckx & K. Grohmann (Eds.), Handbook of biolinguistics (pp. 69–85). Cambridge University Press.
Montrul, S. (2023). The psycholinguistics of heritage language acquisition. In A. Godfroid & H. Hopp (Eds.),
The Routledge handbook of second language acquisition and psycholinguistics (pp. 73–84). Routledge.
Morgan-Short, K. & Ullman, M. T. (2023). Declarative and procedural memory in second language learning:
psycholinguistic considerations. In A. Godfroid & H. Hopp (Eds.), The Routledge handbook of second language
acquisition and psycholinguistics (pp. 322–334). Routledge.
Navarra, J., Sebastián-Gallés, N., & Salvador, S.-F. (2005). The perception of second language sounds in early
bilinguals: New evidence from an implicit measure. Journal of Experimental Psychology Human Perception &
Performance, 31(5), 912.
Nemeth, D., Janacse, K., & Fiser, J. (2013). Age-dependent and coordinated shift in performance between
implicit and explicit skill learning. Frontiers in Computational Neuroscience, 7(147).
Neubauer, K., & Clahsen, H. (2009). Decomposition of infected words in a second language. Studies in Second
Language Acquisition, 31, 403–435.
Pallier, C. (2007). Critical periods in language acquisition and language attrition. In B. Köpke, M. Schmid, M.
Kejzer, & S. Dostert (Eds.), Language attrition: Theoretical perspectives (pp. 155–168). John Benjamins.
Paradis, J. (2011). Individual diferences in child English second language acquisition: Comparing child-internal
and child-external factors. Linguistic Approaches to Bilingualism, 1(3), 213–237.
Paradis, J., & Jia, R. (2017). Bilingual children’s long-term outcomes in English as a second language: Language
environment factors shape individual diferences in catching up with monolinguals. Developmental Science,
20(1), e12433.
Paradis, J., Rusk, B., Duncan, T., & Govindarajan, K. (2017). Children’s second language acquisition of English
complex syntax: The role of age, input, and cognitive factors. Annual Review of Applied Linguistics, 37, 148–167.
Pham, G., & Kohnert, K. (2013). A longitudinal study of lexical development in children learning vietnamese
and English. Child Development, 85(2), 767–782.
Poarch. G. J. (2023). Cognitive efects of bilingualism. In A. Godfroid & H. Hopp (Eds.), The Routledge handbook
Poarch, G. J., & van Hell, J. G. (2012). Cross-language activation in children’s speech production: Evidence from
second language learners, bilinguals, and trilinguals. Journal of Experimental Child Psychology, 111(3), 419–438.
36
Roesch, A. D., & Chondrogianni, V. (2016). “Which mouse kissed the frog?” Efects of age of onset, length
of exposure, and knowledge of case marking on the comprehension of wh-questions in German-speaking
simultaneous and early sequential bilingual children. Journal of Child Language, 43(3), 635–661.
Rothman, J., Long, D., Iverson, M., Judy, T., Lingwall, A., & Chakravarty, T. (2016). Older age of onset in child
L2 acquisition can be facilitative: Evidence from the acquisition of English passives by Spanish natives. Journal
of Child Language, 43(3), 662–686.
Safran, J. R., Johnson, E. K., Aslin, R. N., & Newport, E. L. (1999). Statistical learning of tone sequences by
human infants and adults. Cognition, 70, 27–52.
Schimke, S. (2015). Die Verarbeitung von Markierungen der Diskurskohärenz bei Grundschulkindern mit
Deutsch als Erst- oder Zweitsprache. In G. Pagonis & H. Klages (Eds.), Linguistisch fundierte Sprachförderung
und Sprachdidaktik: Grundlagen, Konzepte, Desiderate (pp. 193–214). De Gruyter.
Schmid, M. S., Soto, C., & Heimann, J. (2023). The psycholinguistics of frst language attrition. In A.
Godfroid & H. Hopp (Eds.), The Routledge handbook of second language acquisition and psycholinguistics (pp.
61–72). Routledge.
Schulz, P., & Grimm, A. (2019). The age factor revisited: Timing in acquisition interacts with age of onset in
bilingual acquisition. Frontiers in Psychology, 9, 2732.
Schwartz, B. D. (2004). On child L2 development of syntax and morphology. Lingue et Linguaggio, 3, 97–132.
Serratrice, L. (2013). Cross-linguistic infuence in bilingual development. Determinants and Mechanisms. Lin-
guistic Approaches to Bilingualism, 3(1), 3–25.
Song, H. S., & Schwartz, B. D. (2009). Testing the fundamental diference hypothesis. Studies in Second Language
Acquisition, 31(2), 323–361.
Tracy, R., & Thoma, D. (2009). Convergence on fnite V2 clauses in L1, bilingual L1 and early L2 acquisition.
In P. Jordens & C. Dimroth (Eds.), Functional categories in learner language (pp. 1–43). Mouton de Gruyter.
Tsimpli, I. (2014). Early, late or very late? Timing acquisition and bilingualism. Linguistic Approaches to Bilingual-
ism, 4(3), 283–313.
Unsworth, S. (2005). Child L1, adult L2, child L2: Diferences and similarities: A study on the acquisition of direct object
scrambling [Doctoral dissertation], Utrecht University. Dissertation series 119.Utrecht: LOT.
van Hell, J. G., & Tanner, D. (2012). Second language profciency and cross-language lexical activation. Language
Learning, 62, 148–171.
Vasić, N., Chondrogianni, V., Marinis, T., & Blom, W. B. T. (2012). Processing of gender in Turkish-Dutch
and Turkish-Greek child L2 learners. BUCLD36: Proceedings of the 36th annual Boston University conference on
language development (pp. 646–659). Cascadilla Press.
Veríssimo, J., Heyer, V., Jacob, G., & Clahsen, H. (2018). Selective efects of age of acquisition on morphological
priming: Evidence for a sensitive period. Language Acquisition, 25(3), 315–326.
Vinter, A., & Perruchet, P. (2000). Implicit learning in children is not related to age: Evidence from drawing
behavior. Child Development, 71, 1223–1240.
von Holzen, K., Fennell, C. T., & Mani, N. (2019). The impact of cross-language phonological overlap on
bilingual and monolingual toddlers’ word recognition. Bilingualism: Language and Cognition, 3, 476–499.
von Holzen, K., & Mani, N. (2012). Language nonselective lexical access in bilingual toddlers. Journal of Experi-
mental Child Psychology, 113, 569–586.
Weber, A., & Cutler, A. (2004). Lexical competition in non-native spoken-word recognition. Journal of Memory
and Language, 50(1), 1–25.
Wulf, S., & Ellis, N. C. (2018). Usage-based approaches to second language acquisition. In D. Miller, F. Bayram,
J. Rothman, & L. Serratrice (Eds.), Bilingual cognition and language: The state of the science across its subfelds
(pp. 37–56). John Benjamins.
37
4
THE PSYCHOLINGUISTICS OF
ADULT SECOND LANGUAGE
ACQUISITION
Carrie N. Jackson
Introduction
Key questions within second language (L2) acquisition research revolve around whether L2
learners—especially those who do not begin learning or acquiring their L2 until after puberty—
can achieve native-like L2 profciency and, in the absence of native-like profciency, what fac-
tors can best explain observed diferences in language comprehension and production between
native (L1) and L2 speakers. At the lexical level, such research revolves around the organization
of the bilingual mental lexicon, and the impact non-selective lexical access (i.e., the phenom-
enon whereby all languages a person speaks are simultaneously active in the brain) has on the
speed and accuracy of L2 word recognition and production (see Tokowicz, 2014; Gor, 2023
[this volume]; Conklin & Thul, 2023 [this volume], for review). At the morphosyntactic level,
these questions often invoke debates surrounding whether the same underlying linguistic and
cognitive mechanisms drive L1 and L2 acquisition, or whether such mechanisms and strategies
are fundamentally diferent in L1 versus L2 acquisition, especially for adult L2 learners who
already have fully developed L1 linguistic systems (e.g., Birdsong, 1992; Bley-Vroman, 1990;
White, 2003). Starting in the mid-1990s, second language acquisition (SLA) researchers began
to adopt research methods common in L1 psycholinguistic research to investigate real-time lan-
guage processing among highly-profcient L2 speakers, to provide novel insight into the nature
of L2 speakers’ L2 linguistic systems (see Jufs & Rodríguez, 2014; Hopp, 2023 [this volume],
for review). At the same time, the use of behavioral and neurophysiological research methods to
investigate L2 sentence processing and L2 production can also advance our understanding of the
underlying linguistic and cognitive mechanisms implicated in L2 acquisition, and what types of
learning contexts may best support L2 learning. To this end, the present chapter reviews psy-
cholinguistic research on L2 sentence comprehension and production, with a focus on research
that investigates the relationship between how adult L2 speakers process input in real-time and
L2 grammar acquisition.
38 DOI: 10.4324/9781003018872-5
The psycholinguistics of adult second language acquisition
Parsing-to-learn: The hypothesis that how L1 and L2 speakers assign grammatical structure and meaning
during real-time sentence comprehension has an impact on how they learn or acquire new grammati-
cal structures (see also Hopp, 2023 [this volume]).
Lexical access: The act of retrieving words from the mental lexicon (see also Conklin & Thul, 2023
[this volume]). In the context of L2 processing, the speed with which lexical retrieval occurs can
have downstream consequences for constructing the structure and meaning of a sentence.
Grammatical/syntactic representations: How grammatical information is structured and stored in a lan-
guage user’s linguistic system.
Predictive processing: When a language user takes advantage of previous and current linguistic cues (e.g.,
phonological, semantic, or morphosyntactic information) to predict upcoming words and phrases in
a sentence.
Structural priming: The phenomenon whereby speakers are more likely to (re)use a particular syntactic
structure that they have recently heard or produced.
Learning to parse and parsing to learn

Several accounts in both the L1 and L2 acquisition literature argue in favor of a close link between
online processing mechanisms and the development of grammatical knowledge, rather than treating
research on real-time language processing and research assessing learners’ grammatical development
as two independent lines of inquiry. In so doing, parsing-to-learn accounts provide critical insight
into what L1 and L2 learners internalize from their linguistic input, hypothesizing that learners’
resulting mental representations (i.e., intake) may difer from the input to which they are exposed,
and that such misinterpretations can impede acquisition (Dekydtspotter & Renaud, 2014; Fodor,
1998b; see also Omaki, 2017; Omaki & Lidz, 2015).
In two seminal articles, Fodor (1998a, 1998b) laid out the groundwork for this line of inquiry by
raising two critical questions: (1) how language learners acquire the processing routines necessary to
successfully parse linguistic input and, conversely, (2) whether and how the processing routines learn-
ers use to parse linguistic input, and whether these routines lead to a correct parse and interpretation
of the input, impact the course of language acquisition and development. In addressing the frst ques-
tion, Fodor (1998a) ultimately concludes that, at least for L1 speakers, processing routines are innate
and do not need to be acquired. Any apparent cross-linguistic diferences in L1 sentence processing
stem from diferences in the underlying grammar that feeds real-time sentence processing, rather than
the processing routines themselves being language-specifc. However, in addressing the second ques-
tion, Fodor (1998b) raises the inherent catch-22 that “the sentences from which the learner should
learn are the ones that the current grammar does not yet license, but they’re not available because
they are the ones that the learner cannot yet parse” (p. 344). Taking a generative perspective, she
addresses this problem by proposing that when child L1 learners encounter sentences that the current
grammar cannot parse, they try out other grammatical parameters, choosing such parameters based
on innate processing preferences (e.g., construct the simplest syntactic structure possible based on the
input), “linguistic markedness considerations” (e.g., prefer a sentence interpretation that privileges
the nonmovement of syntactic constituents over the movement of such constituents; p. 365), and the
frequency of diferent structural alternatives in the input, which, in conjunction, lead child L1 learn-
ers to acquire the correct grammatical parameters for the target grammar slowly over time.
39
Carrie N. Jackson
Working under the assumption that the initial state of adult L2 acquisition is the fully mature
L1 linguistic system (i.e., full transfer/full access model; Schwartz & Sprouse, 1996) as well as the
associated processing routines used to process L1 input, Dekydtspotter and Renaud (2014) adapt
Fodor’s (1998b) parsing-to-learn hypothesis for an L2 context. Like L1 processing, L2 processing
is incremental, meaning that L2 speakers piece together the structure and meaning of a sentence in
real time rather than waiting until the end of the sentence. From this, Dekydtspotter and Renaud
(2014) propose that the incremental processing of L2 input becomes the key process driving L2
grammar acquisition. More specifcally, L2 learners will begin to acquire new L2 grammatical fea-
tures when the L1-based parsing fails for the L2, that is at the point they can no longer rely on L1
specifcations to successfully build and interpret the structure of an L2 utterance during real-time
language processing. However, Dekydtspotter and Renaud (2014) argue that such processing fail-
ures are critical for learning, because they can lead learners to register new L2 grammatical features,
thereby facilitating L2 grammatical development (for parallel arguments in the SLA literature see
VanPatten, 2012).
At the same time, not every instance of L2 processing will trigger changes to the developing
L2 linguistic system, nor will all changes necessarily lead to a more targetlike grammar. This is
because other factors—as outlined in the next sentence—may intervene and prevent learners
from successfully accessing their grammatical intuitions during real-time language processing,
or lead learners to process the input string in a manner that does not generate detailed syntac-
tic representations. Such factors include working memory or other cognitive resource limita-
tions, L1 grammar specifcations that confict with the target language grammar, and learners’
tendency to apply the least efort necessary to process an L2 utterance (see also discussion of
the least efort strategy during L1 processing in Fodor, 1998a). Such a least-efort strategy can
lead an economy-driven parser to rely on default forms, for example by over-relying on default
masculine gender marking in L2 French and thus not detecting grammatical gender mismatches
in adjectival phrases involving feminine nouns and masculine-marked adjectives (e.g., la valise
. . . la plus petit “the-FEM suitcase-FEM . . . the-FEM smallest-MASC”; Renaud, 2014). Along
similar lines, an economy-driven parser may also lead learners to posit the minimal level of syn-
tactic structure necessary to interpret an utterance, even if doing so overlooks key components
of the target language grammar.
These factors outlined by Dekydtspotter and Renaud (2014) highlight potential obstacles to
acquiring L2 grammatical structures via real-time incremental processing. In particular, Clahsen
and Felser’s Shallow Structure Hypothesis (2006, 2018) argues that L2 speakers may privilege lexical-
semantic information over morphosyntactic information when they are parsing a sentence to con-
struct the structure and meaning of an utterance in real time (see Hopp, 2023 [this volume]; Sagarra,
2023 [this volume], for review). For example, in line with the shallow structures hypothesis, Marinis
et al. (2005) reported that L2 English speakers may attach fronted wh-phrases in complex sentences
directly to their lexical subcategorizing verb, rather than taking advantage of syntactic structure to
facilitate processing, leading to a shallower, or structurally underspecifed, parse of the target sentence
(see Hopp, 2023 [this volume], for further details).
Such shallow processing could hinder learners from positing detailed morphosyntactic repre-
sentations during real-time sentence processing—a necessary component for learning according to
Dekydtspotter and Renaud (2014). Similarly, insufciently detailed lexical representations can lead
to slower and less automatic lexical retrieval. In turn, this can lead to delays in syntactic structure
building during sentence comprehension (Dekydtspotter et al., 2006; Hopp, 2018; see also Hopp,
2023 [this volume]). While parsing-to-learn hypotheses hold promise for understanding how online
sentence processing can support adult L2 acquisition, many key questions regarding the precise rela-
tionship between L2 processing and learning—and whether the link between the two is as intercon-
nected as proposed by Fodor (1998b) and Dekydtspotter and Renaud (2014)—remain.
40
Developmental stages of L2 processing

Indirect evidence for parsing-to-learn hypotheses (Dekydtspotter & Renaud, 2014; Fodor, 1998b)
comes from behavioral and event-related potentials (ERP) research on L2 speakers’ use of L2 mor-
phosyntactic information during real-time processing. While such research has not tested whether the
development of L2 morphosyntactic knowledge evolves directly from the success or failure of real-time
language processing, this research has revealed ways in which L2 acquisition may develop via discrete
stages over time. For example, Jackson (2008) showed that more-profcient L1 English–L2 German
speakers used case-marking information to process temporarily ambiguous object-frst wh-questions
(e.g., WelcheNOM/ACC Ingenieurin traf derNOM Chemiker gestern Nachmittag im Café? “Which engineer did
the chemist meet yesterday in the café?”), exhibiting longer reading times immediately at the second
noun phrase (i.e., der Chemiker) when case-marking information disambiguated the sentence to a less-
preferred object-frst sentence rather than a subject-frst sentence, even though case marking is not a
feature in the speakers’ L1 English (see also Gerth et al., 2017; Hopp, 2006). In contrast, less-profcient
L1 English–L2 German speakers only exhibited reading time diferences on the sentence-fnal phrase,
suggesting that while they still processed critical case-marking information, their ability to do so was
delayed. In a follow-up study, Jackson et al. (2012) used more sensitive eye-tracking methods to reveal
that, at intermediate profciency, L1 English–L2 German speakers took longer to process the accusative-
marked determiner den, even when embedded in preferred subject-frst sentences, pointing towards an
intermediate step in L2 acquisition whereby learners recognize the importance of case-marking infor-
mation for identifying grammatical roles, even if they do not yet use this information to incrementally
build the syntactic structure of a sentence in real time (see also Henry, 2015; see Sagarra, 2023 [this
volume], for additional discussion of L1/L2 morphosyntactic diferences and L2 sentence processing).
Related to these fndings, event-related-potentials (ERP) research shows that, especially at lower
profciency levels, L2 learners may exhibit online sensitivity to L2 grammatical information, even
when they remain at chance levels in distinguishing grammatical from ungrammatical sentences
containing the same structures, as measured by ofine grammaticality judgments (e.g., Tokowicz &
MacWhinney, 2005; Osterhout et al., 2006). ERP research collects neurophysiological responses,
measured by scalp-level electrical activity, while participants are listening to or reading sentences.
Critically, the measurement of participants’ brain responses is timed to coincide with the presence
of grammatical or semantic anomalies, and researchers then compare these brain responses to par-
ticipants’ brain responses when reading or listening to comparable sentences that do not contain an
anomaly (see Morgan-Short, 2014; Morgan-Short & Ullman, 2023 [this volume], for review).
Both longitudinal and cross-sectional ERP studies have demonstrated that L2 learners exhibit qualita-
tively diferent brain responses to grammatical versus ungrammatical L2 sentences, but that the nature of
these brain responses can vary based on a variety of diferent factors. Early ERP research revealed that the
presence of similar grammatical structures in the L1 led to more native-like brain responses when process-
ing L2 grammatical anomalies, although even L1/L2 structural similarity did not guarantee native-like
processing (e.g., Tokowicz & MacWhinney, 2005; see van Hell & Tokowicz, 2010, for review). More
recently, studies have shown that immersion-based L2 learning can lead to more native-like brain responses
than classroom-based learning contexts (e.g., Alemán Bañón et al., 2018; Faretta-Stutenberg & Morgan-
Short, 2018). More native-like brain responses are also found among participants who demonstrate greater
explicit knowledge of targeted L2 structures, as compared to participants with less explicit knowledge (e.g.,
Tanner et al., 2013; White et al., 2012; see Morgan-Short, 2014; Roberts et al., 2018, for review). Criti-
cally, such studies also report that knowledge of the specifc L2 structure under investigation is a stronger
predictor of participants’ brain responses than general L2 profciency per se. Together, such research high-
lights ways in which L2 acquisition may develop via discrete stages over time, and that a variety of factors,
including L1/L2 structural similarity and learning context, can impact the trajectory of development.
41
Carrie N. Jackson
Predictive processing
One way in which researchers have begun to draw a more direct connection between learners’
parsing routines and their impact on language learning is by investigating the role of prediction in
sentence comprehension (see Huettig, 2015; Kuperberg & Jaeger, 2016, for review) and whether
predictive sentence processing constitutes a core mechanism in language learning (e.g., Chang et al.,
2006; Dell & Chang, 2014; Phillips & Ehrenhofer, 2015). Under prediction-based learning accounts,
language users make predictions about upcoming linguistic input. When subsequent input does not
match these predictions, users experience prediction error, leading them to change future expecta-
tions. Over time, such adjustments can lead to cumulative changes in the linguistic system, which
constitutes a form of learning. However, as outlined in greater detail below, some researchers propose
that L2 speakers are less likely to engage in certain types of prediction during real-time sentence
comprehension compared to L1 speakers (e.g., Grüter et al., 2017; Hopp, 2015), and this may lead to
reduced abilities to learn from prediction error (e.g., Kaan et al., 2019; Phillips & Ehrenhofer, 2015).
Paralleling earlier psycholinguistic research arguing that L2 speakers may privilege lexical-semantic
over morphosyntactic information during online processing (e.g., Clahsen & Felser, 2006, 2018), a
similar pattern emerges in L2 predictive processing (for review, see Godfroid, 2020). Much of the
research on predictive processing uses the visual world paradigm to track L2 speakers’ eye movements
to images on a computer screen while they listen to target sentences. Multiple studies (e.g., Cham-
bers & Cooke, 2009; Dijkgraaf et al., 2017; Ito et al., 2018) have shown that L2 speakers engage in
semantic prediction, in that when they listen to semantically high-constraint sentences (e.g., Mary
reads the . . .; from Dijkgraaf et al., 2017), they use verb semantics to anticipate an upcoming noun
(e.g., letter) and gaze at an image of this target noun prior to hearing the onset of this noun in the
input.
However, even when L2 speakers successfully use verb semantics to predict upcoming input, they
often have difculty using morphosyntactic cues for predictive processing. For instance, Hopp (2015)
found that when listening to unambiguous subject-frst (SVO) and object-frst (OVS) sentences, like
(1) and (2), L1 English–L2 German speakers used lexical-semantic information to anticipate the post-
verbal noun. After hearing the sentence-initial noun rabbit and the verb eat, they exhibited a strong
subject-frst preference and always looked towards the likely patient of that action (i.e., an image of
a fower).
(1) Der Hase frisst gleich die Blume. (SVO)

the.NOM rabbit eats soon the.ACC fower
“The rabbit soon eats the fower.”
(2) Den Hasen frisst gleich der Fuchs. (OVS)
The.ACC rabbit eats soon the.NOM fox
“The fox soon eats the rabbit.”
In contrast to L1 German speakers, the L2 speakers exhibited this subject-frst preference, regardless
of whether the sentence-initial noun (i.e., Hase “rabbit”) was the nominative-marked subject or an
accusative-marked object. Further, the strength of this subject-frst preference was not modulated by
L2 profciency, suggesting that even at more advanced profciency levels, it may be difcult for L2
speakers to engage in morphosyntactic predictive processing (see also Mitsugi & MacWhinney, 2016,
for similar fndings with L2 Japanese).
Researchers have come to a similar conclusion regarding L2 speakers’ difculty of using gram-
matical gender information to predict upcoming nouns (e.g., Grüter et al., 2012; but see Dussias
et al., 2013, for counterevidence). For example, Hopp (2013) found that among L1 English–L2
German speakers, only those participants with near perfect gender assignment accuracy, as measured
42
by their accuracy when producing target nouns with the correct gender-marked determiner, then
used grammatical gender information predictively in a visual world paradigm task. Similarly, after
training intermediate to advanced L1 English–L2 German speakers on the correct gender assignment
for target German nouns, only those participants with high gender recall accuracy post-training used
gender information predictively at posttest (Hopp, 2016). Such studies highlight that the absence of
L1 grammatical gender does not preclude L2 speakers from using gender information predictively
during L2 online processing, but that the strength of speakers’ gender representations and their accu-
racy in gender assignment modulate their ability to engage in predictive processing.
When viewed together, these (and other) studies underscore that L2 speakers can use both seman-
tic and morphosyntactic cues to anticipate upcoming L2 input, but that linguistic and learner-based
factors modulate the magnitude of L2 predictive processing. Further, factors such as the accuracy and
stability of lexical and syntactic representations (e.g., Hopp, 2013, 2016), and the presence or absence
of parallel L1 information (e.g., Dussias et al., 2013; van Bergen & Flecken, 2017), may play a larger
role in supporting or hindering L2 predictive processing than L2 profciency per se. At the same time,
even native speakers may not always predict upcoming linguistic input in real time, underscoring the
need to better understand how both linguistic and learner-based factors modulate L2 prediction, and
whether reported diferences in L1 versus L2 predictive processing are quantitative or qualitative in
nature (see Kaan, 2014, for review). Especially as future research continues to uncover if and how
predictive processing and prediction error facilitate adult L2 acquisition, advancing our understand-
ing of how individual learner variables may modulate such efects—and whether such factors simi-
larly modulate L1 predictive processing—is critical.
L2 syntactic priming as a form of predictive processing

(and prediction-based learning)
Closely linked to predictive processing is structural priming, since structural priming is often char-
acterized as a form of error-based implicit learning (e.g., Chang et al., 2006; Dell & Chang, 2014;
Kaschak et al., 2014). Structural priming is the phenomenon whereby speakers are more likely to
reuse a particular syntactic structure that they have recently heard or produced, thus producing this
structure more often than they otherwise would. For instance, if someone hears a passive sentence
like, “The church was struck by lightning,” he or she is then more likely to use a passive construction,
like, “The boy was bitten by a mosquito,” in a subsequent utterance than its active alternative, namely,
“A mosquito bit the boy.” Structural priming occurs in both written and oral modalities, across child
and adult L1 and L2 speakers, and in both experimental contexts and in more naturalistic discourse
(see Gries & Kootstra, 2017; Mahowald et al., 2016, for reviews). Among L2 speakers, structural
priming occurs both within the L2 and between a speaker’s two languages (see Jackson, 2018; van
Gompel & Arai, 2018, for reviews).
Initial studies investigating structural priming measured whether L1 and L2 speakers were more
likely to produce a given structural alternative, like a passive versus an active sentence, immedi-
ately after hearing another passive sentence (e.g., Bock, 1986). Working within a lexicalist and
construction-based framework, Pickering and Branigan (1998) proposed that short-term priming
stems from residual activation of abstract syntactic representations, or combinatorial nodes. More
specifcally, combinatorial nodes provide speakers with ways to combine individual words into a
sentence. During language comprehension, these nodes are activated by a prime sentence and they
remain active for several seconds, thereby increasing the chances that this combinatorial node and
its associated syntactic structure remain active long enough to infuence subsequent production.
Alternatively, short-term priming may be driven by explicit memory of the prime sentence, which
in turn facilitates retrieval of the syntactic structure during subsequent production (e.g., Bernolet
et al., 2016).
43
Carrie N. Jackson
However, research has also shown that increased production of a target structure can extend over
a day or even several weeks later (e.g., Branigan & Messenger, 2016; Kaschak et al., 2014; Shin &
Christianson, 2012, among others). Both L1 and L2 speakers also exhibit cumulative priming, in
which the magnitude of priming increases over the course of a priming task because of repeated
exposure to the targeted structure (e.g., Jackson & Ruf, 2017; Kaan & Chun, 2018). These types of
long-term priming are difcult to explain via residual activation, leading researchers to argue that
long-term priming represents a form of implicit error-based learning, paralleling accounts of predic-
tive processing as a form of error-based learning (e.g., Chang et al., 2006; Dell & Chang, 2014; but
for an alternative learning-based account see Reitter et al., 2011). More recently, researchers have
proposed a two-locus account for structural priming, in which short-term priming is best explained
via explicit memory processes but long-term priming represents a form of implicit learning (e.g.,
Bernolet et al., 2016; see also Ferreira & Bock, 2006).
While not directly addressing the role of implicit learning per se, Hartsuiker and Bernolet
(2017) have proposed the developmental model of shared syntax to account for within- and across-
language priming in L2 speakers at diferent L2 profciency levels. As with the research on the
development of L2 morphosyntactic processing summarized earlier, Hartsuiker and Bernolet’s
(2017) model points to ways in which structural priming may play a key role in L2 acquisition.
Early on in L2 acquisition, L2 speakers’ syntactic representations (or combinatorial nodes) are
lexically based, meaning they are connected to individual words rather than being more abstract
and generalized across words in the lexicon. Thus, L2 speakers’ means for combining individual
words into a grammatical utterance are more constrained, leaving them to rely on pre-existing L1
structures or to simply imitate recently heard L2 structures during L2 production. As profciency
increases, stronger links develop between L2 syntactic representations and L2 words, with stronger
links developing earlier for more frequent L2 structures. At this point, L2 speakers may exhibit
short-term priming based on their explicit memory of utterances they have recently heard or pro-
duced. Explicit memory processes—and the short-term priming that results from such processes—
are further boosted when words are repeated between the prime and target sentence, leading to the
lexical boost efect (e.g., Jackson & Ruf, 2017; Kim & McDonough, 2008), although Hartsuiker
and Bernolet (2017) point out that at this intermediate stage, explicit memory may even account
for priming in the absence of lexical repetition. Only at advanced L2 profciency are syntactic
representations truly abstract and generalized across the L2 lexicon (and then eventually between
the L1 and the L2), with both within- and between-language priming driven by either residual
activation or implicit learning.
Hartsuiker and Bernolet’s (2017) model fnds support in both lab- and classroom-based studies
that investigate short-term and long-term priming among L2 speakers at diferent profciency levels.
For instance, less-profcient L2 speakers are more likely to exhibit priming only in the presence of
lexical repetition between prime and target sentences, whereas more profcient L2 speakers exhibit
priming both with and without lexical repetition (Kim & McDonough, 2008; Ruf, 2011). Further,
only when L2 speakers exhibit short-term priming in the absence of lexical repetition does short-
term priming translate into longer-term priming of a target structure, as measured via immediate or
delayed posttest measures (Jackson & Ruf, 2017, 2018; McDonough, 2011). Similarly, for wh-ques-
tions, McDonough and Chaikitmongkol (2010) reported a signifcant positive correlation between
L2 English speakers’ sustained production of target wh-questions on posttest and delayed posttest
measures and the number of target sentences they had produced—but not the number of prime sen-
tences they had heard—during classroom-based priming tasks. Additionally, when intermediate L2
speakers are required to repeat prime sentences aloud rather than simply reading or listening to them
(Jackson & Ruf, 2018; Kim & McDonough, 2016), or if they are forced to produce target structures
with a wider variety of verbs (McDonough & Kim, 2009; but see McDonough & de Vleeschauwer,
44
2012), they exhibit greater short- and long-term priming. Together, these results highlight that actu-
ally producing target structures—and producing them with more varied vocabulary—may be critical
for fostering longer-term learning via structural priming.
While structural priming research shows promise for facilitating L2 acquisition, many questions
remain. Importantly, McDonough and Fulga (2015) showed that among beginning L2 learners,
priming only occurred once learners possessed at least a basic understanding of the target structure,
as measured by high comprehension accuracy on a separate task in which accurate processing of the
target grammatical structure was required for comprehension (see also McDonough, 2006). Such
results suggest that while structural priming can strengthen form-meaning mappings once they have
been established, its impact may be limited in the initial stages of learning. Additionally, while many
L1 researchers assume that L1 priming is implicit in nature, it remains an open question whether
this is always the case in L2 priming (e.g., Grüter et al., 2021; Jackson & Ruf, 2018; Shin & Chris-
tianson, 2012). Future research should more systematically investigate whether the magnitude of
immediate and long-term priming is driven by the use of explicit strategies on the part of learners,
and whether the implicit versus explicit nature of L2 priming modulates learning (McDonough &
Kim, 2016; see Rebuschat, 2015, for a more general review of explicit versus implicit learning in
L2 acquisition).

L2 psycholinguistic research has come a long way in the last twenty years and we are now at the point
at which multiple hypotheses exist that meaningfully connect various facets of L2 sentence process-
ing and L2 production to L2 acquisition. However, we still need more empirical research in both
lab-based settings and classroom-based or other more naturalistic learning contexts (see Textbox 4.2).
Future research must also increase the diversity of languages and learners represented in L2 psycho-
linguistic research, because even the brief review of research here highlights that not all processing
routines lead to equal outcomes for all learners. More research is also needed to test the impact of
diferent instructional methods that are grounded in our current understanding of L2 psycholinguis-
tic constructs and theories.
Finally, it behooves researchers and practitioners alike to focus less on identifying potential difer-
ences between L1 and L2 speakers, and instead to think more about how factors like those reviewed
here—including task demands, speed of lexical retrieval, or the ability to engage in predictive pro-
cessing—modulate language processing and acquisition among L2 and L1 speakers (see also Grüter
et al., 2020; Kaan, 2014). Only then will we gain a more nuanced understanding of the cognitive
and linguistic mechanisms that underlie language acquisition and use across speakers and populations.
How does L2 learners’ ability to process L2 input in real time impact L2 acquisition?
How do diferent instructional methods and learning contexts impact real-time L2 sentence process-
ing and do potential processing diferences translate into diferences in the rate or success of L2
acquisition?
How do lexical retrieval speed and vocabulary development infuence L2 processing routines and the
acquisition of L2 morphosyntax?
How do diferent L1 and L2 populations learn from mis-parses and prediction error?
45
Carrie N. Jackson
Further reading
McDonough, K., & Trofmovich, P. (Eds.). (2011). Applying priming methods to L2 learning, teaching and research.
John Benjamins.
This edited volume presents a collection of empirical research articles that involve various priming method-
ologies (including but not limited to structural priming) and discussion of how psycholinguistic research on
priming is relevant for L2 teaching and learning.
Sanz, C., & Leow, R. (Eds.). (2011). Implicit and explicit language learning: Conditions, processes, and knowledge in
SLA and bilingualism. Georgetown University Press.
This edited volume contains empirical research articles that encompass a wide range of psychological and
neurological constructs and how they are relevant for L2 processing and learning.
VanPatten, B., Keating, G. D., & Wulf, S. (2020). Theories in second language acquisition: An introduction. Routledge.
This edited volume provides an overview of a range of key theories in second language acquisition. It
includes several chapters that probe the relationship between L2 processing and L2 learning.
References
Alemán Bañón, J., Fiorentino, R., & Gabriele, A. (2018). Using event-related potentials to track morphosyntac-
tic development in second language learners: The processing of number and gender agreement in Spanish.
PLoS ONE, 13(7).
Bernolet, S., Collina, S., & Hartsuiker, R. J. (2016). The persistence of syntactic priming revisited. Journal of
Memory and Language, 91, 99–116.
Birdsong, D. (1992). Ultimate attainment in second language acquisition. Language, 68(4), 706–755.
Bock, J. K. (1986). Syntactic persistence in language production. Cognitive Psychology, 18(3), 355–387.
Branigan, H. P., & Messenger, K. (2016). Consistent and cumulative efects of syntactic experience in children’s
sentence production: Evidence for error-based implicit learning. Cognition, 157, 250–256.
Chambers, C. G., & Cooke, H. (2009). Lexical competition during second-language listening: Sentence con-
text, but not profciency, constrains interference from the native lexicon. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 35(4), 1029–1040.
Chang, F., Dell, G. S., & Bock, K. (2006). Becoming syntactic. Psychological Review, 113(2), 234.
Clahsen, H., & Felser, C. (2006). Grammatical processing in language learners. Applied Psycholinguistics, 27(1),
3–42.
Clahsen, H., & Felser, C. (2018). Some notes on the shallow structure hypothesis. Studies in Second Language
Acquisition, 40(3), 693–706.
Conklin, K., & Thul, R. (2023). Word and multiword processing: cognitive approaches. In A. Godfroid & H.
Hopp (Eds.), The Routledge handbook of second language acquisition and psycholinguistics (pp. 203–215). Routledge.
Dekydtspotter, L., & Renaud, C. (2014). On second language processing and grammatical development: The
parser in second language acquisition. Linguistic Approaches to Bilingualism, 4(2), 131–165.
Dekydtspotter, L., Schwartz, B. D., & Sprouse, R. A. (2006). The comparative fallacy in L2 processing research.
In M. O’Brien, C. Shea, & J. Archibald (Eds.), Proceedings of the 8th generative approaches to second language
acquisition conference (GASLA 2006) (pp. 33–40). Cascadilla Proceedings Project.
Dell, G. S., & Chang, F. (2014). The P-chain: Relating sentence production and its disorders to comprehen-
sion and acquisition. Philosophical Transactions of the Royal Society B: Biological Sciences, 369(1634), 20120394.
Dijkgraaf, A., Hartsuiker, R. J., & Duyck, W. (2017). Predicting upcoming information in native-language and
non-native-language auditory word recognition. Bilingualism: Language and Cognition, 20(5), 917–930.
Dussias, P. E., Krof, J. R. V., Tamargo, R. E. G., & Gerfen, C. (2013). When gender and looking go hand in
hand: Grammatical gender processing in L2 Spanish. Studies in Second Language Acquisition, 35(2), 353–387.
Faretta-Stutenberg, M., & Morgan-Short, K. (2018). The interplay of individual diferences and context of
learning in behavioral and neurocognitive second language development. Second Language Research, 34(1),
67–101.
Ferreira, V. S., & Bock, K. (2006). The functions of structural priming. Language and Cognitive Processes, 21(7–8),
1011–1029.
Fodor, J. D. (1998a). Learning to parse? Journal of Psycholinguistic Research, 27(2), 285–319.
Fodor, J. D. (1998b). Parsing to learn. Journal of Psycholinguistic Research, 27(3), 339–374.
46
Gerth, S., Otto, C., Felser, C., & Nam, Y. (2017). Strength of garden-path efects in native and non-native
speakers’ processing of object—subject ambiguities. International Journal of Bilingualism, 21(2), 125–144.
Godfroid, A. (2020). Eye tracking in second language acquisition and bilingualism: A research synthesis and methodological
guide. Routledge.
& H. Hopp (Eds.), The Routledge handbook of second language acquisition and psycholinguistics (pp. 191–202).
Routledge.
Gries, S. T., & Kootstra, G. J. (2017). Structural priming within and across languages: A corpus-based perspec-
tive. Bilingualism: Language and Cognition, 20(2), 235–250.
Grüter, T., Lau, E., & Ling, W. (2020). How classifers facilitate predictive processing in L1 and L2 Chinese: The
role of semantic and grammatical cues. Language, Cognition and Neuroscience, 35(2), 221–234.
Grüter, T., Lew-Williams, C., & Fernald, A. (2012). Grammatical gender in L2: A production or a real-time
processing problem? Second Language Research, 28(2), 191–215.
Grüter, T., Rohde, H., & Schafer, A. J. (2017). Coreference and discourse coherence in L2: The roles of gram-
matical aspect and referential form. Linguistic Approaches to Bilingualism, 7(2), 199–229.
Grüter, T., Yanxin, A. Z., & Jackson, C. N. (2021). Forcing prediction increases priming and adaptation in sec-
ond language production. In E. Kaan & T. Grüter (Eds.), Prediction in second language processing (pp. 207–231).
John Benjamins Press.
Hartsuiker, R. J., & Bernolet, S. (2017). The development of shared syntax in second language learning. Bilin-
gualism: Language and Cognition, 20(2), 219–234.
Henry, N. (2015). Morphosyntactic processing, cue interaction, and the efects of instruction: An investigation of processing
instruction and the acquisition of case marking in L2 German [Unpublished Ph.D. dissertation], Pennsylvania State
University, University Park, PA.
Hopp, H. (2006). Syntactic features and reanalysis in near-native processing. Second Language Research, 22(3),
369–397.
Hopp, H. (2013). Grammatical gender in adult L2 acquisition: Relations between lexical and syntactic variabil-
ity. Second Language Research, 29(1), 33–56.
Hopp, H. (2015). Semantics and morphosyntax in predictive L2 sentence processing. International Review of
Applied Linguistics in Language Teaching, 53(3), 277–306.
Hopp, H. (2016). Learning (not) to predict: Grammatical gender processing in second language acquisition.
Second Language Research, 32(2), 277–307.
Hopp, H. (2023). Sentence processing: cognitive approaches to L2 morphosyntactic and morphological process-
ing. In A. Godfroid & H. Hopp (Eds.), The Routledge handbook of second language acquisition and psycholinguistics
(pp. 216–228). Routledge.
Huettig, F. (2015). Four central questions about prediction in language processing. Brain Research, 1626, 118–135.
Ito, A., Corley, M., & Pickering, M. J. (2018). A cognitive load delays predictive eye movements similarly during
L1 and L2 comprehension. Bilingualism: Language and Cognition, 21(2), 251–264.
Jackson, C. N. (2008). Profciency level and the interaction of lexical and morphosyntactic information during
L2 sentence processing. Language Learning, 58(4), 875–909.
Jackson, C. N. (2018). Second language structural priming: A critical review and directions for future research.
Jackson, C. N., Dussias, P. E., & Hristova, A. (2012). Using eye-tracking to study the on-line processing of case-
marking information among intermediate L2 learners of German. International Review of Applied Linguistics in
Language Teaching, 50(2), 101–133.
Jackson, C. N., & Ruf, H. T. (2017). The priming of word order in second language German. Applied Psycho-
linguistics, 38(2), 315–345.
Jackson, C. N., & Ruf, H. T. (2018). The importance of prime repetition among intermediate-level second
language learners. Studies in Second Language Acquisition, 40(3), 677–692.
Jufs, A., & Rodríguez, G. A. (2014). Second language sentence processing. Routledge.
Kaan, E. (2014). Predictive sentence processing in L2 and L1: What is diferent? Linguistic Approaches to Bilingual-
ism, 4(2), 257–282.
Kaan, E., & Chun, E. (2018). Priming and adaptation in native speakers and second-language learners. Bilingual-
ism: Language and Cognition, 21(2), 228–242.
Kaan, E., Futch, C., Fuertes, R. F., Mujcinovic, S., & de la Fuente, E. Á. (2019). Adaptation to syntactic struc-
tures in native and nonnative sentence comprehension. Applied Psycholinguistics, 40(1), 3–27.
Kaschak, M. P., Kutta, T. J., & Coyle, J. M. (2014). Long and short term cumulative structural priming efects.
Language, Cognition and Neuroscience, 29(6), 728–743.
47
Carrie N. Jackson
Kim, Y., & McDonough, K. (2008). Learners’ production of passives during syntactic priming activities. Applied
Linguistics, 29(1), 149–154.
Kim, Y., & McDonough, K. (2016). Prime repetition and Korean EFL learners’ comprehension and production
of passives. International Review of Applied Linguistics in Language Teaching, 54(4), 319–346.
Kuperberg, G. R., & Jaeger, T. F. (2016). What do we mean by prediction in language comprehension? Lan-
guage, Cognition and Neuroscience, 31(1), 32–59.
Mahowald, K., James, A., Futrell, R., & Gibson, E. (2016). A meta-analysis of syntactic priming in language
production. Journal of Memory and Language, 91, 5–27.
Marinis, T., Roberts, L., Felser, C., & Clahsen, H. (2005). Gaps in second language sentence processing. Studies
McDonough, K. (2006). Interaction and syntactic priming: English L2 speakers’ production of dative construc-
tions. Studies in Second Language Acquisition, 28(2), 179–207.
McDonough, K. (2011). Eliciting wh-questions through collaborative syntactic priming activities during peer
interaction. In P. Trofmovich & K. McDonough (Eds.), Applying priming methods to L2 learning, teaching and
research: Insights from psycholinguistics (pp. 131–151). John Benjamins.
McDonough, K., & Chaikitmongkol, W. (2010). Collaborative syntactic priming activities and EFL learners’
production of wh-questions. Canadian Modern Language Review, 66(6), 817–841.
McDonough, K., & De Vleeschauwer, J. (2012). Prompt-type frequency, auditory pattern discrimination, and
EFL learners’ production of wh-questios. Studies in Second Language Acquisition, 34(3), 355–377.
McDonough, K., & Fulga, A. (2015). The detection and primed production of novel constructions. Language
Learning, 65(2), 326–357.
McDonough, K., & Kim, Y. (2009). Syntactic priming, type frequency, and EFL learners’ production of wh-
questions. The Modern Language Journal, 93(3), 386–398.
McDonough, K., & Kim, Y. (2016). Working memory and L2 English speakers’ primed and subsequent pro-
duction of passives. In G. Granena, D. O. Jackson, & Y. Yilmaz (Eds.), Cognitive individual diferences in second
language processing and acquisition (pp. 205–222). John Benjamins.
Mitsugi, S., & MacWhinney, B. (2016). The use of case marking for predictive processing in second language
Japanese. Bilingualism: Language and Cognition, 19(1), 19–35.
Morgan-Short, K. (2014). Electrophysiological approaches to understanding second language acquisition: A
feld reaching its potential. Annual Review of Applied Linguistics, 34, 15–36.
Omaki, A. (2017). Linking learning and parsing in bilingual sentence processing. Bilingualism: Language and
Cognition, 20(4), 706–707.
Omaki, A., & Lidz, J. (2015). Linking parser development to acquisition of syntactic knowledge. Language
Acquisition, 22(2), 158–192.
Osterhout, L., McLaughlin, J., Pitkänen, I., Frenck-Mestre, C., & Molinaro, N. (2006). Novice learners, longi-
tudinal designs, and event-related potentials: A means for exploring the neurocognition of second language
processing. Language Learning, 56(S1), 199–230.
Phillips, C., & Ehrenhofer, L. (2015). Learning obscure and obvious properties of language. Linguistic Approaches
to Bilingualism, 5(4), 545–555.
Pickering, M. J., & Branigan, H. P. (1998). The representation of verbs: Evidence from syntactic priming in
language production. Journal of Memory and Language, 39(4), 633–651.
Rebuschat, P. (Ed.). (2015). Implicit and explicit learning of languages (Vol. 48). John Benjamins Publishing Company.
Reitter, D., Keller, F., & Moore, J. D. (2011). A computational cognitive model of syntactic priming. Cognitive
Science, 35(4), 587–637.
Renaud, C. (2014). A processing investigation of the accessibility of the uninterpretable gender feature in L2
French and L2 Spanish adjective agreement. Linguistic Approaches to Bilingualism, 4(2), 222–255.
Roberts, L., González Alonso, J., Pliatsikas, C., & Rothman, J. (2018). Evidence from neurolinguistic meth-
odologies: Can it actually inform linguistic/language acquisition theories and translate to evidence-based
applications? Second Language Research, 34(1), 125–143.
Ruf, H. T. (2011). An investigation of syntactic priming among German speakers at varying profciency levels [Unpub-
lished Ph.D. dissertation], University of Wisconsin, Madison, WI.
Sagarra, N. (2023). Sentence processing: cognitive approaches to L2 morphosyntactic and morphological process-
ing. In A. Godfroid & H. Hopp (Eds.), The Routledge handbook of second language acquisition and psycholinguistics
48
Schwartz, B. D., & Sprouse, R. A. (1996). L2 cognitive states and the full transfer/full access model. Second
Language Research, 12(1), 40–72.
Shin, J. A., & Christianson, K. (2012). Structural priming and second language learning. Language Learning,
62(3), 931–964.
Tanner, D., McLaughlin, J., Herschensohn, J., & Osterhout, L. (2013). Individual diferences reveal stages of L2
grammatical acquisition: ERP evidence. Bilingualism: Language and Cognition, 16(2), 367–382.
Tokowicz, N. (2014). Lexical processing and second language acquisition. Routledge.
Tokowicz, N., & MacWhinney, B. (2005). Implicit and explicit measures of sensitivity to violations in second
language grammar: An event-related potential investigation. Studies in Second Language Acquisition, 27(2),
173–204.
Van Bergen, G., & Flecken, M. (2017). Putting things in new places: Linguistic experience modulates the pre-
dictive power of placement verb semantics. Journal of Memory and Language, 92, 26–42.
Van Gompel, R. P., & Arai, M. (2018). Structural priming in bilinguals. Bilingualism: Language and Cognition,
21(3), 448–455.
Van Hell, J. G., & Tokowicz, N. (2010). Event-related brain potentials and second language learning: Syntactic
processing in late L2 learners at diferent L2 profciency levels. Second Language Research, 26(1), 43–74.
VanPatten, B. (2012). Input processing. In S. Gass & A. Mackey (Eds.), The Routledge handbook of second language
acquisition (pp. 268–281). Taylor and Francis.
White, E. J., Genesee, F., & Steinhauer, K. (2012). Brain responses before and after intensive second language
learning: Profciency based changes and frst language background efects in adult learners. PLoS ONE,
7(12).
White, L. (2003). Second language acquisition and universal grammar. Cambridge University Press.
49
5
OF L3/Ln ACQUISITION
Introduction
Traditionally, the study of non-native languages has been limited, at least conceptually, to second lan-
guage (L2) acquisition and processing. Our aim in this chapter is to introduce, dissect, and address the
reasonable question of whether a separate consideration of third or further language (L3/Ln) acquisi-
tion and processing is justifed on empirical grounds. Is acquiring and processing an L3/Ln the same
as acquiring and processing an L2? If not, are diferences between L2 and Ln merely quantitative or
also qualitative in nature? Does language processing (across the system) change with each additional
language in a manner that is predictable from observing monolingual-to-bilingual diferences?
De Bot and Jaensch (2015) asked a crucial question that subsumes all of the foregoing: What
is special about L3 processing? The evidence base at the time turned out to be less than abundant in
some areas, but the article ofers a comprehensive review of bilingual to multilingual diferences (or
lack thereof) across diferent domains of language and cognition. De Bot and Jaensch concluded
that, while the cognitive machinery underlying online production and comprehension is likely to
be qualitatively similar for any context of language irrespective of timing and order of acquisition,
there is signifcant value in considering multilingual processing as a subject in its own right. If this
general hypothesis is on the right track, the mere presence of surface diferences between L2 and L3/
Ln stands out and requires meaningful explanation. How could the same mechanisms be at play if
developmental trajectories and outcomes seem ostensibly distinct?
Oftentimes, multilingual settings incorporate variables that are either absent, such as multiple
previous languages that could afect the L3 or Ln, or confounded in bilingualism, such as processing
slowdowns due to one additional language. In this sense, multilingualism ofers an under-utilized
natural laboratory to explore signifcant questions impossible to properly frame, much less inves-
tigate within bilingualism (two languages) alone. While it is possible that most researchers do not
consciously assume L2 and L3/Ln to be efectively equivalent, methodological practice has very
frequently confounded the two. Many studies in the guise of “L2” have failed to properly divide or
even acknowledge that many cohorts include individuals who are true L2 learners alongside those
who have previously acquired several non-native languages. As a natural laboratory, L3/Ln allows for
a deeper understanding of how previous linguistic experience shapes subsequent language acquisition
and processing by virtue of the mind having more relevant experience at its disposal than would be
the case in bilingualism alone. More specifcally, how the selection of competing previously acquired
linguistic experience is negotiated in acquisition and processing ofers an opportunity to explore
50 DOI: 10.4324/9781003018872-6
The psycholinguistics of L3/Ln acquisition
the interface of linguistic features and domain general cognition in unique ways. Much of the work
in L3/Ln morphosyntax has focused on how to predict and model which previous language (the
L1 or the L2) constitutes the main source of transfer in L3/Ln acquisition and how this impacts the
developmental sequence. This research embodies the confuence of two lines of inquiry pursued in
the previous literatures of L3/Ln vocabulary acquisition, on the one hand, and L2 morphosyntax,
on the other.
Indeed, even assuming that L2 and L3/Ln processing are fundamentally similar, this does not
mean that we can reliably predict how the presence of more or fewer languages impacts the function-
ing of the overall system, and the processing of each individual language. For example, the precise
nature of changes in processing speed across the diferent languages of a multilingual speaker, which
are likely to appear as a result of acquisition or attrition, depends on factors that go beyond the over-
all cognitive “weight” of the system. Accurately modeling multilingual processing thus requires an
empirical assessment of the relative magnitude of these changes.
In Section 2, we will review some studies that suggest diferent types and rates of impact as a func-
tion of variables such as amount of input/exposure or typological (genealogical) proximity between
languages. Section 3 is dedicated to discussing transfer source selectivity of grammar, as opposed to lexis,
as one of the main diferences between L2 and L3/Ln, with a special focus on the insights and oppor-
tunities for disambiguation that language processing data ofer for addressing this research question.
Multilingual: A person with at least some knowledge, active or passive, of more than two languages.
Sequential multilingualism: The acquisition of one (or more) languages after the onset of acquiring at
least two other languages.
Linguistic transfer: The infuence of one language on another at the level of mental representation within
the multilingual system.
Cognates: Words belonging to diferent languages that, through a common etymology, retain some degree
of semantic and formal (phonological/orthographic) similarity (see Conklin, 2023 [this volume]).
Mirror-image design: Experimental design consisting of testing two groups of L3 learners of the same
language which also have the same previous language combination, but in reverse order of acquisition
(e.g., L1 Hindi–L2 English–L3 French vs. L1 English–L2 Hindi–L3 French).
Nonlinear consequences of sequential multilingualism

It might be tempting to assume that what we know of L2 processing will apply to any subsequently
acquired languages. After all, there is typically much more in common between the L2 and further
languages than there is between the L1 and the L2: Learning context, cognitive maturity, or prior
communicative ability (which is markedly diferent for L2 learners as compared to children acquiring
their L1s), among other factors, contribute to align instances of non-native language acquisition and
processing as a somewhat diferent set of developmental processes from the L1.
A reasonable frst-pass hypothesis would then be that any diferences between L2 and L3/Ln
acquisition and processing are destined to be simply quantitative, e.g., that Ln processing will be
slower. Here, we want to discuss why diferences in the weighting of similar contributing variables
51
are not so simple, and why their study is warranted if we want to accurately characterize multilingual
processing. Processing speed of a given word or a given (sub)domain of grammar is determined by
a complex set of factors that includes, at least, (i) language-specifc exposure/use, which is a large
contributor to the strength and therefore the retrievability of linguistic representations, and (ii) lexi-
cal or structural similarity between the various languages of a multilingual, which may lead to some
degree of overlap in representation across languages (e.g., shared syntax, Hartsuiker et al., 2004). The
interaction of (i) and (ii) is particularly crucial: With markedly separate linguistic representations (e.g.,
structurally diferent ways of verbalizing the same event), exposure to one or another language will
only strengthen the representation for the structure used in that language and therefore likely increase
processing speed. With partially or completely overlapping representations, language-specifc expo-
sure/use may increase processing speed for that structure across the system. In short, linear growth in
the number of languages has nonlinear consequences which merit their own empirical investigation.
Most of the evidence we have about processing in an L3/Ln sufers from a confound well known
from the study of L2 processing: the difculty of disentangling a bilingualism efect from a non-
native language efect. While processing speed is typically slower in a non-native language (a fnding
as old as the seminal work of Cattell, 1887), research with highly profcient sequential bilinguals
has shown that these diferences may be more directly related to diferential amounts of exposure
and use between the native and non-native languages than to anything specifc about the processing
of the latter. This is deducible from the fact that the distance is demonstrably reduced when these
learners are tested under conditions that provide some compensation for L1-to-L2 input/exposure
diferences, e.g., early/simultaneous bilinguals, or learners in an L2 immersion context (Basnight-
Brown & Altarriba, 2007; Duñabeitia et al., 2010; Zhao et al., 2011). Additionally, research com-
paring the lexical processing of monolingual and bilingual speakers has found that bilinguals seem
to be slower than their monolingual peers in their frst and dominant language (e.g., Gollan et al.,
2005; Ivanova & Costa, 2008), but this has been attributed to the relatively lower amount of total
experience with a single language in bi-/monolinguals as compared to monolinguals (the frequency-
lag hypothesis; e.g., Gollan et al., 2005, 2011). Unfortunately, to our knowledge, very few studies
have controlled for such confounds in the L3/Ln processing literature. A notable early exception is
the study by Mägiste (1979), who showed that monolingual German and Swedish speakers outper-
formed age-matched bilingual (German-Swedish) and multilingual speakers (similar to the previous
group, but including another native language) in speed and, sometimes, accuracy in a series of lin-
guistic encoding and decoding tasks—mostly translation and naming. Additionally, bilingual speakers
outperformed multilingual speakers.
In a recent study, Sullivan et al. (2018) compared 200 monolingual, bilingual, and trilingual
speakers, including both young and old adults, in a picture-naming task in English. The aim was
to test the frequency-lag hypothesis, mentioned earlier, against an alternative view whereby slower
lexical retrieval in multilingual speakers (understood as non-monolinguals) may be due to the need
to resolve confict between two (or several) jointly activated alternatives (the competition account;
Abutalebi & Green, 2007). Their results partially supported the predictions of the frequency-lag
hypothesis, in that lower-frequency words amplifed group diferences between the monolinguals,
on the one hand, and the bi- and trilinguals, on the other. However, their data were most compat-
ible with the competition account, since these diferences did not make a distinction between the
bilinguals and the trilinguals, and did not attenuate with age, which would both be predicted by the
frequency-lag hypothesis.
Cognate efects, whereby the processing of a given word is facilitated in bilingual speakers if its
translation equivalent bears resemblance in form in addition to meaning, have been a well-studied
phenomenon in the bilingual literature since Caramazza and Brones (1979). The same has been
shown to hold for both L1–L2 and L1–L3 cognates in L1 lexical processing (van Hell & Dijkstra,
2002). Lemhöfer et al. (2004) investigated these efects in the processing of German words by L1
52
Dutch–L2 English–L3 German multilinguals. In a lexical decision task, participants responded fastest
to triple cognates (e.g., echo), then to double (L1–L3) cognates (e.g., Schuld, “guilt”), and, fnally, to
uniquely German words (e.g., Antrag, “application”). When these results were compared to those
of a monolingual German group, no diferences were observed across these word categories. Similar
results were obtained by Szubko-Sitarek (2011) with L1 Polish–L2 English–L3 German speakers in
their L1 and L3, and by Poarch and van Hell (2014) with three groups of trilinguals: L1 Dutch–L2
English–L3 German, L1 German–L2 English–L3 Dutch, and L1 Russian–L2 English–L3 German,
the latter two groups being immersed in an L3 environment. Poarch and van Hell (2014) found
that cognate efects held across languages even in the face of script diferences (for triple cognates),
although they were modulated by both immersion and profciency.
Cognate efects have been argued to support lexical incorporation (i.e., vocabulary learning)
in multilingualism (e.g., Hall & Ecke, 2003), as it combines two pathways for support of memory
formation in establishing new lexical representations: semantic and formal, i.e., phonological/ortho-
graphic, similarity. In this sense, pseudocognates provide a unique testing ground for such claims,
because they combine formal similarity with dissimilar semantics. González Alonso (2012) tested L1
Polish–L2 English–L3 Russian speakers in a picture naming task containing target words in all three
languages. After hearing an auditory prime word in one of the other two languages, participants
were visually cued to name an upcoming picture in the other. Primes could either be unrelated
noncognate words, or pseudocognates—e.g., English fabric as a prime to the picture of a factory
(Russian фабрика, “fabrika”). For the L3 Russian targets, participants were faster when these fol-
lowed an L1 Polish or an L2 English pseudocognate to the Russian word. These results suggest that
the multilingual lexicon exploits similarity at diferent levels to enhance the incorporation of new
representations, as proposed by the parasitic model (Hall & Ecke, 2003). In this model, similarities
between the target L3 vocabulary item and words from previously acquired languages are sought
and capitalized on beyond semantic (i.e., translation) equivalencies. These may include similarities in
phonology, subcategorization preferences, argument structure and more. The initial links established
through these connections to facilitate retrieval of the incipient L3/Ln lexical representations (para-
sites) are gradually overcome and abandoned as the L3 words receive sufcient reinforcement through
conceptual activation to become independent of their original hosts. Indeed, these pseudocognate
efects were less prominent for the more advanced learners in González Alonso (2012), suggesting
that learners move towards establishing independent representations in the L3.
Crucially, what all these studies are missing is a direct comparison of (pseudo)cognate efects
between a second and a third language, controlling for diferences in profciency and keeping the
amount of cognate overlap constant. However, they are indicative of cumulative processes that reso-
nate across the whole of the multilingual system. In order to illuminate the issue of how the impact
rate of a given factor changes as the system grows, further inquiry should zoom in on the efect size
of these diferences. It would be particularly interesting, for example, to fnd out whether the efect
size for the diference between unique German words and Dutch-German cognates in Lemhöfer
et al. (2004) is larger or smaller than the efect size for the diference between the double and triple
cognates.
As these studies on cognate processing demonstrate, it is not a given that processing speed increases
as a function of the number of languages someone speaks (pace Mägiste, 1979), but rather it seems to
be conditioned by cross-linguistic similarity and distribution of input across these languages. While
the lexical processing literature has a long tradition of examining these variables, some recent work
has also explored the impact of converging syntactic representations in multilingualism. Cedden and
Aydın (2019) used a self-paced reading task to investigate the processing of canonical SOV sentences
in Turkish compared to two types of noncanonical structures: SVO sentences in which the object
was base-generated preverbally but moved to a postverbal position, and SVO-ki sentences, where
the postverbal constituents were base-generated there. Their participants belonged to one of four
53
groups: Turkish monolinguals, L1 Turkish–L2 English bilinguals, L1 Turkish–L2 English–L3 Ger-

man trilinguals, and a L3+ multilingual group that spoke at least four languages. When consider-
ing the data from all sentence types, the monolingual group was the slowest of the four in average
reading speed across all sentences, with the other three groups not difering from each other. Other
than average speed, nothing seemed to change qualitatively across the relative continuum of increas-
ingly multilingual speakers: All groups were faster at processing canonical SOV sentences than non-
canonical SVO sentences of both types. Two aspects from Cedden and Aydın’s (2019) results are
especially relevant to our discussion. The frst is that processing speed, at least for these structures,
reliably decreased beyond monolingual status. The second is that this rate of change was not sustained
throughout further instances of sequential multilingualism. However, there was a 12-year diference
between the mean age of the monolingual group and that of all other groups, and the authors did
not report socioeconomic status (SES) information. Further research should establish whether these
factors may have contributed, and to what extent, to these results.
The studies reviewed in this section demonstrate that processing speed is variably afected by
the incorporation of new languages to the multilingual system. Neither the direction (slowdown or
speedup) nor the size of this impact can be straightforwardly predicted, since these are likely modu-
lated by factors such as cross-linguistic similarity among the languages involved, diferences and simi-
larities in phonology and script, amount and type of exposure to the new language (e.g., immersion
context), and more. Overall, these data present a complex picture in which linguistic representations
at diferent levels interact dynamically across all languages of a multilingual.
Theoretical perspectives on cross-language effects in L3/Ln

In this section, we focus on the literature exploring source selectivity (L1, L2, or both) of transfer
efects specifcally in L3/Ln (see Rothman et al., 2019, for review). While our main focus hones
in on L3/Ln (interlanguage) morphosyntax, we must acknowledge that very similar questions have
existed within the study of the multilingual lexicon since the late 1970s/1980s (e.g., Ringbom,
1978, 1987; Singleton, 1987). In an extension of long-lasting discussions in bilingualism, the relevant
processing literature has typically been concerned with cross-linguistic infuence (CLI), understood
as interference in lexical access and retrieval, and what this phenomenon can reveal about the archi-
tecture of the multilingual lexicon (e.g., Dijkstra, 2003). Since the early 2000s, there has been a sharp
increase in studies focused on L3/Ln learners and their bespoke vulnerabilities to instances of CLI
with respect to the creation and development of L3 lexical representations instead of—or besides—
their processing (e.g., González Alonso, 2012; Hall & Ecke, 2003; Hammarberg, 2001).
Here, the question frst came to light of whether the L1 or the L2 plays a more prominent role in
transfer/CLI—and, if so, why. A number of potential factors were proposed from early on, including
profciency, recency of activation, L2 status, and the typological relationships between the L3 and any
pre-existing languages (e.g., Williams & Hammarberg, 1998). Of these, two factors received special
attention and empirical support: Some evidence pointed at a critical role of (psycho)typology (e.g.,
Cenoz, 2001), whereas other evidence was interpreted as indicating a special status of the L2 as the
default interlingual supplier (e.g., Hammarberg, 2001; Williams & Hammarberg, 1998, p. 304).
The insights of the multilingual vocabulary acquisition literature heavily infuenced the basis for
the theoretical models in the later emerging feld of formal approaches to L3 morphosyntax. In
fact, three of its most dominant models to date advocate either L2 status or typological/structural
similarity as the main variables for selection of a transfer source in L3/Ln: the L2 status factor (L2SF;
Bardel & Falk, 2007, 2012; Falk & Bardel, 2011), the typological primacy model (TPM; Rothman,
2011, 2015; González Alonso & Rothman, 2017; Rothman et al., 2019), and the linguistic proximity
model (LPM; Mykhaylyk et al., 2015; Westergaard et al., 2017). These were preceded by a seminal
paper by Flynn et al. (2004) ofering the cumulative enhancement model (CEM).1
54
The CEM claims that infuence of previous linguistic representations would either facilitate L3/
Ln or efectively be blocked; in other words, transfer efects are only expected if specifc representa-
tions from the L1 or L2 align, property-by-property, with the target L3. The L2SF claims that L2
representations are the default for transfer (Bardel & Falk, 2007), given a proposed psychological
similarity between an L2 and other subsequent languages—following the authors’ interpretation of
Paradis’ (2009) declarative/procedural memory distinction (Bardel & Falk, 2012). Although there are
cases in which this L2 default might be overridden (e.g., Falk et al., 2015), the idea of a privileged
role for L2 selectivity means that, contra the CEM, L3/Ln transfer is envisaged to be replete with
non-facilitative efects predicted to stem from L2 representations.
The TPM and LPM agree with the L2SF that transfer in L3/Ln is not uniquely facilitative; that is,
that linguistic representations from previously acquired languages can sometimes lead to non-target
outcomes in the L3. Instead of the sequence of acquisition of previous languages, however, they posit
that transfer source selection is guided by cognitive economic factors, although they conceptualize
these diferently (see the following). Similarly, both hold that selection between the L1 and the L2 is
linguistically motivated. However, they diverge most signifcantly in their proposals related to tim-
ing and extent of transfer. The TPM is a full transfer proposal, in the sense of Schwartz and Sprouse
(1996), whereby the entirety of one system is transferred as a full copy as an initial (fully specifed)
interlanguage for L3 development. Which complete system copies over is determined after an ini-
tial state of minimal exposure to the target L3 in which the parser holistically determines L1 or L2
selection based on typological similarity, as assessed via an implicational hierarchy of linguistic cues:
lexicon → phonotactics → functional morphology → basic syntactic structure (see Rothman, 2015;
Rothman et al., 2019). This means that representational transfer is predicted to be complete from a
single source and very early in the L3 process.
Conversely, the LPM maintains that transfer is not holistic into an initial L3 interlanguage. While
there is a potential for full transfer, in that all structure from either language is in principle available
for transfer, transfer is argued to happen property-by-property when relevant in the L3/Ln process.
As such, it is unlikely to have a single source: Transfer can come from either the L1 or L2 depend-
ing on what the parser analyzes as the best source for any given domain at the point in development
that such domain is within the focus of acquisition. At frst glance, this seems reminiscent of the
CEM’s claim of property-by-property transfer. Recall, however, that the LPM rejects the notion
that linguistic properties are transferred only to the extent that they are targetlike, claiming (like the
TPM) that structural proximity leads the parser to make its selection. To account for the observation
of non-facilitative transfer efects, the LPM claims that factors such as lexical similarity can override,
at least at an early stage, the parser’s sensitivity to fne grammatical distinctions and contribute to its
misanalyzing whether the L1 or L2 would be a better source of transfer (Westergaard, 2019; Wester-
gaard et al., 2017).
Despite non-trivial diferences between the foregoing models, some things are shared across them
all. Each ofers a view and argumentation whereby not all non-native language acquisition/process-
ing can or should be viewed in the same way. All agree that L2 and L3/Ln must constitute distinct
felds that can only ever partially overlap in their theoretical focus. Furthermore, each model consid-
ers, albeit in distinct ways, psycholinguistic factors such as processing and/or cognitive load factors
as a driving force in transfer selectivity. For example, whereas the L2SF supports its default L2 status
based on claimed diferences in how non-native languages are acquired and the memory systems
implicated, the TPM appeals to notions of cognitive economy pressures to avoid acquisition redun-
dancies and reduce previous language activation loads throughout L3 development to motivate its
claims of full transfer.
Puig-Mayenco et al. (2020) provide a systematic review of the increasing literature that comes
to bear on these models. Bringing together the results of 71 of-line behavioral studies published
between 2004 and 2017, they examined the extent to which macro-variables related to claimed
55
source (L1, L2, or both) and type of transfer (evidence of typology and/or facilitation factors for
source selection) correlate with methodological factors within studies (e.g., profciency in the L3 at
experimental testing, type of methodology [comprehension vs. production], languages tested). The
bird’s eye view from collating the data suggests that some factors are more infuential than others.
The clearest result is the proliferation of non-facilitative transfer efects, present in roughly 92.5%
of the studies reviewed, irrespective of methodology used, timing of testing in the developmental
continuum, or language combination considered. The systematic review also revealed that typologi-
cally motivated transfer had the best coverage of the data—roughly 61% of results, more than twice
the coverage of any other macro-variable. For much of the remaining 39%, the review revealed
methodological factors that might have afected the outcomes or compromise their interpretation,
for example, testing participants at later stages of L3/Ln acquisition, in which earlier transfer efects
might be obscured by the natural course of L3/Ln development. Nevertheless, this review revealed
that available data are, at present, inconclusive with respect to determining what (set of) variable(s)
predict(s) transfer selectivity in adult multilingualism.
The overwhelming majority of work in this line of inquiry has been done with ofine measures,
rarely probing processing directly (but see González Alonso et al., 2020; Lago et al., 2021; Rothman
et al., 2015). Very recently, some studies using processing-based methods have begun to appear. For
example, Lago et al. (2019) used speeded judgments and self-paced reading to examine the processing
of possessive pronouns in L3 German, by late bilingual speakers of Spanish and English. In particular,
they were interested in the potential interference produced during the computation of gender agree-
ment between the subject and a possessive pronoun, when the possessee noun had a diferent gender
(e.g., Mr. Smith called his mother, a mismatch, vs. Ms. Smith called her mother, a match). Since gender
agreement with the possessee is not encoded in Spanish possessives, this phenomenon applies only
to German and English. Comparing mirror-image groups (see Textbox 5.1) of adult English–Span-
ish bilinguals (L1 English-to-L2 Spanish and L1 Spanish-to-L2 English), Lago et al. found that L1
English speakers were overall more sensitive to violations of possessor agreement in both measures.
Additionally, the speeded judgment data provided some evidence of an increased sensitivity in L1
Spanish speakers as a function of L2 profciency. Counter to the predictions of all the transfer models
discussed previously, Lago et al. (2019) argue that the native language seemed to infuence each group
more, although knowledge of an L2 seems to also have an efect on reading comprehension.
To our knowledge, there are exceedingly few studies that have used EEG/ERP in the L3 morpho-
syntax context, although a position paper detailing the benefts of adopting this method along with
a fully articulated methodology was ofered by Rothman et al. (2015). Grey et al. (2018) studied the
L3 learning of artifcial/mini-grammars (hereafter AL, for “artifcial languages”), although it should
be noted that they did not focus on the question of transfer selectivity per se. They examined the
behavioral and neural correlates of learning an AL in early Mandarin–English bilinguals, compared to
English monolinguals as a control. After grammar instruction, participants were tested by means of a
grammaticality judgment task at low and high profciency while EEG was recorded. Most interesting
to the present context are the results at low profciency, in which bilinguals and monolinguals did
not difer on behavioral measures, but showed distinct ERP patterns. Only the bilinguals showed a
P600, a common ERP correlate of combinatorial processing, for sentences that were ungrammatical
in the AL. Grey et al. (2018) interpret this particular fnding as suggestive of diferent trajectories in
L2 and L3 processing, whereby bilinguals may more rapidly adopt native-like processing mechanisms
when engaged in further language learning. An alternative explanation, which puts more weight on
cognitive than linguistic factors, is that the bilinguals’ experience with managing interference from
coactivated languages may allow them to engage earlier and/or more efciently in the processes of
syntactic reanalysis that are thought to underlie the P600 (e.g., Bartolotti & Marian, 2012).
González Alonso et al. (2020) constructed two ALs that were lexically based on English and Span-
ish. Each of these mini-grammars was the target L3 for a separate group of Spanish native L2 English
56
speakers. The domain of grammar in focus was grammatical gender in the L3. Both ALs displayed
gender agreement (which only Spanish has) between determiners, nouns, and adjectives. Agreement
morphemes, as well as the article, were common to both ALs, and equally phonotactically plausible
in both English and Spanish. After receiving implicit training on grammatical sentences only in one
of the ALs, participants were tested behaviorally to a criterion of 80% accuracy in agreement. Once
this threshold was reached, participants’ EEG activity was recorded while they read sentences with
gender agreement errors in a violation paradigm with RSVP.
González Alonso et al. (2020) found a broadly distributed positivity between 300 and 600 ms only for
those speakers exposed to Mini-Spanish. They proposed that this may be an instance of the P300 com-
ponent, typically elicited in connection with the detection and categorization of task-relevant deviant
stimuli. Thus, while both groups were L1 Spanish–L2 English, their ERP responses to gender violations
refect diferential allocation of attentional resources as a consequence of exposure to diferent ALs. They
speculated that typological proximity between L1 Spanish (a morphologically rich language) and Mini-
Spanish seems to focus attention on the morphology shortly after initial exposure, while this is not the
case when the same speaker population is exposed to Mini-English. While this is not transfer per se, they
argue that these data show a stage prior to transfer selection indicating a role of typology in drawing the
parser’s attention to specifc cues that later, they hypothesize, will result in typologically based selection.
The studies reviewed in this section show that L3 processing and acquisition are constrained by
a complex set of interactions between the structural confguration and parsing preferences of previ-
ously acquired languages. Whether these interactions are themselves weighted by L1 or L2 status,
overall or property-specifc similarity between the L3/Ln and each previous language is still an open
empirical question that has so far been addressed mostly through ofine behavioral experiments. The
sparse evidence from online studies suggests that cross-linguistic efects may partially be diferent dur-
ing online comprehension than in ofine studies. Future studies are thus called to weigh in on this
debate with the unique insight that online processing data can provide.

In this chapter, we have introduced and substantiated a unique set of factors that should earn the
study of multilingual acquisition and processing dedicated attention, both in itself and as a window
into larger questions about language and mind more generally. We have discussed two major aspects.
Firstly, the impact of each language added to the system is not always proportional or predictable
from previous instances of acquisition, and interacts with factors such as typological proximity—both
in a genealogical sense, which correlates with overall lexical similarity, and in a strictly structural one.
Secondly, the fact that previous experience in L3/Ln acquisition comprises more than one grammar
brings up issues of selectivity among competing sources of cross-language efects.
Dedicated investigations into L3/Ln acquisition and processing are rapidly becoming an established,
separate line of inquiry from second language acquisition (SLA). The mental processes involved in the
learning and use of more than two languages have been a hot topic in applied linguistics for several
decades now (e.g., Cenoz et al., 2001), but they have rarely been investigated using the psychometric
and brain imaging methodologies currently dominant in psycholinguistics (Van den Noort et al.,
2014; see Schwieter, 2019, for a comprehensive volume gathering some chapters specifcally on multi-
lingualism). Since most debates relevant to the scientifc study of language will revolve around linguis-
tic representation, it is reasonable to expect that the feld soon begin to experience a “processing turn”
similar to that which has taken place in SLA over the past three decades. The motivation for such a
turn is clear, and enjoys notable consensus in the study of L2 acquisition. The simultaneous collection
of real-time and ofine linguistic data strongly suggests that ofine behavioral measures might some-
times, by themselves, under- or misrepresent the linguistic knowledge of nonnative speakers. In this
sense, inquiry into the processing of an L3/Ln will build on the behavioral data collected over the past
57
two decades, to develop fne-grained models of linguistic competence and language trajectories in
sequential multilingualism, and to address the many open questions that remain.
What factors determine the source of infuence from previously acquired languages on the L3/Ln, at the
level of both mental representation and real-time processing?
What impact does each additional language have on the performance of the multilingual system as a
whole, and on the processing of specifc linguistic domains within each language?
What role does similarity between the languages, at diferent levels of linguistic representation and parsing
preferences, play in determining the magnitude of that impact?
Is L3 acquisition, the frst scenario in which previous linguistic experience consists of more than one lan-
guage, representative of Ln acquisition, or does the L3 learning experience lead to diferent, perhaps
more efcient, transfer dynamics in L4/Ln acquisition?
Note
1 Prior to the CEM, there were few examples of studies regarding transfer efects in L3/Ln grammar (e.g. Klein,
1995; Leung, 1998, 2003). Research questions largely revolved around using the context of L3 acquisition
as a new testing ground for existing generative L2 theories about the quality and extent of transfer and the
accessibility to Universal Grammar.
Further reading
Festman, J. (2021), Learning and Processing Multiple Languages: The More the Easier?. Language Learning,
71(S1), 121–162.
Festman (2021) provides an overview of research into the variables that contribute to making the learning
curve steeper or gentler for each additional language in sequential multilingualism.
Frank, S. L. (2021), Toward Computational Models of Multilingual Sentence Processing. Language Learning,
71(S1), 193–218.
Frank (2021) calls for an implementation of multilingual (sentence) processing theories into computational
models, arguing that the dynamics of multilingual processing can equally emerge from current computational
architectures (i.e., without a specifc multilingual design) given the right input conditions.
Lago, S., Mosca, M. and Stutter Garcia, A. (2021), The Role of Crosslinguistic Infuence in Multilingual Pro-
cessing: Lexicon Versus Syntax. Language Learning, 71(S1), 163–192.
Lago et al. (2021) provide a comprehensive review of cross-language efects in L3/Ln, focusing specifcally
on language processing.
Rothman, J., González Alonso, J., & Puig-Mayenco, E. (2019). Third language acquisition and linguistic transfer.
Cambridge University Press.
Finally, Rothman et al. (2019) constitutes an in-depth introduction to the feld of L3/Ln acquisition, with a
focus on the role of linguistic transfer in the genesis and development of an L3/Ln grammar.
References
Abutalebi, J., & Green, D. W. (2007). Bilingual language production: The neurocognition of language represen-
tation and control. Journal of Neurolinguistics, 20, 242–275.
Bardel, C., & Falk, Y. (2007). The role of the second language in third language acquisition: The case of Ger-
manic syntax. Second Language Research, 23(4), 459–484.
58
Bardel, C., & Falk, Y. (2012). Behind the L2 status factor: A neurolinguistic framework for L3 research. In
J. Cabrelli Amaro, S. Flynn, & J. Rothman (Eds.), Third language acquisition in adulthood (pp. 61–78). John
Benjamins.
Bartolotti, J., & Marian, V. (2012). Language learning and control in monolinguals and bilinguals. Cognitive Sci-
ence, 36, 1129–1147.
Basnight-Brown, D. M., & Altarriba, J. (2007). Diferences in semantic and translation priming across languages:
The role of language direction and language dominance. Memory & Cognition, 35, 953–965.
Caramazza, A., & Brones, I. (1979). Lexical access in bilinguals. Bulletin of the Psychonomic Society, 13(4), 212–214.
Cattell, J. M. (1887). Experiments on the association of ideas. Mind, 12, 68–74.
Cedden, G., & Aydın, Ö. (2019). Do non-native languages have an efect on word order processing in frst lan-
guage Turkish? International Journal of Bilingualism, 23(4), 804–816.
Cenoz, J. (2001). The efect of linguistic distance, L2 status and age on cross-linguistic infuence in third lan-
guage acquisition. In J. Cenoz, B. Hufeisen, & U. Jessner (Eds.), Cross-linguistic infuence in third language
acquisition: Psycholinguistic perspectives (pp. 8–20). Multilingual Matters.
Cenoz, J., Hufeisen, B., & Jessner, U. (Eds.), (2001). Cross-linguistic infuence in third language acquisition: Psycholin-
guistic perspectives. Multilingual Matters.
De Bot, K., & Jaensch, C. (2015). What is special about L3 processing? Bilingualism: Language and Cognition,
18(2), 130–144.
Dijkstra, T. (2003). Lexical processing in bilinguals and multilinguals: The word selection problem. In J. Cenoz,
B. Hufeisen, & U. Jessner (Eds.), The multilingual lexicon (pp. 11–26). Kluwer Academic.
Duñabeitia, J. A., Dimitropoulou, M., Uribe-Etxebarria, O., Laka, I., & Carreiras, M. (2010). Electrophysi-
ological correlates of the masked translation priming efect with highly profcient simultaneous bilinguals.
Brain Research, 1359, 142–154.
Falk, Y., & Bardel, C. (2011). Object pronouns in German L3 syntax: Evidence for the L2 status factor. Second
Falk, Y., Lindqvist, C., & Bardel, C. (2015). The role of L1 explicit metalinguistic knowledge in L3 oral produc-
tion at the initial state. Bilingualism: Language and Cognition, 18(2), 227–235.
Festman, J. (2021). Learning and Processing Multiple Languages: The More the Easier?. Language Learning,
71(S1), 121–162.
Flynn, S., Foley, C., & Vinnitskaya, I. (2004). The Cumulative-Enhancement model for language acquisition:
Comparing adults’ and children’s patterns of development in frst, second and third language acquisition.
International Journal of Multilingualism, 1(1), 3–16.
Frank, S.L. (2021). Toward Computational Models of Multilingual Sentence Processing. Language Learning,
71(S1), 193–218.
Gollan, T. H., Montoya, R. I., Fennema-Notestine, C., & Morris, S. K. (2005). Bilingualism afects picture
naming but not picture classifcation. Memory & Cognition, 33(7), 1220–1234.
Gollan, T. H., Slattery, T. J., Goldenberg, D., Van Assche, E., Duyck, W., & Rayner, K. (2011). Frequency drives
lexical access in reading but not in speaking: The frequency-lag hypothesis. Journal of Experimental Psychology:
General, 140(2), 186–209.
González Alonso, J. (2012). Assessing multilingual lexical incorporation hypotheses through a primed picture-
naming task. Linguistic Approaches to Bilingualism, 2(1), 91–107.
González Alonso, J., Alemán Bañón, J., DeLuca, V., Miller, D., Pereira Soares, S. M., Puig-Mayenco, E., Slaats,
S., & Rothman, J. (2020). Event related potentials at initial exposure in third language acquisition: Implica-
tions from an artifcial mini-grammar study. Journal of Neurolinguistics, 56, 100939.
González Alonso, J., & Rothman, J. (2017). Coming of age in L3 initial stages transfer models: Deriving devel-
opmental predictions and looking towards the future. International Journal of Bilingualism, 21(6), 683–697.
Grey, S., Sanz, C., Morgan-Short, K., & Ullman, M. (2018). Bilingual and monolingual adults learning an
additional language: ERPs reveal diferences in syntactic processing. Bilingualism: Language and Cognition,
21(5), 970–994.
Hall, C. J., & Ecke, P. (2003). Parasitism as a default mechanism in L3 vocabulary acquisition. In J. Cenoz, B.
Hufeisen, & U. Jessner (Eds.), The multilingual lexicon (pp. 71–85). Kluwer.
Hammarberg, B. (2001). Roles of L1 and L2 in L3 production and acquisition. In J. Cenoz, B. Hufeisen, &
U. Jessner (Eds.), Cross-linguistic infuence in third language acquisition: Psycholinguistic perspectives (pp. 21–41).
Multilingual Matters.
Hartsuiker, R. J., Pickering, M. J., & Veltkamp, E. (2004). Is syntax separate or shared between languages? Cross-
linguistic syntactic priming in Spanish-English bilinguals. Psychological Science, 15, 409–414.
Ivanova, I., & Costa, A. (2008). Does bilingualism hamper lexical access in speech production? Acta Psychologica,
127(2), 277–288.
Klein, E. C. (1995). Second vs. third language acquisition: Is there a diference? Language Learning, 45(2), 419–465.
59
Lago, S., Mosca, M. and Stutter Garcia, A. (2021). The Role of Crosslinguistic Infuence in Multilingual Pro-
cessing: Lexicon Versus Syntax. Language Learning, 71(S1), 163–192.
Lago, S., Stutter Garcia, A., & Felser, C. (2019). The role of native and non-native grammars in the comprehen-
sion of possessive pronouns. Second Language Research, 35(3), 319–349.
Lemhöfer, K., Dijkstra, T., & Michel, M. (2004). Three languages, one ECHO: Cognate efects in trilingual
word recognition. Language and Cognitive Processes, 19(5), 585–611.
Leung, Y. I. (1998). Transfer between interlanguages. In A. Greenhill, M. Hughes, H. Littlefeld, & H. Walsh
(Eds.), Proceedings of the 22nd Boston University conference on language development (pp. 477–487). Cascadilla Press.
Leung, Y. I. (2003). Failed features versus full transfer full access in the acquisition of a third language: Evidence
from tense and agreement. In J. M. Liceras, H. Zobl, & H. Goodluck (Eds.), Proceedings of the 6th generative
approaches to second language acquisition conference (GASLA 2002) (pp. 199–207). Cascadilla Proceedings Project.
Mägiste, E. (1979). The competing language systems of the multilingual: A developmental study of decoding
and encoding processes. Journal of Verbal Learning and Verbal Behavior, 18(1), 79–89.
Mykhaylyk, R., Mitrofanova, N., Rodina, Y., & Westergaard, M. (2015). The linguistic proximity model: The
case of verb-second revisited. In E. Grillo & K. Jepson (Eds.), Proceedings of the 39th Boston University conference
on language development. Cascadilla Proceedings Project.
Paradis, M. (2009). Declarative and procedural determinants of second languages. John Benjamins.
Poarch, G. J., & van Hell, J. G. (2014). Cross-language activation in same-script and diferent-script trilinguals.
International Journal of Bilingualism, 18(6), 693–716.
Puig-Mayenco, E., González Alonso, J., & Rothman, J. (2020). A systematic review of transfer studies in third
language acquisition. Second Language Research, 31(1), 31–64.
Ringbom, H. (1978). The infuence of the mother tongue on the translation of lexical items. Interlanguage Studies
Bulletin—Utrecht, 3(1), 80–101.
Ringbom, H. (1987). The role of the frst language in foreign language learning. Multilingual Matters.
Rothman, J. (2011). L3 syntactic transfer selectivity and typological determinacy: The typological primacy
model. Second Language Research, 27, 107–127.
Rothman, J. (2015). Linguistic and cognitive motivations for the typological primacy model (TPM) of third
language (L3) transfer: Timing of acquisition and profciency considered. Bilingualism: Language and Cogni-
tion, 18(2), 179–190.
Rothman, J., Alemán Bañón, J., & González Alonso, J. (2015). Neurolinguistic measures of typological efects in
multilingual transfer: Introducing an ERP methodology. Frontiers in Psychology, 6(1087), 1–14.
Rothman, J., González Alonso, J., & Puig-Mayenco, E. (2019). Third language acquisition and linguistic transfer.
Schwartz, B. D., & Sprouse, R. A. (1996). L2 cognitive states and the full transfer/full access model. Second
Schwieter, J. W. (Ed.). (2019). The handbook of the neuroscience of multilingualism. Wiley-Blackwell.
Singleton, D. (1987). Mother and other tongue infuence on learner French. Studies in Second Language Acquisi-
tion, 9(3), 327–345.
Sullivan, M., Poarch, G., & Bialystok, E. (2018). Why is lexical retrieval slower for bilinguals? Evidence from
picture naming. Bilingualism: Language and Cognition, 21(3), 479–488.
Szubko-Sitarek, W. (2011). Cognate facilitation efects in trilingual word recognition. Studies in Second Language
Learning and Teaching, 1(2), 189–208.
Van den Noort, M., Struys, E., Kim, K., Bosch, P., Mondt, K., van Kralingen, R., Lee, M., & van de Craen, P.
(2014). Multilingual processing in the brain. International Journal of Multilingualism, 11(2), 182–201.
Van Hell, J. G., & Dijkstra, T. (2002). Foreign language knowledge can infuence native language performance
in exclusively native contexts. Psychonomic Bulletin & Review, 9(4), 780–789.
Westergaard, M. (2021). Microvariation in multilingual situations: The importance of property-by-property
acquisition. Second Language Research, 37(3), 379–407.
Westergaard, M., Mitrofanova, N., Mykhaylyk, R., & Rodina, Y. (2017). Crosslinguistic infuence in the acqui-
sition of a third language: The linguistic proximity model. International Journal of Bilingualism, 21(6), 666–682.
Williams, S., & Hammarberg, B. (1998). Language switches in L3 production: Implications for a polyglot speak-
ing model. Applied Linguistics, 19(3), 295–333.
Zhao, X., Li, P., Liu, Y., Fang, X., & Shu, H. (2011). Cross-language priming in Chinese-English bilinguals with
diferent second language profciency levels. In L. Carlson, C. Hölscher, & T. Shipley (Eds.), Proceedings of
the 33rd annual conference of the cognitive science society (pp. 801–806). Cognitive Science Society.
60
6
FIRST LANGUAGE ATTRITION
Monika S. Schmid, Concepción Soto,
and Julia Heimann
Introduction
While research on bilingualism has traditionally been biased towards investigations of L1-to-L2 trans-
fer, there is increasing recognition that cross-linguistic infuence (CLI) is bidirectional, impacting
the processing mechanisms underpinning the frst language (L1) as well as the second language (L2)
(e.g., Kroll & Bialystok, 2013; Schmid, 2011). The co-existence and interaction of languages in the
bilingual mind afects the representation, organization, processing, and use of both (or all) of them.
Over the past forty years, L1 attrition—conceived here as any change to the representation, use,
and processing of the native language as a consequence of the acquisition and use of another one—
has been studied across all linguistic levels (for a broad range of linguistic features and in a variety of
contexts and linguistic settings, see Schmid & Köpke, 2019). Cumulatively, the fndings from this
research demonstrate that the native language of late bilinguals, i.e., adult L2 learners,1 is afected by
the development of the second language at all levels and all stages of development, and that bilinguals
may therefore be expected to difer from the monolingual baseline in their L1 as well as in their L2.
One of the fundamental premises of language attrition research has always been that the principles
governing attrition as a developmental process are an integral part of the human language capacity in
the same way as the principles underpinning acquisition. Evidence from this process should therefore
similarly be used to inform theoretical models of the architecture of the linguistic system and psycho-
linguistic models of language development. Psycholinguistic research comparing both the acquisition
and the attrition of linguistic systems or features therefore has the potential to reveal processes which
allow unique insights into the underlying nature of developmental processes (Hopp & Schmid, 2013;
Schmid, 2014).
However, theoretical and psycholinguistic models of bilingual language development have been
slow to adapt to the recognition that transfer can be both forward (from an established to a newly
developing system) and backward, and continue to be based mainly, if not exclusively, on observa-
tions drawn from the development of languages acquired after or, at most, at the same time as the L1
(Schmid & Köpke, 2017). Similarly, outside investigations of language pathology, “language develop-
ment” is almost invariably taken to refer to an increase in profciency and fuency, while the progressive
inaccessibility of language knowledge is rarely studied under this umbrella. Despite the growing body
of empirical evidence documenting L1 attrition, the implications of these fndings for theory forma-
tion have often not sufciently been taken into account (Hicks & Domínguez, 2020). In particular,
the fnding that attrition phenomena in late bilinguals tend to be confned to issues of co-activation
DOI: 10.4324/9781003018872-7 61
and processing but rarely, if ever, extend to representation (Schmid & Köpke, 2017) appears to have
led to a reluctance in wide parts of the research community to accept attrition as a real and tangible
process which is part and parcel of bilingual development, not just a set of “glitches” that should thus
inform theoretical approaches on an equal footing to acquisition data (e.g., Meisel, 2017).
First language (L1) attrition: Any change to the processing, production, or understanding of a native lan-
guage as a consequence of the acquisition and use of a diferent language.
Language development: The process by which the profciency a language user has in any language either
increases or decreases
Reversibility of attrition efects: A change in processing, production or understanding of a native language
that more closely aligns an attriter with the monolingual baseline upon re-immersion.
Historical perspectives
Interest in L1 attrition has existed for most of the twentieth century in the form of speculative,
impressionistic, and/or qualitative contributions, the most relevant of which is Haugen’s (1938) essay
“Language and Immigration.” To the best of our knowledge, it contains the frst use of the term
attrition to describe this particular linguistic phenomenon. Based on his extensive observations on
American Norwegian, particularly in the lexical domain, Haugen describes phenomena such as the
loss of L1–L2 lexical distinctions, the decline in L1 lexical accessibility, and cross-linguistic efects in
the mental lexicon. He concludes that the infuence exerted by the L2 on the L1 starts “from the day
of immigration” (p. 12).
Aside from such relatively isolated publications, language attrition as a research feld in its own
right is commonly taken to have been initiated with a UPenn conference on “The Loss of Language
Skills” in 1980 (and the publication of the proceedings in Lambert & Freed, 1982), treating a range
of developmental processes including L1 and L2 attrition, language death, pathological language loss,
etc., alongside some more practical issues related to matters of policy and methodology.
This inaugural conference was followed by a number of scientifc gatherings, edited volumes, and
special issues of journals throughout the 1980s, accompanied by some small-scale case studies attempt-
ing to link the scarce body of literature to preliminary fndings and insights (see Köpke & Schmid,
2004). Throughout this phase, the realization grew that research would need to diferentiate between
the diferent types of phenomena that had previously been lumped together under the cover term
“language loss,” such as processes taking place at the level of society across generations (language shift
and death, creolization, pidginization, etc.), processes that are the outcome of some kind of pathologi-
cal condition (e.g., aphasia, dementia), and non-pathological L1 attrition among long-term migrants.
In contrast to the widely held view that L1 attrition is probably a phenomenon that occurs rarely
and under extreme circumstances, when it comes to second or foreign languages, it is generally
assumed by researchers and laypeople alike that, after even comparatively short periods of disuse
(e.g., the summer vacation), L2 attrition is probably inevitable even in advanced speakers. This view
was the original driving force behind the eforts in the early 1980s to establish attrition as a research
feld: that is, assessing whether the time and efort invested by American college students in foreign
language education was ultimately wasted (Köpke & Schmid, 2004).
However, and to the considerable surprise of the research community, large-scale studies of L2
attrition established quite early on that rumors of the demise of the L2 in the mind of the non-user
62
The psycholinguistics of frst language attrition
were greatly exaggerated (e.g., Bahrick, 1984a, 1984b). While these studies did indeed fnd an initial
dip, largely in the accessibility of vocabulary items, during the frst three to fve years of non-use,
profciency was shown to be extremely stable for up to 50 years after this. This resilience of L2
knowledge acquired in an instructed setting to erosion through non-use has since been confrmed in
a range of investigations (see Mehotcheva & Köpke, 2019, and Schmid, 2022, for overviews).
As language attrition emerged as a more clearly defned and independent research feld, it began to
encompass a wider range of theoretical approaches. In the early phases, attrition had been seen as a phe-
nomenon mainly interesting to theoretical/generative linguistics—in particular in light of its potential to
provide additional insights into the distinction between competence vs. performance and the develop-
ment of grammatical core vs. periphery features (e.g., Sharwood Smith, 1983; Seliger & Vago, 1991), with
the prevailing view being that “once we can establish general principles for both types of change separately,
then we can investigate the systematic relationships between the two” (Sharwood Smith, 1983, p. 49).
However, the later 1980s saw a surge of investigations that adopted a more sociolinguistic perspective and
tried to establish what circumstances would facilitate or delay attrition (e.g., Jaspaert & Kroon, 1989),
while during the 1990s, studies further branched out into psycholinguistic approaches (e.g., de Bot, 1999).
These shifts in theoretical perspective also implied a re-thinking of research methods and designs: Insights
into sociolinguistic and psycholinguistic questions required larger populations and diferent empirical
methods (e.g., surveys, online experiments) than those conducted in the earlier phases of attrition research.
Cumulatively, these investigations thus led to the recognition that, like any other developmental
phenomenon, language attrition takes place at the intersection of (minimally) linguistic, social, and
cognitive/psychological processes and that a comprehensive understanding can only emerge when
these are all taken into consideration.
The scope of attrition

Until recently, it was customary for studies to set inclusion thresholds such as long periods of resi-
dence (typically longer than 10–15 years) and/or advanced to near-native profciency in the L2 (e.g.,
Tsimpli et al., 2004) for their participants. However, the assumption that attrition will only occur
in “extreme circumstances” has more recently been questioned, based on fndings which show that
L2-to-L1 CLI afects even beginning L2 learners with low levels of profciency as a result of the
permanent co-activation of various linguistic systems (for an overview, see Schmid & Köpke, 2017).
In a similar vein, many attempts have been made to separate efects that are indicative of underly-
ing restructuring from more superfcial and temporary instances of CLI. For some researchers, the
line between the two phenomena refects the division between underlying knowledge/representation
on the one hand and online processing efects, which should not be considered attrition, on the other
(e.g., Bardovi-Harlig & Stringer, 2017; Meisel, 2017). This debate reaches back to the early stages of
language attrition research, e.g., Seliger and Vago’s dictum that “efects of performance (accessing,
processing, control) need to be sorted out from those of competence (tacit knowledge)” and that
“it is erosion that reaches the level of competence . . . which allows for interesting claims about and
meaningful insight into the attrition process” (Seliger & Vago, 1991, p. 7).
As Schmid and Köpke (2017) point out, the attempt to separate phenomena of L2-to-L1 transfer
into attrition at the level of representation vs. purely processing-based online phenomena is fraught
with a number of problems. The frst of these relates to methodological and practical issues: empirical
research so far has not been able to conclusively establish that representational changes among late
bilinguals are possible at all: Even the most strongly attritted participants, tested across a large range
of studies, languages and linguistic phenomena, tend to exhibit error rates that compare very favor-
ably with those of near-native adult second language learners (e.g., Montrul, 2008; Schmid, 2014).
63
The very few exceptions either concern individual speakers of typologically very closely related
languages (e.g., Iverson, 2012) or use a methodology that does not fully probe all aspects of the phe-
nomenon (e.g., de Leeuw et al., 2018).
Beyond this failure of empirical research to conclusively establish evidence for representational
erosion, however, Schmid and Köpke (2017) argue that the distinction between attrition at the
level of processing and at the level of representation is neither meaningful nor useful at a more
fundamental level. They propose that all stages of the attrition continuum, whether they “merely”
afect online processing or extend to representational changes, should be viewed as instances of
the same psycholinguistic process, and that attrition thus encompasses “any of the phenomena that
arise in the native language of a sequential bilingual as the consequence of the co-activation of
languages, cross-linguistic transfer or disuse, at any stage of L2 development and use” (Schmid &
Köpke, 2017, p. 2). From the frst occurrence of cross-linguistic co-activation in production or
processing, linguistic features, or items of the two diferent languages inevitably compete with one
another for selection, resulting in the kinds of phenomena that can be observed at the diferent
stages of the attritional process.
Reversibility of attrition
A relevant question in the context of the scope of attrition is how transient or permanent the efect
of CLI might be: Even in the (hypothetical) case of an attriter who has indeed lost a grammatical
phenomenon to the extent that they are unable to use it in a target-like fashion at any given time, it
would then have to be established whether re-exposure in the absence of any conscious relearning
efort might trigger recovery of the target structure at a faster rate than mere incidental learning in
non-instructed learners with no previous experience of the language. If that were shown to be the
case, it would constitute a strong argument against the assumption that erosion had reached the level
of representation.
The efect of re-exposure has only been investigated in a handful of case studies. However, these
studies unanimously fnd a surprisingly quick reversion towards the monolingual target, for example
in the area of pronunciation (Sancier & Fowler, 1997) or the use and processing of null vs. overt
pronouns (Chamorro et al., 2016). One study followed the further development of a participant who
had shown a similar recovery-to-normal pattern in null and overt pronoun use after a three-week
stay in her home country (Bulgaria). It found that she maintained the target-like distribution of these
features during several years of re-immersion in the same L2 context which had previously led to the
observed pre-return attrition. Having experienced such a recovery may thus facilitate language main-
tenance and subsequently protect against attrition (Köpke & Genevska-Hanke, 2018). These fndings
suggest a complex interaction of factors such as length and amount of exposure and re-exposure,
which we are barely beginning to understand, and the picture is complicated further by fndings that
age at migration and re-exposure play an important role (see Flores, 2019 for an overview of returnee
heritage speakers).
Selective vulnerability of linguistic features

Diferent linguistic structures and linguistic levels are diferentially afected by language attrition.
For instance, it is widely accepted that the L1 lexicon is comparatively susceptible to change under
a situation of L2 acquisition and sustained usage. In practice, lexical attrition often means that L1
lexical items become less accessible, leading to phenomena such as a lower productivity on naming
tasks, lower accuracy and slower responses on picture naming tasks, and less sophisticated vocabulary
and more disfuencies in free speech (Schmid & Jarvis, 2014; for overviews, see Jarvis, 2019). More
elusive phenomena, such as subtle shifts in word meaning and the scope and delimitation of semantic
64
felds (Pavlenko, 2004, 2010; Pavlenko & Malt, 2010) or event conceptualization (Bylund & Jarvis,
2011) have also been reported. The underlying cause for such lexical attrition phenomena can be
ascribed to an interaction of a wide range of factors, including the spread of activation through lexical
networks, the interaction of lexical items at various levels of representation, and the accessibility of
items (see Conklin, this volume).
Where the more structural areas of language—phonetics, phonology, and morphosyntax—are
concerned, it has been more difcult to isolate typical attrition phenomena and identify common
trends across diferent contexts of attrition. As in the lexicon, it seems clear that similarity between
languages and features is conducive to attrition, while more clearly distinct phenomena between the
two languages are less susceptible to L2 infuence.
With respect to speech sounds, similarity has long been taken to be one of the main driving fac-
tors in cross-linguistic infuence, both from L1 to L2 and vice versa, as is set out, for example, in the
speech learning model (SLM, e.g., Flege, 1995, 2002). A number of studies have explored L2-to-L1
CLI in phonetics, beginning with Flege’s own seminal study of voice onset time (VOT) in bilingual
speakers of American English and French (Flege, 1987) which showed a tendency for the settings
in both languages to shift in the direction of the other language, modulated by length of exposure
and level of L2 profciency. These fndings on L2-to-L1 CLI for VOT have been replicated numer-
ous times for other languages and populations. Other phonetic factors such as rhoticity (Ulbrich &
Ordin, 2014), tongue position in laterals leading to an alteration in formant values (e.g., de Leeuw
et al., 2013), vowel formants (Bergmann et al., 2016), and suprasegmentals (Mennen, 2004) have
been shown to be similarly susceptible to cross-linguistic infuence. It has also repeatedly been estab-
lished that long-term attriters are often no longer clearly identifed as native speakers (e.g., de Leeuw
et al., 2010; Hopp & Schmid, 2013), although attempts to link changes in specifc articulatory fea-
tures in the speech production of individuals to their perceived native-likeness have not, so far, been
successful (Bergmann et al., 2016).
Where morphosyntactic features are concerned, the two most important principles in terms of
the diferential susceptibility to change and adaptation in a language attrition setting appear to be (a)
cross-linguistic similarity and (b) the nature and characteristics (e.g., complexity, regularity, transpar-
ency) of the feature. Overall, the tendency seems to be that rule-governed infectional processes
(such as grammatical case and gender concord) tend to be relatively stable among late attriters (e.g.,
Bergmann et al., 2015; Schmid, 2002; Stolberg & Münch, 2010) while speakers with younger ages
of onset show substantial erosion even of features typically fully acquired before the speaker becomes
bilingual (e.g., Schmitt, 2004, 2010). In contrast, features that vary pragmatically, such as depending
on information status (null vs. overt pronouns) or aspects related to meaning (for instance, diferen-
tial object marking, tense/aspect/mood) appear to be susceptible to change (e.g., Chamorro et al.,
2016; Montrul, 2002; Tsimpli et al., 2004). Similarly, word order phenomena that are governed by
contextual/pragmatic factors are more prone to L2-to-L1 infuence than purely syntactic phenomena
(as was shown by Perpiñán, 2011 for inversion in Spanish).
Investigations of language attrition thus tend to fnd reliable, although usually not dramatic, indi-
cations of CLI at all linguistic levels. These appear to be linked with factors such as reduced activa-
tion, interference from the L2 and cross-linguistic competition. They mainly afect knowledge that
is (a) sustained by declarative memory in the lexicon, (b) shared across languages (similarity) and (c)
governed by constraints of context and discourse. Knowledge that resides in procedural memory,
difers between languages, and is subject to rule-governed constraints appears to be more resilient.
Extralinguistic factors
Since one of the most consistent fndings on L1 attrition has been that attriting populations tend to
have more variance in their performance on a range of tasks than monolinguals, the question has
65
often been asked what factors will play a role in determining individual levels of attrition or main-
tenance. Factors that have often been investigated in this context include exposure-related variables
such as frequency and context of language use and length of residence (for an overview see Schmid,
2019), attitudes and motivation (Yılmaz, 2019), and age at onset (Bylund, 2019). Schmid (2007)
further points to the importance of language mode and suggests that an important factor in this
equation may be whether bilingual individuals are routinely required to use their L1 in situations in
which it is necessary for them to inhibit the L2 (e.g., in a professional context)—the “intermediate
mode” in terms of Grosjean’s (2001) language mode model. Subsequent studies have provided further
support for an important role of inhibitory efciency in attenuating language attrition phenomena,
in accordance with Green’s behavioral ecology of bilingualism (Green, 2011).
Cumulative evidence from a range of studies suggests a complex picture, in particular not pointing
to any straightforward or unifactorial impact of exposure-related factors, as had often been assumed.
Schmid and Yılmaz (2018) ascribe the lack of a coherent overall picture to the fact that there may be
complex and non-linear interactions between factor sets unable to be captured by the statistical mod-
els usually applied in such studies. In particular, they propose that a high level of L1 exposure may
be necessary to allow maintenance of L1 profciency at or near monolingual levels for speakers with
lower levels of language aptitude, while high-aptitude migrants may be able to sustain L1 profciency
in the absence of such exposure.
The only factor that has consistently been shown to have a strong and overriding impact on L1
attrition appears to be age at onset, with pre-puberty attriters consistently having lower levels of
L1 profciency than older migrants. After puberty, age efects no longer appear to be relevant for L1
maintenance (Schmid, 2012; Bylund, 2019).

Language attrition is an interesting testing ground for a range of theoretical approaches to bilingual
development, since it allows a diferent perspective on questions about bilingual development, relat-
ing to factors such as the role of age at onset, cross-linguistic similarity and competition, and rep-
resentation vs. processing, than is possible in studies focusing only on L2 acquisition (e.g., Hopp &
Schmid, 2013; Schmid, 2014). Both L1 attriters and L2 learners have to contend with a number of
processing hurdles absent in monolinguals, such as resisting intrusions or automatisms from the non-
selected language, less input in either language, weaker representations and lower resting activation of
linguistic features and lexemes. It is currently assumed that both attrition and acquisition are expres-
sions of the human capacity for language development and hence should be guided and constrained
by the same principles. Over the past decade, this has led to the increasing recognition that evidence
from language attrition should feed into psycholinguistic models of bilingualism and be used to vali-
date them in the same way as observations based on L2 acquisition.
The neurolinguistic perspective: activation and inhibition

There are a number of accounts of language development and attrition which place their explana-
tory focus mainly on the neurolinguistic mechanisms that underpin and facilitate language process-
ing. The most prominent is Paradis’ (2004, 2007) neurolinguistic theory of bilingualism, which is
most often cited in the context of language attrition in terms of the activation threshold hypothesis.
Paradis proposes an account of bilingual development that includes both acquisition and deteriora-
tion, and in particular also draws on his studies of language decline and recovery under pathological
conditions (such as aphasia). He proposes that the phenomena that can be witnessed in all of these
developmental processes can be unifed through an understanding of how the activation and main-
tenance of knowledge takes place in the brain: In order to retrieve any information from memory,
66
efort is needed in the form of electrical impulses. The amount of efort required (i.e., the activation
threshold) is a function of the frequency and recency with which the information has previously
been called upon, modulated by the type of memory (declarative vs. procedural) which subserves
the information and by other factors, such as afective ones. For attrition, these assumptions make a
number of predictions, including that language disuse causes loss, in particular to less frequent ele-
ments of the language and elements that are sustained by declarative memory (lexicon) rather than
procedural memory (phonology, morphosyntax), and that the rate of attrition will be contingent on
factors such as motivation (Paradis, 2007, p. 121f).
Usage-based approaches and the unifed competition model

Usage-based approaches see language as an emergent and dynamic system whose development is
mainly driven factors such as frequency, saliency, and markedness of forms in the input. The unifed
competition model (UCM, MacWhinney, 2015, 2018) attempts to provide an account that brings
together these factors. Following the UCM, knowledge is based upon cognitive associations (“map-
pings”), which become strengthened (“entrenched”) or weakened (“disentrenched”) as a function of
the frequency with which they are encountered or used. Sufciently entrenched patterns receive a
“unit status,” that is, they become integrated wholes and exist purely in dynamic and recurring pat-
terns of neurological activity. In theory, as frst language attrition occurs when frequency of use of the
L1 decreases and the use of similar, competing L2 structures increases, L1 units should be amenable
to disentrenchment. As such, disentrenchment (and in this sense, attrition) could be directly opera-
tionalized as a function of frequency of language use (Steinkrauss & Schmid, 2016).
The UCM is very efective in explaining a range of developmental phenomena. For instance, the
diminished susceptibility to attrition with increased age can be ascribed to the fact that the L1 is rep-
resented in cortical maps, which become more stable (and thus more resistant to modifcation) with
increasing age and decreasing brain plasticity. As the monolingual speaker gets older before begin-
ning to learn a second language, it becomes increasingly hard for the L2 to compete with the already
deeply entrenched structures. Furthermore, more experienced learners can achieve L2 encoding
through resonance, that is, linking L2 knowledge into existing L1 knowledge. Resonance is built
up through “repeated coactivation of reciprocal connections” (MacWhinney, 2005, p. 61), that is,
through meaningful use of language in context during which form becomes ever more strongly asso-
ciated with context which eventually lead to the language system forming a coherent coactivating
neural circuit. Inhibition of the L1 knowledge in conjunction with the resonant activation of the L2
consequently leads to decoupling of the two types of knowledge (Schmid & Köpke, 2017).
When it comes to the diferential susceptibility of linguistic levels, structures, and features, the
UCM holds that similarity plays a key role. Structures that are sufciently similar cross-linguistically
are more prone to attrition than entirely diferent structures. Again, this efect can be ascribed to
resonance, with more similar features resonating with each other more readily across languages, while
those L2 structures that are more diferent do not fnd a compatible L1 counterpart to link to. In the
same way, structures that are completely identical across the two languages will allow rapid linking to
the L1, so that no cross-linguistic competition, and thus no L1 inhibition, ensues.
However, not all phenomena that have been observed in language attrition are capable of being fully
explained by the UCM. For example, in terms of age efects, the entrenchment account would predict
a stable, gradual, and more or less linear diminishment of susceptibility to attrition from early child-
hood all the way into and beyond middle age. Findings suggest, however, that susceptibility remains
uniformly high more or less until the onset of puberty (e.g., Schmid, 2012; Bylund, 2019) and then
drops quite suddenly to a much lower level, where it then holds steady (see also Schimke, 2023 [this
volume]). Similarly, the fact that adoptee studies report catastrophic and wholesale loss of L1 knowledge
that is resistant to reactivation or recovery (e.g., Pallier, 2007) also presents a puzzle to the UCM. Under
67
such circumstances, it seems as though, rather than undergoing the process of resonance and decou-
pling, the L1 units become entirely replaced by L2 units in a relatively short time, leaving no recov-
erable trace—a fnding at odds with this model. There is also very little work to date attempting to
disentangle factors such as frequency, similarity, and cue reliability and their impact on the attrition
of individual linguistic features.
Generative approaches: interface hypothesis and feature reassembly hypothesis

The key assumption relating to language attrition from the point of view of generative grammar is
that—beyond aspects of recency and input frequency—linguistic distinctions matter in explaining
diferential attrition. Currently, several approaches to bilingual development have been applied to
attrition. One such approach, the Interface Hypothesis (Sorace, 2011), holds that narrowly gram-
matical structures, which reside primarily within the remit of one linguistic domain (such as syntax
or phonology), should be resilient when it comes to attrition, while those linguistic structures whose
function is sensitive to conditions involving an interface—particularly an external interface with
pragmatics—are more difcult to process for bilinguals and may change in an attrition setting (e.g.,
Sorace, 2011; for a recent overview see Chamorro & Sorace, 2019).
Another framework that has been applied to the analysis of data from language attrition as well
as heritage languages is the Feature Reassembly Hypothesis (henceforth FRH; Lardiere, 2009). This
hypothesis allows a somewhat more fne-grained perspective on developmental processes than the
Interface Hypothesis, because it proceeds on the assumption that during language development,
forms (morphemes) are mapped to underlying (bundles of) features of grammatical function(s) (such
as tense). In L2 development, and presumably also in L1 attrition, the developmental process is com-
plicated if the feature confguration difers between languages, that is, if each language maps diferent
feature bundles to a morpheme. Second language acquisition is seen as an incremental process in
which the learner starts with the assumption that the L2 maps the same features to forms as the L1,
and the learner gradually adjusts this mapping to the L2 mappings based on evidence encountered in
the input. For language attrition, the assumption would then be that the mapping of features in the
native language may shift over time, based on evidence encountered in the L2. Hicks and Domín-
guez (2020) hypothesize that such a process may occur if the “inference engine” used to compile
the grammar in the process of acquisition may be “reinvoked” under certain circumstances. In other
words, the processes and functions used in language acquisition to assemble feature sets in the frst
instance may kick in again under such circumstances, leading to changes in a grammar that had previ-
ously reached what would normally be considered the end-state. Hicks and Domínguez (2020) point
to a range of criteria and circumstances that are necessary in order for mature (or “end-state”) gram-
mars to become subject to modifcation, including sufcient cross-linguistic similarity and signifcant
changes in input/intake conditions.

Over the past decades, language attrition has emerged from a niche status to a point where it is
considered a ubiquitous and natural part of all language development. The potential contribution of
evidence from attriters for validating and refning psycholinguistic models of bilingual development
is becoming more and more recognized, as diferent approaches to and views on the human capacity
for language begin to take it into account.
However, empirical research on bilingual development still tends to adopt a largely categorical
view of diferent populations, such as L2 learners, heritage speakers, and L1 attriters; and to see
development itself in terms of either acquisition or attrition. It is to be hoped that these largely artif-
cial distinctions will be overcome in future: We feel that a broader perspective will help us gain better
68
and more fundamental insights by recognizing that the diferences between populations as well as
between developmental processes in which the language user approaches the idealized target as well
as those in which they move away from it are gradual and quantitative.
More broadly, considering language attrition a natural development of language in the context
of bilingualism allows us to gain additional insight into the nature of the human capacity for lan-
guage. Investigations of development which consider the resilience or vulnerability of linguistic
knowledge once it has been acquired have the potential to make a fundamental contribution to
theoretical models of the developmental process that have so far been based only on observations
about how difcult or easy they are to acquire in the frst place under conditions of cross-linguistic
interaction.
What does the shape of the age of learning/ultimate attainment function seen in L1 attrition, in particular
the discontinuity often found around puberty, tell us about language representation and maturational
constraints?
What is the scope of attrition efects in mature morphosyntactic systems? If it can be confrmed that
attrition is limited mainly or entirely to online/processing phenomena even in the absence of input
and exposure, what insights can we draw from this about the underlying representation of morpho-
syntactic systems?
What is the scope of attrition efects in mature sound systems? If it can be confrmed that attrition is
limited mainly or entirely to phonetic assimilation, what insights can we draw from this about the
underlying representation of phonological systems?
What is the diferential susceptibility of various phonological/phonetic and grammatical features to attri-
tion? What role does cross-linguistic similarity play in this context? What can we learn from this about
the underlying representation of the structure of language?
How reversible are attrition phenomena in diferent populations and what does this tell us about the
representation of language in long-term memory?
Note
1 Where not explicitly indicated otherwise, this chapter considers cross-linguistic transfer among mature late
bilinguals who were monolingual in an L1 until the onset of L2 acquisition after puberty.
Further reading
Montrul, S. A. (2008). Incomplete acquisition in bilingualism: Re-examining the age factor (Vol. 39). John Benjamins.
Montrul reports on a range of diferent studies of various types of Spanish-English and English-Spanish
bilinguals in order to arrive at a better understanding of the diferential susceptibility of language systems to
CLI at diferent points of the lifespan.
Schmid, M. S. (2011). Language attrition. Cambridge University Press.
The only textbook on language attrition to date, this volume contains sections on linguistic factors in lan-
guage attrition and the extralinguistic variables which determine its progression. The second half consists of
a detailed treatment of research methods and tools for analysis, introducing the Language Attrition Test Bat-
tery, a collection of experiments and resources that is also made available on the Language Attrition Website
(https://languageattrition.org).
69
Schmid, M. S., & Köpke, B. (Eds.). (2019). The Oxford handbook of language attrition. Oxford University Press.
This volume documents the state of the art of language attrition research, providing over forty contributions
on theoretical models, psycho- and neurolinguistic aspects, and linguistic and extralinguistic factors, as well
as on L2 attrition and on heritage languages.
References
Bahrick, H. P. (1984a). Fifty years of second language attrition: Implications for programmatic research. The
Modern Language Journal, 68(2), 105–118.
Bahrick, H. P. (1984b). Semantic memory content in permastore: Fifty years of memory for Spanish learned in
school. Journal of Experimental Psychology: General, 113(3), 1–29.
Bardovi-Harlig, K., & Stringer, D. (2017). Beyond steady-state models of ultimate attainment. Linguistic
Approaches to Bilingualism, 7(6), 673–677.
Bergmann, C., Meulman, N., Stowe, L. A., Sprenger, S. A., & Schmid, M. S. (2015). Prolonged L2 immersion
engenders little change in morphosyntactic processing of bilingual natives. Neuroreport, 26(17), 1065–1070.
Bergmann, C., Nota, A., Sprenger, S. A., & Schmid, M. S. (2016). L2 immersion causes non-native-like L1
pronunciation in German attriters. Journal of Phonetics, 58, 71–86.
Bylund, E. (2019). Age efects in language attrition. In M. S. Schmid & B. Köpke (Eds.), The Oxford handbook
of language attrition (pp. 277–287). Oxford University Press.
Bylund, E., & Jarvis, S. (2011). L2 efects on L1 event conceptualization. Bilingualism: Language and Cognition,
14(1), 47–59.
Chamorro, G., & Sorace, A. (2019). The interface hypothesis as a framework for studying L1 attrition. In M.
S. Schmid & B. Köpke (Eds.), The Oxford handbook of language attrition (pp. 25–35). Oxford University Press.
Chamorro, G., Sorace, A., & Sturt, P. (2016). What is the source of L1 attrition? The efect of recent L1 re-
exposure on Spanish speakers under L1 attrition. Bilingualism: Language and Cognition, 19(3), 520–532.
Chamorro, G., Sturt, P., & Sorace, A. (2016). Selectivity in L1 attrition: Diferential object marking in Spanish
near-native speakers of English. Journal of Psycholinguistic Research, 45(3), 697–715.
de Bot, K. (1999). The psycholinguistics of language loss. In G. Extra & L. Verhoeven (Eds.), Bilingualism and
migration (pp. 345–362). De Gruyter.
de Leeuw, E., Mennen, I., & Scobbie, J. M. (2013). Dynamic systems, maturational constraints and L1 phonetic
attrition. International Journal of Bilingualism, 17(6), 683–700.
de Leeuw, E., Schmid, M. S., & Mennen, I. (2010). The efects of contact on native language pronunciation in
an L2 migrant setting. Bilingualism: Language and Cognition, 13(1), 33–40.
de Leeuw, E., Tusha, A., & Schmid, M. S. (2018). Individual phonological attrition in Albanian—English late
bilinguals. Bilingualism: Language and Cognition, 21(2), 278–295.
Flege, J. E. (1987). A critical period for learning to pronounce foreign languages? Applied Linguistics, 8(2),
162–177.
Flege, J. E. (1995). Second language speech learning: Theory, fndings, and problems. In W. Strange (Ed.), Speech
perception and linguistic experience: Issues in cross-language research (pp. 233–277). York Press.
Flege, J. E. (2002). Interactions between native and second-language phonetic systems. In P. Burmeister, T.
Piske, & A. Rohde (Eds.), An integrated view of language development: Papers in honor of Henning Wode (pp.
217–244). Wissenschaftlicher Verlag.
Flores, C. (2019). Language development in bilingual returnees. In M. S. Schmid & B. Köpke (Eds.), The Oxford
handbook of language attrition (pp. 493–502). Oxford University Press.
Green, D. W. (2011). Language control in diferent contexts: The behavioral ecology of bilingual speakers.
Frontiers in Psychology, 2, 103.
Grosjean, F. (2001). The bilingual’s language modes. In J. Nicol (Ed.), One mind, two languages: Bilingual language
processing (pp. 1–22). Blackwell.
Haugen, E. (1938). Language and immigration. Norwegian-American Studies, 10(1), 1–43.
Hicks, G., & Domínguez, L. (2020). A model for L1 grammatical attrition. Second Language Research, 36(2),
143–165.
Hopp, H., & Schmid, M. S. (2013). Perceived foreign accent in frst language attrition and second language
acquisition: The impact of age of acquisition and bilingualism. Applied Psycholinguistics, 34(2), 361–394.
Iverson, M. (2012). Advanced language attrition of Spanish in contact with Brazilian Portuguese [Ph.D. thesis], Uni-
versity of Iowa.
Jarvis, S. (2019). Lexical attrition. In M. S. Schmid & B. Köpke (Eds.), The Oxford handbook of language attrition
(pp. 241–250). Oxford University Press.
70
Jaspaert, K., & Kroon, S. (1989). Social determinants of language loss. Review of Applied Linguistics (ITL), 83(84),
75–98.
Köpke, B., & Genevska-Hanke, D. (2018). First language attrition and dominance: Same same or diferent?
Köpke, B., & Schmid, M. S. (2004). Language attrition: The next phase. In M. S. Schmid, B. Köpke, M. Kei-
jzer, & L. Weilemar (Eds.), First language attrition: Interdisciplinary perspectives on methodological issues (pp. 1–43).
John Benjamins.
Kroll, J. F., & Bialystok, E. (2013). Understanding the consequences of bilingualism for language processing and
cognition. Journal of Cognitive Psychology, 25(5), 497–514.
Lambert, R. D., & Freed, B. (Eds.). (1982). The loss of language skills. Newbury House.
Lardiere, D. (2009). Some thoughts on the contrastive analysis of features in second language acquisition. Second
MacWhinney, B. (2005). A unifed model of language acquisition. In J. F. Kroll & A. M. de Groot (Eds.), Hand-
book of bilingualism: Psycholinguistic approaches (pp. 49–61). Oxford University Press.
MacWhinney, B. (2015). Multidimensional SLA. In S. Eskilde & T. Cadierno (Eds.), Usage-based perspectives on
second language learning (pp. 22–45). Oxford University Press.
ziano, & H. Jisa (Eds.), Sources of variation in frst language acquisition. Languages, contexts, and learners (pp.
287–312). John Benjamins.
Mehotcheva, T. H., & Köpke, B. (2019). Introduction to L2 attrition. In M. S. Schmid & B. Köpke (Eds.), The
Oxford handbook of language attrition (pp. 331–348). Oxford University Press.
Meisel, J. M. (2017). On frst language attrition in second language learners. Linguistic Approaches to Bilingualism,
7(6), 734–738.
Mennen, I. (2004). Bi-directional interference in the intonation of Dutch speakers of Greek. Journal of Phonetics,
32(4), 543–563.
Montrul, S. A. (2002). Incomplete acquisition and attrition of Spanish tense/aspect distinctions in adult bilin-
guals. Bilingualism: Language and Cognition, 5(1), 39–68.
Montrul, S. A. (2008). Incomplete acquisition in bilingualism: Re-examining the age factor (Vol. 39). John Benjamins
Publishing.
Pallier, C. (2007). Critical periods in language acquisition and language attrition. Language Attrition: Theoretical
Perspectives, 33, 155–168.
Paradis, M. (2004). A neurolinguistic theory of bilingualism (Vol. 18). John Benjamins Publishing.
Paradis, M. (2007). L1 attrition features predicted by a neurolinguistic theory of bilingualism. In B. Köpke,
M. S. Schmid, M. Keijzer, & S. Dostert (Eds.), Language attrition: Theoretical perspectives (pp. 121–133). John
Benjamins.
Pavlenko, A. (2004). L2 infuence and L1 attrition in adult bilingualism. In M. S. Schmid, B. Köpke, M. Kei-
jzer, & L. Weilemar (Eds.), First language attrition: Interdisciplinary perspectives on methodological issues (pp. 47–59).
John Benjamins.
Pavlenko, A. (2010). Verbs of motion in L1 Russian of Russian—English bilinguals. Bilingualism: Language and
Cognition, 13(1), 49–62.
Pavlenko, A., & Malt, B. C. (2010). Kitchen Russian: Cross-linguistic diferences and frst-language object nam-
ing by Russian—English bilinguals. Bilingualism: Language and Cognition, 14(1), 19–45.
Perpiñán, S. (2011). Optionality in bilingual native grammars. Language, Interaction, Acquisition, 2(2), 312–341.
Sancier, M. L., & Fowler, C. A. (1997). Gestural drift in a bilingual speaker of Brazilian Portuguese and English.
Journal of Phonetics, 25(4), 421–436.
Schimke, S. (2023). The psycholinguistics of child second language acquisition. In A. Godfroid & H. Hopp
Schmid, M. S. (2002). First language attrition, use and maintenance: The case of German Jews in Anglophone countries
(Vol. 24). John Benjamins Publishing.
Schmid, M. S. (2007). The role of LI use for LI attrition. Language attrition: Theoretical perspectives (pp. 135–154).
John Benjamins.
Schmid, M. S. (2011). Language attrition. Cambridge University Press.
Schmid, M. S. (2012). The impact of age and exposure on bilingual development in international adoptees and
family migrants: A perspective from Holocaust survivors. Linguistic Approaches to Bilingualism, 2(2), 177–208.
Schmid, M. S. (2014). The debate on maturational constraints in bilingual development: A perspective from frst
language attrition. Language Acquisition, 21(4), 386–410.
Schmid, M. S. (2019). The impact of frequency of use and length of residence on L1 attrition. In M. S.
Schmid & B. Köpke (Eds.), The Oxford handbook of language attrition (pp. 288–303). Oxford University Press.
71
Schmid, M. S. (2022). The fnal frontier? – Why we have been ignoring L2 attrition, and why it is time we
stopped. Language Teaching (First View).
Schmid, M. S., & Jarvis, S. (2014). Lexical access and lexical diversity in frst language attrition. Bilingualism:
Language and Cognition, 17(4), 729–748.
Schmid, M. S., & Köpke, B. (2017). The relevance of frst language attrition to theories of bilingual develop-
ment. Linguistic Approaches to Bilingualism, 7(6), 637–667.
Schmid, M. S., & Köpke, B. (2019). Introduction. In M. S. Schmid & B. Köpke (Eds.), The Oxford handbook of
language attrition (pp. 1–6). Oxford University Press.
Schmid, M. S., & Yılmaz, G. (2018). Predictors of language dominance: An integrated analysis of frst language
attrition and second language acquisition in late bilinguals. Frontiers in Psychology, 9, 1306.
Schmitt, E. (2004). No more reductions!—To the problem of evaluation of attrition data. In M. S. Schmid, B.
Köpke, M. Keijzer, & L. Weilemar (Eds.), First language attrition: Interdisciplinary perspectives on methodological
issues (pp. 299–316). John Benjamins.
Schmitt, E. (2010). When boundaries are crossed: Evaluating language attrition data from two perspectives.
Seliger, H. W., & Vago, R. M. (1991). The study of frst language attrition: An overview. In H. W. Seliger & R.
M. Vago (Eds.), First language attrition (pp. 1–16). Cambridge University Press.
Sharwood Smith, M. (1983). On explaining language loss. In S. Felix & H. Wode (Eds.), Language development
at the crossroads (pp. 49–69). Gunter Narr.
Sorace, A. (2011). Pinning down the concept of “interface” in bilingualism. Linguistic Approaches to Bilingualism,
1(1), 1–33.
Steinkrauss, R., & Schmid, M. S. (2016). Entrenchment and language attrition. In H.-J. Schmid (Ed.), Entrench-
ment and the psychology of language learning: How we reorganize and adapt linguistic knowledge (pp. 367–384).
De Gruyter Mouton.
Stolberg, D., & Münch, A. (2010). “Die Muttersprache vergisst man nicht”—or do you? A case study in L1
attrition and its (partial) reversal. Bilingualism: Language and Cognition, 13(1), 19–31.
Tsimpli, I., Sorace, A., Heycock, C., & Filiaci, F. (2004). First language attrition and syntactic subjects: A study
of Greek and Italian near-native speakers of English. International Journal of Bilingualism, 8(3), 257–277.
Ulbrich, C., & Ordin, M. (2014). Can L2-English infuence L1-German? The case of post-vocalic/r. Journal of
Phonetics, 45, 26–42.
Yılmaz, G. (2019). L1 attrition, L2 development, and integration. In M. S. Schmid & B. Köpke (Eds.), The Oxford
handbook of language attrition (pp. 304–313). Oxford University Press.
72
7
OF HERITAGE LANGUAGE
ACQUISITION
Silvina Montrul
Introduction
Colonization and migration lead to societal bilingualism, which afects the sociopolitical status of
the languages and the way the languages are acquired and eventually mastered in these situations.
Heritage language speakers are early bilinguals who start acquiring both languages before puberty,
yet whose native language or one of their native languages is a minority language in the wider speech
community (Montrul, 2016; Polinsky, 2018).
Heritage speakers exposed to the majority language and the minority language since birth and
before age three years are simultaneous bilinguals, those exposed to the majority language after
age three are sequential bilinguals. Although the boundary between simultaneous and sequential
bilinguals has been delineated about age three (Meisel, 2013), fve to six years old is the most criti-
cal age in heritage language acquisition because it marks school entry in most parts of the world.
Many heritage speakers are monolingual or dominant in the heritage language before age fve, but
language dominance shifts dramatically after that age, especially in the United States (Carreira &
Kagan, 2011) where the majority of the heritage speakers are schooled exclusively in English. In
addition to age of onset of bilingualism, other factors that lead to unbalanced bilingualism and
majority language dominance are the status of the heritage language in the host society, the size of
the speech community, and the vitality of the language beyond the home. In general, profciency
in the heritage language varies greatly across speakers, covering the full spectrum from mere recep-
tive knowledge in illiterate speakers to full fuency in production and comprehension in speech and
writing. Most of the studies of heritage speakers have focused on those with less than advanced
profciency. The language of these speakers exhibits systematic structural diferences from (a) the
language of monolingual and native language dominant bilingual individuals of the same age and
socio-economic status (SES) and language variety, and (b) the language of the parental generation
who are presumably the input providers to the young heritage speakers living in a situation of lan-
guage contact (Polinsky, 2018).
DOI: 10.4324/9781003018872-8 73
Silvina Montrul
Heritage language (also ethnic, minority, community, non-ofcial language): Socio-politically minority lan-
guage acquired as a frst or one of the frst languages. The term heritage language was initially coined
in Canada (Cummings & Danesi, 1990) to refer to languages other than the aboriginal languages of
native and Inuit people and the two ofcial languages of Canada (French and English). The term rap-
idly expanded in the United States to refer to children of immigrants and immigrant children. Despite
the diferent labels used in diferent countries, the terms heritage language and heritage speakers have been
gaining ground in Europe and in other countries as well.
Majority language is the language that has ofcial status and recognition in a territory. It is also the lan-
guage of education, government, and the media.
Minority language: A language with no ofcial status or with less perceived value than another language
spoken within the same territory.
Unbalanced bilingualism: When bilinguals know two languages, but typically use one language more than
the other or are more profcient in one of the two languages.
The study of bilingual speakers of minority languages has been common in sociolinguistic, educa-
tional, and anthropological linguistic traditions (Haugen, 1953; Pfaf, 1991; Valdés, 1995), but inter-
est in heritage languages and their speakers from a psycholinguistic perspective began in earnest in
the mid- and late 1990s (Håkansson, 1995; Polinsky, 1997; Song et al., 1997). This chapter focuses
on studies of language representation, access, and use in heritage speakers and second language learn-
ers as measured by online methodologies.
Two chief factors propelled the study of heritage languages and heritage speakers from a psycho-
linguistic perspective. The frst one is practical and highly consequential: The continuous movement
of people across national borders creates situations of bi/multilingualism. The education of children
whose home language is not the majority language is a challenge for many societies. When the bilingual
children’s home language is not supported during the school age period, heritage language children
risk incomplete command and attrition of aspects of their heritage language as well as broken com-
munication and language transmission in the family. For instance, in the United States, the ideology
that exclusive focus on English is critical for immigrant children to become native-like in English and
assimilated to American culture quickly has had highly detrimental efects for bilingual development.
As adults, many heritage speakers seek to learn their language in the classroom, and are often placed in
second/foreign language classes because there are no other options. Foundational linguistic and psycho-
linguistic research on heritage speakers’ linguistic knowledge has been critical to inform evidence-based
specialized pedagogical practices and efective teaching methodologies for the specifc needs of heritage
language learners, and to promote the creation of special classes and programs (Kagan, 2010).
The second factor is theoretical. Theories of human language representation, processing, and
use are based on monolingual models, as epitomized in Chomsky’s (1965) emphasis on the ideal
speaker-listener who knows language perfectly. Only recently has there been recognition that such
theories of language are not representative of the majority of the world’s population, and that they
must be revised to account for bilingual and multilingual representation and use (Benmamoun et al.,
2013), including signed languages (Lillo-Martin et al., 2016). Similarly, psycholinguistic theories
of input processing and language use in real time, and theories of language acquisition, have also
been biased toward monolingual, middle-class, and even more prominently, English-based models.
74
The psycholinguistics of heritage language acquisition
Heritage languages are critical to expand the research base because they bring novel data from his-
torically diverse and underrepresented groups to theories of language representation, acquisition, and
processing.
The study of heritage languages and heritage speakers has brought a diferent perspective to long-
standing debates in linguistics and psycholinguistics, such as the relative contributions of nature and
nurture in language development, the role of input and exposure in language growth and maintenance,
the wax and wane of linguistic profciency in the frst language in response to input factors, the role of
age of acquisition in frst and second language development, the role of literacy in acquisition, repre-
sentation, and use, types of linguistic knowledge (implicit, explicit), and dissociations in receptive and
productive linguistic abilities in written and auditory modalities. In research on heritage language acqui-
sition, studies of diverse languages have uncovered that diferent areas of language (phonology, morphol-
ogy, syntax, semantics, discourse) are mastered unevenly by child and adult heritage speakers (Montrul,
2008, 2016; Polinsky, 2018). The extension of psycholinguistic theories and experimental tools that
measure language processing to investigate the linguistic knowledge of heritage speakers is just getting
started, but two key issues that have emerged are (1) the methodological issue of the identifcation of
appropriate research methods to best characterize heritage language knowledge, and (2) the empirical
issues of how the linguistic knowledge and use of the heritage language varies from those of mono-
lingually raised native speakers and of adult second language learners. The rest of the chapter addresses
these issues and covers representative research in phonological, morphological, semantic, syntactic, and
discourse processing, in oral production and in online lexical access and sentence comprehension.

Because heritage speakers start as frst langauge (L1) learners but resemble second langauge (L2)
learners in their uneven non-native ultimate attainment when they reach adulthood, their special
language learning situation, which combines early exposure to the heritage language during the
critical period with suboptimal and limited exposure and use of the heritage language during that
same period, provides a testing ground for theories that propose very diferent mechanisms for L1
and L2 learning on the basis of age (nature) and/or experience (nurture). The key questions are (1)
what aspects of language are sensitive to early exposure for eventual full development, (2) which ones
depend more heavily on experience and language use, and (3) which ones require both for native
mastery? These questions cannot be addressed by just comparing L1 and L2 development in SLA
because age of acquisition and amount and quality of input are typically related and confounded in
these two learning situations. However, because these factors are dissociated in heritage speakers,
studying heritage speakers may bring us closer to answering these questions.
Theoretical models of L1, early bilingual, and L2 acquisition can be used to frame, formulate hypoth-
eses about, and understand the psycholinguistic processes of heritage language representation and use
(Montrul, 2008, 2016). For example, the Shallow Structure Hypothesis (Clahsen & Felser, 2006) holds that
there is a fundamental diference in how native speakers and non-native speakers (postpuberty L2 learn-
ers) engage in deep versus shallow processing during morphological and syntactic processing. Although
native speakers pay attention to fne grammatical distinctions, L2 learners are allegedly predominantly
guided by extragrammatical (lexical, pragmatic) information. In morphology, Pinker and Prince (1988)
proposed a dual mechanism model of infection for native language processing. Following this model,
Ullman (2001) and Clahsen et al. (2010) contend that, unlike native speakers, postpuberty L2 learners do
not engage in morphological decomposition due to critical period efects. DeKeyser (2000) also assumes
that the key diference between L1 and L2 learners lies in maturational efects (Johnson & Newport,
1989) and the fact that L1 acquisition is primarily guided by implicit learning whereas L2 acquisition by
adults takes place through explicit learning. Maturational efects on diferent memory systems have been
claimed to have consequences for implicit but not for explicit learning (Paradis, 2009).
75
Silvina Montrul
Phonological perception and production

Studies of phonological perception and production support the view that early onset of acquisition
and childhood experience brings linguistic advantages (i.e., more native-like knowledge) to heritage
speakers relearning their language in adulthood. In their pioneering study of heritage speakers with
receptive knowledge of Spanish, Au et al. (2002) found heritage speakers to be more native-like
than L2 learners of similarly low profciency level in phonological production. Lukyanchenko and
Gor (2011) report that heritage speakers of Russian performed better than L2 learners on conso-
nant discrimination and phonological discrimination in word recognition under noisy conditions,
and those with high profciency were close to native speakers (Gor, 2014). Chang (2016) examined
Korean heritage speakers’ phonological perception of syllable fnal stops in their Korean and English.
Not only were the heritage speakers indistinguishable from the Korean native speakers, but they also
surpassed the perceptual abilities of English native speakers in English. Kim (2019) tested perception
and production of lexical stress in Spanish, which is contrastive with some verb forms (canto “I sing,”
cantó, “she/he sang”). In perception, the heritage speakers did not difer from the native speakers,
while the L2 learners performed at chance. Studies of international adoptees (Pierce et al., 2014;
Singh et al., 2011), whose access to input is completely interrupted, confrm that the perceptual basis
of a discontinued native language is latent in phonological processing later in life.
Although advantages of early language experience on speech production are commonly reported,
the degree of native-likeness in phonological production depends on actual heritage language use.
Oh et al. (2003) compared two groups of Korean heritage speakers (receptive bilinguals whom they
termed overhearers and bilingual speakers who spoke Korean until age fve) and L2 learners of Korean
on perception and production of Korean stops. Their results confrmed perception advantages for
the two heritage language groups compared to the L2 group, and production advantages only for
the heritage speakers who actually spoke some Korean. Chang et al. (2011) investigated the produc-
tion of English and Mandarin back vowels and short voice onset time (VOT) stop consonants in
the two languages. Although the heritage speakers did not produce all categories in the same way as
the native Mandarin speakers, they were more native-like than the L2 learners, supporting the early
exposure advantage hypothesis for phonology. Finally, Kim (2019) found that, in production of stress
in Spanish, the heritage speakers patterned with the L2 learners instead of with the native speakers
as they did in perception. Together, these fndings suggest that early acquisition afects phonological
perception more than production. Both may be related to age efects but production requires actual
language use to achieve mastery. Heritage speakers display an accent which is non-native but still
more native than L2 accent (Kupisch et al., 2014).
Lexical access and morphological processing

Lexical access during comprehension and production is an under-researched area in heritage lan-
guage studies. Compared to fuent native speakers, heritage speakers’ lexicons are smaller, underde-
veloped in some semantic areas, and slower to access (Hulsen, 2000).
Studies of lexical processing involve receptive online tasks of lexical decision (with or without
priming) and translation judgment, and production tasks with naming. There are many word-related
variables that contribute to how quickly and accurately words are identifed and produced in a frst,
second, or less dominant language, including concreteness, imageability, frequency, length, morpho-
logical complexity, semantic relatedness, phonological relatedness, cognate status, and the age at which
words are frst learned (age of acquisition or AoA; see Conklin & Thul, 2023 [this volume]). In general,
studies of lexical processing emphasize the role of experience rather than age of acquisition because
76
vocabulary grows over time. Hulsen (2000) investigated lexical access and retention of nouns in
three generations of Dutch immigrants in New Zealand in their two languages. An oral picture-
naming task of objects and a picture-word matching task measured accuracy and speed of lexical
access in production and comprehension. Results of both the Dutch and the English experiments
showed main efects for generation, cognate status of words, and frequency. First-generation speak-
ers were less accurate and slower in Dutch than controls from the Netherlands, but only in pro-
duction. Second-generation speakers were signifcantly slower and less accurate in Dutch than in
English in both measures, and much slower and more inaccurate in production than in comprehen-
sion. Finally, third-generation speakers were unable to complete the production task, suggesting that
production is afected more than comprehension with language disuse. Montrul and Foote (2014)
investigated the AoA of nouns, verbs, and adjectives using a lexical decision task and a visual transla-
tion judgment task in heritage speakers and L2 learners of Spanish in the US. AoA of words was a
signifcant factor in speed of lexical access and translation recognition for both groups. The heritage
speakers were faster and more accurate in accessing words learned early in Spanish (their L1) and
early in English (their L2) (pañal “diaper,” perro “dog”), than accessing words typically acquired later
in life (correo “mail”).
An important question in lexical processing is the organization of the mental lexicon and the vari-
ables that afect speed of lexical retrieval and access. The revised hierarchical model (Kroll & Stewart,
1994) and the bilingual interaction activation plus model (Dijkstra & van Heuven, 2002) emphasize
the mediating role of the L1 in L2 lexical access and the role of frequency, respectively. Etxebarria
(2020) extended these models to an investigation of Spanish heritage speakers and examined the
semantic relationship of high-frequency non-cognate words in heritage speakers, L2 learners, and
Spanish native speakers who were L2 learners of English. In two lexical decision tasks with masked
priming, target words were preceded either by a translation equivalent prime (hand—MANO;
mano—HAND) or a semantically unrelated prime (year—MANO; año—HAND). In Experi-
ment 1, primes were presented in Spanish followed by target words in English; in Experiment 2,
English primes preceded Spanish target words. Like native speakers (who knew some English), the
heritage speakers were faster and more accurate when processing Spanish targets with English primes
than the L2 learners of Spanish, yet also patterned with the native speakers of English in the English
experiment. Priming efects were stronger from their dominant to the non-dominant language.
Etxebarria found that profciency, length of exposure, and language use were more predictive than
order and onset of L2 acquisition in the heritage speakers in establishing and enhancing bidirectional
facilitatory translation priming efects.
In terms of the morphological structure of words, a prevalent view has been that whereas L1
speakers engage in morphological decomposition (Pinker & Prince, 1988; Taft & Forster, 1975),
L2 learners store morphologically complex words as one unit (Clahsen et al., 2010; Ullman, 2001).
Kırkıcı and Clahsen (2013) tested L1 and L2 speakers of English and Turkish and found that native
English speakers showed morphological priming, yet L2 speakers showed only partial priming for
derived words and none for infected words. A few recent studies with heritage speakers (Gor &
Cook, 2010; Gor & Vdovina, 2010; Jacob & Kırkıcı, 2016, Mason, 2019) report that, like native
speakers, heritage speakers show evidence of morphological decomposition of regular forms. For
instance, Jacob and Kırkıcı (2016) and Jacob et al. (2019) found signifcant morphological and ortho-
graphic priming in heritage speakers of Turkish. Studies of nonce verb production with monolin-
gually raised native, heritage, and L2 speakers of Russian at diferent profciency levels (Gor & Cook,
2010; Gor & Vdovina, 2010) concluded that all speakers can productively use many patterns, with
the degree of productivity depending on the type frequency of the pattern, productivity, and com-
plexity. Gor and Cook (2010) suggest that the patterns of processing found in the diferent types of
speakers are related to their diferent learning environments and access to written and auditory input.
In three experiments, Mason (2019) examined whether evidence of morphological decomposition
77
Silvina Montrul
of regular and irregular Spanish verbs in native, heritage, and L2 speakers of Spanish varied by kind
of language use, productive or receptive response, and task implicitness, and found that, overall, lan-
guage experience and age of acquisition determine how native, non-native, and heritage speakers
organize and access their mental lexicons and engage in morphological decomposition of complex
words. Therefore, lexical access and extent of morphological decomposition in heritage speakers and
L2 learners is heavily dependent on their specifc experience with the weaker language in terms of
amount of exposure and type of exposure.
Morphosyntactic processing
Infectional morphology is an area of great vulnerability in adult L2 acquisition, and it is similarly
prone to simplifcation and erosion in heritage languages (Montrul, 2018; Polinsky, 2018). Even
though infectional morphology (agreement, case, aspect, mood) is acquired early, it is not mastered
or it undergoes attrition through the lifespan in heritage speakers. In psycholinguistic studies, sensi-
tivity to ungrammaticality in morphosyntax is measured by the time it takes to process strings with
morphological or syntactic errors. In general, studies that used online methodologies and take into
account task modality fnd that heritage speakers show native processing patterns in morphosyntax.
Jiang (2007) suggested that L2 learners have persistent diffculties with the automatic access and
use of infectional morphology arguing that they lack underlying integrated knowledge. Foote
(2011) investigated whether heritage speakers and L2 learners of Spanish of advanced and near-native
profciency have integrated knowledge of agreement by examining their sensitivity to number and
gender agreement errors while reading for comprehension in sentences manipulating agreement in
local and long-distance contexts. The results of a self-paced reading task indicated that both heritage
speakers and L2 learners were as sensitive as the native speakers to subject-verb number agreement
and noun-adjective gender agreement violations in Spanish, demonstrating integrated knowledge of
agreement. Intervening materials in the long-distance conditions led to diminished sensitivity to all
groups, including the native speaker control group.
Fuchs et al. (2015) found that native Spanish speakers exhibited agreement attraction with num-
ber but not with gender, treating the two features diferently. Scontras et al. (2018) applied the
same agreement attraction paradigm to diagnose the structure and content of number and gender
agreement morphology in noun phrases in heritage speakers of Spanish with low and intermediate
profciency. Unlike the native speakers tested by Fuchs et al. (2015), the heritage speakers tested by
Scontras et al. perceived errors in number as equally (un)grammatical as errors in both number and
gender, suggesting that low-profciency heritage speakers process gender and number diferently
from baseline speakers. Taken together, the results of Foote (2011) and Scontras et al. (2018) suggest
that increased profciency critically correlates with native-like processing of infectional morphology
in heritage grammars.
In addition to profciency, psycholinguistic studies of gender agreement also highlight the role of
experience on input processing. They report a much higher incidence of native ability in the heritage
speakers than in the L2 learners, even in grammatical areas that are problematic for both. Montrul
et al. (2008) found task and modality efects: Heritage speakers of Spanish had an advantage over L2
learners with gender agreement in oral production compared to performance in written tasks. To
further investigate the role of explicitness of tasks independent of modality, Montrul et al. (2013)
and Montrul et al. (2014) focused on the processing of spoken language. In Montrul et al. (2014)
Spanish native speakers, L2 learners, and heritage speakers of intermediate to advanced profciency
completed three word recognition experiments that varied in the degree of explicitness of the task: a
gender monitoring task, a grammaticality judgment task, and a repetition task with grammatical and
ungrammatical noun phrases in gender congruent and gender incongruent conditions. All groups
demonstrated sensitivity to gender agreement violations in Spanish noun phrases in general, but the
78
heritage speakers displayed better performance in the more implicit task (repetition), while the L2
learners showed the reverse response.
Montrul et al. (2013) examined the interaction of gender marking in nouns with diminutive
formation to address whether type of input contributes to the acquisition of gender. Diminutives are
a hallmark of Child Directed Speech in early language development (Kempe & Brooks, 2001) and a
highly productive morphological mechanism that regularizes gender marking in nouns with a non-
canonical ending. For example, in Spanish, el elefante “the elephant” ends in “e,” which is opaque, but
in its diminutive form (+ito) it becomes el elefantito, and the regular masculine -o ending surfaces after
the diminutive. In an elicited oral production task with simplex and diminutive canonical and non-
canonical ending nouns, the heritage speakers were signifcantly more accurate with non-canonical
ending nouns, whose learning could have been facilitated through exposure to diminutive formation
in early childhood compared to the L2 learners, who are rarely exposed to diminutives in the L2.
This study confrms that early auditory input confers some advantages to heritage speakers over L2
learners with early-acquired aspects of language.
Case marking is another area of signifcant difculty wherein erosion and simplifcation are com-
mon in L2 learners and heritage speakers (Laleko & Polinsky, 2016), and wherein experience and
input modality seem to play a prominent role. Gor et al. (2019) investigated recognition of case-
infected nouns in Russian and their representation in the mental lexicon of native speakers, heri-
tage speakers, and L2 learners of Russian. In two lexical decision tasks, the processing of noun
targets infected with nominative, genitive, or instrumental cases preceded by adjective primes were
examined to test the case type frequency hierarchy. All groups showed a processing advantage for
the citation form, but only native speakers and highly profcient L2 learners were sensitive to the
oblique-case type frequency hierarchy (Gor et al., 2017). The performance of the heritage speakers
was related to a literacy efect, because Russian heritage speakers had had limited formal instruction
and behaved like preliterate children in relying on surface-form similarity-based associations. Overall,
the heritage speakers were signifcantly slower than L2 learners in accessing visually presented targets.
Several studies have reported that heritage speakers, like L2 learners, have difculty with Diferen-
tial Object Marking (DOM), i.e., the overt marking of objects that are semantically or pragmatically
prominent. In Spanish, animate, specifc direct objects are marked with the preposition a (the object
marker), while inanimate objects are unmarked. Montrul (2004) and Montrul and Bowles (2009)
found that heritage speakers omitted DOM in production and judged ungrammatical sentences with
DOM omission to be grammatical. Jegerski (2015) looked at real-time sensitivity of DOM with
Spanish heritage speakers using a self-paced reading task. Participants read sentences with use or
omission of DOM with animate and inanimate objects, and the reading times suggested no sensitivity
to the omission of DOM with animate objects (ungrammatical), confrming representational prob-
lems with DOM. To test the potential dissociation between online and ofine processing of DOM,
Jegerski (2018a) tested ditransitive and transitive verbs with inanimate or animate direct objects in
a self-paced reading task. As in the ofine grammaticality judgment task, reading times revealed a
robust grammaticality efect for the ungrammatical segment in ditransitive sentences and no efect for
DOM with animate direct objects. Arechabaleta (2020) investigated sensitivity to DOM in Spanish
in SVO and VSO sentences using eye-tracking during reading. With SVO sentences, the heritage
speakers were less sensitive to DOM omission than the native speaker group. With VSO sentences,
the heritage speaker showed sensitivity to the ungrammaticality of DOM, like the native speakers.
With diferent word order, case appears to be more critical in identifying the object, and in these
circumstances, heritage speakers seem to process the DOM marker. In a Visual World paradigm study
with auditory stimuli presentation, Jegerski and Sekerina (2020) found that Spanish heritage speakers
could notice DOM on animate objects when the marker a appeared at the beginning of the sentence.
Therefore, the salience of the DOM marker in the sentence plays a role in the heritage speakers’
ability to notice it and process it.
79
Silvina Montrul
Sagarra et al. (2019) further confrmed the importance of salience for morphological processing in
heritage speakers and L2 learners. They examined efects of word order (SVO, OSV, OVS), optional-
ity (obligatory vs. optional contexts), and saliency (bound vs. unbound morphology) in the processing
of DOM in Spanish relative clauses via a self-paced reading task. Both the heritage speakers and the L2
learners were more accurate but slower in subject than object relative clauses, and in OSV than OVS
clauses. Slower RTs in subject RCs were due to the presence of DOM, and in OSV clauses due to a
misanalysis of the clause as subject initial to interpreting OVS as SVO. All participants were also more
accurate and faster with unbound (a la) than bound morphology (al). Overall, the heritage speakers
patterned more with native speakers and were more accurate and faster than L2 learners.
Syntax, semantics, and discourse

Other studies of sentence processing report native patterns in heritage speakers. Montrul (2006)
investigated Spanish heritage speakers’ processing of English and Spanish unaccusative and unerga-
tive verbs, comparing their performance to that of native speakers of both languages in a self-paced
reading task. The heritage speakers were slower than the native speakers overall. However, the rela-
tive diferences which heritage speakers made in processing Spanish and English unaccusatives and
unergatives were similar to that of the native speakers.
Relative clause attachment preferences with coordinated noun phrases have received signifcant
attention in psycholinguistics. Sentences like Alguien disparó contra el criado de la actriz que estaba en el
balcón/“Somebody shot the servant of the actress who was on the balcony,” typically have diferent
interpretations in English and in Spanish. In Spanish, the head of the relative clause is preferentially
taken to be the frst NP (NP1) whereas in English it is the second NP (NP2). Thus, Spanish is con-
sidered a high attachment preference language while English is low attachment (see Hopp, 2023 [this
volume]). Jegerski et al. (2016) found that L2 learners of Spanish preferred low attachment in Spanish
in an ofine comprehension task as in English, whereas heritage speakers and native speakers pre-
ferred high attachment. Jegerski (2018b) also reports that intermediate profciency Spanish heritage
speakers preferred high attachment in ofine comprehension. In a self-paced reading task, Jegerski
et al. (2016) tested high-profciency heritage speakers and found that their ofine comprehension
was better and their RTs faster for the high-attaching conditions, as in Spanish native speakers.
In the area of discourse-pragmatics, Arslan et al. (2015) investigated the processing of evidential-
ity in Turkish heritage speakers using the Visual World eye-tracking paradigm. Turkish encodes the
evidentiality of events with a special sufx on the verb in the past tense. The past tense -DI is used
when somebody witnessed an event frst-hand; the past tense form -mIş is used when the event is
indirectly reported through hearsay. In general, direct evidentials are acquired before indirect evi-
dentials (Aksu-Koç, 1988). In ofine and online measures, heritage speakers with diferent onset of
exposure to Dutch in the Netherlands were less accurate and slower in selecting the target picture,
and looked at it less in the direct evidentiality condition with -DI than the native controls. No dif-
ferences were found among the three groups in the indirect evidentiality condition with -mIş, which
in a way is contrary to what would be predicted if -mIş is acquired after -DI. Arslan et al. attribute
the fndings to restructuring and neutralization of the evidentiality system in Turkish as a heritage
language, in which the direct evidential sufx has become a marker of both types of evidentiality
contexts. In general, except for Arslan et al. (2015), who found diferences between heritage speakers
and native speakers, most other studies suggest native-like syntactic processing in heritage speakers.

This chapter presented a selective overview of psycholinguistic studies of heritage speakers of
diferent languages, some of which also included groups of L2 speakers. Current theoretical
80
models of linguistic representations and processing make predictions for diferences between
native and non-native processing, and heritage speakers straddle the boundaries between native
and non-native grammatical knowledge and processing. In general, heritage speakers are rela-
tively understudied in psycholinguistics, and there are many possible avenues for future research.
We have seen that, in the area of phonological perception and to some extent production, age of
acquisition is a signifcant factor in native-like processing, and heritage speakers show native-like
abilities. Language production, lexical knowledge, and morphological and morphosyntactic pro-
cessing are subject to strong efects of language experience, as determined by degree of language
use, frequency of input and type of input, and especially the modality in which the input was
experienced, including literacy. Although heritage speakers are often not native-like in their lexi-
cal access and in their production and judgment of infectional morphology, they perform closer
to native speakers in tasks involving oral production and auditory processing that tap implicit
knowledge. In areas of syntactic processing, such as relative clause attachment, unaccusativity,
and long-distance dependencies, heritage speakers have also shown native processing, and little
interference or transfer from the dominant language. Processing studies that examine how speak-
ers parse the input and turn it into intake are needed to understand grammatical learning and use
in diferent types of speakers.
In terms of methodology, there is need for further developments to tap the implicit linguistic
knowledge of heritage speakers, both in terms of establishing profciency and language dominance
and going deeper into specifc aspects of grammatical knowledge. While many studies have used
background questionnaires, translation lists or written profciency tests and oral interviews devel-
oped for L2 learners, O’Grady et al. (2009) designed a psycholinguistic assessment of language
strength in bilinguals based on independently established psycholinguistic principles. The HALA
(Hawai’i Assessment of Language Access) battery exploits the fact that the relative strength of two
(or more) languages is refected in the speed with which their words and structure-building routines
are accessed in the course of speech. The HALA instrument consists of three tasks: a body-part
naming task, a nature-image naming task, and a structure-building production task, in which the
items in each task are graded for frequency to capture diferent profciency levels. Accuracy and
response latencies are measured in the two languages of the bilinguals. Because many heritage
speakers lack experience with literacy in their heritage language, the oral nature of this battery is
ideal to evaluate heritage speakers’ linguistic abilities and speed of processing and can be adapted in
any language.
Finally, eye-tracking and neurolinguistic methodologies will be ideal to tap implicit language
knowledge, and thus perhaps more suitable to understand the linguistic knowledge and processing of
heritage speakers, native speakers, and L2 learners.
Beyond elucidating the psycholinguistics of heritage language acquisition, research in heritage
language processing will contribute to evidence-based pedagogical practices and theoretical advance-
ments. Looking ahead, an area that needs further research is lexical knowledge and lexical access in
heritage speakers, especially in comparison with L2 learners. Lexical knowledge relies on experi-
ence, and heritage speakers have smaller lexicons in their heritage language. Lexical knowledge also
interfaces with phonology, morphology, and syntax, which are all areas of grammar that are sensitive
to age efects. Recent studies of gender agreement have documented strong relationships between
lexical storage and access and processing of morphosyntactic dependencies (Montrul et al., 2013;
Hopp, 2017), as well as phonological perception and lexical access (Gor, 2014). Finally, research
on the heritage language population can also contribute to topics of current interest in mainstream
psycholinguistics and L2 acquisition. For instance, heritage language processing during reading in
heritage speakers with diferent degrees of reading ability can inform how literacy contributes to
language acquisition and processing.
81
Silvina Montrul
How is the bilingual lexicon organized in heritage speakers?

How do heritage speakers process auditory and visual input during lexical, morphological, and syntactic
processing?
How does language experience modulate heritage language processing?
Is morphological processing related to the acquisition and representation of infectional morphology in
heritage speakers?
How do heritage language speakers engage in analysis, reanalysis, cue-based memory retrieval, and pre-
diction during processing?
What can online technologies (eye-tracking and brain imaging) reveal about the nature of language pro-
cessing in heritage speakers, compared to native speakers and L2 learners, that other existing methods
have been unable to elucidate?
Further reading
Montrul, S. (2008). Incomplete acquisition in bilingualism: Reexamining the age factor. John Benjamins.
This is the frst treatment of heritage language speakers compared to second language learners from a psy-
cholinguistic perspective.
This is a comprehensive treatment of linguistic and experimental research on heritage language speakers from
a global perspective.
Polinsky, M. (2018). Heritage languages and their speakers. Cambridge University Press.
A recent theoretical and methodological contribution to understanding recent research in heritage languages.
References
Aksu-Koç, A. (1988). Simultaneity in children’s narratives: The development of cohesion in discourse. Studies
in Turkish Linguistics, 55–78.
Arechabaleta, B. (2020). The processing of diferential object marking by heritage speakers of Spanish. In A.
Mardale & S. Montrul (Eds.), The acquisition of diferential object marking (pp. 237–260). John Benjamins.
Arslan, S., Bastiaanse, R., & Felser, C. (2015). Looking at evidence in visual world: Eye-movements reveal how
bilingual and monolingual Turkish speakers process grammatical evidentiality. Frontiers in Psychology, 6, Art. 1387.
Au, T., Knightly, L., Jun, S., & Oh, J. (2002). Overhearing a language during childhood. Psychological Science,
13, 238–243.
Benmamoun, E., Montrul, S., & Polinsky, M. (2013). Heritage languages and their speakers: Opportunities and
challenges for linguistics. Theoretical Linguistics, 46, 129–181.
Carreira, M., & Kagan, O. (2011). The results of the national heritage language survey: Implications for teach-
ing, curriculum design, and professional development. Foreign Language Annals, 4, 40–64.
Chang, C. B. (2016). Bilingual perceptual benefts of experience with a heritage language. Bilingualism: Language
Chang, C. B., Yao, Y., Haynes, E. F., & Rhodes, R. (2011). Production of phonetic and phonological contrast
by heritage speakers of Mandarin. Journal of the Acoustical Society of America, 129(6), 3964–3980.
Chomsky, N. (1965). Aspects of the theory of syntax. The MIT Press.
Clahsen, H., & Felser, C. (2006). Grammatical processing in language learners. Applied Psycholinguistics, 7, 3–42.
Clahsen, H., Felser, C., Neubauer, K., Sato, M., & Silva, R. (2010). Morphological structure in native and non-
native language processing. Language Learning, 60, 21–43.
Conklin, K. & Thul, R. (2023). Word and multiword processing: cognitive approaches. In A. Godfroid & H.
Cummings, J., & Danesi, M. (1990). Heritage languages. The development and denial of Canada’s linguistic resources.
La maîtresse d’école inc.
82
DeKeyser, R. (2000). The robustness of critical period efects in second language acquisition. Studies in Second
Dijkstra, T., & Van Heuven, W. J. (2002). The architecture of the bilingual word recognition system: From
identifcation to decision. Bilingualism: Language and Cognition, 5(3), 175–197.
Etxebarria, E. (2020). Masked translation priming efects in profcient heritage speakers and L2 learners [Unpublished
manuscript], Department of Spanish and Portuguese, University of Illinois at Urbana-Champaign.
Foote, R. (2011). Integrated knowledge of agreement in early and late English—Spanish bilinguals. Applied
Psycholinguistics, 32(1), 187–220.
Fuchs, Z., Polinsky, M., & Scontras, G. (2015). The diferential representation of number and gender in Spanish.
The Linguistic Review, 32(4), 703–737.
Gor, K. (2014). Raspberry, not a car: Context predictability and a phonological advantage in early and late learn-
ers’ processing of speech in noise. Frontiers in Psychology, 5, Art. 1449.
Gor, K., Chrabaszcz, A., & Cook, S. (2017). Processing of native and nonnative infected words: Beyond afx
stripping. Journal of Memory and Language, 93, 315–332.
Gor, K., Chrabaszcz, A., & Cook, S. (2019). A case for agreement: Processing of case infection by early and late
learners. Linguistic Approaches to Bilingualism, 9(1), 6–41.
Gor, K., & Cook, S. (2010). Nonnative processing of verbal morphology: In search of regularity. Language Learn-
ing, 60(1), 88–126.
Gor, K., & Vdovina, T. (2010). Frequency, regularity, and input in second language processing of Russian verb
infection. The Slavic and East European Journal, 7–31.
Håkansson, G. (1995). Syntax and morphology in language attrition. A study of fve bilingual expatriate Swedes.
International Journal of Applied Linguistics, 5, 153–171.
Haugen, E. (1953). The Norwegian language in American, vol. 1: The bilingual community. Harvard University Press.
Hopp, H. (2017). Cross-linguistic lexical and syntactic co-activation in L2 sentence processing. Linguistic
Hopp, H. (2023). Sentence processing: linguistic approaches. In A. Godfroid & H. Hopp (Eds.), The Routledge
handbook of second language acquisition and psycholinguistics (pp. 216–228). Routledge.
Hulsen, M. (2000). Language loss and language processing: Three generations of Dutch migrants in New Zealand [Doc-
toral dissertation], University of Nijmegen, The Netherlands.
Jacob, G., & Kırkıcı, B. (2016). The processing of morphologically complex words in a specifc speaker group:
A masked-priming study with Turkish heritage speakers. The Mental Lexicon, 11(2), 308.
Jacob, G., Safak, D. F., Demir, O., & Kırkıcı, B. (2019). Preserved morphological processing in heritage speakers:
A masked priming study on Turkish. Second Language Research, 35(2), 173–194.
Jegerski, J. (2015). The processing of case in near-native Spanish. Second Language Research, 31(3), 281–307.
Jegerski, J. (2018a). The processing of the object marker a by heritage Spanish speakers. International Journal of
Bilingualism, 22(6), 585–602.
Jegerski, J. (2018b). Sentence processing in Spanish as a heritage language: Relative clause attachment in early
bilinguals. Language Learning, 68(3), 598–634.
Jegerski, J., Keating, G. D., & VanPatten, B. (2016). On-line relative clause attachment strategy in heritage speak-
ers of Spanish. International Journal of Bilingualism, 20(3), 254–268.
Jegerski, J., & Sekerina, I. A. (2020). The processing of input with diferential object marking by heritage Span-
ish speakers. Bilingualism: Language and Cognition, 23(2), 274–282.
Jiang, N. (2007). Selective integration of linguistic knowledge in adult second language learning. Language Learn-
ing, 57, 1–33.
Johnson, J., & Newport, E. (1989). Critical period efects in second language learning. The infuence of matu-
rational state on the acquisition of English as a second language. Cognitive Psychology, 21, 60–99.
Kagan, O. (2010). Russian heritage language speakers in the U.S.: A profle. Russian Language Journal, 60,
213–218.
Kempe, V., & Brooks, P. J. (2001). The role of diminutives in the acquisition of Russian gender: Can elements
of child-directed speech aid in learning morphology? Language Learning, 51(2), 221–256.
Kim, J. Y. (2019). Discrepancy between heritage speakers’ use of suprasegmental cues in the perception and
production of Spanish lexical stress. Bilingualism: Language and Cognition, 1–18.
Kırkıcı, B., & Clahsen, H. (2013). Infection and derivation in native and non-native language processing:
Masked priming experiments on Turkish. Bilingualism: Language and Cognition, 16(4), 776–791.
Kroll, J., & Stewart, E. (1994). Category interference in translation and picture naming: Evidence for asymmet-
ric connections between bilingual memory representations. Journal of Memory and Language, 33(2), 149–174.
Kupisch, T., Barton, D., Hailer, K., Stangen, I., Lein, T., & van de Weijer, J. (2014). Foreign accent in adult
simultaneous bilinguals. Heritage Language Journal, 11(2), 123–150.
83
Silvina Montrul
Laleko, O., & Polinsky, M. (2016). Topic and case marking in heritage speakers and L2 learners of Japanese and
Korean. Linguistic Approaches to Bilingualism, 6(4), 396–439.
Lillo-Martin, D., Müller de Quadros, R., & Chen Pichler, D. (2016). The development of bimodal bilingualism.
Implications for linguistic theory. Linguistic Approaches to Bilingualism, 6, 719–755.
Lukyanchenko, A., & Gor, K. (2011). Perceptual correlates of phonological representations in heritage speak-
ers and L2 learners. In Proceedings of the 35th annual Boston University conference on language development (pp.
414–426). Cascadilla Press.
Mason, S. (2019). The infuence of task type and speaker background on morphological processing in Spanish [Unpublished
Ph.D. dissertation], Department of Spanish and Portuguese, University of Illinois, Urbana-Champaign.
Meisel, J. (2013). Sensitive phases in successive bilingual acquisition. The critical period hypothesis revisited. In
C. Boeckx & K. Gromann (Eds.), The Cambridge handbook of biolinguistics (pp. 69–85). Cambridge University
Press.
Montrul, S. (2004). Subject and object expression in Spanish heritage speakers: A case of morpho-syntactic
convergence. Bilingualism: Language and Cognition, 7, 125–142.
Montrul, S. (2006). On the bilingual competence of Spanish heritage speakers: Syntax, lexical-semantics and
processing. International Journal of Bilingualism, 10, 37–69.
Montrul, S. (2008). Incomplete acquisition in bilingualism: Reexamining the age factor. John Benjamins.
Montrul, S. (2018). The Bottleneck Hypothesis and heritage language acquisition. In J. Cho, T. Judy, & T. Leal
Mendez (Eds.), Meaning and structure in second language acquisition (pp. 149–177). John Benjamins.
Montrul, S., & Bowles, M. (2009). Back to basics: Diferential object marking under incomplete acquisition in
Spanish heritage speakers. Bilingualism: Language and Cognition, 12(3), 363–383.
Montrul, S., Davidson, J., de la Fuente, I., & Foote, R. (2014). Early language experience facilitates gender
agreement processing in Spanish heritage speakers. Bilingualism: Language and Cognition, 17(1), 118–138.
Montrul, S., de la Fuente, I., Davidson, J., & Foote, R. (2013). The role of experience in the acquisition and
production of diminutives and gender in Spanish: Evidence from L2 learners and heritage speakers. Second
Montrul, S., & Foote, R. (2014). Age of acquisition interactions in bilingual lexical access: A study of the weaker
language of L2 learners and heritage speakers. The International Journal of Bilingualism, 18(3), 274–303.
Montrul, S., Foote, R., & Perpiñán, S. (2008). Gender agreement in adult second language learners and Spanish
heritage speakers: The efects of age and context of acquisition. Language Learning, 58(3), 503–553.
O’Grady, W., Schafer, A. J., Perla, J., Lee, O. S., & Wieting, J. (2009). A psycholinguistic tool for the assessment
of language loss: The HALA project. Language Documentation & Conservation, 3(1).
Oh, J. S., Jun, S. A., Knightly, L. M., & Au, T. K. F. (2003). Holding on to childhood language memory. Cogni-
tion, 86(3), B53–B64.
Paradis, M. (2009). Procedural and declarative determinants of second languages. John Benjamins.
Pfaf, C. (1991). Turkish in contact with German. Language maintenance and loss among immigrant children in
West Berlin. International Journal of the Sociology of Language, 90, 77–129.
Pierce, L. J., Klein, D., Chen, J. K., Delcenserie, A., & Genesee, F. (2014). Mapping the unconscious mainte-
nance of a lost frst language. Proceedings of the National Academy of Sciences, 111(48), 17314–17319.
Pinker, S., & Prince, A. S. (1988). On language and connectionism: Analysis of a parallel distributed processing
model of language acquisition. Cognition, 28, 73–193.
Polinsky, M. (1997). American Russian. Language loss meets language acquisition. In Proceedings of the annual
workshop on formal approaches to slavic linguistics (pp. 370–406). Slavic Publications.
Polinsky, M. (2018). Heritage languages and their speakers. Cambridge University Press.
Sagarra, N., Sánchez, L., & Bel, A. (2019). Processing DOM in relative clauses: Salience and optionality in early
and late bilinguals. Linguistic Approaches to Bilingualism, 9(1), 120–160.
Scontras, G., Polinsky, M., & Fuchs, Z. (2018). In support of representational economy: Agreement in heritage
Spanish. Glossa, 3(1).
Singh, L., Liederman, J., Mierzejewski, R., & Barnes, J. (2011). Rapid reacquisition of native phoneme contrasts
after disuse: You do not always lose what you do not use. Developmental Science, 14(5), 949–959.
Song, M., O’Grady, W., Cho, S., & Lee, M. (1997). The learning and teaching of Korean in community schools.
In Y.-H. Kim (Ed.), Korean language in America 2 (pp. 111–127). American Association of Teachers of Korean.
Taft, M., & Forster, K. I. (1975). Lexical storage and retrieval of prefxed words. Journal of Verbal Learning and
Verbal Behavior, 14, 638–647.
Ullman, M. (2001). The neural basis of lexicon and grammar in frst and second language, the declarative/pro-
cedural model. Bilingualism: Language and Cognition, 4, 105–122.
Valdés, G. (1995). The teaching of minority languages as foreign languages: Pedagogical and theoretical chal-
lenges. Modern Language Journal, 79(3), 299–328.
84
8
LEARNING A NEW LANGUAGE
IN OLDER AGE
Psycholinguistic and cross-disciplinary approaches
Introduction
Aging can be defned as permanent gradual cellular decay common to most living things. Unlike an
illness, like Alzheimer’s or Parkinson’s disease, which has an onset, aging is defned as a continuum
and represents the accumulation of physical, psychological, and social changes in humans over time.
Aging does not have to be synonymous with reduced capacity: Information processing mechanisms
may slow with age, but knowledge of the world and wisdom expand. However, while development,
including second language development, and aging are synonymous with change, in Western societ-
ies, the frst is seen as positive while the second has negative connotations, and this perception may
afect cognitive performance.
All humans experience multiple aging symptoms, e.g., hair turns grey and wrinkles develop.
At age 30, humans start losing body mass, and muscles lose their strength, so recovery from injury
becomes more difcult. Some changes start early in life. For example, the ability to hear high-
frequency sounds starts disappearing with puberty, but it is not until age 75 that almost half of
humans sufer from hearing loss, afecting their capacity to communicate. Most people over 50 need
corrective lenses as the muscles in the eye weaken, which leads to difculty in reading. And as arter-
ies lose their elasticity, blood fow in the brain decreases. It is not surprising that just as most muscles
change with age, the brain changes too. Parts of the brain involved in complex mental activities grow
smaller, and communication between neurons can be reduced (Sonntag et al., 2007).
Even in healthy older people, changes in the brain afect the cognitive resources available at any
given time to process, store, retrieve, and transform information, pay attention, and process sensory
information, i.e., working memory and information processing mechanisms. It is a common belief,
even among seniors, that older adults do not do as well as younger people on complex memory or
learning tests; however, given enough time, they can. This is in part because older adults develop
compensation strategies and have rich schemata (networks of knowledge) as well as insight on which
to rely (Tomaszewski Farias et al., 2018). Also, there is growing evidence that the brain remains plas-
tic as people age, and that it can compensate for age-related brain changes and health conditions that
afect it; this is known as cognitive reserve, which is subject to individual variation, and is enhanced
by education and by practice with certain cognitive tasks. Bilingualism has been associated with cog-
nitive reserve (Craik & Bialystok, 2006). Importantly, older adults can still learn new things, create
new memories, improve vocabulary, and maintain and develop language skills, including learning
and maintaining a second language, but the range of variation is too important to disregard. In fact,
DOI: 10.4324/9781003018872-9 85
variability across learners both in rate of acquisition and in ultimate attainment is a key diference
between frst and second language development: Explaining such variability is one of SLA research’s
primary goals.
Most work in second language acquisition (SLA) and bilingualism has focused on age, and spe-
cifcally on maturational constraints, to explain variability. Note that it is age of acquisition, and not
aging, that has been under focus, because the key research question in SLA really has been whether
native-like success at language learning is precluded post-puberty (Lenneberg, 1967). Of interest for
us in this chapter is an alternative position (for review, see Birdsong, 1999) according to which age
efects are not categorically bounded but are characterized instead by a linear decline that persists
throughout the lifespan without the salient onset and ofset that accompany any critical period. Also,
by moving away from maturational constraints and loss of neural plasticity as the sole cause of failure
to attain native-like second langauge (L2) profciency, this position allows for other types of age-
related efects, both internal and external to the learner, to add to the equation.
What is generally observed is that very young learners regularly show homogeneous learning out-
comes, comparable to those found for monolinguals. As age of onset increases, scatterplots of gram-
matical attainment data show increasing dispersion that can be explained by diferences in testing task
(e.g., perception vs. production), measure (behavioral vs. brain-based), linguistic domain (phonology vs.
lexicon), and generalizability (regular vs. irregular forms) (Hakuta et al., 2003). Demographic variables
also play a role: years of education, especially access to bilingual education or at least education in the L2,
and length of exposure, which usually but not always correlates with length of residence for acquisition in
naturalistic contexts. Socio-cognitive variables such as motivation and attitude towards the language and
the community that speaks it will vary with age, because older learners have more established, stronger
identities and their learning goals may be more refective of an instrumental, rather than integrative, view
of the L2. And as their goals vary, so will the learning strategies older learners apply to achieve those
goals. Success will also vary depending on domain-general cognitive skills such as working memory and
cognitive control, the bilingual learner’s ability to suppress the frst language (L1) as competing informa-
tion, as well as procedural and declarative memory (see Birdsong, 2018, for review).
Not only the mind but also the brain has been posited to account for increased variability across
the lifespan in learners’ rate of acquisition and ultimate attainment. The last ten years have seen a
growth in interest in neurogenetics and their role in L2 learning. Because procedural learning of
grammar is mediated by genetically encoded dopaminergic reception and transmission, diferences
in genetic profles will result in individual diferences in rate of development and fnal attainment.
Because aging is associated with a decline in dopamine reception and binding, neurogenetics can
also explain the dispersion in L2 outcomes observed with increases in age of onset of L2 learning
(Wong et al., 2017).
Even though for the layperson, cognitive decline is the reason for the perceived lack of efciency
at language learning, the reality of L2 development in older learners is more complex and will vary
greatly depending on learning context, i.e., classroom or naturalistic language learning, as well as
population, including migrants aging in place, non-migrant retirees, and new arrivals. While each
of these internal and external factors—cognitive aging and neurogenetics, language contact, access
to education, motivation—plays a role in explaining individual diferences in language attainment
across the lifespan, it is clear that they do not work in isolation but rather interact. For example, age
of migration will determine years of schooling in the L2, which itself will facilitate access to rich,
frequent input while also promoting the ability to look at language as an object that comes with
literacy; schooling also facilitates integrative motivation. Nature and nurture interactions are also
exemplifed in the relationship between general health, healthier cognitive aging, and educational
levels, since higher levels of education are associated with better general health in adults and thus
healthier cognitive aging: Neither neurogenetics alone, nor any other single factor for that matter,
can fully explain the heterogeneity in cognitive aging and in L2 learning outcomes in older learners.
86
Learning a new language in older age
We need to know much more about how all these factors work to facilitate or hinder second lan-
guage development across the lifespan; to meet this goal, we need an interdisciplinary approach. In
this chapter, we will outline such an approach by frst reviewing key theoretical models for under-
standing the psycholinguistics of older adult learners, especially those that integrate psycholinguistics
with social and afective perspectives. We then turn to an overview of current critical issues and the
principal contributions to the feld to date. We conclude by proposing future directions for research
to take in this area.
Cognitive reserve: What allows some people to cope better than others with brain damage (Craik &
Bialystok, 2006).
Compensation strategies: A set of behaviors aimed at mitigating or adapting to loss (either actual or per-
ceived) (Tomaszewski Farias et al., 2018).
Linear decline: Age efects are not categorically bounded but are characterized instead by a linear decline
that persists throughout the lifespan without the salient onset and ofset that accompany any critical
period (Birdsong, 1999); it has been proposed as an alternative to the Critical Period Hypothesis.
Metalinguistic awareness: Awareness of language as an object that can be manipulated (Sanz, 2012).
Neurogenetics: a branch of genetics dealing with the nervous system and especially with its development
(Wong et al., 2017).
Permastore: Long-term or permanent memory that develops after extensive learning, training, or experi-
ence (Bahrick, 1984).
Stereotype threat: Cognitive performance can be lowered in individuals who belong to groups that soci-
ety judges to be poor at the cognitive ability being tested (Levy & Langer, 1994).

Neurolinguistic approaches to SLA center on diferences and similarities between L1 and L2 access
and representation in the brain (e.g., Paradis, 2004; Ullman, 2004; Bates & MacWhinney, 1982).
While most neurolinguistic theories of L2 do not directly address older adulthood, researchers are
currently exploring links between what we know about cognitive aging in the brain and what we
know about language in the brain, for L1 (e.g., Diaz et al., 2016) and for L2 (Reifegerste, 2020).
Importantly, even though cognitive aging includes declines in some cognitive abilities, at least some
neuroplasticity remains in older age (e.g., Bubbico et al., 2019).
One neurolinguistic approach, the declarative/procedural model (DP model; see Morgan-Short &
Ullman, 2023 [this volume]) hypothesizes that two memory systems that are important for human
learning in general, declarative memory and procedural memory, are also used for language learning
and processing, but with some diferences depending on language status: For both L1 and L2, lexical
and semantic processing rely on declarative memory. However, L1 morphosyntactic processing relies
primarily on procedural memory; for adults learning an L2, morphosyntactic processing initially relies
on declarative memory and can come to occur in a more L1-like manner, using procedural memory,
with higher levels of practice and profciency. These predictions have been tested in studies that compare
the extent to which neurological substrates associated with declarative and procedural memory respec-
tively are activated when performing diferent linguistic tasks in L1 and L2. A recent meta-analysis of
neuroimaging studies supports the DP model’s predictions for adult L2 learners (Tagarelli et al., 2019).
87
Tagarelli et al. (2019) noted that, although participants’ age at testing for studies included in
their meta-analysis ranged from 12 to 68 years, no study’s mean participant age was above 36 years;
in other words, including older adults in neurolinguistic research on L2 has been very much the
exception to the rule to date. Reifegerste (2020) posits that, according to the DP model, older adult
bilinguals’ morphosyntactic processing would likely rely on procedural memory to a greater extent
than is the case for young adult bilinguals, due both to the extended amount of exposure older adults
would have with the L2 and to age-related declines in declarative memory. Importantly, this predic-
tion is centered on older adults who are already profcient in two languages, not novice learners at an
older age. For the latter case, we can extrapolate that language learning in older adulthood, especially
morphosyntactic learning, would be afected by age-related declines in declarative memory (Ullman,
2015; Clahsen & Reifegerste, 2017) but without the compensatory efects of increased experience
with L2 that Reifegerste (2020) hypothesized for older bilinguals. Therefore, as Reifegerste notes in
her conclusion, we do not yet know how the neuro- and psycholinguistic profles of older adults at
early stages of L2 learning would compare to those of children and young adults in early stages of L2
learning, or to older adults’ L2 use following many years of immersion.
Usage-based approaches to SLA contend that adult L2 learners process target-language input using
domain-general cognitive processes (not processes dedicated to language); that is, that “language has
the shape that it does because of the way that it is used, not because of an innate bio-program or
internal mental organ” (Larsen-Freeman, 2012, p. 75). As with neurolinguistic approaches, to the
best of our knowledge, no usage-based model directly addresses the question of processing and learn-
ing L2 in older age, but their hypotheses can be relevant for older adult learners.
Combining neurolinguistic with behavioral factors, the unifed competition model (UCM; Mac-
Whinney, 2017, 2018) builds on the original competition model (Bates & MacWhinney, 1982),
which posited that learners build form-meaning mappings by evaluating diferent cues that are in
competition with each other. For example, for a simple sentence with two noun phrases, cues such
as word order, subject-verb agreement, and noun case endings might indicate which noun phrase has
the semantic role of agent and which has the role of patient. Cue strength is determined by cue avail-
ability (how often the cue is present when needed) and cue reliability (how often using the cue leads
one to the intended interpretation); children and adults difer in their dependence on cue availability
versus reliability, thus the diferences in child versus adult language learning. To this foundation, the
UCM adds four risk and four support factors that infuence cue strength and processing. The frst
risk factor is entrenchment: adult learners have already learned at least one language and their neural
connections are built accordingly. L1 entrenchment can be countered through resonance when learn-
ing L2, by connecting new L2 lexical items and morphosyntactic patterns to the existing L1 net-
works, e.g., mnemonics and spaced retrieval practice. A second risk factor is negative transfer, which
can be limited by decoupling L2 from L1 and strengthening links between L2 items and structures.
Third, adults tend to overanalyze L2 input by learning isolated items instead of extracting patterns
from longer co-occurring phrases, as children learning L1 do. Nevertheless, adults can proceduralize
L2 patterns when cues are “consistent, simple, and reliable” (MacWhinney, 2017, p. 357). Lastly, the
UCM notes that adults tend to be learning L2 in relative social isolation (tending to have stronger
connections to L1 communities, greater fear of making mistakes, and less deftness incorporating
feedback), the efects of which can be reduced by promoting participation in L2 communities and
practices.
Specifc potential diferences between young adult and older adult learners in terms of most of
these factors have not yet been tested, although some hypotheses can be generated based on what we
know about general neural and memory changes due to cognitive aging. Moreover, we also know
that adults tend to experience greater social isolation in older age (e.g., Cox, 2019), so promoting
L2 participation may be particularly key for L2 learning for this age group. For example, older adults
may be willing to communicate but lack the social opportunities to practice the target language
88
if their social networks primarily consist of L1-profcient individuals (Cox, 2019) and that lack of
practice potentially then increases language anxiety and avoidance (Arxer et al., 2017). Then, when
anxious learners do have target-language interactions, L2 input might be less deeply processed and
therefore more easily forgotten. Additionally, MacWhinney (2017) notes that “each of the sup-
port factors requires conscious manipulation of input, metalinguistic refection, good linguistic and
pedagogical analysis, study of L2 materials, and ongoing spoken interactions in the new language”
(p. 358); therefore, older adults may beneft from instruction in refecting on and analyzing their L2.
Complex dynamic systems theory (CDST; e.g., de Bot et al., 2007; Larsen-Freeman, 2012) also
takes a usage-based approach that brings together psycholinguistic and social factors to build a com-
prehensive account of language (L1, L2, and beyond) across the lifespan. In this approach, learners
build mental representations through iterative and adaptive practice, which is shaped by the commu-
nicative environment and the learner’s social background and cognitive abilities. Importantly, social
and cognitive factors are also seen as complex and dynamic: Not only do factors like motivation or
working memory vary over time, so also do their efects on the developing L2. In sum, according
to CDST, “Each individual is unique because he or she has developed his or her physical, afective,
and cognitive self from a diferent starting point and through difering experience and history. Each
individual thus acts as a unique learning context, bringing a diferent set of systems to a learning
event, responding diferently to it, and therefore, learning diferently as a result of participating in it”
(Larsen-Freeman, 2012, p. 78–79).
Applying CDST to the situation of older adult learners, aging might at frst seem less prone to
dynamic change than socio-afective factors such as motivation, but aging is social as well as bio-
logical (de Bot & Makoni, 2005) and the efects of aging on L2 learning and processing will vary
depending on the nature of the communicative task, as CDST predicts. In addition, older adults
come to their L2 with vast amounts of “experience and history” beyond that of young adult learners,
including developed cognitive abilities, accumulated linguistic experiences (potentially in multiple
languages), and afective experiences related to their languages and to learning contexts. All of that
history will interact to infuence the shape of their L2 development and in turn, also create the socio-
cultural environment in which their learning happens. Thus, in this view, multiple types of factors
(e.g., socio-afective, cognitive) should be studied simultaneously, both as factors that infuence L2
learning and as elements that may change as a result of L2 learning.

Critical issues to be addressed for older adult learners center on identifying what learning contexts
(that is, diferent types of instructional conditions in classrooms or laboratory settings) best support
older adults’ processing and learning L2. To some extent, this involves asking the same questions that
we ask of young adult learners while expecting that the answers might not be the same for a mul-
tiplicity of reasons. These questions might include: Is task-essential practice (i.e., when processing
the target form is necessary to complete the task) sufcient for developing targetlike L2 processing?
To what extent does processing L2 semantics and morphosyntax rely on declarative and procedural
memory, respectively? To what extent does social interaction in L2 communities promote targetlike
L2 processing? One reason that we might expect at least some older adults to difer from young adults
is due to biological changes in the brain that in turn afect cognitive abilities, as outlined in the intro-
duction and refected in multiple psycholinguistic theories, such as the DP model and the UCM.
Additionally, activation of stereotype threat may play a role: Studies of older adults show that, in
societies with negative stereotypes about aging and cognition, older adults’ performance on memory
tasks is lower than that of older individuals from societies with more positive stereotypes of aging
(e.g., Levy & Langer, 1994). Thus, cultural context also afects cognitive performance, and since pro-
cessing and learning L2 utilizes domain-general cognitive abilities, psycholinguists should take into
89
account the cultural context of their older adult participants and, where relevant, consider treatments
that reduce the potential activation of stereotype threat. Lastly, older adults may difer from young
adults in their goals and motivations for undertaking language study (e.g., Ramírez Gómez, 2016).
Although goals and motivations are socio-afective variables, holistic approaches such as CDST posit
that social factors likewise contribute to learners’ internal “cognitive ecosystem” (de Bot et al., 2007),
fuctuating over time and in response to the developing L2 system.
Finally, a pending issue is whether beginning a new language in older age afects other aspects of
cognition, such as executive function. In brief, this research is based on the hypothesis that, due to
the additional demands on language control from switching between languages, at least some forms
of bilingualism can have positive efects on domain-general cognition and that these efects may be
particularly apparent in older adults who are no longer at peak cognitive capacity (see Poarch, this
volume, for a full treatment of the topic). A related claim is that bilinguals may experience a later
onset of dementia than monolinguals (e.g., Bialystok et al., 2016). However, it has been difcult to
empirically test this claim, given that “bilingualism” in fact encompasses a range of language expe-
riences that vary in age of acquisition, frequency and context of use of each language, and more.
Some studies of older adults show positive efects on general cognition after even a short period of
L2 instruction (Bak et al., 2016; Wong et al., 2019) while others show no such efects (Berggren
et al., 2018; Ramos et al., 2016). Less frequently investigated, but still highly relevant for integrative
approaches such as the UCM and CDST, is whether and how studying L2 as an older adult impacts
social dimensions. Older learners interact with teachers and students in the classroom and, as they
gain profciency, potentially broaden their social networks and enhance their social integration (e.g.,
Pot et al., 2018), which in turn may improve cognitive performance, both in L2 and more broadly.

The last few decades have seen the publication of a growing number of laboratory studies on aging
and language learning. These studies include cognitive variables relevant to SLA and either only older
adults or compare two ages. The studies can be grouped into two types according to their goals and
methods: While one strand uses language to understand cognition and cognitive aging with an eye
to clinical applications, the other is interested in understanding second language acquisition in older
adults and generates results of interest for cognitive scientists but also for language practitioners and
the education of the general older population, including migrants and refugees.
Within cognitively oriented research, researchers look at lexical learning as a way of understand-
ing the efects of aging in memory, specifcally episodic memory (e.g., van der Hoeven & de Bot,
2012); another strand investigates the robustness in old age of statistical learning, i.e., the detection of
regularities in the input and generalization to novel items (e.g., Wong et al., 2019).
Experimental studies in cognitive psychology that have relied on lexical acquisition and retrieval
to uncover age-related efects in certain memory functions generally fnd that episodic memory
(conscious awareness of past events), speed of processing, short-term memory, and working memory
decline to a similar degree, whereas semantic memory (world knowledge) shows an increase across
the lifespan (Park et al., 2002). In line with Naveh-Benjamin’s (2000) Associative Defcit Hypoth-
esis, Hoyer and Verhaeghen’s (2006) review of cross-sectional studies of episodic memory concludes
that decline in older adults’ ability to form new associations between concepts is the consequence
of age-related associative defcits or binding defcits, or the age-sensitive consequences of attentional
demands at either encoding or retrieval.
While younger participants generally learn and retrieve new lexical information more efec-
tively than older learners, a diferent question is whether, after successful memorization in the past,
vocabulary is retained after years of disuse, a question that allows cognitive psychologists to investigate
memory consolidation, and specifcally synaptic consolidation, a slow process that appears to result in
90
a permanent memory trace in the neocortex (Squire et al., 2015). In his cross-sectional study of 773
classroom learners of Spanish, Bahrick (1984) found that older learners experience a greater advan-
tage in relearning old over learning new words than younger learners. He also found that portions
of what was learned was lost within years after learning; other portions were retained for 25 years
and another for beyond 50 years. His results were replicated by van der Hoeven and de Bot (2012),
whose young, middle-aged, and older adults performed similarly on relearning of old words they had
previously learned in their French classes but behaved diferently learning new words, with young
participants outperforming older participants. Lenet et al. (2011) also found evidence of the existence
of a permastore, because their participants aged 65–92 demonstrated retention over fve decades of
limited high school exposure to L2 Latin.
The studies that investigate the robustness of the systems involved in picking up subtle, sequential
probabilistic regularities show older adults are capable of it but learn less detail than their younger
counterparts about the oral input they were briefy exposed to (Schwab et al., 2016). To understand
these results, we need to distinguish between two types of learning: explicit learning, which is efort-
ful, verbalizable, and associated with awareness, and implicit learning, which is efortless, happens
without awareness, and allows a speaker to identify a sentence as ungrammatical, even though they
are unable to articulate the grammatical rule (see Godfroid, 2023 [this volume]). At frst, it was pro-
posed that in older adults, only explicit learning declines while implicit learning mechanisms remain
spared, but more recent research suggests both systems are afected, albeit to diferent degrees, with
aging efects more pronounced for explicit than for implicit learning. Specifcally, implicit learning
of probabilistic sequential relationships declines over the adult lifespan; this may in part relate to age-
related striatal dysfunction (Howard & Howard, 2013). Moreover, efects of aging on explicit learn-
ing are present under time constraints, which do not afect implicit learning, but disappear when
older adults are given more time to apply the explicit knowledge they have (Verneau et al., 2014).
The designs of the research on age and cognition summarized previously—training, input, assess-
ment—make them hardly comparable to research on L2 development, whether in the classroom or
“in the wild,” because they depart from SLA research in signifcant ways. For example, since it is
characteristic of laboratory research, control over the input is key, so the designs typically include lists
of words and/or non-words or short strings of often-repeated syllables presented aurally before they
go on to test whether learners recognize them, produce them, or accept them as possible. One study
stands apart for its use of a natural language: Ristin-Kaufman and Gullberg (2014) used seven min-
utes of Mandarin to investigate implicit learning across the lifespan (ages 10–80+). The researchers
concluded that after only seven minutes of exposure to natural language, all learners could develop
knowledge of syllable structure that they could generalize to novel stimuli; further, age did not have
an efect on participants’ ability.
There is a strand of laboratory research within SLA that uses mini-grammars based on natural
languages (e.g., The Latin Project, Sanz et al., 2014) and manipulates learning variables to study
the interaction between cognitive demands set by the external tasks, such as presence or absence
of information about how the language works (i.e., grammar explanations), and individual difer-
ences, such as aptitude, prior language experience, and age. A subset of this research has looked into
the relationship between aging and pedagogical conditions, comparing younger and older learners,
learning either artifcial grammars (Midford & Kirsner, 2005) or a Latin-based mini-grammar (Cox,
2017; Cox & Sanz, 2015; Lenet et al., 2011).
Lenet et al. (2011) compared a sample of young adults with a group of learners aged 65–92; both
groups were monolingual and completed one of two conditions: feedback that included only explicit
correction (right/wrong), and correction with metalinguistic information (grammar rules). Results
showed a clear interaction between age and type of feedback in that aging learners outperformed
their college-aged counterparts when the feedback did not provide information on how the language
worked; it seems that for aging learners, feedback that includes grammar explanations, instead of
91
serving as a scafold, overloads the learners’ system. These results are compatible with those in Mid-
ford and Kirsner (2005) and their artifcial grammar study, since they show that older learners beneft
the most from more implicit conditions which do not require conscious rule formulation. However,
aging efects were much less pronounced in Lenet et al. (2011) because they identifed that overall
learning for the older group was not signifcantly less than for their younger college-aged group.
Cox (2017) and Cox and Sanz (2015) expanded this research in two ways: They included younger
and older bilingual groups and they also provided grammar rules, but this time they appeared prior
to practice. When combined, Cox (2017) and Cox and Sanz (2015) show that for bilinguals over 60
years old, learning a new language can be as successful as for young bilinguals, and that practice, not
grammar explanation, plays a key role as the driver of success. Importantly, and in contrast with Lenet
et al., their results also show that older bilinguals can take advantage of grammar explanations when
they are provided as an advanced organizer, prior to practice, possibly due to the older bilinguals’
potentially higher metalinguistic awareness. While these results cannot directly address the efects of
aging on implicit learning mechanisms involved in L2 development, they support the notion that
older adults can rely on explicit mechanisms to learn complex morphosyntax, i.e., assignment of
semantic functions in transitive sentences, involving competing cues of word order, case, and subject-
verb agreement, in ab initio development of Latin, even in a short period of time.
Researchers have also started to extend what we have learned from laboratory studies, which mostly
study retirees aging in place, to other populations, using cross-disciplinary approaches. Blumenfeld et al.
(2017) found that older migrants’ gains in English were predicted by their orientation to time and place
and digit span (cognitive factors), as well as by their L1 skills, previous experiences with English, and
amount of exposure to English (linguistic factors). This extends the advantage for multilingual older
adults reported by Cox and Sanz (2015) to a new population. Kliesch et al. (2018) found that the L1
German/L2 English learners who had the highest L1 verbal fuency and working memory scores had
the greatest L2 gains, a result similar to that of Mackey and Sachs (2012) in the US. Pfenninger and
Polz (2018) conducted a follow-up with German speakers, half of whom were monolingual and half of
whom also spoke Slovenian. Interestingly, participants showed moderate linguistic gains with no difer-
ence between the monolingual and bilingual groups, a result that contrasts to Blumenfeld et al. (2017)
and Cox and Sanz (2015). Participants also showed signifcant improvement on the Stroop task (execu-
tive function) and reported increased social engagement as a result of their participation; these results
speak to older adults’ cognitive reserve (see Poarch, 2023 [this volume]) and to the interrelationships
between social engagement and language learning at an older age (e.g., MacWhinney, 2017).

In sum, the study of older adult language learners is rapidly expanding; we are beginning to have a
critical mass of studies that inform us as to older learners’ capabilities while there is also much left to
explore. Holistic models of language learning such as the UCM and CDST remind us that societal
context and learners’ individual trajectories afect their L2 performance at any given time; therefore,
studies should not limit themselves to solely linguistic performance or gains; instead, they should also
consider cognitive and socio-afective outcomes (Pot et al., 2019). For similar reasons, it is essential
that studies be replicated with diferent populations and learning contexts. An important underpin-
ning coming from cognitive psychology (e.g., Amer et al., 2016) is to target the language gains of
older adults to understand what and how they learn, instead of only highlighting defcits like failing
to reach native-speaker norms or failing to learn as quickly as young adults. One way to achieve this
is to study older adults without including a young adult comparison group. Such a study would not
necessarily compromise validity, since it would still compare participants; for example, those who
receive a certain type of instruction and those who do not (control group), or the range of partici-
pants’ performance on a test of a relevant cognitive ability.
92
Incorporating diverse older adult populations into the psycholinguistics of SLA will also make our
research relevant to a wider variety of practitioners and learners, and many researchers are working on
doing so. For example, informed by the study of stereotype threat on processing, Ramírez Gómez (2016)
proposed critical foreign language geragogy as a methodology for teaching world languages to older adults
through two principal components: “how to teach and how to teach to learn” (p. 86). Guiding principles
of “how to teach” include adapting instruction to older adults’ information processing abilities without
patronizing or underestimating them, granting learners as much autonomy as possible, and proactively
countering negative stereotypes about aging to minimize their efect on learners. “How to teach to
learn,” then, leads learners to consciously re-evaluate negative stereotypes they may hold and guides
learners to adopt efective learning strategies, which may not be the strategies that were successful for
them in the past, perhaps due to changes brought about by cognitive aging. In Ramírez Gómez’s (2016)
study of an L2 Spanish course for L1 Japanese older adults, learners reported that the course changed
their ideas about what is involved in language study and they no longer believed that older age was an
obstacle to learning a new language. In other words, following critical foreign language geragogy low-
ered barriers to L2 processing and retention. Notably, this pilot study focused on lexical learning, and so
similar programs for other linguistic domains and other instructional contexts are open to further study.
Importantly, “best practices” for teaching will vary by whether a classroom is dedicated to older
adults or incorporates a wide range of adult ages (as is the case of many continuing education pro-
grams). Additionally, “best practices” will vary based on population, because older adults learning
the dominant language following migration likely require diferent approaches than those learning
something new in retirement in their home country, given expected diferences in attention, work-
ing memory, motivation, and identity. Thus, there will never be a single set of “best practices” for
older learners, but rather examinations of best practices for each given situation.
How does older adult learners’ use of declarative and procedural memory when processing the target
language in early stages of learning compare to (a) young adult learners and (b) older adult bilinguals
with long periods of L2 immersion?
How do socio-afective and cognitive factors interact to determine the linguistic outcomes of older adults’
language study, and how do any interactions vary by population (e.g., retirees in their country of origin
versus minoritized migrants in a new country)? Conversely, what efect does increasing target language
profciency have on cognitive performance and social integration in diferent older adult populations?
How can the activation of stereotype threat be minimized for older adult learners in societies with nega-
tive stereotypes of aging and cognition, so as to maximize their processing and retention of target
language input?
Informed by research in second language acquisition and cognitive psychology, what set of pedagogical
principles would guide language instruction for older learners? How would those principles vary
depending on population, such as newly arrived migrants and non-migrant retirees?
Further reading
de Bot, K., & Makoni, S. (2005). Language and aging in multilingual contexts. Multilingual Matters.
One of the frst works on aging in SLA/bilingualism, this volume uses the complex dynamic systems theory
to explore how multilinguals’ languages fuctuate over the course of a lifespan.
93
Kemper, S. (2006). Language in adulthood. In E. Bialystok & F. I .M. Craik (Eds.), Lifespan cognition: Mechanisms
of change (pp. 223–238). Oxford University Press.
An essential introduction to age-related changes in L1 and L2 and their potential causes.
Reifegerste, J. (2020). The efects of aging on bilingual language: What changes, what doesn’t, and why. Bilin-
gualism: Language and Cognition, 1–17.
This recent review article summarizes research on language and cognition in older bilinguals in comparison
to older monolinguals and to young adult bilinguals.
References
Amer, T., Campbell, K. L., & Hasher, L. (2016). Cognitive control as a double-edged sword. Trends in Cognitive
Science, 20(12), 905–915.
Arxer, S. L., Ciriza, M. D. P., & Shappeck, M. (2017). Aging in a second language: A case study of aging, immigration,
and an English learner speech community. Springer.
Bahrick, H. P. (1984). Semantic memory content in permastore: Fifty years of memory for Spanish learned in
school. Journal of Experimental Psychology: General, 113(1), 1–29.
Bak, T. H., Long, M. R., Vega-Mendoza, M., & Sorace, A. (2016). Novelty, challenge, and practice: The impact
of intensive language learning on attentional functions. PLoS ONE, 11(4), e0153485.
Bates, E., & MacWhinney, B. (1982). Functionalist approaches to grammar. In E. Wanner & L. Gleitman (Eds.),
Language acquisition: The state of the art (pp. 173–218). Cambridge University Press.
Berggren, R., Nilsson, J., Brehmer, Y., Schmiedek, F., & Lövdén, M. (2018). No evidence that foreign language
learning in old age improves cognitive function: A randomized controlled study, Psychology of Aging, 35(2),
212–219.
Bialystok, E., Abutalebi, J., Bak, T. H., Burke, D. M., & Kroll, J. F. (2016). Aging in two languages: Implications
for public health. Ageing Research Reviews, 27, 56–60.
Birdsong, D. (Ed.). (1999). Second language acquisition and the critical period hypothesis. Lawrence Erlbaum Associ-
ates Publishers.
Birdsong, D. (2018). Plasticity, variability and age in second language acquisition and bilingualism. Frontiers in
Psychology, 9, 81.
Blumenfeld, H., Quinzon, S. J. R., Alsol, C., & Riera, S. A. (2017). Predictors of successful learning in multi-
lingual older adults acquiring a majority language. Frontiers in Communication, 2, 23.
Bubbico, G., Chiacchiaretta, P., Parenti, M., di Marco, M., Panara, V., Sepede, G., Ferretti, A., & Perrucci, M.
G. (2019). Efects of second language learning on the plastic aging brain: Functional connectivity, cognitive
decline, and reorganization. Frontiers in Neuroscience, 13, 423.
Clahsen, H., & Reifegerste, J. (2017). Morphological processing in old-age bilinguals. In M. Libben, M. Goral, &
G. Libben (Eds.), Bilingualism: A framework for understanding the mental lexicon (pp. 217–248). Benjamins.
Cox, J. G. (2017). Explicit instruction, bilingualism, and the older adult learner. Studies in Second Language
Acquisition, 39(1), 29–58.
Cox, J. G. (2019). Multilingualism in older age: A research agenda from the cognitive perspective. Language
Teaching, 52(3), 360–373.
Cox, J. G., & Sanz, C. (2015). Deconstructing PI for the ages: Explicit instruction v. practice in young and older
bilinguals. Language Teaching Research, 53(2), 225–248.
Craik, F. I. M., & Bialystok, E. (2006). Cognition through the lifespan: Mechanisms of change. Trends in Cogni-
tive Sciences, 10(3), 131–138.
De Bot, K., Lowie, W., & Verspoor, M. (2007). A dynamic systems approach to second language acquisition.
De Bot, K., & Makoni, S. (2005). Language and aging in multilingual contexts. Multilingual Matters.
Diaz, M. T., Rizio, A. A., & Zhuang, J. (2016). The neural language systems that support healthy aging: Inte-
grating function, structure, and behavior. Language and Linguistics Compass, 10(7), 314–334.
Hakuta, K., Bialystok, E., & Wiley, E. W. (2003). Critical evidence: A test of the critical-period hypothesis for
second-language acquisition. Psychological Science, 14, 31–38.
Howard, J. H., Jr., & Howard, D. V. (2013). Aging mind and brain: Is implicit learning spared in healthy aging?
94
Hoyer, W. J., & Verhaeghen, P. (2006). Memory aging. In J. E. Birren & K. W. Schaire (Eds.), Handbook of the
psychology of aging (pp. 209–232). Elsevier.
Kliesch, M., Giroud, N., Pfenninger, S. E., & Meyer, M. (2018). Research on second language acquisition in old
adulthood: What we have and what we need. In D. Gabryś-Barker (Ed.), Third age learners of foreign languages
(pp. 48–75). Multilingual Matters.
Larsen-Freeman, D. (2012). Complexity theory. In S. Gass & A. Mackey (Eds.), The Routledge handbook of second
language acquisition (pp. 73–87). Routledge.
Lenet, A. E., Sanz, C., Lado, B., Howard, J. H., Jr., & Howard, D. V. (2011). Aging, pedagogical conditions, and
diferential success in SLA: An empirical study. In C. Sanz & R. P. Leow (Eds.), Implicit and explicit language
learning: Conditions, processes, and knowledge in SLA and bilingualism (pp. 73–84). Georgetown University Press.
Lenneberg, E. H. (1967). The biological foundations of language. Hospital Practice, 2(12), 59–67.
Levy, B., & Langer, E. (1994). Aging free from negative stereotypes: Successful memory in China and among
the American deaf. Journal of Personality and Social Psychology, 66(6), 989–997.
Mackey, A., & Sachs, R. (2012). Older learners in SLA research: A frst look at working memory, feedback, and
L2 development. Language Learning, 62(3), 704–740.
MacWhinney, B. (2017). Entrenchment in second-language learning. In H.-J. Schmid (Ed.), Entrenchment and
the psychology of language learning: How we reorganize and adapt linguistic knowledge (pp. 343–366). American
Psychological Association.
ziano, & H. Jisa (Eds.), Sources of variation in frst language acquisition: Languages, contexts, and learners (pp.
Midford, R., & Kirsner, K. (2005). Implicit and explicit learning in aged and young adults. Aging, Neuropsychol-
ogy and Cognition, 12(4), 359–387.
Morgan-Short, K. & Ullman, M.T. (2023). Declarative and procedural memory in second language learning:
Naveh-Benjamin, M. (2000). Adult age diferences in memory performance: Tests of an associative defcit
hypothesis. Journal of Experimental Psychology: Learning, Memory, and Cognition, 26(5), 1170–1187.
Paradis, M. (2004). A neurolinguistic theory of bilingualism. John Benjamins.
Park, D. C., Lautenschlager, G., Hedden, T., Davidson, N. S., Smith, A. D., & Smith, P. K. (2002). Models of
visuospatial and verbal memory across the adult life span. Psychology and Aging, 17(2), 299–320.
Pfenninger, S. E., & Polz, S. (2018). Foreign language learning in the third age: A pilot feasibility study on
cognitive, socioafective and linguistic drivers and benefts in relation to previous bilingualism of the learner.
Journal of the European Second Language Association, 2(1), 1–13.
Poarch, G.J. (2023). Cognitive efects of bilingualism. In A. Godfroid & H. Hopp (Eds.), The Routledge handbook
Pot, A., Keijzer, M., & de Bot, K. (2018). The language barrier in migrant aging. International Journal of Bilingual
Education and Bilingualism, 1–19.
Pot, A., Porkert, J., & Keijzer, M. (2019). The bidirectional in bilingual: Cognitive, social and linguistic efects
of and on third-age language learning. Behavioral Sciences, 9(9), 98.
Ramírez Gómez, D. (2016). Language teaching and the older adult: The signifcance of experience. Multilingual Matters.
Ramos, S., Fernández García, Y., Eneko, A., Casaponsa, A., & Duñabeita, J. A. (2016). Does learning a language
in the elderly enhance switching ability? Journal of Neurolinguistics, 43(A), 39–48.
Reifegerste, J. (2020). The efects of aging on bilingual language: What changes, what doesn’t, and why. Bilin-
gualism: Language and Cognition, 1–17.
Ristin-Kaufmann, N., & Gullberg, M. (2014). The efects of frst exposure to an unknown language at diferent
ages. Bulletin Suisse de Linguistique Appliquée, 99, 17–29.
Sanz, C. (2012). Multilingualism and metalinguistic awareness. In C. A. Chapelle (Ed.), The encyclopedia of applied
linguistics (pp. 3933–3942). Wiley-Blackwell.
Sanz, C., Lin, H. J., Lado, B., Bowden, H. W., & Staford, C. A. (2014). One size fts all? Learning conditions
and working memory capacity in ab initio language development. Applied Linguistics, 37(5), 669–692.
Schwab, J. F., Schuler, K. D., Stillman, C. M., Newport, E. L., Howard, J. H., Jr., & Howard, D. V. (2016). Aging
and the statistical learning of grammatical form classes. Psychology and Aging, 31(5), 481–487.
Sonntag, W. E., Eckman, D. M., & Ingraham, J. (2007). Regulation of cerebrovascular aging. In D. R. Riddle
(Ed.), Brain aging: Models, methods, and mechanisms (pp. 279–304). CRC Press/Taylor & Francis.
Squire, L. R., Genzel, L., Wixted, J. T., & Morris, R. G. M. (2015). Memory consolidation. Cold Spring Harbor
Perspectives in Biology, 7(8), a021766.
Tagarelli, K. M., Shattuck, K. F., Turkeltaub, P. E., & Ullman, M. T. (2019). Language learning in the adult
brain: A neuroanatomical meta-analysis of lexical and grammatical learning. NeuroImage, 193, 178–200.
95
Tomaszewski Farias, S., Schmitter-Edgecombe, M., Weakley, A., Harvey, D., Denny, K. G., Barba, C., Gravano,
J. T., Giovannetti, T., & Willis, S. (2018). Compensation strategies in older adults: Association with cognition
and everyday function. American Journal of Alzheimer’s Disease & Other Dementias, 3(33), 184–191.
Ullman, M. (2004). Contributions of memory circuits to language: The declarative/procedural model. Cogni-
tion, 92(1–2), 231–270.
Ullman, M. (2015). The declarative/procedural model: A neurobiologically motivated theory of frst and second
language. In B. VanPatten & J. Williams (Eds.), Theories in second language acquisition: An introduction (2nd ed.,
pp. 135–158). Routledge.
Van der Hoeven, N., & de Bot, K. (2012). Relearning in the elderly: Age-related efects on the size of savings.
Language Learning, 62, 42–67.
Verneau, M., van der Kamp, J., Savelsbergh, G. J., & de Looze, M. P. (2014). Age and time efects on implicit
and explicit learning. Experimental Aging Research, 40(4), 477–511.
Wong, P. C. M., Ou, J., Pang, C. W. Y., Zhang, L., Tse, C. S., Lam, L. C. W., & Antoniou, M. (2019). Lan-
guage training leads to global cognitive improvement in older adults: A preliminary study. Journal of Speech,
Language, and Hearing Research, 62(7), 2411–2424.
Wong, P. C. M., Vuong, L. C., & Liu, K. (2017). Personalized learning: From neurogenetics of behaviors to
designing optimal language training. Neuropsychologia, 98, 192–200.
96
9
SYNTHESIS
Psycholinguistics across the lifespan
Theres Grüter
Introduction
The development of our ability to produce and comprehend language, which for most people world-
wide means more than one language, is a dynamic process that continues across the lifespan. Con-
trary to the popular understanding of learning and acquisition as processes of additive, cumulative
growth towards a stable target, language development is a process of ongoing adaptation to an envi-
ronment that itself may change. Development in general, and language development in particular,
involves both waxing and waning, and there are some developmental steps, such as infants’ loss of the
ability to discriminate contrasts not present in their ambient language(s) towards the end of the frst
year of life, that are difcult to classify at all in terms of increase or decrease of linguistic skills (see also
Potter & Lew-Williams, 2023 [this volume]).
The complex, non-linear nature of language development across the lifespan becomes particu-
larly evident in the context of bilingualism, as illustrated by the seven chapters in this section of the
handbook. Collectively, these chapters present a state-of-the-art review of what we have learned over
the past decades about the human ability to process multiple languages, the development of this abil-
ity across the lifespan from infancy (Potter & Lew-Williams, 2023 [this volume]) through childhood
(Schimke, 2023 [this volume]) and adulthood (Jackson, 2023 [this volume] on second language (L2);
González Alonso & Rothman, 2023 [this volume] on third or further language (L3/Ln)) into older age
(Cox & Sanz, 2023 [this volume]), and the complex waxing-and-waning dynamics of this develop-
ment in changing environments that often result from global migration (Montrul, 2023 [this volume]
on heritage language acquisition; Schmid et al., 2023 [this volume] on frst language (L1) attrition).
In this synthesis chapter, I will begin by drawing eclectically from the summative reviews pro-
vided by these chapters to illustrate how the inclusion of methodological and theoretical paradigms
from psycholinguistics and cognitive science into Second Language Acquisition (SLA) research has
led to a more nuanced understanding of the linguistic abilities that various types of multilingual
language users possess, and the similarities and diferences between them. These insights have
contributed to the now well-established recognition that “bilinguals” come in many diferent
shapes and sizes that each deserve to be studied in their own right. This recognition has led to the
establishment of a number of subfelds within SLA focused on the study of a particular type of
language learner or user, as refected by chapters in this volume specifcally dedicated to the study
of, for example, L3/Ln learners or L1 attriters. While applauding the recognition of these diferent
learner types, I would like to take this opportunity to also point to the potential risks that come
with the increasing compartmentalization of the feld, and the consequences this may have for
DOI: 10.4324/9781003018872-10 97
Theres Grüter
our ability to work towards a more comprehensive understanding of language development across
the lifespan. I will suggest that our next steps must be towards communication and theorizing
across subfelds, both within SLA and across the boundaries between psycholinguistic approaches
to SLA and what is sometimes referred to as “mainstream psycholinguistics” (e.g., Montrul, 2023
[this volume]), that is, the study of psycholinguistics that has traditionally focused on monolingual
native speakers (Francis, 2021). I will conclude by addressing some of the challenges that come
with SLA and bilingualism claiming ground in mainstream psycholinguistics, and outline some
steps that our feld can take, and has already started to take, towards addressing these challenges and
growing into a cumulative quantitative discipline (Cumming, 2014; Bolibaugh et al., 2021) with
reach beyond the study of the WEIRDest (Western, Educated, Industrialized, Rich, and Demo-
cratic) people in the world (Henrich et al., 2010).
The contribution of psycholinguistic paradigms to the study of

SLA and multilingualism across the lifespan
The authors of each chapter discuss how the inclusion of methodological tools and theoretical frame-
works from the wider felds of psycholinguistics and cognitive science has increased our understanding
of the linguistic abilities of various learner groups. For example, the use of eye-tracking methods to
capture language processing during real-time listening, as in the looking-while-listening and visual
world paradigms, has opened the door to investigating language processing efciency among popula-
tions of language users with limited expressive and/or literacy skills. This includes infants and young
children, but also adult heritage speakers, whose reading and writing skills in their heritage L1 are often
much more limited than their speaking and listening abilities (Montrul, 2023 [this volume]). Recent
studies employing such methods have shown, for example, that child second language (L2) learners in
majority L2 contexts tend to pattern more closely with child L1 than with adult L2 speakers in online
processing tasks, despite the fact that in ofine tasks and production they often show similar errors as
adult L2 speakers (Schimke, 2023 [this volume]). Similarly, the use of methodological paradigms from
speech perception research has revealed that heritage speakers are more likely to pattern with mono-
lingual native speakers, and diferent from L2 speakers, in the perception of phonological contrasts in
their heritage L1 (Montrul, 2023 [this volume]). At a more general level, as both Montrul and Schimke
highlight, psycholinguistic measures that aim to tap into more implicit language knowledge and pro-
cessing skills tend to show diferent patterns of alignment between diferent groups of language users
than the ofine comprehension, judgment, and production tasks used in earlier SLA research.
Studies including more implicit psycholinguistic measures have revealed that traditional ofine
measures often underrepresent the abilities of certain types of language users, especially those who
had limited formal educational experience with the language in question, and may overestimate the
abilities of highly educated learners. This is relevant for theoretical questions such as those pertain-
ing to the role of age of acquisition in the attainment of “native-like” linguistic abilities, but it also
has important implications for applied concerns such as language assessment in educational contexts.
At the same time, it is important to bear in mind that psycholinguistic measures do not provide us
with truer, better, more accurate, or more important evidence of a learner’s linguistic knowledge and
abilities. Instead, the inclusion of such measures allows us to probe a greater range of knowledge and
abilities involved in successful language use. It is the inclusion of both on- and ofine measures that
has helped bring to light diferent profles of strengths and weaknesses in diferent learner groups,
and the SLA feld will beneft from continuing to include and value the full array of measures that
we now have at our disposal.
In addition to providing a more comprehensive picture of learners’ various abilities at an empiri-
cal level, the investigation of incremental language processing among language learners has also led
to increasing attention to key theoretical questions about the relation between language processing
98
Synthesis
and learning, an issue of central concern in SLA and Applied Linguistics. As discussed in more depth
by Jackson (2023 [this volume]; see also Hopp, 2021), there are close links between how learners use
their current linguistic knowledge to assign structure and meaning to the input they receive (parsing),
and how they modify their existing linguistic knowledge in light of that input and the parse they
have assigned to it (learning). A mechanism that has received considerable attention in this context is
that of prediction, more specifcally the notion of learning as a result of prediction error (e.g., Chang
et al., 2006; for discussion in the context of SLA, see Kaan, 2014; Jackson & Hopp, 2020; Kaan &
Grüter, 2021). The extension of theoretical perspectives and mechanisms such as these into the realm
of SLA provides an illustration of how theoretical frameworks from the wider feld of psycholinguis-
tics and cognitive science have begun to inform theorizing on L2 processing and learning in ways
that transcend traditional (sub)disciplinary boundaries.
The coming of age of subfelds and the risks of compartmentalization

Several authors express well-justifed pride in the coming of age of subdisciplinary felds focusing on
specifc types of multilingual learners. A good example is the feld of L3/Ln acquisition, which has
held dedicated annual conferences for more than a decade now (e.g., the International Conference
on Third Language Acquisition and Multilingualism) and has been the topic of two recent keynote
papers in leading SLA journals (Schwartz & Sprouse, 2021; Westergaard, 2021). As González Alonso
and Rothman (2023 [this volume], p. 55) rightly note, there are “factors that are unique to the L3/
Ln acquisition setting, such as the existence of several grammars within the speaker’s prior linguistic
experience (versus only the L1 in L2 acquisition).” This leads them to argue that “these characteristics
warrant a consideration of language processing in sequential multilingualism as a feld of study in
its own right.” In line with this perspective, multiple theoretical models of L3 acquisition have been
proposed in the recent literature, reviewed in González Alonso and Rothman (2023 [this volume]),
who conclude that “all [of these accounts] agree that L2 and L3/Ln must constitute distinct felds that
can only ever partially overlap in their theoretical focus” (p. 55).
While I agree with González Alonso and Rothman that the dedicated attention to L3/Ln acquisi-
tion has led to important insights, including the uncovering of confounds in earlier studies that failed
to consider that the “L2” learners under investigation were really L3 learners (as in my own work;
Grüter and Conradie, 2006, see White, 2021), I am also concerned by the potential consequences
that the increasing compartmentalization into ever more specifc subfelds may have for our ability to
retain sight of the forest for the trees. That forest, ultimately, is the human capacity for language and
its unfolding across the human lifespan in the plethora of environments that humans fnd themselves
in. Indeed, González Alonso and Rothman situate the study of L3/Ln acquisition within this broader
context when they state that it has the potential to provide “a window into larger questions about
language and mind more generally” (p. 57). Yet for insights from specifc subfelds to really contribute
to these larger questions, we need to open González Alonso and Rothman’s metaphorical window
and communicate across and not just within (sub)felds.
This need was recognized already more than thirty years ago. In their introductory chapter to
an edited volume entitled Bilingualism Across the Lifespan, Hyltenstam and Obler (1989) described
research on “bilingualism” at the time as “a dynamic, rapidly developing feld” and refected on the
increasing need for specialization leading to “increasing isolation between those who work in difer-
ent subfelds” (p. 1). They noted that a key purpose of their 1989 volume was to “stimulate research-
ers to extend their thinking beyond their own specialities” (p. 2). Hyltenstam and Obler’s call for
thinking beyond one’s own specialty remains as relevant for the feld today as it was in 1989. While
giving the unique characteristics of a specifc learning context their due empirical and theoretical
attention, it is equally important to ask if and how the mechanisms proposed to explain language pro-
cessing within these specifc contexts scale up to other learning contexts, to fnd out how researchers
99
Theres Grüter
in neighboring subfelds may have already developed empirical or theoretical tools for investigating
related questions, and to refect on how any proposed mechanisms may ft into what we know about
the larger picture of human cognition.
The importance of experiential and environmental factors

Across subfelds, recent fndings highlight the role of environmental and experiential, sometimes
called (learner) “external” factors in multilingual development across the lifespan. For example, as
reviewed in more detail by Potter and Lew-Williams (2023 [this volume]), research by Hurtado,
Marchman, and colleagues has shown that the amount of exposure a bilingual toddler has to each
language in their life afects not just the number of words they know in that language, but also the
speed with which they can access these known words during real-time listening (Marchman et al.,
2010; Hurtado et al., 2014). At the same time, these studies observed no consistent relation between
bilingual children’s lexical processing speed across their two languages. In other words, bilingual chil-
dren were not overall speedy or slow processors; instead, their processing efciency in each language
was tied to the amount of experience they had had with that language. These fndings are important
because they highlight the substantial externally conditioned variability in language processing efciency
that exists within a group of language learners that would traditionally be categorized as a single type
of learner, in this case simultaneous bilingual children.
Similar externally conditioned variability was recently highlighted in work by De Wilde and
colleagues with sequential child L2 learners in a non-majority L2 context (De Wilde et al., 2020a,
2020b). Using traditional measures of language skills in a large randomly sampled group of Dutch-
speaking elementary school students in Flanders prior to the onset of formal English instruction in
school, these authors found that many children had already acquired notable English language skills
through informal out-of-school exposure, such as gaming and social media. This exposure explained
a signifcant amount of the substantial variability within this at frst blush relatively homogeneous
group of incipient sequential child L2 learners of English. Very few studies to date have used psycho-
linguistic measures to investigate real-time processing among learners such as these, i.e., sequential
child L2 learners with exposure primarily through classroom instruction in a societal context in
which the target language is not a majority language (for a notable exception, see Lew-Williams,
2017, on English-speaking children in Spanish immersion in California). In order to better under-
stand the role of external factors related to classroom instruction and the role of the societal language
on the development of L2 processing skills, it would be desirable for future work to include psy-
cholinguistic measures in studies of language development with child L2 learners in such contexts.
Along these lines, several chapters list greater exploration of the role of external factors, such as better
scrutiny of children’s daily experiences (Potter & Lew-Williams), the role of instructed versus unin-
structed learning contexts (Cox & Sanz; Jackson; Schimke), and the infuence of broader societal and
cultural assumptions about language learning and age (Cox & Sanz) among key directions for future
psycholinguistic research in SLA.
One of the many reasons why more thorough attention to environmental and experiential vari-
ables is likely to move the feld and its subfelds forward is that it will force us to question, as Schmid
et al. (2023 [this volume], p. 68) recommend, the “largely categorical view of diferent populations”
that still predominates in current work on bilingualism, and to refect on the criteria by which we
assign learners to a specifc group or learner type. These criteria are to a large extent based on envi-
ronmental factors: in what context a learner was exposed to a language (home, classroom) and when
in their life that exposure began (birth, childhood, adulthood). From these at least partially continu-
ous variables, the feld has derived coarse-grained categories of learner types, such as simultaneous
vs. sequential bilinguals, heritage vs. non-heritage native speakers, L2 vs. L3 learners, instructed vs.
uninstructed learners, etc. Such categories can be useful for studies that seek to investigate language
100
Synthesis
learning within a single, clearly defned context; they can also serve as a useful frst step in investiga-
tions that wish to compare between diferent learning contexts. It must be kept in mind, however,
that many of these categories are neither exclusive nor exhaustive. They are not exclusive in that
an individual learner can fall into multiple categories: Every L1 attriter, for example, is also an L2
learner, perhaps also an L3/Ln learner, and some might also be heritage speakers. They are not
exhaustive in that many individuals do not ft easily into any of the categories along a given dimen-
sion. For example, Hartshorne et al. (2018), who aimed to investigate the critical period hypothesis
in an unprecedentedly large sample of more than 600,000 English speakers, divided their partici-
pants into (i) monolinguals, (ii) immersion learners, and (iii) non-immersion learners; yet more
than 100,000 participants, about 18% of their sample, had to be excluded from analyses because
they could not be classifed into any of these groups (see also van der Slik et al., 2022). While lack
of available background information was likely responsible for some of the exclusions, the point
remains: The boundaries between these learner types are not entirely categorical, and when they
are implemented in categorical terms, a substantial number of individuals are excluded and thus not
represented in the knowledge base that informs the discipline. One way to address this concern is
to go beyond comparisons between predefned learner groups, where individual learners are seen
primarily as representatives of a specifc type (e.g., simultaneous bilingual, L3 learner, native speaker),
towards exploration of experiential and environmental factors as individual-diference variables that
shape language processing across learner types and across the lifespan more broadly (see also Dussias,
2023 [this volume]; Jackson, 2023 [this volume]; Bice & Kroll, 2019; Gullifer & Titone, 2020). Such
an approach can further serve to bring together separate subfelds and allow the larger feld to ask
broader questions about psycholinguistics and cognitive psychology.
Claiming ground in “mainstream” psycholinguistics and cognitive science

In a recent volume aimed at an audience of cognitive psychologists, entitled Bilingualism Across the
Lifespan: Opportunities and Challenges for Cognitive Research in a Global Society (Francis, 2021), the
editor tasked the authors of each chapter to address the question, “What should every cognitive psy-
chologist know about bilingualism?” Several contributors noted that despite cognitive psychologists’
general and largely implicit assumption of monolingual cognition as the default state, bi- and mul-
tilinguals have long been included in “mainstream” cognitive (including psycholinguistic) research,
without their bilingual status being taken note of. As Tiv et al. (2021, p. 33) pointed out, cognitive
psychologists often report participants being “native speakers” of the language in which the experi-
ment was conducted, yet they do not ask about, let alone measure, knowledge of other languages. It
is therefore quite likely that such groups included simultaneous bilinguals, heritage speakers and L1
attriters. (The same, incidentally, applies to so-called native speaker control groups in SLA research.)
Moreover, native speaker status in cognitive psychology is typically determined solely through par-
ticipants’ self-identifcation, often without researchers providing any specifc defning criteria of what
constitutes a native speaker. Indeed, such criteria are not straightforward to formulate and there are
no commonly agreed-upon standards, neither in cognitive psychology nor any other feld that I
am aware of, including bilingualism and SLA. Critical discourse around the construct of the native
speaker has a long tradition in applied linguistics and SLA (Cook, 1999; Davies, 2003; Rampton,
1990; among many others), yet the monolithic notion of the native speaker was, until recently,
rarely questioned or discussed in “mainstream” psycholinguistics and cognitive science (Bice & Kroll,
2019; Gullifer & Titone, 2020; for a notable earlier exception, see Pakulak & Neville, 2010). This
discrepancy is a noteworthy refection of the limited cross-disciplinary discourse that has taken place
between these felds so far.
For researchers in SLA and bilingualism, the growing recognition in cognitive science that mono-
lingual native speakers are neither easy to defne nor easy to fnd in many parts of the world, and that
101
Theres Grüter
consequently “the treatment of monolingual cognition as the default and bilingualism as a modifca-
tion of the monolingual system is not an adequate approach” (Francis, 2021, p. 2) is very encourag-
ing. It opens the door from the side of “mainstream” psycholinguistics and cognitive science, which
is sometimes perceived as unwelcoming or unappreciative of bilingualism and SLA research, to invite
precisely such research to the table. The emerging exchange at this larger table is already refected
in some of the chapters in this handbook. Potter and Lew-Williams (2023 [this volume]), for exam-
ple, explicitly frame their discussion of early childhood bilingualism in terms of understanding key
insights about early bilingualism by viewing them through the lens of cognitive mechanisms that have
been shown to support (presumed) monolingual development.
Note that investigating whether models constructed for one type of learner scale up to other
learner types critically requires comparisons between groups. In the case of models from “main-
stream” psycholinguistics originally constructed to capture data from (presumed) monolingual
speakers, this comparison must be between monolingual and bilingual speakers, a comparison often
eschewed in SLA and applied linguistics because it can lead to the perception of monolinguals as the
gold standard, i.e., the target that bilinguals are to be measured against (e.g., Ortega, 2013). Indeed,
and unfortunately, it is not uncommon for research on L2 processing to be framed in such terms, ask-
ing for example whether bilinguals can “achieve native-like processing efciency.” There are multiple
reasons to be cautious about framing research on L2 processing in such terms, including the question
of whether what is most efcient for a native speaker is also necessarily most efcient for a non-native
speaker (see Grüter et al., 2020, for discussion), but also because it reafrms and perpetuates the view
of monolingual cognition as the default, a view that both cognitive psychologists and (applied) lin-
guists have now come to recognize as inadequate. Such approaches also tend to gloss over the fact that
a monolingual and a native speaker are not the same thing: While all monolinguals are native speakers
of their single language, it is not uncommon for speakers to have more than one native language,
making the juxtaposition of “native speakers” vs. “bilinguals” non-sensical.
It is equally important, however, to recognize that comparing two groups does not necessarily
require that one must present the standard against which the other is assessed. As an (imperfect)
analogy, consider the case of a hypothetical drug being tested among mice and among humans. Let
us assume the outcome of the trials with mice shows the drug is highly efective at preventing dis-
ease among mice; yet in clinical trials with humans, the same drug turns out not to be efective at
preventing disease among humans. It seems unlikely that researchers will interpret such an outcome
as humans having failed to achieve mouse-like standards, nor that eforts will then be directed at
modifying or training humans to make them more mouse-like. Instead, researchers will presumably
conclude that the drug’s efectiveness did not scale up from a simpler to a more complex organism,
and thus is not a useful contribution to managing human health. By the same logic, in cases in which
we fnd that bilinguals’ processing does not look the same as that previously observed among mono-
linguals, this does not have to be interpreted as a failure on the bilinguals’ part to achieve monolin-
gual standards; it simply provides evidence that a particular behavior observed in a simpler organism
(monolinguals) does not scale up to more complex ones (bilinguals), and hence is not likely to be a
generalizable property of human language processing.
When viewed through this lens, comparisons between monolinguals or native speakers on the
one hand and bi/multilinguals or non-native speakers on the other allow us to probe to what
extent hard-won and valuable insights on language processing from cognitive research with (pre-
sumed) monolinguals are generalizable to multilingual speakers. Critically, such comparisons must
be informed by and framed within the recognition that cognitive processes observed among native
or monolingual speakers do not constitute a default or gold standard against which bi- and multi-
linguals are to be measured. Instead, the performance of this select group can serve as proof of con-
cept, i.e., as illustration that certain mechanisms can be operative in human cognition in the special
context when that human has only one language. The more signifcant and interesting question
102
Synthesis
then is if and how these mechanisms operate in more complex (and more typical) humans, that is
bi- and multilinguals.
Alternatively and more radically, one could argue, as did Prior and van Hell (2021, p. 51), that
“this question should be turned on its head. Namely, why would researchers make the choice to
exclusively study monolingual language speakers and then aim to make generalizations about a
population that is mostly non-monolingual?” This is an excellent point, especially with regard to
the authors’ recommendation that future model building in cognitive science and psycholinguistics
should take a more “inclusive approach,” i.e., work with data more representative of the heterogene-
ity of typical language users. With regard to the current knowledge base in these felds, however,
the fact remains that it has been built mostly on evidence from (presumed) monolingual speakers so
far. On the null hypothesis that bilinguals are not radically diferent from monolinguals, I believe the
SLA feld still has much to learn from the valuable insights that this literature has provided, despite its
original monolingual bias. Thus, testing the generalizability of models originally built for (presumed)
monolinguals to various non-monolingual populations remains a valuable endeavor in psycholinguis-
tic research on SLA and bilingualism, as long as we keep in mind that what is being assessed is the
usefulness of the models to capture bilingual language processing, not bilinguals’ ability to behave like
“the dwindling minority of monolingual speakers” (Prior & van Hell, 2021, p. 59).
Stepping up our game

Claiming a place at the table of mainstream psycholinguistics and cognitive science requires us to
recognize the methodological standards of those wider felds. This can be challenging, especially
when it comes to standards for statistical power. In the context of the so-called “replication crisis” in
psychology, limitations in statistical power due to small participant and item samples have become a
key focus of attention in psycholinguistics (e.g., Brysbaert & Stevens, 2018). Similar concerns about
lack of generalizability due to limited power have also been documented in SLA research (Plonsky,
2013), with Brysbaert (2021, p. 816) recently calling upon bilingualism researchers “to step up our
game if we want research on bilingualism to be more than an endless quarrel about exciting, new,
signifcant observations that others fnd difcult to replicate.”
Such calls can be perceived as intimidating especially by those working in subfelds of SLA that
focus on very specifc or “rare” types of bilinguals, such as L1 attriters or L3/Ln learners. To be sure,
it is one thing to recruit a large sample of Dutch-French bilinguals in Flanders, or Catalan-Spanish
bilinguals in Barcelona, but quite another to do the same with, for example, Bulgarian-speaking
expats in Germany or L3 learners of Portuguese with L1 English and L2 Spanish. The fact that
it is often simply not feasible for research on special types of multilinguals to achieve sample sizes
comparable to those common in “mainstream” psycholinguistics is sometimes used as an argument,
implicitly or explicitly, to justify the need for separate subfelds, with diferent standards that accom-
modate these unavoidable challenges when working with these specifc populations. This is under-
standable, but equally unfortunate because it will prevent insights from research on these populations
to be perceived as meaningful contributions to the larger enterprise of cognitive science, to which
researchers in these subfelds explicitly wish to contribute (e.g., González Alonso & Rothman, 2023
[this volume]; Schmid et al., 2023 [this volume]).
Importantly, recruiting more participants for a single experiment is not the only way to increase
statistical power. Brysbaert (2021) concludes with a very helpful list of other steps that can be taken
to increase or preserve power. These include keeping experimental designs as simple as possible.
In other words, while the researcher might have good theoretical reasons to formulate hypotheses
that require a fully crossed 2 x 3 experimental design, they can potentially make a more meaningful
contribution by limiting themselves to testing just one of these factors in a properly powered study
than by conducting a more complex but hopelessly underpowered experiment that will not allow
103
Theres Grüter
for frm conclusions regarding either factor (let alone their interaction). Another recommendation
on Brysbaert’s list is to increase collaboration between labs and researchers in diferent locations to
pool their data for analysis. This is already being done on a large scale in research on early childhood
bilingualism (e.g., Byers-Heinlein et al., 2021), driven by the larger enterprise of the ManyBabies
Consortium (Frank et al., 2017). To date, only few multisite replication studies have been undertaken
in SLA; for a notable exception with adult L2 learners, see Morgan-Short et al. (2018). Perhaps the
time is ripe for the establishment of ManyAttriters and ManyHeritageSpeakers consortia?
One might object that the diversity of individual social and linguistic contexts in, for example,
heritage language or L3 acquisition make it difcult or even impossible to aggregate data over mul-
tiple research sites. Yet even if multi-site research is not an option, we can step up our game by
increasing the transparency and accessibility of our data and analyses, especially when they come from
harder-to-fnd populations, to enable meta-analyses that will contribute to a better understanding of
the robustness of patterns across specifc contexts. The need for this is apparent: In a recent system-
atic review of research on L3 transfer, Puig-Mayenco et al. (2020) noted that they were unable to
conduct a proper meta-analysis because “[u]nfortunately, a majority of the studies reviewed here do
not meet the requirements to conduct a meta-analysis of the type just described: efect sizes are not
reported, and they often cannot be directly or indirectly estimated from the information reported in
the studies” (p. 32). If we raise our standards for reporting our data fully and honestly, especially when
they come from small samples of “rare” populations, future researchers will not have to grapple with
this problem. These are critical steps towards becoming a cumulative quantitative discipline (Cumming,
2014), in which data do not just contribute to the generation of knowledge through the results of
a single (potentially underpowered) study but through being reported and made available, fully and
honestly, to the research community for further use in larger analyses. This point is made forcibly
and convincingly in a recent paper by Bolibaugh et al. (2021), who include concrete guidance for
how bilingualism researchers can contribute towards a more transparent and cumulative scientifc
enterprise.
In sum, stepping up our game in terms of methodological rigor and research integrity is both
necessary and in many ways doable, and must constitute a central focus as we move forward towards a
cumulative discipline that views psycholinguistic approaches to SLA as part and parcel of the broader
enterprise of psycholinguistics and cognitive science that seeks to understand the human capacity for
language across learning contexts and life stages.
Beyond Eurocentrism
Last but not least, it is notable that almost all research reviewed in the chapters in this section of the
handbook involved European languages and was conducted in Europe or North America (see Mon-
tanari & Nicoladis, 2016, for a similar concern about bilingualism research more generally). Thus,
while multilingualism is obviously a global phenomenon, its study from psycholinguistic and cogni-
tive perspectives does not appear to have been a globally inclusive enterprise so far.
The reasons for this Eurocentric focus are not difcult to explain: Psycholinguistic experiments
typically require participants to come to a laboratory on a university campus with specialized and
often expensive equipment. The efects that are of interest in psycholinguistic experiments are typi-
cally subtle and small, involving diferences in reaction times or electrophysiological responses on a
millisecond scale. Analysis of such data requires generalization over participants and items; thus there
is a need for a substantial number of participants and items in each experiment. Preparing materials
for such experiments in languages that the principal investigator is not fuent in presents an addi-
tional hurdle in an already time- and resource-intensive process. The same is true for the recruitment
and testing of participants who are speakers of languages that are not widely spoken and learned
in the vicinity of well-equipped research laboratories. The challenges are obvious, but so are the
104
Synthesis
consequences: Our current understanding of language processing and learning across the lifespan is
limited largely to well-educated speakers of two (or more) typologically related European languages
in the Global North. This raises questions not just about inclusiveness and social justice (Ortega,
2019), but about the empirical foundations of our theoretical models. Seminal models of the bilin-
gual mental lexicon, for example, were built on data from Dutch-German, Dutch-English, and
Spanish-Catalan bilinguals, with processing of cross-language cognates playing a critical role. How
informative are these models for understanding lexical access in speakers of typologically unrelated
languages with few or no cognates? Similarly, prediction in L2 sentence processing has been investi-
gated largely in the context of grammatical gender, a relatively exotic linguistic phenomenon from a
global perspective. How do our insights about prediction in L2 processing generalize beyond gram-
matical gender? Extending future psycholinguistic SLA research more substantially beyond speakers
of closely related European languages will force us to expand experimental investigations to a wider
range of linguistic phenomena, which in turn will provide more solid empirical foundations for the
development of theoretical models of bilingual processing.
Indeed, the authors of several chapters point to increasing the diversity of languages and learn-
ers represented in L2 psycholinguistic research as an important future direction for the feld. Yet
for these wishful statements to turn into action, we need to begin by acknowledging the practical
challenges this will involve and the concrete actions that will be required on the part of those with
access to resources and knowledge to overcome them. Specifcally, the inclusion of underrepresented
languages in psycholinguistic research will require that those with access to resources make proactive
eforts to collaborate with and include researchers who are speakers of those languages and members
of the communities in which they are spoken and learned. Working with languages that the principal
investigator is not fuent in presents yet another step outside one’s comfort zone, but if statements
about increasing the diversity in bilingualism research are to go beyond paying lip service to an ideal,
it will be an inevitable one.
Conclusion
As refected in the seven chapters in this section of the handbook, the inclusion of methodological
and theoretical tools from psycholinguistics and cognitive science into SLA research has led to excit-
ing new insights and perspectives on language processing and learning in diferent learning contexts
and at diferent stages of life. In this synthesis chapter, I have tried to outline some of the directions to
take if we want these insights on language processing in individual learner types to contribute towards
a more comprehensive and integrative understanding of language development across learning con-
texts and across the life span. In particular, I have highlighted a need for increased exchange between
SLA subfelds focused on specifc learner types, as well as for stepping beyond the (perceived and
crumbling) boundaries between SLA and “mainstream” psycholinguistics. Contributions to these
endeavors can come from comparisons between diferent types of language users with the goal of
testing models developed for one group for their ability to scale up to data from the other. Such
comparisons have the potential to uncover generalizable properties of human language development,
and to reveal properties that may be uniquely associated with language learning under specifc cir-
cumstances (e.g., monolingualism). Another approach to broadening our investigations of language
development across subfelds and learner types is to set aside categorical classifcations of learners
altogether and instead include environmental and experiential factors as (potentially continuous)
individual-diference variables, thus acknowledging the substantial variability that typically still exists
within traditionally defned learner types (including monolinguals).
Finally, I have sought to highlight the benefts of taking concrete steps towards becoming a
cumulative quantitative discipline that values collaboration, transparent reporting and data sharing,
and moving beyond easily accessible languages and populations to include more diverse multilingual
105
Theres Grüter
participants (and researchers) representative of languages and language users worldwide. Meeting
these challenges will require us to step beyond our individual comfort zones, linguistically, geograph-
ically, socially, methodologically, and epistemologically, and build on the insights we have gained so
far on psycholinguistic processes in various individual learner types and attempt to integrate and
broaden these insights to arrive at a more comprehensive understanding of psycholinguistic develop-
ment across learning contexts and across the lifespan.
References
Bice, K., & Kroll, J. F. (2019). English only? Monolinguals in linguistically diverse contexts have an edge in
language learning. Brain and Language, 196, 104644.
Bolibaugh, C., Vanek, N., & Marsden, E. J. (2021). Towards a credibility revolution in bilingualism research:
Open data and materials as stepping stones to more reproducible and replicable research. Bilingualism: Lan-
guage and Cognition, 24(5), 801–806.
Brysbaert, M. (2021). Power considerations in bilingualism research: Time to step up our game. Bilingualism:
Brysbaert, M., & Stevens, M. (2018). Power analysis and efect size in mixed efects models: A tutorial. Journal
of Cognition, 1(1), 9.
Byers-Heinlein, K., Tsui, A. S. M., Bergmann, C., Black, A. K., Brown, A., Carbajal, M. J., Durrant, S.,
Fennell, C. T., Fiévet, A.-C., Frank, M. C., Gampe, A., Gervain, J., Gonzalez-Gomez, N., Hamlin, J. K.,
Havron, N., Hernik, M., Kerr, S., Killam, H., Klassen, K. . . ., & Wermelinger, S. (2021). A multilab study
of bilingual infants: Exploring the preference for infant-directed speech. Advances in Methods and Practices in
Psychological Science, 4(1), 1–30.
Chang, F., Dell, G. S., & Bock, K. (2006). Becoming syntactic. Psychological Review, 113(2), 234–272.
Cook, V. (1999). Going beyond the native speaker in language teaching. TESOL Quarterly, 33, 185–209.
Cox, J.C. & Sanz, C. (2023). Learning a new language in older age: psycholinguistic and crossdisciplinary
approaches. In A. Godfroid & H. Hopp (Eds.), The Routledge handbook of second language acquisition and psy-
cholinguistics (pp. 85–96). Routledge.
Cumming, G. (2014). The new statistics: Why and how. Psychological Science, 25(1), 7–29.
Davies, A. (2003). The native speaker: Myth and reality. Multilingual Matters.
De Wilde, V., Brysbaert, M., & Eyckmans, J. (2020a). Learning English through out-of-school exposure: How do
word-related variables and profciency infuence receptive vocabulary learning? Language Learning, 70(2), 349–381.
De Wilde, V., Brysbaert, M., & Eyckmans, J. (2020b). Learning English through out-of-school exposure. Which
levels of language profciency are attained and which types of input are important? Bilingualism: Language and
Cognition, 23(1), 171–185.
Dussias, P.E. (2023). Synthesis: transdisciplinary approaches to understanding variability in bilingual processing
and use. In A. Godfroid & H. Hopp (Eds.), The Routledge handbook of second language acquisition and psycholin-
guistics (pp. 268–280). Routledge.
Francis, W. S. (2021). Introduction to the volume. In W. S. Francis (Ed.), Bilingualism across the lifespan: Opportu-
nities and challenges for cognitive research in a global society (pp. 1–6). Routledge.
Frank, M. C., Bergelson, E., Bergmann, C., Cristia, A., Floccia, C., Gervain, J., Hamlin, J. K., Hannon, E. E.,
Kline, M., Levelt, C., Lew-Williams, C., Nazzi, T., Panneton, R., Rabagliati, H., Soderstrom, M., Sullivan,
J., Waxman, S., Yurovsky, D., & Yurovsky, D. (2017). A collaborative approach to infant research: Promoting
reproducibility, best practices, and theory-building. Infancy, 22(4), 421–435.
González Alonso, J. & Rothman, J. (2023). The psycholinguistics of L3/Ln acquisition. In A. Godfroid & H.
Grüter, T., & Conradie, S. (2006). Investigating the L2 initial state: Additional evidence from the production and
comprehension of Afrikaans-speaking learners of German. In R. Slabakova, S. Montrul, & P. Prévost (Eds.),
Inquiries in linguistic development: In honor of Lydia White (pp. 89–114). John Benjamins.
Grüter, T., Lau, E., & Ling, W. (2020). How classifers facilitate predictive processing in L1 and L2 Chinese: The
role of semantic and grammatical cues. Language, Cognition and Neuroscience, 35(2), 221–234.
Gullifer, J. W., & Titone, D. (2020). Characterizing the social diversity of bilingualism using language entropy.
Hartshorne, J. K., Tenenbaum, J. B., & Pinker, S. (2018). A critical period for second language acquisition:
Evidence from 2/3 million English speakers. Cognition, 177, 263–277.
Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest people in the world? Behavioral and Brain Sci-
ences, 33(2–3), 61–135.
106
Synthesis
Hopp, H. (2021). Prediction and grammatical learning in second language sentence processing. In E. Kaan & T.
Grüter (Eds.), Prediction in second language processing and learning (pp. 168–185). John Benjamins.
Hurtado, N., Grüter, T., Marchman, V. A., & Fernald, A. (2014). Relative language exposure, processing efciency
and vocabulary in Spanish—English bilingual toddlers. Bilingualism: Language and Cognition, 17(1), 189–202.
Hyltenstam, K., & Obler, L. K. (1989). Bilingualism across the lifespan: An introduction. In K. Hyltenstam &
L. K. Obler (Eds.), Bilingualism across the lifespan: Aspects of acquisition, maturity, and loss (pp. 1–12). Cambridge
University Press.
Jackson, C.N. (2023). The psycholinguistics of adult second language acquisition. In A. Godfroid & H. Hopp
Jackson, C. N., & Hopp, H. (2020). Prediction error and implicit learning in L1 and L2 syntactic priming.
International Journal of Bilingualism, 24(5–6), 895–911.
Kaan, E. (2014). Predictive sentence processing in L2 and L1: What is diferent? Linguistic Approaches to Bilingual-
ism, 4(2), 257–282.
Kaan, E., & Grüter, T. (2021). Prediction in second language processing and learning: Advances and directions.
In E. Kaan & T. Grüter (Eds.), Prediction in second language processing and learning (pp. 2–24). John Benjamins.
Lew-Williams, C. (2017). Specifc referential contexts shape efciency in second language processing: Three
eye-tracking experiments with 6- and 10-year-old children in Spanish immersion schools. Annual Review of
Applied Linguistics, 37, 128–147.
Montanari, S., & Nicoladis, E. (2016). Concluding remarks and future directions. In E. Nicoladis & S. Mon-
tanari (Eds.), Bilingualism across the lifespan: Factors moderating language profciency (pp. 325–330). American
Psychological Association and Walter de Gruyter.
Montrul, S. (2023). The psycholinguistics of heritage language acquisition. In A. Godfroid & H. Hopp (Eds.),
Morgan-Short, K., Marsden, E., Heil, J., Issa, B., Leow, R. P., Mikhaylova, A., Mikołajczak, S., Moreno, N.,
Slabakova, R., & Szudarski, P. (2018). Multi-site replication in SLA research: Attention to form during lis-
tening and reading comprehension in L2 Spanish. Language Learning, 68(2), 392–437.
Ortega, L. (2013). SLA for the 21st century: Disciplinary progress, transdisciplinary relevance, and the bi/mul-
tilingual turn. Language Learning, 63(S1), 1–24.
Ortega, L. (2019). SLA and the study of equitable multilingualism. The Modern Language Journal, 103(S1), 23–38.
Pakulak, E., & Neville, H. J. (2010). Profciency diferences in syntactic processing of monolingual native speak-
ers indexed by event-related potentials. Journal of Cognitive Neuroscience, 22(12), 2728–2744.
Plonsky, L. (2013). Study quality in SLA: An assessment of designs, analyses, and reporting practices in quantita-
tive L2 research. Studies in Second Language Acquisition, 35(4), 655–687.
Potter, C.E. & Lew-Williams, C. (2023). The psycholinguistics of early bilingualism. In A. Godfroid & H. Hopp
Prior, A., & van Hell, J. G. (2021). Including bilingual participants in cognitive research. In W. S. Francis (Ed.), Bilin-
gualism across the lifespan: Opportunities and challenges for cognitive research in a global society (pp. 47–65). Routledge.
Puig-Mayenco, E., González Alonso, J., & Rothman, J. (2020). A systematic review of transfer studies in third
language acquisition. Second Language Research, 36(1), 31–64.
Rampton, M. B. H. (1990). Displacing the “native speaker”: Expertise, afliation, and inheritance. ELT Journal,
44(2), 97–101.
Schimke, S. (2023). The psycholinguistics of child second language acquisition. In A. Godfroid & H. Hopp
Schmid, M.S., Soto, C., & Heimann, J. (2023). The psycholinguistics of frst language attrition. In A. Godfroid &
H. Hopp (Eds.), The Routledge handbook of second language acquisition and psycholinguistics (pp. 61–72). Routledge.
Schwartz, B. D., & Sprouse, R. A. (2021). The full transfer/full access model and L3 cognitive states. Linguistic
Tiv, M., Kutlu, E., & Titone, D. (2021). Bilingualism moves us beyond the ideal speaker narrative in cognitive
psychology. In W. S. Francis (Ed.), Bilingualism across the lifespan: Opportunities and challenges for cognitive research
in a global society (pp. 29–46). Routledge.
van der Slik, F., Schepens, J., Bongaerts, T., & van Hout, R. (2022). Critical period claim revisited: Reanalysis
of Hartshorne, Tenenbaum, and Pinker (2018) suggests steady decline and learner-type diferences. Language
Learning, 72(1), 87–112.
Westergaard, M. (2021). Microvariation in multilingual situations: The importance of property-by-property
acquisition. Second Language Research, 37(3), 379–407.
White, L. (2021). Choosing between theories: Can L3 acquisition really decide? Second Language Research, 37(3),
429–433.
107
SECTION II
Research methods in second

language psycholinguistics
10
STUDYING SECOND LANGUAGE
COMPREHENSION
Introduction
Second language comprehension includes the understanding of written and spoken language and
covers a range of component processes, from the perception of orthographic or phonological units,
through the recognition of words, to the interpretation of sentences and discourse structures. The
focus of this chapter is on experimental research methods used to explore the recognition of words as
the basic building blocks of language, and the parsing and understanding of sentences. For approaches
to the study of second language production, see Bernolet (2023 [this volume]).
For skilled readers and listeners, real-time comprehension is normally seamless, and researchers
need to probe the processes of comprehension through careful manipulation of experimental materi-
als. Increasingly, time-sensitive techniques have been used to tap into comprehension processes as
they occur. Although comprehension of a second language (L2) may be more efortful, the same
psycholinguistic techniques are generally applicable.
Textbox 10.1 Key methods
Reaction time (RT) studies measure the time taken by participants to register an overt response (e.g., a
button-press or the onset of a voice response).
Priming paradigms involve responses to one stimulus (the target) preceded (or, in backward priming, followed)
by a related stimulus (the prime); the relationship typically involves either meaning or form.
Eye-movement studies record participants’ eye movements during reading or listening.
Event-related potential (ERP) studies record electrophysiological brain responses to visually or auditorily
presented stimuli.
Methods and paradigms

The experimental methods measure efects of systematic manipulations of independent variables on
dependent (outcome) variables, while controlling for confounding efects of other variables. Since the cog-
nitive processes involved in comprehension are not directly observable, researchers draw inferences
DOI: 10.4324/9781003018872-12 111

about the nature and time-course of these cognitive processes using proxy outcome measures, the most
common of which is RT. RTs are used in studies of sub-lexical (letter/phone identifcation), word-
level (formal-lexical, lexical semantic, morphological), and sentence-level (syntactic) representations
and their processing (see also Hamrick, 2023 [this volume]). RT diferences between experimental
conditions indicate how theoretically and empirically motivated predictors afect comprehension.
Further temporal measures not requiring an overt response include the duration of eye fxations
during reading in eye-movement research, or the time-course of electrophysiological brain responses
following a visual or auditory stimulus in ERP studies. Other measures include response accuracy,
motor behavior tracking, physiological responses (e.g., pupil dilation as a measure of processing efort
and attention), or patterns of blood fow in brain-imaging studies.
In the next sections we consider key aspects of time-sensitive experimental methods used to
study comprehension (see Textbox 10.1). Selecting the right methodology is critical for making
inferences about the phenomena in question, e.g., whether semantic representations of L2 words
are activated automatically in visual input processing (Silverberg & Samuel, 2004), or whether L1
and L2 readers use the same or diferent mechanisms in morphological processing (Clahsen et al.,
2010) or syntactic ambiguity resolution (Clahsen & Felser, 2006). The diferent paradigms will be
described in some detail when they are frst introduced in the next section. Some are more readily
employed in word recognition research, and some more in sentence comprehension research. Many
can be used in both.
Recognition and comprehension of words
Lexical decision
Probably the most widely used behavioral paradigm is the lexical decision task (LDT). It is used in
visual or auditory modalities, either by itself or in combination with other methods, such as priming
and ERP (see below). In LDTs, participants decide whether a stimulus is a word, and their RT and
accuracy are recorded. LDTs are easy to administer; they can be conducted in and out of the lab (e.g.,
on a laptop computer) and can be administered remotely, over the internet. LDTs have been used
extensively in L1, L2, and bilingual studies to understand which word characteristics afect lexical
processing. Results from LDTs provide insight into how words are represented and/or organized in
memory.
One of the most robust fndings from LDTs is the word frequency efect, i.e., more frequent
words are identifed more quickly. Other lexical characteristics that afect speed (and accuracy)
include word length, number of orthographic and phonological neighbors, age of acquisition, famil-
iarity (based on untimed ratings), and prevalence (how widely the word is known in the population).
RTs are also afected by semantic variables, such as concreteness, imageability, and meaningfulness
(number of dictionary meanings). Many of these efects are stronger with less skilled than more
skilled readers and listeners; for example, frequency efects tend to be larger in L2 than L1 processing
(Brysbaert et al., 2017). This is because language processing is less automatic for less skilled language
users. However, other processing diferences can have an impact as well, such as the nature of the
morphological analysis of complex words in L1 and L2 (Neubauer & Clahsen, 2009).
LDTs are used extensively in SLA research to study how multiple languages are represented and
processed (see also Conklin & Thul, 2023 [this volume]). LDTs with cognates (words similar in form
and meaning in L1 and L2, e.g., ring or winter in English and Dutch), interlingual homographs (similar
spellings but diferent meanings, e.g., angel, which means “sting” in Dutch), and homophones (similar
pronunciation but diferent meanings, e.g., Dutch paar “couple” and French part “leave/part”) have
shown that the non-target language is activated during target language processing (Dijkstra et al.,
1999). Cognates consistently facilitate word recognition for profcient bilinguals, even when the task
112
Studying second language comprehension
is completely in the L1 (van Hell & Dijkstra, 2002). Processing efects for interlingual homographs
range from facilitation to inhibition to “no efect,” depending on the task used and the salience of
the non-target language (Dijkstra & van Heuven, 2002). For example, in a study with Dutch learn-
ers of English, van Heuven et al. (2008) found interference efects for interlingual homographs in
an L2 LDT but not in a generalized LDT (in which participants had to respond “yes” to both L1
and L2 words). Interestingly, the same study showed enhanced neurological activity for homographs
compared to language-unique control words, under both experimental conditions. This result shows
the value of combining behavioral and neurological evidence in SLA research.
In SLA research, LDTs need to be tailored to the L2 population, since L2 lexical representations
are frequently less precise (see also Gor, 2023 [this volume]) and lexical processing is less automatic
(Segalowitz, 2000). An alternative RT task is a go/no-go procedure in which participants only
respond if the stimulus is of a particular type, e.g., a word. This task removes the need to register
a response for each stimulus, while retaining the stimulus identifcation component (Gómez et al.,
2007), and may facilitate a focus on words for developing readers (Moret-Tatay & Perea, 2011). This
task may also be useful in L2 lexical decision studies with lower-profciency participants.
RT studies may involve a speed-accuracy trade-of, with participants responding more slowly
when pursuing accuracy. Such trade-ofs may depend on language ability (Draheim et al., 2019).
Accuracy is generally high in L1 studies, and the small number of incorrect responses are normally
excluded from RT analyses. In L2 studies, the number of incorrect trials is often too high for this
approach, and so response accuracy is included as an additional predictor in the RT analysis. It is also
useful to report error rates and analyze error data alongside RTs in L2 studies, as indicators of the
precision and specifcity of lexical representations (Elgort, Candry, et al., 2018).
Semantic decision
There are various semantic decision paradigms. In semantic categorization, participants decide whether
a word refers to a member of a specifed category, e.g., whether car is a vehicle. In semantic relatedness,
participants decide whether pairs of expressions are related in meaning and, in sense judgments, they
decide whether they ft together, i.e., “make sense” (e.g., wrapping paper paired with shredded paper
or with liberal paper). Such decisions cannot be made without accessing and processing in depth the
meaning senses of the stimulus. This more advanced processing of meaning results in slower responses
to semantically ambiguous items in semantic decisions than in LDTs (Hino et al., 2006).
Semantic decisions (often combined with priming) may reveal how L1 and L2 semantic repre-
sentations are stored and accessed. L2–L1 translation priming studies, for instance, show a diference
in the nature of lexical and semantic decisions, with semantic priming obtained in semantic but not
lexical decisions (Finkbeiner et al., 2004). This is because L2 lexical semantic representations may be
less specifed and, therefore, require a task more narrowly focused on semantic processing in order to
efectively prime L1 translation equivalents.
Semantic decision tasks have also been used in SLA research to study efects of profciency and
L1/L2 diferences on the storage and access of L2 word meanings. Participants show faster and more
efcient semantic processing in L1 than in L2, and highly profcient bilinguals show higher process-
ing automaticity than less profcient bilinguals (Phillips et al., 2004). Higher processing efciency in
more profcient bilinguals was also observed in a follow-up L2 semantic categorization task, using the
event related potentials N400 component (see below).
Priming
Priming involves presentation of two (or more) successive stimuli: a prime and a target, with the
prime creating a context for the processing of the target. Researchers are interested in how processing
113
of the target varies as a function of its relationship (e.g., orthographic, phonological, syntactic, or
semantic) with the prime. Primes can support the processing of related targets (by activating repre-
sentational knowledge of targets, e.g., doctor primes nurse but bread does not), leading to facilitation of
responses compared with an unprimed condition, or it may disrupt it, leading to inhibition. Priming
is often combined with an overt response task, such as lexical decision, semantic decision, or naming
(reading the target aloud). Priming in SLA research can involve primes and targets in the learners’
diferent languages, e.g., a target may be preceded by its translation equivalent (chair preceding silla,
meaning chair in Spanish).
In form-priming experiments with skilled readers, lexical inhibition is observed between ortho-
graphic neighbors (words difering by one letter, e.g., angel—ANGER) both within-language
(Davis & Lupker, 2006) and across a bilingual’s languages (Bijeljac-Babic et al., 1997; van Heuven
et al., 2001). In semantic priming with monolinguals and profcient bilinguals, on the other hand,
facilitation is observed for targets preceded by related primes (dog—cat), compared to a baseline con-
dition (table—cat) (Favreau & Segalowitz, 1983; Frenck-Mestre & Prince, 1997; Hutchison, 2003).
A variant of priming, masked priming, is often used in studies of lexical access because it taps into
early stages of word processing. The prime is presented for a very short time (under 0.1 of a second)
and is preceded (and, sometimes, followed) by a mask (such as a string of hashes, ####). Participants
may be completely unaware of the presence of the prime; therefore, masked priming dramatically
reduces the use of consciously controlled strategies when responding to the target word and encour-
ages automated processing. Masked priming experiments provide a window into implicit lexical
knowledge and processing that can be used to test and fne-tune theoretical and computational mod-
els of L1, L2, and bilingual memory.
Masked priming has been used extensively to investigate activation of the non-target language
during target-language processing (see also Conklin & Thul, 2023 [this volume]). Using cross-lan-
guage masked homophone priming, Brysbaert et al. (1999) showed that both L1 and L2 phonologi-
cal codes are accessed by Dutch-English bilinguals, even when participants were unaware of the L2
primes and believed that they were engaged in a monolingual L1 task. Masked translation-equivalent
priming has even been observed in lexical and semantic decision tasks involving an L1 and L2 that
do not share a script (e.g., Gollan et al., 1997, for Hebrew and English).
In morphological masked priming, prime and target are morphologically related, e.g., regular
(walk-walked) or irregular infections (send-sent), or derived word forms (govern-government). Using
unmasked priming, Clahsen and Neubauer (2010) found evidence suggesting that both L2 and L1
participants accessed whole-word representations, which militated against a morphological decom-
position hypothesis. Using masked priming, however, they found reliable morphological stem-
priming efects for L1 but not L2 participants, suggesting that only L1 speakers analyze these complex
forms as stem and afx.
In SLA studies, canonical priming efects refect ease of access to L2 representations and acces-
sibility of L2 knowledge (Godfroid, 2020b). Such priming efects, however, are more likely with
participants who started learning the L2 earlier in life (Silverberg & Samuel, 2004). When the
quality of lexical knowledge is low, form-priming may result in facilitation instead of inhibi-
tion (Andrews & Hersch, 2010). When lexical semantic representations are weak or poorly inte-
grated, semantic priming studies show inhibition rather than facilitation (Dagenbach et al., 1990).
Priming experiments can also be used to test and select the most efective learning and teaching
approaches (Elgort, 2011). However, obtaining masked priming efects in L2 research sometimes
requires modifying the canonical paradigms, e.g., accommodating slower L2 processing by increas-
ing the interval between prime and target (the SOA or stimulus onset asynchrony). Importantly,
masked priming may be less robust (or may not be observed) in L2 processing, even with known
words (Qiao & Forster, 2017).
114
Eye-movement research
Eye-movement methodology has made a unique contribution to the study of L1 and L2 processing
and comprehension in real time (Staub & Rayner, 2007). It has been used to study various efects
(e.g., word length, text spacing, word frequency, predictability, various linguistic constraints) in L2
comprehension with millisecond precision (Keating, 2014), as well as the efects of individual difer-
ences on L2 acquisition and processing (e.g., Godfroid et al., 2013). A key advantage of using eye-
movement techniques in studies investigating spoken and written comprehension is that they provide
a window into cognitive processes taking place naturally, as participants read or listen to L2 input.
Eye-movements in reading consist of saccades (or jumps) separated by fxations, during which the
reader extracts and encodes information from the text. It is assumed that the duration and number of
fxations on a region of interest in a written text refect difculty of processing. Eye-tracking allows
researchers to study real-time comprehension processes noninvasively under conditions close to natu-
ral reading, and modern portable eye-tracking equipment allows in-situ data collection (for example,
in schools or libraries). Diferent stages of cognitive processing have been associated with diferent
eye-movement measures (see Godfroid, 2020a, for a review).
Eye-tracking provides insights not available from ofine measures. For instance, Clahsen et al.
(2013) compared the processing of English morphological constraints by native speakers and advanced
Dutch learners of English. Regular English plurals, such as mites, should not feed compounding or
derivation (*mitesless) and such derived words should therefore be processed more slowly than nouns
with a singular base (fealess) or an irregular plural base (liceless). Although L2 learners performed
native-like in an untimed grammaticality judgment task, only L1 readers showed an efect of the
morphological constraint in eye-movement measures.
Eye-movement research has contributed to a more nuanced understanding of cross-language lexi-
cal activation, such as interlingual homograph interference (Duyck et al., 2007). Thus, while naming
tasks (Schwartz & Kroll, 2006) have shown that a semantically constraining context can reduce cross-
linguistic interference (e.g. reading gossip in an English sentence reduces subsequent interference of
French chat (cat) on English cat), eye-tracking studies have allowed researchers to locate this context
efect in late processing stages and show that cross-language activation is still present at the early pro-
cessing stages (Libben & Titone, 2009).
Eye-tracking has also been used to explore spoken language comprehension, through the visual
world paradigm. Eye fxations are measured as participants look at pictures or objects while listening
to speech. The earliest such study in SLA research showed activation of words in both L1 and L2
during a single-language task (Spivey & Marian, 1999). In this study, L1 Russian learners of English
saw a set of objects that included a stamp and a marker-pen, and their eye-movements towards these
objects were measured as they listened to a Russian instruction Poloji marku nije krestika “Put the
stamp (=marku) below the cross.” In this condition, participants’ eyes frequently moved briefy to
the interlingual distractor (the marker-pen) when hearing marku. Amongst its advantages for SLA
research, the visual world paradigm has high ecological validity, since participants are carrying out
a natural response to spoken instructions and typically do not have to press buttons or make gram-
maticality judgments; because this paradigm focuses on spoken language comprehension, it is also not
dependent on participants’ literacy skills.
In SLA research, longer total reading times on novel words have been associated with better con-
textual word learning outcomes (Godfroid et al., 2013; Pellicer-Sánchez, 2016). Eye-movement
experiments have been used to trace the developmental trajectories of diferent components of
L2 (English) word knowledge for Dutch university students by Elgort, Brysbaert, et al., (2018),
using diferent eye-movement measures associated with the component processes involved in word
comprehension during reading. They found that frst-fxation duration on unfamiliar words and
115
on high-frequency controls became similar within the frst fve to seven encounters in the text,
suggesting a rapid learning trajectory for orthographic form. A less rapid learning trajectory was
found for the learning of meanings, as measured by the summed fxation duration on the word
and regressive fxations on preceding context, taken to refect integration of the word with its
supporting context.
One of the limitations of eye-movement research is the cost of specialized equipment and
software. Another is extra time needed for calibration of the equipment for each participant, and
for analyzing the large amounts of data that result from the tracking of eye-movements. Further,
in L2 eye-tracking research in particular, a higher number of participants may be needed, due to
higher individual variability exhibited by L2 participants. SLA researchers also need to be aware
of possible pitfalls of associating individual eye-movement measures directly with specifc com-
ponent processes of L2 comprehension. It is critical for eye-movement studies to be guided by
theoretically or empirically motivated hypotheses because the same eye-movement measure may
be associated with more than one cognitive process (see Godfroid & Hui, 2020, for fve common
pitfalls of eye-tracking).
Event-related brain potentials

Another technique used to study the time course of L2 comprehension is event-related brain poten-
tials (ERP), whereby brain activity is recorded non-invasively using electrodes placed on the scalp.
This method ofers real-time measures of neural activation during language processing, includ-
ing sub-lexical, lexical, semantic, and syntactic processing in written and spoken comprehension.
Typically, the timing of changes in the electroencephalographic (EEG) signal is related to a lan-
guage event, such as the onset of a word. EEG signals are averaged across multiple experimental
trials, giving an ERP wave with positive and negative peaks, and latent components are associated
with diferent cognitive functions, such as perception, attention, and meaning processing. In L1
processing, for example, N400 (the letter N indicates the direction of the peak, the number 400
is the latency) is a negative-going wave that is commonly associated with semantic processing and
typically occurs between 300 and 500 milliseconds from the onset of a stimulus, while P600 is a
positive-going wave associated with morphosyntactic violation, syntactic complexity, and syntactic
ambiguity (for overviews of what the components of an ERP wave can show us, see Duñabeitia
et al., 2016; Kaan, 2007; Morgan-Short et al., 2015). Note though that in L2 and bilingual pro-
cessing, the latency, amplitude, morphology (i.e., wave width), and topography (the mapping of
electrical recordings over the scalp) of ERP components may difer from their L1 monolingual
counterparts (see Mueller, 2005, for a review of ERP correlates of L2 processing).
Like eye-tracking, ERP experiments do not require decision-based tasks. However, the ERP
measurements are commonly taken during behavioral tasks that are used to create conditions neces-
sary for critical comparisons (e.g., related/unrelated; grammatical/ungrammatical). Because aspects
of the behavioral responses, such as eye or hand movements, might afect the ERP trace, the behav-
ioral tasks are often modifed. For example, Elgort et al. (2015) reduced eye-movements by present-
ing L2 reading texts one word at a time at the same screen location, rather than unfolding across the
screen, and in a study comparing L1 and L2 sentence processing, Kotz et al. (2008) used a go/no-go
paradigm, whereby critical stimuli appeared in the no-go condition so they did not require the hand
movements associated with an overt response.
ERPs can be used to examine covert processes (i.e., processes that are not refected in explicit
behavioral responses that listeners or readers might exhibit). In bilingual research, ERP evidence
indicates cross-language activation in tasks requiring the use of only one language, even in the
absence of such evidence in RT data (van Heuven & Dijkstra, 2010). For example, Thierry and
Wu (2007) asked Chinese-English bilinguals to perform a semantic relatedness task with pairs of
116
related (post–mail) and unrelated (train–ham) English words. The Chinese translations (not shown to
participants) of some pairs contained a shared Chinese character (e.g., train, 火车, and ham, 火腿).
Although the shared character had no efect on the behavioral response, it was refected in ERP data.
Similarly, while Phillips et al. (2004) found semantic priming efects for both high and low L2 prof-
ciency participants, a corresponding N400 efect was only observed for high profciency participants.
One of the constraints of ERP research is the need for strong theoretical models for interpreting
results, because the same ERP component may be associated with more than one cognitive process,
particularly in L2 comprehension (e.g., Frenck-Mestre, 2005). Additionally, relatively high costs of
facilities and equipment and a steep technological learning curve make ERP less broadly accessible
to SLA researchers, and because ERP data are quite noisy, more stimuli are needed for ERP than for
behavioral experiments.
Comprehension of sentences
Many of the methods used to study word recognition and comprehension can also be used to research
sentence comprehension. Where methods are not further explained below, the reader should refer
to earlier sections.
Self-paced reading
Self-paced reading (SPR) is a straightforward task (Jegerski, 2014; Keating & Jegerski, 2015), in
which participants silently read text broken down into display segments (words or phrases). They ini-
tiate each display by pressing a button, and button press latencies provide a measure of the time taken
to process each display (i.e., segment). For instance, in L1 sentence comprehension research, slower
SPR times are found for text areas that contain syntactic and semantic anomalies. The assumption is
that anomalies interrupt normal, automatic sentence processing, increasing RT on or immediately
after the anomaly. The most widely used version of SPR is the noncumulative moving-window
paradigm, with dashes replacing each non-space character in the text, making the length of each
word visually obvious. Each button press hides the current word and reveals the next one. Although
SPR is closer to normal reading than decision tasks, it still requires the additional action of pressing
a button to display each word or phrase.
Jufs and Harrington (1995) were the frst SLA researchers to use SPR as a measure of performance
alongside grammaticality judgments as a measure of linguistic competence. Over recent decades the
range of L2 studies using SPR has increased dramatically, e.g., investigating L1 efects on L2 process-
ing of relative clause ambiguities (Havik et al., 2009), and testing the efects of learning conditions
on the acquisition and processing of L2 grammatical structures (Bordag et al., 2019), words (Elgort
et al., 2020), and multi-word expressions (Obermeier & Elgort, 2021).
A challenge for SPR as a measure of L2 sentence processing, however, is that L2 readers may fnd
word recognition and sentence reading generally more efortful than L1 readers, and such efects may
swamp syntactic processing diferences (see Marsden, Thompson et al., 2018 for further discussion).
Cross-modal priming
In cross-modal priming (CMP) the prime is presented in one modality (e.g., spoken) and the tar-
get in the other (visual, either text or a picture). This is often used to explore aspects of spoken
sentence processing, such as the activation of lexical, semantic, and syntactic information during
processing. One fnding is that L2 participants maintain activation of words encountered in a
sentence for longer than L1 participants (e.g., Felser & Roberts, 2007), which probably refects
diferences in the speed with which the meanings of words are integrated with the sentence
117
meaning in L2 compared to L1. The CMP paradigm is used in bilingual lexical ambiguity resolu-
tion (Heredia & Cieślicka, 2020) to assess activation of multiple languages in bilingual processing
(see also Marinis, 2018).
Eye movements and sentence processing

Because eye-tracking is a time-course-sensitive technique, it has been used to investigate aspects of
L2 sentence processing such as anomaly detection, ambiguity resolution, and the establishment of
syntactic dependencies. For a review of the use of eye-tracking to measure L2 sentence processing,
see Frenck-Mestre (2005). In one example of its use in L2 processing, Felser et al. (2012) used early
(frst-pass) and late (second-pass or rereading) measures of reading time to get a more detailed picture
of the relative timing of diferent types of processing in L2 versus L1 comprehension. Diferences in
the patterning of earlier and later eye-movement measures in the study were key to uncovering the
reliance on semantic processing in L2 rather than on structural preferences found in L1.
Event-related brain potentials

ERP research has been used to investigate whether L1 and L2 processing are qualitatively or quantita-
tively diferent (Hahne & Friederici, 2001) and to understand the efect of target language profciency
on comprehension. Weber-Fox and Neville (1996) investigated the efect of age of L2 acquisition
(AoA) for Chinese-English bilinguals on acceptability judgments. They found that all participants dis-
played an N400 efect in response to semantic anomalies, although it was delayed for participants who
learned English after the age of 11–13 years. Conversely, for phrase structure violations, the canonical
ELAN (early left anterior negativity that normally peaks at less than 200 ms after presentation of a vio-
lation) efect was not observed for L2 participants regardless of their AoA; only early bilinguals showed
this efect in the 300–500 millisecond window, and the P600 efect was present for early but not late
learners. More recent studies suggest that L2 profciency (rather than AoA) may be the primary reason
for these diferences. Kotz (2009), Morgan-Short et al. (2015), and Steinhauer (2014) provide useful
critical reviews of ERP evidence on L2 syntactic processing; see also Duñabeitia et al. (2016) on the
use of electrophysiological measures of sentence processing by multilingual readers.
ERP experiments have also been used to evaluate the efect of diferent learning contexts. Batter-
ink and Neville (2013) compared ERP responses to novel L2 syntactic structures learned under either
explicit or implicit instructions. After training, participants performed grammaticality judgments on
visually presented sentences while their ERP responses were recorded. Explicit instructions resulted
in superior explicit judgments of grammaticality. Participants’ P600 responses to morphosyntactic
violations were modulated by their gained L2 profciency; participants who produced more accurate
behavioral responses also showed the P600 efect, in both implicit and explicit treatment groups. In
contrast, Morgan-Short et al. (2012) reported diferent neurocognitive response patterns to syntactic
violations in explicit and implicit learning conditions. This important line of SLA research is still
in its early stages and further ERP work on explicit and implicit learning and instruction is clearly
warranted.
Innovations and future directions
Issues in experimental psycholinguistic research with L2 participants

Because many component processes in L1 comprehension are efortless and automated, estab-
lished priming efects are easy to replicate in new L1 studies, and efects of semantic ambiguity
118
and syntactic anomaly in sentence processing/comprehension are relatively straightforward to

interpret. In L2 comprehension, however, fuency and automaticity of processing cannot be
assumed, which impacts our interpretation of efects observed and may also give rise to null
efects due to increased variability with L2 participants. It is prudent, therefore, to verify that the
efect of interest can be observed with known L2 items with the target L2 population, and to use
multiple measures that ofer diferent types of evidence. It is also critical to pilot experimental
procedures from L1 research with L2 participants, to confrm the time-course of the expected
efect and adjust the experimental settings accordingly. Higher numbers of participants and items
are also likely needed in L2 experimental research due to higher L2 processing variability (see
Textbox 10.2).
Future directions and challenges

A rapid increase in the use of psycholinguistic research paradigms to investigate questions in L2
learning and processing creates possibilities for innovative research. One exciting development is
open-science approaches to conducting SLA research. In psycholinguistic research, adopting existing
experimental paradigms and stimuli is a widely accepted and well-regarded practice because it afords
verifcation, extension, and meta-analysis of previous research fndings. Until recently, this practice
has not existed in SLA research to the same extent (see, e.g., Marsden, Morgan-Short et al., 2018).
Important steps towards such open practices are being made, as evidenced by increased availability of
materials and data on publicly accessible fle sharing sites, such as Open Science Framework (OSF) and
Digital repository of instruments and materials for research into second languages (IRIS, Marsden et al., 2016),
and the introduction by some key journals (e.g., Language Learning, Bilingualism: Language and Cogni-
tion) of registered reports, in which the literature review motivating a study, its methods, and proposed
analyses are peer-reviewed prior to data collection.
A key challenge is increasing the volume of in-situ SLA studies that use the new research
tools provided by psycholinguistic methods (see also Sachs, Baralt, & Gurzynski-Weiss, 2023
[this volume]). Compared with laboratory research, classroom-based studies investigating self-
directed contextual L2 learning do not give the same degree of control of experimental con-
ditions, stimuli, or procedures; also, true randomized assignment of participants to conditions
is often not possible or ethical. Yet, such in-situ studies ofer unique opportunities for more
ecologically valid tests of frameworks and hypotheses developed in laboratory studies, measur-
ing efect sizes of predictor variables in real-life settings. One way of doing this is to combine
naturalistic learning scenarios with highly controlled testing. For example, in a contextual L2
word learning study, Elgort and Warren (2014) gave participants a book to read at home follow-
ing a recommended reading schedule; to ascertain the development of tacit formal and semantic
word knowledge acquired through reading, participants were tested using priming experiments
in controlled laboratory conditions. Sachs et al. (2023 [this volume]) review tasks and methods
suitable in classroom-based psycholinguistic research. Another challenge is to carefully calibrate
equipment and software used for data collection, following benchmarks and standards appropri-
ate for L2 studies, and to report relevant settings in published research (see Hamrick, 2023 [this
volume] for a discussion in the context of reaction time).
This chapter has provided an overview of psycholinguistic methods and experimental paradigms
used to study real-time L2 comprehension processes in SLA and bilingual research. Although we
have had to limit discussion to most common methods used at the word and sentence levels, we
have aimed to provide sufcient information to assist SLA researchers in selecting and deploy-
ing psycholinguistic research techniques appropriate for the study of component processes in L2
comprehension.
119
What adjustments need to be made to stimulus display duration, SOA, and response deadline in behav-
ioral L2 tasks?
How many participants and items per experimental condition do researchers need to have in experimen-
tal L2 studies?
What experimental paradigms are most efective with L2 participants, who are more likely to engage
task-related metacognitive strategies?
How can researchers verify that an ERP efect observed in response to L2 stimuli is the same as the corre-
sponding L1 efect, if the latency, location, and shape of the ERP response are diferent in L2 and L1?
Further reading
Duñabeitia, J., Dimitropoulou, M., Molinaro, N., & Martin, C. (2016). The electrophysiology of the bilingual
brain. In R. R. Heredia, J. Altarriba, & A. B. Cieślicka (Eds.), Methods in bilingual reading comprehension research
(pp. 265–312). Springer.
A summary of key ERP studies investigating the time course of the cognitive processes underlying bilingual
comprehension.
guide. Routledge.
A methodological guide for designing, conducting, and interpreting eye-tracking research in L2 studies.
Jegerski, J., & VanPatten, B. (Eds.), (2014). Research methods in second language psycholinguistics. Routledge.
An overview of methods, designs, and techniques used in psycholinguistics research with L2 learners, includ-
ing SPR, eye-tracking, ERPs, LDTs, and priming.
Jiang, N. (2012). Conducting reaction time research in second language studies. Routledge.
An introduction and practical guide for conducting reaction time research in L2 studies.
McDonough, K., & Trofmovich, P. (2009). Using priming methods in second language research. Routledge.
An introduction to the use of auditory, semantic, and syntactic priming in L2 research.
References
Andrews, S., & Hersch, J. (2010). Lexical precision in skilled readers: Individual diferences in masked neighbor
priming. Journal of Experimental Psychology: General, 139, 299–318.
Batterink, L., & Neville, H. (2013). Implicit and explicit second language training recruit common neural
mechanisms for syntactic processing. Journal of Cognitive Neuroscience, 25, 936–951.
Bernolet, S. (2023). Studying second language production. In A. Godfroid & H. Hopp (Eds.), The Routledge
Bijeljac-Babic, R., Biardeau, A., & Grainger, J. (1997). Masked orthographic priming in bilingual word recog-
nition. Memory and Cognition, 25, 447–457.
Bordag, D., Kirschenbaum, A., Rogahn, M., Opitz, A., & Tschirner, E. (2019). Misbehaved masculines: Inci-
dental acquisition of grammatical gender in L2 German during reading. Bilingualism: Language and Cognition,
22, 130–141.
Brysbaert, M., Lagrou, E., & Stevens, M. (2017). Visual word recognition in a second language: A test of
the lexical entrenchment hypothesis with lexical decision times. Bilingualism: Language and Cognition, 20,
530–548.
Brysbaert, M., Van Dyck, G., & Van de Poel, M. (1999). Visual word recognition in bilinguals: Evidence from
masked phonological priming. Journal of Experimental Psychology: Human Perception and Performance, 25, 137–148.
120
Clahsen, H., Balkhair, L., Schutter, J.-S., & Cunnings, I. (2013). The time course of morphological processing
in a second language. Second Language Research, 29, 7–31.
Clahsen, H., Felser, C., Neubauer, K., Sata, M., & Silva, R. (2010). Morphological structure in native and non-
native language processing. Language Learning, 60, 21–43.
Clahsen, H., & Neubauer, K. (2010). Morphology, frequency, and the processing of derived words in native and
non-native speakers. Lingua, 120, 2627–2637.
Conklin, K. & Thul, R. (2023). Word and multiword processing: cognitive approaches. In A. Godfroid & H. Hopp
Dagenbach, D., Carr, T. H., & Barnhardt, T. M. (1990). Inhibitory semantic priming of lexical decisions due to
failure to retrieve weakly activated codes. Journal of Experimental Psychology: Learning, Memory, and Cognition,
16, 328–340.
Davis, C. J., & Lupker, S. J. (2006). Masked inhibitory priming in English: Evidence for lexical inhibition. Jour-
nal of Experimental Psychology: Human Perception and Performance, 37, 668–687.
Dijkstra, T., Grainger, J., & van Heuven, W. J. B. (1999). Recognition of cognates and interlingual homographs:
The neglected role of phonology. Journal of Memory and Language, 41, 496–518.
Dijkstra, T., & van Heuven, W. J. B. (2002). The architecture of the bilingual word recognition system: From
identifcation to decision. Bilingualism: Language and Cognition, 5, 175–197.
Draheim, C., Mashburn, C. A., Martin, J. D., & Engle, R. W. (2019). Reaction time in diferential and develop-
mental research: A review and commentary on the problems and alternatives. Psychological Bulletin, 145, 508–535.
Duñabeitia, J., Dimitropoulou, M., Molinaro, N., & Martin, C. (2016). The electrophysiology of the bilingual
brain. In R. R. Heredia, J. Altarriba, & A. B. Cieślicka (Eds.), Methods in bilingual reading comprehension research
Duyck, W., Van Assche, E., Drieghe, D., & Hartsuiker, R. (2007). Visual word recognition by bilinguals in
a sentence context: Evidence for nonselective lexical access. Journal of Experimental Psychology: Learning,
Memory, and Cognition, 33, 663–679.
Elgort, I. (2011). Deliberate learning and vocabulary acquisition in a second language. Language Learning, 61,
367–413.
Elgort, I., Beliaeva, N., & Boers, F. (2020). Contextual word learning in the frst and second language: Defni-
tion placement and inference error efects on declarative and nondeclarative knowledge. Studies in Second
Elgort, I., Brysbaert, M., Stevens, M., & Van Assche, E. (2018). Contextual word learning during reading in a
second language. Studies in Second Language Acquisition, 40, 341–366.
Elgort, I., Candry, S., Boutorwick, T. J., Eyckmans, J., & Brysbaert, M. (2018). Contextual word learning with
form-focused and meaning-focused elaboration. Applied Linguistics, 39, 646–667.
Elgort, I., Perfetti, C. A., Rickles, B., & Stafura, J. Z. (2015). Contextual learning of L2 word meanings: Second
language profciency modulates behavioural and event-related brain potential (ERP) indicators of learning.
Language, Cognition and Neuroscience, 30, 506–528.
Elgort, I., & Warren, P. (2014). L2 vocabulary learning from reading: Explicit and tacit lexical knowledge and
the role of learner and item variables. Language Learning, 64, 365‒414.
Favreau, M., & Segalowitz, N. (1983). Automatic and controlled processes in the frst and second language read-
ing of fuent bilinguals. Memory and Cognition, 11, 565–574.
Felser, C., Cunnings, I., Batterham, C., & Clahsen, H. (2012). The timing of island efects in nonnative sentence
processing. Studies in Second Language Acquisition, 34, 67–98.
Felser, C., & Roberts, L. (2007). Processing wh-dependencies in a second language: A cross-modal priming
study. Second Language Research, 23, 9–36.
Finkbeiner, M., Forster, K., Nicol, J., & Nakamura, K. (2004). The role of polysemy in masked semantic and
translation priming. Journal of Memory and Language, 51, 1–22.
Frenck-Mestre, C. (2005). Eye-movement recording as a tool for studying syntactic processing in a second language:
A review of methodologies and experimental fndings. Second Language Research, 21, 175–198.
Frenck-Mestre, C., & Prince, P. (1997). Second language autonomy. Journal of Memory and Language, 37, 481–501.
Godfroid, A. (2020a). Eye tracking in second language acquisition and bilingualism: A Research synthesis and method-
ological guide. Routledge.
Godfroid, A. (2020b). Sensitive measures of vocabulary knowledge and processing: Expanding Nation’s
framework. In S. Webb (Ed.), The Routledge handbook of vocabulary studies (pp. 433–453). Routledge.
Godfroid, A., Boers, F., & Housen, A. (2013). An eye for words: Gauging the role of attention in inci-
dental L2 vocabulary acquisition by means of eye-tracking. Studies in Second Language Acquisition, 35,
483‒517.
121
Godfroid, A., & Hui, B. (2020). Five common pitfalls in eye-tracking research. Second Language Research, 36,
277–305.
Gollan, T. H., Forster, K. I., & Frost, R. (1997). Translation priming with diferent scripts: Masked priming with
cognates and noncognates in Hebrew-English bilinguals. Journal of Experimental Psychology: Learning, Memory,
and Cognition, 23, 1122–1139.
Gómez, P., Ratclif, R., & Perea, M. (2007). A model of the go/no-go task. Journal of Experimental Psychology:
General, 136, 389–413.
Routledge.
Hahne, A., & Friederici, A. D. (2001). Processing a second language: Late learners’ comprehension mechanisms
as revealed by event-related brain potentials. Bilingualism: Language and Cognition, 4, 123–141.
Hamrick, P. (2023). Conducting reaction time research in second language psycholinguistics. In A. Godfroid
Routledge.
Havik, E., Roberts, L., van Hout, R., Schreuder, R., & Haverkort, M. (2009). Processing subject-object ambi-
guities in the L2: A self-paced reading study with German L2 learners of Dutch. Language Learning, 59,
73–112.
Heredia, R. R., & Cieślicka, A. B. (2020). The cross-modal lexical priming paradigm and bilingual exhaustive
access. In R. R. Heredia & A. B. Cieślicka (Eds.), Bilingual lexical ambiguity resolution (pp. 3–16). Cambridge
University Press.
Hino, Y., Pexman, P. M., & Lupker, S. J. (2006). Ambiguity and relatedness efects in semantic tasks: Are they
due to semantic coding? Journal of Memory and Language, 55, 247–273.
Hutchison, K. A. (2003). Is semantic priming due to association strength or feature overlap? A microanalytic
review. Psychonomic Bulletin and Review, 10, 785–813.
Jegerski, J. (2014). Self-paced reading. In J. Jegerski & B. VanPatten (Eds.), Research methods in second language
Jufs, A., & Harrington, M. (1995). Parsing efects in second language sentence processing: Subject and object
asymmetries in wh-extraction. Studies in Second Language Acquisition, 17, 483–516.
Kaan, E. (2007). Event-related potentials and language processing: A brief overview. Language and Linguistics
Compass, 1, 571–591.
Keating, G. D. (2014). Eye-tracking with text. In J. Jegerski & B. VanPatten (Eds.), Research methods in second
language psycholinguistics (pp. 69–92). Routledge.
Keating, G. D., & Jegerski, J. (2015). Experimental designs in sentence processing research. Studies in Second
Kotz, S. A. (2009). A critical review of ERP and fMRI evidence on L2 syntactic processing. Brain and Language,
109, 68–74.
Kotz, S. A., Holcomb, P. J., & Osterhout, L. (2008). ERPs reveal comparable syntactic sentence processing in
native and non-native readers of English. Acta Psychologica, 128, 514–527.
Libben, M. R., & Titone, D. A. (2009). Bilingual lexical access in context: Evidence from eye-movements dur-
ing reading. Journal of Experimental Psychology: Learning, Memory, and Cognition, 35, 381–390.
Marinis, T. (2018). Cross-modal priming in bilingual sentence processing. Bilingualism: Language and Cognition,
21, 456–461.
Marsden, E., Mackey, A., & Plonsky, L. D. (2016). The IRIS repository: Advancing research practice and meth-
odology. In A. Mackey & E. Marsden (Eds.), Advancing methodology and practice: The IRIS repository of instru-
ments for research into second languages (pp. 1–21). Routledge.
Marsden, E., Morgan-Short, K., Thompson, S., & Abugaber, D. (2018). Replication in second language research:
Narrative and systematic reviews and recommendations for the feld. Language Learning, 68, 321–391.
Marsden, E., Thompson, S., & Plonsky, L. (2018). A methodological synthesis of self-paced reading in second
language research. Applied Psycholinguistics, 39, 861–904.
Moret-Tatay, C., & Perea, M. (2011). Is the go/no-go lexical decision task preferable to the yes/no task with
developing readers? Journal of Experimental Child Psychology, 110, 125–132.
Morgan-Short, K., Faretta-Stutenberg, M., & Bartlett-Hsu, L. (2015). Contributions of event-related potential
research to issues in explicit and implicit second language acquisition. In P. Rebuschat (Ed.), Implicit and
explicit learning of languages (pp. 349–384). John Benjamins.
Morgan-Short, K., Steinhauer, K., Sanz, C., & Ullman, M. T. (2012). Explicit and implicit second language
training diferentially afect the achievement of native-like brain activation patterns. Journal of Cognitive Neu-
roscience, 24, 933–947.
122
Mueller, J. L. (2005). Electrophysiological correlates of second language processing. Second Language Research,
21, 152–174.
Neubauer, K., & Clahsen, H. (2009). Decomposition of infected words in a second language. Studies in Second
Obermeier, A., & Elgort, I. (2021). Deliberate and contextual learning of L2 idioms: The efect of learning
conditions on online processing. System, 97, 1–11.
Pellicer-Sánchez, A. (2016). Incidental L2 vocabulary acquisition from and while reading. Studies in Second Lan-
guage Acquisition, 38, 97‒130.
Phillips, N. A., Segalowitz, N., O’Brien, I., & Yamasaki, N. (2004). Semantic priming in a frst and second
language: Evidence from reaction time variability and event-related brain potentials. Journal of Neurolinguistics,
17, 237–262.
Qiao, X., & Forster, K. I. (2017). Is the L2 lexicon diferent from the L1 lexicon? Evidence from novel word
lexicalization. Cognition, 158, 147–152.
Sachs, R., Baralt, M. & Gurzynski-Weiss (2023). Psycholinguistics for the second language classroom. In
A. Godfroid & H. Hopp (Eds.), The Routledge handbook of second language acquisition and psycholinguistics
Schwartz, A. I., & Kroll, J. F. (2006). Bilingual lexical activation in sentence context. Journal of Memory and
Language, 55, 197–212.
Segalowitz, N. (2000). Automaticity and attentional skill in fuent performance. In H. Riggenbach (Ed.), Perspec-
tives on fuency (pp. 200–219). University of Michigan Press.
Silverberg, S., & Samuel, A. G. (2004). The efect of age of second language acquisition on the representation
and processing of second language words. Journal of Memory and Language, 51, 381–398.
Spivey, M. J., & Marian, V. (1999). Cross talk between native and second languages: Partial activation of an
irrelevant lexicon. Psychological Science, 10, 281–284.
Staub, A., & Rayner, K. (2007). Eye movements and on-line comprehension processes. In M. G. Gaskell (Ed.),
The Oxford handbook of psycholinguistics (pp. 326–342). Oxford University Press.
Steinhauer, K. (2014). Event-related Potentials (ERPs) in second language research: A brief introduction to the tech-
nique, a selected review, and an invitation to reconsider critical periods in L2. Applied Linguistics, 35, 393–417.
Thierry, G., & Wu, Y. J. (2007). Brain potentials reveal unconscious translation during foreign-language com-
prehension. Proceedings of the National Academy of Sciences, 104, 12530–12535.
van Hell, J. G., & Dijkstra, T. (2002). Foreign language knowledge can infuence native language performance
in exclusively native contexts. Psychonomic Bulletin & Review, 9, 780–789.
van Heuven, W. J. B., & Dijkstra, T. (2010). Language comprehension in the bilingual brain: fMRI and ERP
support for psycholinguistic models. Brain Research Reviews, 64, 104–122.
van Heuven, W. J. B., Dijkstra, T., Grainger, J., & Schriefers, H. (2001). Shared neighborhood efects in masked
orthographic priming. Psychonomic Bulletin and Review, 8, 96–101.
van Heuven, W. J. B., Schriefers, H., Dijkstra, T., & Hagoort, P. (2008). Language confict in the bilingual brain.
Cereb Cortex, 18, 2706–2716.
Weber-Fox, C., & Neville, H. J. (1996). Maturational constraints on functional specializations for language
processing: ERP and behavioral evidence in bilingual speakers. Journal of Cognitive Neuroscience, 8, 231–256.
123
11
STUDYING SECOND LANGUAGE
PRODUCTION
Sarah Bernolet
Production of words
Introduction
The production of words in a second language (L2) has mostly been investigated via naming tasks. In
most studies, the speech onset latencies (i.e., the time between the onset of the target display and the
onset of the verbal response uttered by the participant) are the most important dependent variable,
though naming errors are in some cases collected and analyzed as well (Friesen et al., 2014; Smits
et al., 2006). Behavioral measures like naming latencies and errors are sometimes combined with
neurophysiological measures (ERP, fMRI, MEG) measuring brain activity. Importantly, naming tasks
include both a lexical access component and a lexical production component: speakers have to select
the correct lexical representation corresponding to the item to be named (e.g., dog), but they also need
to activate and pronounce the phonological segments that make up the sound of the word (/dɒg/).
Like naming tasks investigating L1 production, L2 naming tasks aim to investigate the mental represen-
tation of words, and the access to their semantics, written word form, and phonology. In L2 studies, how-
ever, the focus is on how lexical access and production are infuenced by the presence and the activation of
another language in the language production system, the L1. To this purpose, the materials of L2 naming
tasks often include words with a specifc relation to their L1 translation equivalents, so-called language-
ambiguous words (see Conklin & Thul, 2023 [this volume], and Elgort & Warren, 2023 [this volume]):
• cognates: translation equivalents that have the same or a similar form in the L1 and the L2, in
terms of orthography and phonology, e.g., flm-flm (identical cognates for Dutch-English bilin-
guals), rood-rot [red] (color words that are phonologically nearly identical and orthographically
similar for Dutch-German bilinguals).
• interlingual homographs: words with the same orthographic word form, but a diferent mean-
ing in both languages. These can be “false friends” that also sound similar (e.g., pet [cap]—pet
for Dutch-English bilinguals; bellen [to call]—bellen [to bark] for Dutch-German bilinguals), or
words that look the same, but have a diferent pronunciation (e.g., room [cream]—room for Dutch-
English bilinguals; four [oven]—four for French-English bilinguals).
• interlingual homophones: words with a similar phonology across languages that difer in mean-
ing and orthography (e.g., koek [cookie]—cook for Dutch-English bilinguals; bague [ring]—bag for
French-English bilinguals).
124 DOI: 10.4324/9781003018872-13

Studying second language production
Naming latencies for L2 words like these are compared with naming latencies for L2 words with a
comparable length and frequency that are unrelated to their translation equivalents in the L1 (e.g.,
emmer—bucket for Dutch-English bilinguals). If processing of the ambiguous words is diferent from
that of the control words, this is considered as evidence for the co-activation of the ambiguous word
in the diferent languages. This, in turn, suggests that words of both languages are stored in one
integrated lexicon, and that accessing those words is language non-selective. Conversely, if L1 and
L2 words were stored in separate lexicons, language-ambiguous words would behave like words that
unambiguously belong to one of both languages (see Gor, 2023 [this volume] and Conklin & Thul,
2023 [this volume]), and there would be no diference between ambiguous words and control words.
Naming errors in which, for example, the L1 pronunciation is used for an L2 interlingual homograph
(a French-English bilingual pronouncing the English word “pain” as /pɛ/), ̃ or cross-language blend
errors (elchother as a blend of the Dutch word elkaar and the English word each other, Poulisse & Bon-
gaerts, 1994) are also attributed to the co-activation of the non-target language—in this case, the L1.
Naming tasks: Tasks in which participants are asked to read out a given word as quickly as possible (word
naming), or to name a color (color naming) or a picture (picture naming) as quickly as possible
Language switching tasks: Tasks in which participants are asked to switch between naming in their frst
(L1) and second (L2) language
Interference tasks: Tasks in which a secondary stimulus—a distractor—is presented simultaneously with
the target stimulus, or shortly before or after its onset
Sentence completion tasks: Tasks in which sentence fragments (prompts, sentences containing gaps, indi-
vidual words) are used to elicit full sentences
Picture description tasks: Tasks that use pictures of scenes or actions to elicit sentences

The diferent naming tasks that are used in studies investigating L2 word production can be classifed
according to the stimuli that are used to elicit naming responses. These can be either printed words,
pictures, or colors.
Word naming
The word naming paradigm was one of the earliest paradigms to be used in research on L2 production.
In fact, Catell already observed in 1887 that “it takes longer to see and name words in one’s foreign
language than in one’s native language” (p. 531). Lambert et al. (1959) used word naming to investigate
language dominance in French-English bilinguals. They subtracted reaction times in English from reac-
tion times in French and discovered that this measure was positively correlated with the bilinguals’ degree
of bilingualism: balanced bilinguals had scores around zero; negative values (i.e., longer naming latencies
in English than in French) were observed for French-dominant bilinguals, positive values for English-
dominant ones. In later studies on L2 word naming, language-ambiguous words (see previous) were
used in order to investigate lexical access (the activation of words in the mental lexicon) and production.
Smits et al. (2006) used Dutch and English interlingual homographs with diferent pronunciations
(e.g., room [cream]—room) to investigate the co-activation of word forms of the L1 (Dutch) during L2
(English) word naming. The test items in their study were divided into four frequency categories, on
125
Sarah Bernolet
the basis of their relative frequencies: high frequency in both English and Dutch (arts—arts [doctor]),
high frequency in English and low frequency in Dutch (fee—fee [fairy]), low frequency in English and
high frequency in Dutch (rug—rug [back]), and low frequency in both languages (wig—wig [wedge]). In
a stimulus list containing items from both languages, interlingual homographs with a high-frequency
Dutch reading (arts and rug) yielded longer naming latencies, more Dutch (L1) responses, and more
other errors in both response languages, because Dutch responses were allowed if only a Dutch reading
was possible (so not for interlingual homographs like arts and rug). When the stimulus list contained
only English (L2) words and homographs, there was no homograph efect in naming latencies, but
homographs did elicit more errors than control words. The reduced efects in the L2-only list cor-
roborate fndings from several lexical decision experiments, which found a similar reduction of the
homograph interference efect in L2-only versus mixed language lists (Dijkstra & Van Heuven, 1998;
De Groot et al., 2000), indicating that L1 words interfere with L2 naming especially if the L1 reading
has a high frequency. In a later study, Friesen and colleagues (2014) showed that similar efects can
be found in L1 production (e.g., an interference efect for interlingual homographs in onset latencies
and errors), indicating that cross-language activation is a two-way street (L1 production infuences L2
production and vice versa), and not just due to incomplete or imperfect acquisition of the L2.
In the last few decades, researchers moved from word naming studies with factorial designs to
megastudies on large sets of stimuli. Balota et al. (2007), for example, collected speeded word nam-
ing data for 2,428 English words in young and older healthy adults in order to investigate the unique
contributions of specifc phonological, lexical, and semantic variables. More recently, Cortese et al.
(2018) collected conditional reading-aloud data (participants only read aloud words they recognize as
existing words, like in a go/no-go task) for 2,145 English words in order to establish the diferences
with regular reading and lexical decision. The advantage of such large datasets is that they provide
researchers with a common set of items, on which they can test experimental hypotheses, either via
large-scale regression approaches or via more specifc tests on selected items.
Picture naming
Picture naming experiments have been used quite extensively in studies investigating L2 production.
There are several variants of the picture naming paradigm. In a regular picture naming experiment,
participants name a series of pictures as quickly as possible (see Figure 11.1a for a possible stimulus;
the base pictures used in examples 11.a-c are taken from the MultiPic database [Duñabeitia et al.,
2018]). In a diferent variant the naming task is combined with an interference paradigm. In the pic-
ture-word interference paradigm the distractor that is presented together with the target picture (or
shortly before or after it) is a word (presented visually or auditorily, see Figure 11.1b); in the picture-
picture interference paradigm, the distractor is a picture that is superimposed on the target picture
(see Figure 11.1c).
Figure 11.1 Examples of possible stimuli in (a) regular picture naming experiments, (b) picture-word interfer-
ence experiments, and (c) picture-picture interference experiments (pictures taken from the MultiPic database
[Duñabeitia et al., 2018]).
126
One of the frst studies on L2 picture naming was a regular picture naming study by Costa et al. (2000).
The pictures in their study were either pictures whose names were cognates in Catalan and Spanish
(gat—gato [cat]) or pictures whose names were unrelated in these two languages (taula—mesa [table]).
Catalan-Spanish bilinguals showed a cognate facilitation efect in Spanish (their less dominant language
or L2): Naming latencies in L2 were shorter for words with cognate names than for words with non-
cognate names. Monolingual speakers of Spanish showed no such efect, indicating that the efect
arose in the bilinguals due to the co-activation of the target words’ translations in Catalan. Costa et al.
(2000) argued that this facilitation efect could arise at two diferent levels. One possibility is that
it arises at the level at which speakers have to select the phonological segments for the target word.
The cognate words in the experiment (e.g., gato—Spanish [cat]), share most of their segmental ele-
ments (/g/, /a/, /t/) with their translations in the other language (e.g., gat—Catalan [cat]). The seg-
ments that need to be activated for pronunciation thus receive extra activation via the co-activation
of their translation equivalent. This, it is argued, makes it easier to select the phonology of cognate
words than of non-cognate words. Another possibility is that the phonological segments of the non-
selected translation word send feedback to the lexical level, and thus help bilinguals in selecting the
correct word (and not related words like dato [fact; phonologically related] or perro [dog; semantically
related]). With this study, Costa et al. (2000) did not only provide evidence for the co-activation
of lexical items in the L1 during L2 production, they also showed that these co-activated lexemes
activate their phonology (see also Hoshino and Kroll [2008] for co-activation across diferent scripts).
Picture naming experiments without distractors have further been used in language switching experi-
ments, in which participants have to switch between naming pictures in their L1 and their L2. In
these experiments, the target language is often indicated by a language cue, which can be a word in
the target language (Hernandez et al., 2001) or a picture referring to the target language (Verhoef
et al., 2009) that is presented before the picture onset. Alternatively, the cue is the color of the picture
to be named (Costa & Santésteban, 2004). After Meuter and Allport (1999) had used the language
switching paradigm with digit naming, Costa and Santésteban (2004) used a version with pictures to
investigate activation and inhibition mechanisms in L2 production. Like Meuter and Allport’s study,
their study showed (a) that reaction times were slower on switch trials than on trials in which par-
ticipants did not need to switch languages, and (b) that this switching cost was larger when switching
from the weaker L2 to the dominant L1 than in the opposite direction. This asymmetrical switching
cost refects bilingual speakers’ need to suppress their dominant language during production in their
weaker language to such an extent that it makes reactivation of the dominant language harder than
it is to reactivate the weaker one (Costa & Santésteban, 2004). Consistent with this hypothesis, the
researchers found no asymmetrical switching cost in Spanish-Catalan bilinguals who were highly
profcient in both of their languages.
In the last few decades, the language switching paradigm has been used with diferent language
combinations, and with adult speakers, children, and patients to investigate the ins and outs of inhibi-
tory control in bilingual language production (see Declerck & Philipp, 2015, for an overview). Many
of these studies made use of cues to induce language switching (see previous). In recent years, the
focus shifted towards voluntary switching and switching “in the wild.” As Gollan and Ferreira (2009)
noted, the switching costs observed in experiments featuring cued language switches seem at odds
with what happens in spontaneous conversations, in which bilinguals easily and commonly switch
between languages, even if there is no obvious reason to do so. In Gollan and Ferreira’s switching
experiments, Spanish-English participants (either English-dominant or balanced) named pictures in
an “only-English” condition, an “only-Spanish” condition, and in an “either-language” condition, in
which they could switch whenever they felt like doing so. The results indicated that language switch-
ing was still costly, even if participants switched voluntarily. In contrast with what happens in cued
language switching experiments, however, no asymmetric switching cost was found: It was equally
costly to switch to the dominant and the non-dominant language. Recently, Blanco-Elorrieta and
127
Sarah Bernolet
Pylkkänen (2017) conducted a language switching experiment (combined with MEG recordings)
in which they compared colors as switching cues with pictures of monolingual (Arabic or English)
or bilingual (Arabic-English) interlocutors as switching cues. In the case of a bilingual interlocutor,
participants were free to choose the language of production. Both the MEG and the behavioral data
of their study suggest that language switching is harder when the language cue is artifcial (i.e., a
color) and gets easier as the paradigm becomes more natural (i.e., on the basis of the language(s) used
by the interlocutor). In fact, voluntary switching did not engage prefrontal control regions or elicit
behavioral switch costs, suggesting that the bilinguals in this study switched quite efortlessly on the
basis of their interlocutor. Together, these studies suggest that cued language switching might induce
additional inhibitory processes that are not necessarily active in everyday bilingual language use.
Finally, many studies investigating L2 word production have made use of an interference paradigm
(picture-word or picture-picture interference), in which a secondary stimulus—a distractor—is pre-
sented simultaneously with the target picture, or shortly before or after its onset (see Figure 11.1b
and c). By varying the relative timing of the distractor and the target stimulus, researchers can investi-
gate the activity of the distractor in the diferent stages of production. This technique has shown that
distractors presented before picture onset produce interference when they are semantically related
to the picture (e.g., sheep–goat), but do not infuence naming if they are phonologically related (e.g.,
sheep–ship); distractors that are presented together with or after the picture facilitate naming if they
are phonologically related to the picture to be named, but have no efect if they are semantically
related because semantic information is activated before phonological information during produc-
tion (see Schriefers et al., 1990).
The frst study to use the picture-word interference paradigm to study word production in L1
and L2, was a study by Ehri and Ryan (1980). They made Spanish-English bilinguals name pictures
in their L1 (Spanish) and their L2 (English). Three diferent types of distractor words were superim-
posed on the pictures: the name of a diferent object in the same language as the naming language
(e.g., ghost for a picture of an arm), the picture’s translation in the other language (brazo), and a set of
Xs (neutral condition). They measured interference by comparing naming latencies in the two frst
conditions with latencies in the neutral condition. The results of the L1 naming trials unambiguously
pointed in the direction of integrated lexicons across languages: In L1 Spanish, both Spanish and
English distractors caused a comparable delay in naming latencies due to semantic interference. For
the L2 naming trials with L2 English, a learning efect was observed: On the frst trial, Spanish words
produced more interference than English words (i.e., more interference was observed for between-
language than within-language distractors), whereas the pattern was reversed afterwards.
In 2011, Hoshino and Thierry combined this paradigm with event-related potentials (ERPs),
which measure brain activity during language processing. For the same group of bilinguals as the
ones tested by Ehri and Ryan (Spanish as L1 and English as L2), they used semantically related
distractors (leg for a picture of an arm), distractors that were phonologically related to the English
picture name (arrow), distractors that were phonologically related to the Spanish picture name (bridge,
which is a phono-translation distractor for brazo [arm]), and unrelated distractors (table). They found
semantic, phonological, and phono-translation interference efects between 200 and 400 ms after
picture onset, suggesting that, even for bilinguals living in an L2-dominant (English) context, L1
lexical alternatives (i.e., the pictures’ L1 names) are activated and compete for selection in the pro-
cess of L2 speech planning. Because the observed interference efects still occur in the 350–400 ms
time window, Hoshino and Thierry argued that competition between languages appears to continue
when the correct word is selected.
One of the most challenging aspects of designing a picture naming experiment is the selection
of the pictures. Recently, the results of a Bayesian meta-analysis performed on 14 picture norming
studies (Perret & Bonin, 2019) indicated that not all factors that are usually controlled for are equally
important: Image agreement, name agreement, image variability/imageability, age of acquisition,
128
and conceptual familiarity had a strong infuence on naming speed; visual complexity and length
yielded null efects; for lexical frequency the results were inconclusive. Note that the studies included
in the meta-analysis only used black-and-white line pictures.
Color naming
Although the picture-word interference paradigm is now used more frequently, the Stroop paradigm
was the frst interference paradigm used in research on L1 and L2 word production. The paradigm is
named after John Ridley Stroop, who frst used it in 1935. In the task, participants are asked to name
the color in which words are presented to them. Crucially, the critical stimuli in the experiment
consist of color words that are either congruent (“red” presented in red ink) or incongruent with the
color in which they are presented (“red” presented in green ink). Naming the color of the word takes
longer and is more prone to errors when the color of the ink does not match the name of the word.
This is because participants need to actively inhibit the word names in order to be able to name the
word color instead. Word activation is an automatic process.
When bilinguals have to name colors in their L2, the Stroop task can have many diferent set-ups:
Distractor words can either be words in the L1 (investigating between-language competition) or in
the L2 (investigating within-language competition), and in both cases the name of the distractor can
be semantically related or phonologically related to the color name. In the former case, the distractor
is a diferent color in the L1 or the L2.
Costa and colleagues (2008) used a Stroop task in which Catalan-Spanish bilinguals had to name
the color of frst-language words (Catalan) in their L2 (Spanish). The results of the experiment
showed a Stroop efect across languages both in naming latencies and error rates: Participants were
slower to name the color names in L2 Spanish (e.g., AZUL) when the L1 Catalan word was in fact
another color term (GROC “yellow”) than when it corresponded to an unrelated word (LENT
“slow”). Additionally, they observed a robust cross-language identity efect: Naming latencies in L2
Spanish (e.g., AZUL “blue”) were sped up when the distractor corresponded to the L1 Catalan trans-
lation of the color name in which the word was printed (e.g., BLAU “blue”) compared to when it
corresponded to an unrelated word (e.g., LENT “slow”). Finally, phono-translation distractors (e.g.,
BRAU “ferce” instead of BLAU “blue”) did not interfere more than unrelated control distractors
(e.g., LLEST “clever”). Costa et al. (2008) considered this latter fnding to be the most important
fnding in their study, because interference efects caused by phono-translation distractors had been
reported in several picture-word interference experiments (Costa et al., 2003; Hermans et al., 1998).
While the results of the picture-word interference experiments are interpreted as evidence for com-
petition at the lexical level, the results of the Stroop task suggest a combination of competition at the
lexical level and facilitation at the semantic level. The authors thus argue that their study highlights
the importance of assessing the reliability of efects such as the phono-translation interference in
diferent paradigms before strong theoretical conclusions can be drawn regarding the dynamics of
bilingual language production.
Word production tasks: evaluation and future directions

Studying word production in L2 by means of naming tasks has a number of advantages: Since
the output is necessarily language-specifc, one can be sure that latencies are—at least in part—
determined by the activation of the L2 word form at diferent levels of production (e.g., the lemma
level, the phonological level), which is not necessarily the case in L2 lexical decision tasks (see
De Groot et al., 2002; Smits et al., 2006). The fact that many diferent paradigms have been used
to elicit L2 word production is in fact a strength, because the reliability of a specifc efect can
be scrutinized by assessing it in many diferent ways (see Costa et al., 2008; Smits et al., 2006).
129
Sarah Bernolet
As such, word naming tasks have produced many important insights in the mechanisms involved
in L2 word production.
A big drawback of using naming tasks to investigate L2 word production is that it is not trivial
to collect speech onset latencies. In early naming studies, onset latencies were often automatically
registered using a voice-key, which monitors the input from the microphone and is triggered when
the input exceeds a certain amplitude threshold. Because this method has its problems (data loss
due to hesitations, diferences in sensitivity for diferent onsets due to acoustic diferences between
sounds), as a result, researchers have reverted to measuring speech onsets in the actual recorded wave-
forms, which can be very labor- and time-intensive, especially for larger experiments. In an answer
to this, Roux et al. (2017) developed a tool, CHRONSET, which allows researchers to automatically
estimate speech onsets from waveforms, on the basis of multiple acoustic features extracted via mul-
titaper spectral analysis. This tool, which is publicly available online at www.bcbl.eu/databases/chronset
has been used in diferent studies investigating L2 word production (Broos et al., 2018; de Bruin
et al., 2018) and will greatly facilitate further L2 research using naming latencies.
Another way to deal with the problem of speech onset latencies is to investigate naming latencies
in written naming. Muylle et al. (2022) recently took this approach by comparing the cognate efect
in the spoken and typewritten L2 production of Dutch-English bilinguals. Despite the apparent
diferences between the two production modalities, the cognate facilitation efect was very similar
in spoken and written picture naming, suggesting that the efect already occurs during lexical selec-
tion. The existence of similar co-activation processes during spoken and written naming opens up
the possibility to further investigate L2 production in written naming, in which there is no need for
overt pronunciation. This might, in some cases (e.g., for participants with a dominant native accent),
lead to more accurate measures of word production. Investigating the details of L2 word production
in the (type)written modality can also provide the necessary information to investigate the written
production of larger texts (see open questions in Textbox 11.2). In this day and age, producing writ-
ten texts in a language that is not our native language has become a common experience for many
of us. Consequently, investigating L2 text writing can provide an ecologically valid way to investigate
written word production in context (see van den Bergh et al., this volume).
Production of sentences
Introduction
Studies focusing on sentence production in L2 mostly investigate to which extent co-activated words
also activate the syntactic and morphological information they are associated with and how this infu-
ences production. Nouns, for example, can have the same gender in two languages, but they can
also difer in gender and in the way gender is marked: A beautiful painting has a neuter noun and no
marking on the adjective in English, whereas una hermosa pintura, its translation in Spanish, features a
feminine noun and a sufx marking femininity on the adjective. Likewise, verbs can allow the same
syntactic constructions across languages, but they can also difer in the constructions with which they
combine. For example, in general, both Dutch and German allow prepositional object dative and
double object dative constructions, but the verb gönnen [“not to begrudge”; gunnen in Dutch] can
occur only with a double object dative in German. The question is not only whether co-activation
of morphological and syntactic information occurs across languages/across translation equivalents,
but also whether this helps or hinders the learning and the production of the L2.
Note that not all models for bilingual sentence production predict interference or facilitation
caused by the co-activation of morphological or syntactic features in the non-target language. De Bot
(1992), for example, assumes that bilinguals’ lexicons are shared between languages, but that there
are separate formulators for each language. Because the process of grammatical encoding (as well as
130
production processes following it) is fully encapsulated (i.e., modular or non-communicating) for
the diferent languages, co-activation of syntactic information of the non-target language during L2
production is assumed to be minimal. Co-activation can only happen via links between the diferent
formulators. Ullman’s Declarative-Procedural model for L2 production (2001) makes a distinction
between early and late bilinguals. Early bilinguals use their procedural memory system in L2 to pro-
cess and produce grammatical structures, just as they do in their L1. Late bilinguals, however, cannot
rely on this memory system to acquire L2 grammar, and are assumed to store sentence forms in their
entirety or represent grammatical rules in declarative memory, where lexical information is stored
as well. Consequently, co-activation of grammatical rules can only occur in early bilinguals or late
bilinguals with a very high L2 profciency, who then show native-like processing (see Morgan-Short
and Ullman, 2023 [this volume]).

Methods that are used in L2 sentence production studies include sentence completion and picture
description.
Sentence completion
Various kinds of stimuli and prompts have been used in order to make participants produce full sen-
tences in their L2 in the absence of a picture: tasks in which participants have to form a sentence on
the basis of a prompt (Antón-Méndez, 2010), tasks in which participants fll in the gaps in a sentence
(Ellis et al., 2012), and tasks in which participants need to form a sentence out of a number of given
words (Golestani et al., 2006). In all cases, the production is (partly) constrained (e.g., by giving the
translation of the words to be used), in order to limit the number of response alternatives.
Ellis et al. (2012) used a sentence completion task to investigate the acquisition of grammatical
gender in German in three diferent groups of learners: L1 speakers of Afrikaans, English, and Ital-
ian. German uses three diferent genders for nouns (masculine, feminine, and neuter) and marks
gender on defnite and indefnite determiners, as well as adjectives. In Italian, gender is marked in
the same way, but Italian uses only two diferent genders (masculine and feminine). Though Ital-
ian learners cannot directly transfer the grammatical gender system of their L1 to German, they
might beneft from the fact that their L1 at least expresses grammatical gender in the same way. In
English as well as in Afrikaans, nominal expressions do not display grammatical gender. Afrikaans,
on the other hand, is very similar to German in terms of vocabulary and syntax, which could make
it easier for Afrikaans-speaking learners than for English-speaking learners to acquire grammatical
gender in German.
The sentence completion task the researchers used was designed to test for gender agreement:
Participants were asked to fll in the appropriate form of the determiner and the adjective based
on the preceding German sentence, as well as the Afrikaans, English, or Italian translation of the
determiner and adjective, given in brackets (see example 1 in which the appropriate answer is ein
schönes). The results of this experiment showed that, overall, the Italian group outperformed the
Afrikaans and English groups, and the Afrikaans group similarly outperformed the English group
(the between-group diferences were not signifcant for all categories). The researchers concluded
that the advantage for Italian learners is due to the deep transfer of grammatical properties of their
L1 during L2 acquisition (surface transfer is not possible in this case, since the grammatical gender
system of Italian is not congruent with that of German). Thus, these results suggest that transfer of
grammatical information from the L1 can assist in L2 acquisition. Interestingly, the Italian group
showed no advantage on the other two learner groups when the task merely involved the assignment
of gender in a picture naming task.
131
Sarah Bernolet
(1) Das Mädchen ist schön.

Ich liebe (a beautiful) __________ Mädchen.
The girl is beautiful.
I love (a beautiful) __________ girl.
Golestani et al. (2006) combined a sentence completion task with fMRI imaging to examine the
production of syntax in the L1 and L2 languages of non-profcient, late bilinguals. During scan-
ning, subjects viewed a series of either three or fve words in L1 or in L2. In the baseline condition,
participants were asked to simply read out these words, but in the experimental condition, they
needed to produce a simple grammatical sentence from these same words. By subtracting the neural
activity in the baseline condition from the activity in the experimental condition, Golestani et al.
(2006) aimed to isolate the activation caused by syntactic processing. Their results were in line with
the predictions of Ullman’s Declarative-Procedural model (see earlier): During sentence production
in the L2, more profcient bilinguals engaged a component of the procedural memory system more
than less profcient individuals did, suggesting that more profcient bilinguals show more rule-based,
native-like processing of L2 syntax. Moreover, the left prefrontal activation peaks—which signal syn-
tactic processing—were found to be much closer during syntactic production in L1 and L2 in more,
compared to less grammatically profcient bilinguals.
Picture description
In the last ten years, many studies investigating L2 sentence production have used pictures to elicit
sentences. Many of these experiments made use of a structural priming paradigm, a paradigm that has
become increasingly popular in L1 studies investigating syntactic processing since the late 1980s.
Picture description experiments involving structural priming use pictures of actions or scenes that
can be described by using either one of two syntactic alternatives, which are semantically identical.
A picture showing a pirate chasing a boxer, for example (see Figure 11.2), can be described by using
an active sentence (e.g., “The pirate is chasing the boxer”) or by using its passive counterpart (e.g.,
Figure 11.2 Target picture eliciting active and passive transitive sentences (used in Bernolet et al., 2009, and
reprinted with permission from Elsevier).
132
“The boxer is being chased by the pirate”). Though participants are free to choose either of both
alternatives to describe the picture, they appear to be heavily infuenced by the structure of the prime
sentence preceding the picture: If the prime sentence used the passive form (“The nun is being kissed
by the sailor”), participants are more likely to produce a passive target description than if the prime
sentence used the active form (“The sailor kisses the nun”), and vice versa, even if prime and target
sentences are otherwise completely unrelated (see Bock, 1986, for the frst experimental study on
the structural priming efect). Participants not only have the tendency to re-use syntactic structures
in the primed condition, they are also quicker to initiate sentences and to comprehend sentences if
these sentences re-use the syntax of the previous sentence. It is generally assumed that these efects
occur because prime and target structure share a common underlying representation. Whereas L1
production studies have used structural priming to investigate what kind of representations are active
during syntactic processing, L2 and bilingual production studies mainly use the paradigm in order
to investigate whether and/or under which conditions syntactic structures are shared between the
diferent languages of a bilingual.
One of the frst studies investigating structural priming between the L1 and the L2 of bilinguals
was a study by Hartsuiker and colleagues (2004), investigating the production of English transitive
sentences by Spanish-English bilinguals. This study used a dialogue version of the syntactic prim-
ing paradigm, in which participants described pictures to each other. One of the participants was in
fact a confederate, a co-worker of the experimenter who pretended to be a fellow participant and
who produced scripted prime sentences. Prime sentences were Spanish active and passive transitive
sentences (El taxi persigue el camión—El camión es perseguido por el taxi [a taxi chasing a truck]), as well
as intransitive sentences (El taxi acelera [the taxi accelerates]) and active sentences in which the object
came before the verb and the subject after the verb (El camion lo persigue un taxi [The truck[chasee]
it chases a taxi[chaser], a form not found in English). Target pictures had to be described in partici-
pants’ L2, English. The results showed cross-linguistic syntactic priming in dialogue: Spanish-English
bilinguals tended to produce English passive sentences more often following a Spanish passive sen-
tence than following a Spanish intransitive or active sentence. The fnding of cross-linguistic syn-
tactic priming between Spanish and English transitives suggests that Spanish-English bilinguals use a
common underlying representation for the comprehension and the production of transitives in their
L1 and their L2. On the basis of this result, Hartsuiker et al. (2004) proposed a model for bilingual
sentence production, in which lemmas in an integrated bilingual lexicon are connected to combi-
natorial nodes, unspecifed for language. Contrary to De Bot (1992), Hartsuiker et al. assume that
syntactic structures are shared between languages as much as possible.
Sentence production tasks: evaluation and future directions

From this small sample of L2 sentence production studies it can be concluded that not only words
of the non-target language are co-activated, but also the morphological features and the syntactic
information associated with these words. It is not fully clear what the limits of this co-activation
are: Does transfer of grammatical features occur for all kinds of features? How similar do syntactic
structures need to be in order to be shared? It is also unclear whether the activation occurs for all
bilinguals (see Jackson, 2023 [this volume] this issue, for a discussion on the limits of L2 syntactic
priming). A study by Bernolet et al. (2013) showed, for example, that, for some structures, cross-
linguistic priming does only occur for bilinguals with a high L2 profciency, suggesting that syntax
might not be shared in earlier stages of L2 syntactic development (in line with assumptions from
Ullman’s [2001] Declarative-Procedural model).
Picture description experiments using syntactic priming could be useful to further investigate
the development of L2 syntax: The task is not too complex for L2 learners, since more than one
response is possible, and one of both responses is primed. Recently, a number of studies have shown
133
Sarah Bernolet
that the paradigm can also be used to assess the learning of an artifcial language (Fehér et al., 2016;
Muylle et al., 2020). A possible problem with syntactic priming studies investigating L2 production
is, however, that the paradigm allows learners to (strategically) avoid a specifc syntactic alterna-
tive (e.g., the more complex one), in which case the absence of priming might refect problems in
producing a particular structure, rather than the absence of a representation for this structure (see
open questions in Textbox 11.2). By measuring priming efects in L2 production and comprehen-
sion within learners at the same level of profciency, or by combining choice data with latency data,
future studies can try to get a clearer picture of the representation and the development of L2 syntax.
To what extent does cross-language activation infuence the production of L2 words in context (during
L2 conversations or in the production of longer L2 texts)?
How do spoken L2 naming tasks relate to tasks requiring written naming in L2? Does this mapping difer
between languages?
What is strategic in L2 learners’ syntactic choices in syntactic priming experiments and which part of the
syntactic priming efect is caused by the implicit learning of L2 syntax?
Further reading
Kormos, J. (2014). Speech production and second language acquisition. Routledge.
This book provides a discussion of issues in L1 and L2 speech production research, and contains chapters on
monitoring and fuency in L2 speech production, topics that are not covered in the current chapter.
McDonough, K., & Trofmovich, P. (2011). Using priming methods in second language research. Taylor & Francis.
This book presents a broader overview of the use of priming methods to explore the comprehension,
production, and acquisition of L2 phonology, syntax, and lexicon. The focus is not purely theoretical; the
volume also contains studies investigating applied, practical questions relevant to L2 teaching and learning.
References
Antón-Méndez, I. (2010). Gender bender: Gender errors in L2 pronoun production. Journal of Psycholinguistic
Research, 39(2), 119–139.
Balota, D. A., Yap, M. J., Hutchison, K. A., Cortese, M. J., Kessler, B., Loftis, B., ... & Treiman, R. (2007). The
English lexicon project. Behavior research methods, 39(3), 445–459.
Bernolet, S., Hartsuiker, R. J., & Pickering, M. J. (2009). Persistence of emphasis in language production: A
cross-linguistic approach. Cognition, 112(2), 300–317.
Bernolet, S., Hartsuiker, R. J., & Pickering, M. J. (2013). From language-specifc to shared syntactic represen-
tations: The infuence of second language profciency on syntactic sharing in bilinguals. Cognition, 127(3),
287–306.
Blanco-Elorrieta, E., & Pylkkänen, L. (2017). Bilingual language switching in the laboratory versus in the wild:
The spatiotemporal dynamics of adaptive language control. Journal of Neuroscience, 37(37), 9022–9036.
Bock, J. K. (1986). Syntactic persistence in language production. Cognitive Psychology, 18(3), 355–387.
Broos, W. P., Duyck, W., & Hartsuiker, R. J. (2018). Are higher-level processes delayed in second language word
production? Evidence from picture naming and phoneme monitoring. Language, Cognition and Neuroscience,
33(10), 1219–1234.
Cattell, J. M. (1948). Experiments on the association of ideas, 1887. In W. Dennis (Ed.), Century psychology series.
Readings in the history of psychology (pp. 329–335). Appleton-Century-Crofts.
134
Cortese, M. J., Yates, M., Schock, J., & Vilks, L. (2018). Examining word processing via a megastudy of condi-
tional reading aloud. Quarterly Journal of Experimental Psychology, 71(11), 2295–2313.
Costa, A., Albareda, B., & Santesteban, M. (2008). Assessing the presence of lexical competition across lan-
guages: Evidence from the Stroop task. Bilingualism: Language and Cognition, 11(1), 121–131.
Costa, A., Caramazza, A., & Sebastian-Galles, N. (2000). The cognate facilitation efect: Implications for models
of lexical access. Journal of Experimental Psychology: Learning, Memory, and Cognition, 26(5), 1283.
Costa, A., Colomé, À., Gómez, O., & Sebastián-Gallés, N. (2003). Another look at cross-language competition
in bilingual speech production: Lexical and phonological factors. Bilingualism: Language and Cognition, 6(3),
167–179.
Costa, A., & Santésteban, M. (2004). Lexical access in bilingual speech production: Evidence from language
switching in highly profcient bilinguals and L2 learners. Journal of Memory and Language, 50(4), 491–511.
De Bot, K. (1992). A bilingual production model: Levelt’s “speaking” model adapted. Applied Linguistics, 13,
1–24.
de Bruin, A., Samuel, A. G., & Duñabeitia, J. A. (2018). Voluntary language switching: When and why do
bilinguals switch between their languages? Journal of Memory and Language, 103, 28–43.
Declerck, M., & Philipp, A. M. (2015). A review of control processes and their locus in language switching.
Psychonomic Bulletin & Review, 22(6), 1630–1645.
De Groot, A. M., Borgwaldt, S., Bos, M., & Van den Eijnden, E. (2002). Lexical decision and word naming in
bilinguals: Language efects and task efects. Journal of Memory and Language, 47(1), 91–124.
De Groot, A. M., Delmaar, P., & Lupker, S. J. (2000). The processing of interlexical homographs in translation
recognition and lexical decision: Support for non-selective access to bilingual memory. The Quarterly Journal
of Experimental Psychology Section A, 53(2), 397–428.
Dijkstra, T., & Van Heuven, W. J. B. (1998). The BIA model and bilingual word recognition. In J. Grainger & A.
Jacobs (Eds.), Localist connectionist approaches to human cognition (pp. 189–225). Lawrence Erlbaum Associates.
Duñabeitia, J. A., Crepaldi, D., Meyer, A. S., New, B., Pliatsikas, C., Smolka, E., & Brysbaert, M. (2018).
MultiPic: A standardized set of 750 drawings with norms for six European languages. Quarterly Journal of
Experimental Psychology, 71(4), 808–816.
Ehri, L. C., & Ryan, E. B. (1980). Performance of bilinguals in a picture-word interference task. Journal of Psy-
cholinguistic Research, 9(3), 285–302.
Elgort, I. & Warren, P. (2023). Studying second language comprehension. In A. Godfroid & H. Hopp (Eds.).
Ellis, C., Conradie, S., & Huddlestone, K. (2012). The acquisition of grammatical gender in L2 German by
learners with Afrikaans, English or Italian as their L1*. Stellenbosch Papers in Linguistics, 41, 17–27.
Fehér, O., Wonnacott, E., & Smith, K. (2016). Structural priming in artifcial languages and the regularisation
of unpredictable variation. Journal of Memory and Language, 91, 158–180.
Friesen, D. C., Jared, D., & Haigh, C. A. (2014). Phonological processing dynamics in bilingual word naming.
Canadian Journal of Experimental Psychology/Revue canadienne de psychologie expérimentale, 68(3), 179.
Golestani, N., Alario, F. X., Meriaux, S., Le Bihan, D., Dehaene, S., & Pallier, C. (2006). Syntax production in
bilinguals. Neuropsychologia, 44(7), 1029–1040.
Gollan, T. H., & Ferreira, V. S. (2009). Should I stay or should I switch? A cost—beneft analysis of voluntary
language switching in young and aging bilinguals. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 35(3), 640.
Routledge.
Hartsuiker, R. J., Pickering, M. J., & Veltkamp, E. (2004). Is syntax separate or shared between languages? Cross-
linguistic syntactic priming in Spanish-English bilinguals. Psychological Science, 15(6), 409–414.
Hermans, D., Bongaerts, T., De Bot, K., & Schreuder, R. (1998). Producing words in a foreign language:
Can speakers prevent interference from their frst language? Bilingualism: Language and Cognition, 1(3),
213–229.
Hernandez, A. E., Dapretto, M., Mazziotta, J., & Bookheimer, S. (2001). Language switching and language
representation in Spanish—English bilinguals: An fMRI study. NeuroImage, 14(2), 510–520.
Hoshino, N., & Kroll, J. F. (2008). Cognate efects in picture naming: Does cross-language activation survive a
change of script? Cognition, 106(1), 501–511.
Hoshino, N., & Thierry, G. (2011). Language selection in bilingual word production: Electrophysiological evi-
dence for cross-language competition. Brain Research, 1371, 100–109.
135
Sarah Bernolet
Lambert, W. E., Havelka, J., & Gardner, R. C. (1959). Linguistic manifestations of bilingualism. The American
Journal of Psychology, 72(1), 77–82.
Meuter, R. F., & Allport, A. (1999). Bilingual language switching in naming: Asymmetrical costs of language
selection. Journal of Memory and Language, 40(1), 25–40.
Muylle, M., Bernolet, S., & Hartsuiker, R. (2020). The role of case marking and word order in cross-linguistic
structural priming in late L2 acquisition. Language Learning, 70(S2), 194–220.
Muylle, M., Van Assche, E., & Hartsuiker, R. (2022). Comparing the cognate efect in spoken and written
second language word production. Bilingualism: Language and Cognition, 25(1), 93–107.
Perret, C., & Bonin, P. (2019). Which variables should be controlled for to investigate picture naming in adults?
A Bayesian meta-analysis. Behavior Research Methods, 51(6), 2533–2545.
Poulisse, N., & Bongaerts, T. (1994). First language use in second language production. Applied Linguistics, 15(1),
36–57.
Roux, F., Armstrong, B. C., & Carreiras, M. (2016). Chronset: An automated tool for detecting speech onset.
Behavior Research Methods, 49(5), 1864–1881.
Schriefers, H., Meyer, A. S., & Levelt, W. J. (1990). Exploring the time course of lexical access in language
production: Picture-word interference studies. Journal of Memory and Language, 29(1), 86–102.
Smits, E., Martensen, H., Dijkstra, T., & Sandra, D. (2006). Naming interlingual homographs: Variable competi-
tion and the role of the decision system. Bilingualism: Language and Cognition, 9(3), 281–297.
procedural model. Bilingualism: Language and Cognition, 4, 105–122.
Verhoef, K., Roelofs, A., & Chwilla, D. J. (2009). Role of inhibition in language switching: Evidence from
event-related brain potentials in overt picture naming. Cognition, 110(1), 84–99.
136
12
STUDYING MULTIMODAL
LANGUAGE PROCESSING
The integration of speech and gestures
Marianne Gullberg
Introduction
The processing of language in real-time is a remarkable feat both in production and comprehension.
However, the vast psycholinguistic literature generally underestimates the challenge of language pro-
cessing, ignoring the fact that language use is a multimodal endeavor—both in language production and
comprehension. In normal language use, the processing system deals not only with auditory/acoustic
speech signals, but also with language-related visual signals involving several articulators (e.g., mouth,
head, arm/hand movements, etc.) and movements we often call gestures. These signals must be
assigned relevance since not all visual signals are language-related, and must be integrated under
the time constraints of face-to-face interaction both in production and comprehension (cf. Holler
& Levinson, 2019). There is now a substantial amount of evidence suggesting that language users
produce and comprehend multimodal language signals more or less automatically. With language
use being multimodal, important new challenges arise for psycholinguistics. Although the study of
multimodal language processing is gaining ground in frst language (L1) psycholinguistic studies, it
remains deeply under-researched in second language (L2) psycholinguistics.
This chapter will frst present some evidence for the multimodal nature of language processing
in production, comprehension, and acquisition. Multimodality is a complex term. This chapter will
only deal with auditory and language-related bodily visual signals, that is, with speech and gestures.
It will not address audiovisual signals that involve writing (e.g., subtitles on TV) or pictorial repre-
sentations. Moreover, the chapter will focus on adult L2 acquisition (for multimodal processing in L1
acquisition, see Goldin-Meadow, 2003; multimodal studies of child L2 hardly exist, but see Andrä
et al., 2020; Tellier, 2008). After a brief historical overview, the chapter surveys some key methods
used to study multimodal processing, and provides examples of applications to L2 studies, and then
discusses (dis-)advantages of the methods. The chapter closes with an outline of some open questions
and future directions for research.
Evidence for the multimodal nature of language processing

Psycholinguistic evidence now abounds supporting a multimodal view of language, showing that
speech and gestures—defned as actions and movements related to ongoing talk, and recognized as
communicatively relevant by onlookers—form a tightly linked composite signal in language pro-
duction and comprehension (Bavelas, 1994; Clark, 1996; Holler & Levinson, 2019; Kendon, 2004;
McNeill, 2015; Özyürek, 2017).
DOI: 10.4324/9781003018872-14 137

Marianne Gullberg
In language production, the cross-modal integration is seen in that speakers produce movements
that are semantically, prosodically, discursively, and temporally coordinated with speech with mil-
lisecond precision. The modalities express semantically similar, prosodically and discursively high-
lighted (focused) meaning at the same time (Kendon, 1980; Loehr, 2007; McNeill, 1992). This
fne-grained coordination at several linguistic levels also implies that gestures refect cross-linguistic
variation in which semantic elements are expressed in speech and their morphosyntactic organiza-
tion, leading to language-specifc gestural repertoires (cf. Kita, 2009). The cross-modal integration
is further refected in the parallel development of the modalities in childhood (Colletta et al., 2015),
parallel breakdown in stuttering (Mayberry & Jaques, 2000), disfuency (Graziano & Gullberg, 2018),
and aphasia (Rose, 2006). The cross-modal interdependence is further seen in the (negative) efects
of restricted gesture production on speech fuency and content (Frick-Horbury & Guttentag, 1998).
In language comprehension, language users integrate gesture information into meaning representations
whether they overtly attend to it or not (Gullberg & Kita, 2009), and they show cross-modal interference
if audiovisual signals express conficting information (Langton et al., 1996) and improved comprehension
when they align (Hostetter, 2011). Gestures facilitate lexical ambiguity resolution (Holle & Gunter, 2007),
as well as syntactic (Holle et al., 2012) and pragmatic interpretation (Kelly et al., 1999). Critically for L2
processing and learning, gestures improve recall and learning (Cook & Fenn, 2017 for overviews).
The cross-modal integration has a neurological basis. The brain treats gesture and speech similarly.
A gesture whose content does not ft the preceding context or co-occurring word modulates the so-
called N400 component in event-related potential (ERP) studies, which is assumed to refect the ease
of semantic integration (Özyürek, 2014). Moreover, the processing of speech and gesture activates
the brain network associated with semantic processing of speech (the left inferior frontal gyrus, the
medial temporal gyrus, and the superior temporal gyrus/sulcus).
The theories and models attempting to account for these relationships mostly focus on speech
production, and only on so-called representational gestures (depicting objects and actions). A con-
ceptual link between language and gesture is generally assumed, but models difer in their views of
(a) the source of gesture representations (e.g., Chu & Kita, 2016; Hostetter & Alibali, 2008; Krauss
et al., 2000), (b) the precise role of motoric, imagistic, and linguistic infuences on gestures, and (c)
how far down in the speech production process these aspects infuence each other (De Ruiter, 2017;
McNeill, 1992; Özyürek, 2017). Comprehension models are divided across similar issues (Kelly
et al., 2010), reinforced by recent neurocognitive evidence. Not surprisingly, all models and theories
focus on stable, adult, native linguistic systems and minds.
These facts of multimodal language use obviously have implications for L2 psycholinguistics.
The study of speech and gesture has a long history (cf. Kendon, 2004) but has matured into a
research feld only relatively recently. Pioneering work examined multimodal coordination in
adult interaction (Efron, 1941/1972; Kendon, 1972), in clinical settings (Freedman, 1972), and in
child language acquisition (Bates et al., 1977). In the 1990s, cognitive psychology started to probe
the relationships between cognition, language, and gesture (Goldin-Meadow, 2003; McNeill,
1992), launching new research examining the multimodal refexes of cross-linguistic variation
(cf. Kita, 2009), speaker-directed functions of gestures, and the notion that gestures refect con-
ceptual representations, reduce cognitive load, and contribute to learning (Alibali et al., 2000;
Cook et al., 2012).
In psycholinguistic L2 studies, interest in gestures has been cooler. Although multimodality was
discussed early on as a pedagogical tool for improving L2 comprehension in classrooms (e.g., Kell-
erman, 1992), the empirical work was largely lacking. But the advances in L1 studies in the 1990s
launched L2 work examining L2 learners’ use of gesture to resolve linguistic challenges (Gullberg,
138
Studying multimodal language processing
1998), the internalization of L2 knowledge through private speech and gesture (McCaferty, 1998),
and multimodal cross-linguistic infuence (Stam, 2006). Multimodal L2 research has since diversifed,
focusing either on gesture as a tool to probe L2 development, or as a tool for learning and memoriza-
tion (Gullberg, 2010; Stam & Buescher, 2018).

Before introducing the methods, three remarks on gesture fundamentals are in order. Firstly, ges-
tures can be structurally characterized in terms of articulators (hands, eyebrows, head, etc.), place
of articulation, and movement with internal structure. Analyses often target the stroke phase, the
most efortful and meaningful part of the movement (Kendon, 1980; Lücking et al., 2013). Con-
trol over phases and information onset is vital, for example in ERP paradigms. Secondly, gestures
are functionally or semiotically divided into conventionalized (e.g., the thumbs-up emblem), and
non-conventionalized movements (gestures or co-speech gestures). Co-speech gestures in turn are
frequently grouped into those depicting properties of (concrete and abstract) objects or actions
(representational [iconic/metaphoric] gestures), gestures indicating real or imagined things (deictic
gestures/pointing), those marking rhythm (beats), representing non-propositional meaning such
as attitude (pragmatic gestures), or some aspect of interaction itself (interactive gestures) (Kendon,
2004 for taxonomies). Importantly, these labels are not categories or types since functions often
overlap. Thirdly, gestures are deeply multi-functional. They serve addressee-directed communi-
cative functions, shifting depending on interlocutors’ visibility, degree of common ground, and
speech informativeness (Gullberg, 2006; Holler & Bavelas, 2017; Melinger & Levelt, 2004). Ges-
tures also serve speaker-directed cognitive functions, possibly facilitating organization of informa-
tion for speaking (Alibali et al., 2000), lexical retrieval (Krauss et al., 2000), or reducing cognitive
load (Cook et al., 2012).
In the following, methods used to study multimodal processing in production, comprehension,
and acquisition will be surveyed (cf. Textbox 12.1 for a summary of methods applied in L2 studies).
Elicited production tasks: Picture or video retellings/descriptions in referential communication or

director-matcher tasks. Production data are recorded, transcribed, and coded in multimodal annota-
tion software. Gesture coding schemes often include gesture meaning, timing, frequency, and co-
expressivity with speech (cf. Müller et al., 2013, pp. 837–1124).
Judgment and rating tasks: Participants watch videos with speech-gesture stimuli, followed by forced/
multiple-choice tasks, Likert scales, questionnaires, etc.
Mismatch paradigms: Participants watch videos with ± congruent speech-gesture pairs (e.g., chop +
chopping/whisking gesture), sometimes with further control conditions (no gesture, meaningless
movement), or added noise. Viewing is followed by decision tasks (lexical, Stroop, cross-modal prim-
ing paradigms) yielding reaction times, accuracy scores. EEG/ERP and/or fMRI may also be used.
Intervention studies with pre-test/post-test designs: Participants are trained on linguistic items ± gesture
input, ± gesture output. Post-tests include perceptual tasks (e.g., forced/multiple-choice tasks, word
recognition tasks) and production tasks (e.g., free/cued recall, translation, imitation tasks). EEG/ERP
and/or fMRI may also be used at post-test.
139
Marianne Gullberg
Multimodal production processing

Multimodal production studies explore both cognitive and communicative aspects of multimodal
processing. One line of work examines what gestures reveal about linguistic conceptualization, build-
ing on McNeill’s (1992) proposal that utterances arise from a “growth point”, i.e., a newsworthy
element of thought containing both imagistic and linguistic content, jointly expressed in speech
and gesture. McNeill initiated a paradigm of elicited production tasks involving video or picture
retellings/descriptions, originally without any further experimental manipulation. Participants watch
a stimulus, then retell or describe it from memory to an interlocutor while being video-recorded,
unaware that gestures are of interest until the task is completed. To promote gesture production,
settings are mostly interactive with confederate or naive interlocutors in referential communication
or director-matcher tasks. Recordings are transcribed and coded in multimodal annotation software
(e.g., ELAN; Wittenburg et al., 2006). Gesture coding often targets gesture rate (Nicoladis, 2007),
gesture form/meaning (Gullberg, 2009), gesture timing relative to spoken elements (Stam, 2006),
and degree of semantic overlap (co-expressivity) between speech and gesture (Brown & Gullberg,
2008). Interrater reliability measures are vital. The paradigm has been used to explore cross-linguistic
diferences in how native speakers of diferent languages select and package information multimod-
ally while describing the same stimuli (Özyürek et al., 2005), often discussing fndings in terms of
Slobin’s (1996) notion of “thinking for speaking”.
Experimental manipulations include additional (memory) tasks to probe whether gesture produc-
tion in one task afects performance in the second (Cook et al., 2012). Other variations manipulate
task difculty to explore whether gestures aid conceptual planning, for example asking participants
to describe (simpler) or explain (harder) a stimulus (Goldin-Meadow, 2003). Further variations probe
whether gestures aid linguistic encoding, for example by restricting gesture production during speech
(Krauss et al., 2000), or inducing tip-of-the-tongue states (Frick-Horbury & Guttentag, 1998).
Examples of L2 multimodal production processing

The elicited production paradigm has been used to probe three L2 domains: cross-linguistic infu-
ence; reference tracking; and the relationship between profciency, fuency, and lexical access.
Following the observation that L1 speakers of diferent languages gesture diferently as a func-
tion of how their languages encode and organize meaning morphosyntactically (Kita, 2009), studies
started to investigate what gestures could reveal about cross-linguistic infuence (CLI) in L2. Under
the assumption that gestures refect meaning deemed relevant for expression, L1-like gesture patterns
would suggest an L1-like representation, and L2-like gesture patterns a shift to target-like representa-
tions. The vast majority of studies investigate the semantic domain of voluntary and caused motion,
drawing on Talmy’s (1991) distinction between verb- and satellite-framed languages whereby verb-
framed languages prefer to express the path of motion in the verb stem (e.g, descendre “descend” in
French) whereas satellite-framed languages do so in satellite particles or adverbials (e.g., down in
English).
For instance, Stam (2006) showed an animated cartoon to native speakers of (verb-framed) Span-
ish and (satellite-framed) English, and intermediate and advanced Spanish L2 learners of English,
who then retold the story to a naïve interlocutor while being flmed. The learners performed the
task both in L1 and L2 with orders counterbalanced. Speech was coded for parts of speech expressing
path, and number of spoken path elements per clause. Gesture strokes (and post-stroke holds) were
coded for the motion element expressed (path, e.g., up, down, vs. manner, e.g., crawl, slither). Speech-
gesture synchrony was established by frame-by-frame analyses of the video to establish what spoken
elements co-occurred exactly with the gesture stroke. Baseline comparisons of the native groups
identifed two diferent patterns. L1 Spanish speakers produced one spoken path element per clause
140
expressed in path verbs, whereas L1 English speakers produced several path elements per clause in
adverbial satellites. Moreover, Spanish speakers predominantly aligned their path gestures with path
verbs (one gesture per clause), whereas English speakers aligned path gestures with many diferent
speech elements, often accumulating several path gestures per clause. Both L2 groups produced fewer
path elements per clause than L1 English speakers, and did not consistently express path in satellites.
Interestingly, the L2 learners aligned their path gestures in English with several spoken categories,
showing some shift towards an English pattern, but also continued to align path gestures with verbs
in L1 style. Overall, the learners displayed an in-between pattern. Crucially, the multimodal data
provided a more fne-grained view of which meanings were relevant to L2 learners. The association
of an action (verb) with path in L2 production suggests that L1-like representations drive linguistic
planning and conceptualization also in the L2, even if spoken forms are target-like. The paradigm has
productively been used to study infuence from L1 on L2 (Gullberg, 2009), and from L2 on L1 and
convergence (Brown & Gullberg, 2008).
Elicited narrative production has also been used to examine how L2 learners organize information
about who does what to whom (reference tracking) to create coherent discourse, and especially to
probe whether their gesture production is a strategy to disambiguate non-cohesive L2 speech. For
instance, Gullberg (2006) asked Dutch learners of L2 French to retell printed cartoons from memory
to native speaker confederates in L1 and L2 with or without visual access to the interlocutor, coun-
terbalanced for order. A comparison of L1 and L2 production under full visibility showed that L2
speakers used more full lexical NPs for maintained referents (instead of pronouns or zero anaphora)
and typically gestured with each mention. The comparison of L2 production across visibility condi-
tions showed no efect of visibility on speech or gesture rate, suggesting that the L2 pattern was not
a disambiguation strategy since nothing changed when gestures could not be seen. However, a third
analysis revealed that gestures located entities in space with greater spatial precision and consistency
when addressees could see them. The results therefore suggested that the incidence of gesture in
L2 discourse is not driven by disambiguation concerns, whereas the spatial articulation is. When
they can be seen, gestures are used for disambiguation. This highlights the simultaneous addressee-
directed and self-directed function of L2 gestures.
Finally, elicited production data have been used to probe the relationship between fuency, prof-
ciency, and gesture rates. Gesture rates are computed as the number of gesture strokes per time unit,
or per 100 words, or per clause (the latter two options are preferable to avoid confounds with speech
rate). The expectation is that the lower the speakers’ profciency, the higher the gesture rate (Krauss
et al., 2000). Many studies indeed show that L2 speakers typically produce more gestures overall than
native/monolingual speakers (Gullberg, 2012; Nicoladis, 2007). However, the link between gesture
rate, fuency, and profciency is not straightforward. For example, when only representational ges-
tures (linked to vocabulary) are considered, studies fnd both fewer and more numerous gestures in
L2 production for unclear reasons (Nicoladis, 2007 for an overview). Moreover, gestures are much
more likely to occur with fuent than with disfuent speech, both in L1 and L2 production (e.g.,
Graziano & Gullberg, 2018). Also, gesture rates are modulated by task demands (e.g., Aziz & Nico-
ladis, 2019), and individual communicative style (e.g., Nagpal et al., 2011). No studies control for
all these aspects.
Multimodal comprehension processing

Studies of multimodal language comprehension typically ask whether and how multimodal informa-
tion afects language processing—especially in cases of degraded signal quality and ambiguity—and
whether it facilitates or hinders comprehension.
The fundamental paradigm involves showing a video of someone speaking and gesturing, in
which gesture production may be spontaneous or scripted. In the simplest case, participants watch
141
Marianne Gullberg
a video and then perform explicit judgment or ranking tasks (forced/multiple-choice, open-ended
questionnaires), predicted behavior/inference tasks, or recall tasks (Kelly et al., 1999; Krahmer &
Swerts, 2007; Rogers, 1978). Occasionally, new retellings of the played-back materials are coded
for what semantic information is retained, and whether its source was speech or gesture (Cassell
et al., 1999).
In more experimental versions, participants may be presented with audiovisual vs. audio-only
versions, with gestures vs. meaningless movements, with speech-gesture pairs that match semantically
or not (e.g., chop + chopping vs. whisking gesture), or with stimuli presented under diferent levels
of noise (Kelly et al., 2004; Rogers, 1978). Response tasks often involve reaction times and accuracy
scores, as in lexical decision tasks, Stroop tasks in which one modality is to be ignored, or cross-modal
priming (e.g., Langton et al., 1996; Yap et al., 2011). Eye-tracking has been used to measure overt
visual attention to gestures (Gullberg & Holmqvist, 2006), but it is less reliable for testing multimodal
information processing since there is no strong association between information uptake and fxations
in this context (Gullberg & Kita, 2009). Neurocognitive methods are also used, such as event-related
potentials (ERPs), which measure the brain’s electrophysiological responses to a certain stimulus with
high temporal accuracy, and functional magnetic resonance imaging (fMRI), which measures neural
activity based on changes in blood fow to diferent brain areas (Özyürek, 2014).
Examples of L2 multimodal comprehension processing

L2 multimodal comprehension is much less studied than L2 multimodal production, despite the
wide-spread conviction that gesture input should improve L2 comprehension.
In a study investigating the role of gestures, lip movements, and profciency for L2 comprehen-
sion, Sueyoshi and Hardison (2005) presented intermediate and advanced learners of L2 English with
videos of a lecture in English on ceramics for beginners. The lecturer’s speech and gestures were
spontaneous and non-scripted. Participants were randomly assigned to one of three groups presented
with stimuli showing gesture + face, face-only, or audio-only. Immediately after viewing each clip,
participants answered multiple-choice comprehension questions. The audio-only group did worse
than both of the audiovisual groups, who did not difer. However, profciency modulated the results
such that the face-only condition yielded the highest scores for the advanced learners, whereas the
gesture + face condition proved most helpful to the intermediate learners. To account for the prof-
ciency efects, the authors argue that manual gestures may be more saliently informative at lower pro-
fciency levels than lip movements which require a certain linguistic knowledge base to be exploited.
In a study probing the role of audiovisual information for L2 comprehension in noise Drijvers and
Özyürek (2020) manipulated access to lip movements, representational gestures, and audio quality.
They showed 220 short video clips in which a native Dutch speaker uttered a single verb with or
without a spontaneously produced representational gesture matching the verb meaning (e.g., type +
a typing gesture), and with or without the speaker’s lips being blurred. Audio was presented as clear
speech, moderately, or severely distorted. Importantly, the gestures were independently checked for
their semantic ft to the words, their ambiguity in the absence of speech, their degree of iconic-
ity (i.e., how well a gesture represented an action), and the temporal overlap between the gesture
stroke and speech. Two visual-only control conditions (no sound) were also created showing lips +
gesture or lips-only to check what information was detectable from the visual signal alone. German
learners of L2 Dutch were presented with videos and wrote down the verb perceived after each
trial. Their knowledge of the L2 verbs was also checked, and trials with unknown L2 verbs were
excluded from analysis. The results showed that L2 comprehenders did worse with increasing noise,
and did better the more visual articulators were present (lips + gestures being optimal). Crucially,
however, the enhancement was greatest at moderate noise levels. At the worst noise level, L2 com-
prehenders could presumably not link lip or gesture information to any linguistic representation
142
(cf. Sueyoshi & Hardison, 2005). Similar results were also found in an ERP study using a mismatch
paradigm and with scripted gestures (Drijvers & Özyürek, 2018). The ERP results showed a more
negative N400 amplitude (greater processing cost) to mismatched than matched gestures in clear
audio in both groups. However, L2 comprehenders showed no diference between the match/mis-
match conditions in noise, whereas L1 comprehenders did, suggesting that gestures could not be
integrated with anything in the absence of sufcient auditory information in the L2. Both studies
highlight that co-speech gestures do not automatically make meaning transparent in the absence of
sufcient speech information. Gesture comprehensibility was tested in the absence of speech and
was found to be low. Hence, gestures and speech are co-dependent for their interpretation, and L2
processors cannot make use of gestural semantic information if the auditory cues are too weak. Visual
input is not a general solution to boost all L2 comprehension.
Multimodal L2 learning
Beyond language production and comprehension, a large body of research has shown that seeing and
producing gestures afects memory and learning (Cook & Fenn, 2017 for overviews). For example,
learners who see and produce gestures learn more about math than those who do not, and those with
lower working memory capacity beneft more. Gesture input helps, but efects are typically greater
when learners also produce gesture output. It is argued that gestures boost learning and recall since
richer information during encoding by virtue of the engagement of more modalities creates greater
depth of processing and more stable memory representations (Clark & Paivio, 1991). Recent neuro-
cognitive evidence lends support to this idea showing that sensorimotor brain networks are activated
and grow the more sensory modalities are connected to a word (Macedonia et al., 2019).
In contrast to the relative paucity of studies investigating multimodal L2 comprehension, studies
examining multimodal efects on L2 language learning have proliferated. The fundamental paradigm
is a pre-test/post-test design in which interventions involve explicit training of linguistic items with or
without accompanying gestures in the input, and with or without learners’ themselves producing ges-
tures during training (Allen, 1995; Tellier, 2008). Further manipulations include input with semanti-
cally matching/mismatching gestures, and meaningless control movements (García-Gámez & Macizo,
2019; Kelly et al., 2009). Interventions are followed by immediate and/or delayed post-tests to probe
learning efects. Post-tests include both perception tasks (forced/multiple-choice, word-meaning asso-
ciations, word recognition, discrimination tasks) and production tasks (free/cued recall, L2-L1/L1-L2
translations, imitation tasks). ERPs and fMRI are sometimes recorded during post-testing.
Studies show benefts of gesture on the learning of vocabulary (Kelly et al., 2009; Morett, 2014),
phonology (Baills et al., 2019), and L2 pronunciation (Kushch et al., 2018). Importantly, benefts
depend on semantic speech-gesture congruity, i.e., on movements being meaningfully related to lan-
guage (Kelly et al., 2009; Macedonia et al., 2011). Benefts may also difer across linguistic domains,
with greater efects for the lexicon than for phonology and phonetics, for example (Kelly, 2017).
Various gestures contribute: fully conventionalized gestures known as emblems (e.g., a thumb-index-
circle gesture meaning “good”; Allen, 1995), representational gestures depicting concrete content
(Kelly et al., 2009; Morett, 2014), abstract content (e.g., a lateral sweep to depict duration) (Baills
et al., 2019; Hirata et al., 2014), and rhythmic beats (Kushch et al., 2018).
Examples of L2 multimodal learning

Allen (1995) investigated whether American frst-year students of L2 French would learn L2 idiom-
atic expressions better if they were accompanied by French conventional gestures (emblems) in train-
ing. The treatment consisted of 50 French expressions with English translations and 50 emblematic
gestures (e.g., Il est complètement rond, celui-là. [Lit. translation: He is completely round, that one.]
143
Marianne Gullberg
“That guy is drunk.” with a closed fst rotating on the nose). All items were pre-tested to ensure
that they were not known to American students. Five 30-minute video sessions, each consisting of
a pretest (translation), a treatment (ten expressions each), and a post-test (translation), were adminis-
tered over a period of ten weeks in a between-subjects design. The experimental group saw gestures
in treatment and in post-test; the no-treatment group saw no gestures; and a comparison group saw
gestures in the post-test only. In the treatment, participants frst saw the written French and English
sentences. The no-treatment/comparison groups then heard and repeated the sentences three times.
The experimental group heard and saw the sentences with gestures, and they repeated the gestures
themselves three times without speaking. The experimental group also repeated the gesture before
writing the translation at post-test. A delayed post-test was administered 11 weeks after the frst ses-
sion. The experimental group did better both at immediate and delayed post-tests than the other
groups. Although the results were explained as an efect of added depth of processing from gestures,
the nature of that processing advantage is unclear. Since emblems are not iconically linked to the spo-
ken expressions, it is unlikely that the greater depth of processing was related to semantic processing.
Kelly et al. (2009) used a mismatch paradigm to investigate the importance of the semantic speech-
gesture link. They tested whether representational gestures that depict semantic content facilitate
English learners’ acquisition of Japanese vocabulary. In the stimuli an instructor introduced twelve
Japanese “action words” and translations (e.g., “The next word is nomu. Nomu means drink. Nomu
means drink.”, p. 317). Items were presented as speech-only; speech-gesture congruent pairs with
scripted gestures (drink + drinking gesture), speech-gesture incongruent pairs (drink + face-washing
gesture), or repeated-speech. The immediate post-tests included an English translation task and a
multiple-choice task. Delayed post-tests were administered two days and a week later. The results
showed that, at all post-tests, congruent gesture input yielded better performance than speech-only
and repeated-speech. Incongruent gesture input led to the worst performance of all. The suggested
explanation was that congruent co-speech gestures establish a conceptual link to the meaning of
speech, thus creating a stronger multimodal memory representation (cf. Allen, 1995). To explore the
neural mechanisms of such a proposal, Kelly et al. also conducted an ERP experiment with stimuli
presented as speech-only, or speech-congruent gesture pairs. ERPs were recorded during a post-test
word recognition task, followed by an ofine translation task. The behavioral results replicated the
frst experiment. The ERP results showed a larger bilateral parietal Late Positive Complex (indexing
recollection) for words learned with congruent gestures than for words learned without. There were
no diferences for the N400 component (here assumed to index familiarity). The authors interpreted
the results as supporting the idea that gestures add imagistic detail to multimodal representations, thus
creating deeper and stronger memory traces. These early ERP results have been extended in fMRI
studies, adding details on the sensorimotor and language-related neural networks implicated in learn-
ing with and without gestures (Macedonia et al., 2019).
Methodological refections
The elicited production paradigms involving gestures clearly ofer a richer view of L2 users’ capaci-
ties and conceptualization processes than speech alone. They help address the vexing suspicion that
learners’ speech may be under-specifed with regard to what information they attend to or try to
express. The paradigms are simple to perform and provide some control in allowing several speakers
to talk about the same content under the same circumstances. They also provide rich data for several
types of analyses. On the downside, multimodal transcription and coding is laborious and time-
consuming, leading to small samples and concerns about generalizability. Moreover, unlike speech,
gesture production is not obligatory—speakers do not always gesture or may chose not to. This may
lead to additional “data loss”, uneven sample sizes across conditions, etc. Replicability and reliability
are also at stake given that gesture coding schemes are not standardized.
144
Multimodal comprehension paradigms also ofer more detailed views of L2 users’ receptive language
capacities than speech-only tasks. Being more experimental in nature, they ofer better control. That
said, the experimental paradigms used to study multimodal comprehension often operate with artifcial,
scripted gestures (sometimes with heads blocked) whose properties relative to spontaneous gestures
remain unclear. Discussions of the ecological validity of stimuli are rare. Fortunately, more sophisticated
and natural/realistic manipulations are now gaining ground (cf. Drijvers & Özyürek, 2020).
Overall, the most glaring general concern—both for production and comprehension studies—is
that not enough details are provided on gestures themselves, regarding spatiotemporal properties
(location in space, number of hands used, (a-)symmetry of movement, etc.), degree of interpretabil-
ity, detailed temporal alignment relative to speech, etc. This is especially problematic when analyses
are not supported by detailed descriptions and interrater reliability measures. We need more detailed
descriptions—and better training—in gesture analysis.

As should be clear by now, most issues remain wide open in L2 multimodal processing. It is strik-
ing that much of the existing L2 multimodal processing work is divorced from and not grounded in
SLA concerns. There is no reference to any of the big L2 issues such as the adult capacity for input
processing, explicit/implicit/statistical learning and processing, the role of individual diferences
(working memory, executive control, etc.), cross-linguistic infuence in comprehension, parsing-to-
learn issues, etc. Conversely and equally striking, “traditional” L2 studies in these domains have not
addressed multimodal language. It is clear that speech and gesture jointly contribute meaning and
prominence, refect and help construct linguistic representations, but almost everything remains to
be discovered concerning the development and shifts of such representations in learning, attrition,
and multilingualism. It is also time to move beyond lexical processing and explicit instruction, and
consider other linguistic domains and implicit processing. There are thus enormous theoretical and
empirical knowledge gaps waiting to be flled.
The biggest challenge of all and the most important future direction for L2 processing studies is
therefore to actually include multimodality. Language use is a multimodal endeavor. To ignore that is
to ignore half of our language use—whatever our skills in the language at hand. As psycholinguists it
behooves us to shift our theories and models away from monomodal (and monolingual) perspectives
towards multimodal (and multilingual) ones.
What is the role of gestures for implicit or statistical input processing, e.g., promoting word segmenta-
tion, linking meaning to word forms in continuous input, or promoting the discovery of grammatical
constructions (cf. Gullberg et al., 2012)?
Why does speech shift to the target more readily than gestures in L2 production (gestures show L1 pat-
terns even when speech does not)? When, and why, does gesture production shift?
How do L1 comprehenders process L2 users’ multimodal output (e.g., manual “foreign accents”)?
How do diferences in cognitive task load, linguistic complexity, and gesture production afect L2 speech
production, given that it has been suggested that gesture production reduces cognitive load in L1
contexts?
Do gestures help the parsing of speech predictively, given that visual signals are often preposed relative to
the auditory signal (Holler & Levinson, 2019)? Would this help L2 parsing too?
145
Marianne Gullberg
Further reading
Cook, S., & Fenn, K. M. (2017). The function of gesture in learning and memory. In R. Breckinridge Church,
M. W. Alibali, & S. D. Kelly (Eds.), Why gesture?: How the hands function in speaking, thinking and communicating
An overview of the efects of gestures on learning.
Holler, J., & Levinson, S. C. (2019). Multimodal language processing in human communication. Trends in Cogni-
A model of multimodal language processing in production and comprehension.
Kendon, A. (2004). Gesture. Visible action as utterance. Cambridge University Press.
McNeill, D. (2015). Why we gesture. Cambridge University Press.
Two foundational texts about speech and gesture in language use.
References
Alibali, M. W., Kita, S., & Young, A. J. (2000). Gesture and the process of speech production: We think, there-
fore we gesture. Language and Cognitive Processes, 15(6), 593–613.
Allen, L. Q. (1995). The efect of emblematic gestures on the development and access of mental representations
of French expressions. Modern Language Journal, 79(4), 521–529.
Andrä, C., Mathias, B., Schwager, A., Macedonia, M., & von Kriegstein, K. (2020). Learning foreign language
vocabulary with gestures and pictures enhances vocabulary memory for several months post-learning in
eight-year-old school children. Educational Psychology Review, 32(3), 815–850.
Aziz, J. R., & Nicoladis, E. (2019). “My French is rusty”: Profciency and bilingual gesture use in a majority
English community. Bilingualism: Language and Cognition, 22(4), 826–835.
Baills, F., Suárez-González, N., González-Fuente, S., & Prieto, P. (2019). Observing and producing pitch ges-
tures facilitates the learning of Mandarin Chinese tones and words. Studies in Second Language Acquisition,
41(1), 33–58.
Bates, E., Benigni, L., Bretherton, I., Camaioni, L., & Volterra, V. (1977). From gesture to the frst word: On
cognitive and social prerequisites. In M. Lewis & L. A. Rosenblum (Eds.), Interaction, conversation, and the
development of language (pp. 247–307). Wiley.
Bavelas, J. B. (1994). Gestures as part of speech: Methodological implications. Research on Language and Social
Interaction, 27(3), 201–221.
Brown, A., & Gullberg, M. (2008). Bidirectional crosslinguistic infuence in L1-L2 encoding of Manner in
speech and gesture: A study of Japanese speakers of English. Studies in Second Language Acquisition, 30(2),
225–251.
Cassell, J., McNeill, D., & McCullough, K.-E. (1999). Speech-gesture mismatches: Evidence for one underlying
representation of linguistic and nonlinguistic information. Pragmatics & Cognition, 7(1), 1–33.
Chu, M., & Kita, S. (2016). Co-thought and co-speech gestures are generated by the same action generation
process. Journal of Experimental Psychology: Learning, Memory, and Cognition, 42(2), 257–270.
Clark, H. H. (1996). Using language. Cambridge University Press.
Clark, J. M., & Paivio, A. (1991). Dual coding theory and education. Educational Psychology Review, 3(3),
149–210.
Colletta, J.-M., Guidetti, M., Capirci, O., Cristilli, C., Demir, O. E., Kunene-Nicolas, R. N., & Levine, S.
(2015). Efects of age and language on co-speech gesture production: An investigation of French, American,
and Italian children’s narratives. Journal of Child Language, 42(1), 122–145.
Cook, S., & Fenn, K. M. (2017). The function of gesture in learning and memory. In R. Breckinridge Church,
M. W. Alibali, & S. D. Kelly (Eds.), Why gesture?: How the hands function in speaking, thinking and communicating
Cook, S., Yip, T. K., & Goldin-Meadow, S. (2012). Gestures, but not meaningless movements, lighten working
memory load when explaining math. Language and Cognitive Processes, 27(4), 594–610.
De Ruiter, J. P. (2017). The asymmetric redundancy of gesture and speech. In R. Breckinridge Church, M.
W. Alibali, & S. D. Kelly (Eds.), Why gesture?: How the hands function in speaking, thinking and communicating
Drijvers, L., & Özyürek, A. (2018). Native language status of the listener modulates the neural integration of
speech and iconic gestures in clear and adverse listening conditions. Brain and Language, 177–178, 7–17.
146
Drijvers, L., & Özyürek, A. (2020). Non-native listeners beneft less from gestures and visible speech than native
listeners during degraded speech comprehension. Language & Speech, 63(2), 209–220.
Efron, D. (1941/1972). Gestures, race and culture. Mouton.
Freedman, N. (1972). The analysis of movement behavior during the clinical interview. In A. W. Siegman & B.
Pope (Eds.), Studies in dyadic communication (pp. 153–175). Pergamon.
Frick-Horbury, D., & Guttentag, R. E. (1998). The efects of restricting hand gesture production on lexical
retrieval and free recall. American Journal of Psychology, 111(1), 43–62.
García-Gámez, A. B., & Macizo, P. (2019). Learning nouns and verbs in a foreign language: The role of gestures.
Goldin-Meadow, S. (2003). Hearing gesture: How our hands help us think. The Belknap Press.
Graziano, M., & Gullberg, M. (2018). When speech stops, gesture stops: Evidence from developmental and
crosslinguistic comparisons. Frontiers in Psychology, 9(879).
Gullberg, M. (1998). Gesture as a communication strategy in second language discourse. A study of learners of French and
Swedish. Lund University Press.
Gullberg, M. (2006). Handling discourse: Gestures, reference tracking, and communication strategies in early
L2. Language Learning, 56(1), 155–196.
Gullberg, M. (2009). Reconstructing verb meaning in a second language: How English speakers of L2 Dutch
talk and gesture about placement. Annual Review of Cognitive Linguistics, 7, 222–245.
Gullberg, M. (2010). Methodological refections on gesture analysis in SLA and bilingualism research. Second
Gullberg, M. (2012). Bilingualism and gesture. In T. K. Bhatia & W. C. Ritchie (Eds.), The handbook of bilingual-
ism and multilingualism (2nd ed., pp. 417–437). Wiley-Blackwell.
Gullberg, M., & Holmqvist, K. (2006). What speakers do and what listeners look at. Visual attention to gestures
in human interaction live and on video. Pragmatics and Cognition, 14(1), 53–82.
Gullberg, M., & Kita, S. (2009). Attention to speech-accompanying gestures: Eye movements and information
uptake. Journal of Nonverbal Behavior, 33(4), 251–277.
Gullberg, M., Roberts, L., & Dimroth, C. (2012). What word-level knowledge can adult learners acquire after
minimal exposure to a new language? International Review of Applied Linguistics, 50(4), 239–276.
Hirata, Y., Kelly, S. D., Huang, J., & Manansala, M. (2014). Efects of hand gestures on auditory learning of
second-language vowel length contrasts. Journal of Speech, Language, and Hearing Research, 57, 2090–2101.
Holle, H., & Gunter, T. C. (2007). The role of iconic gestures in speech disambiguation: ERP evidence. Journal
of Cognitive Neuroscience, 19(7), 1175–1192.
Holle, H., Obermeier, C., Schmidt-Kassow, M., Friederici, A., Ward, J., & Gunter, T. (2012). Gesture facilitates
the syntactic analysis of speech. Frontiers in Psychology, 3(74).
Holler, J., & Bavelas, J. (2017). Multi-modal communication of common ground: A review of social functions.
In R. B. Church, M. W. Alibali, & S. D. Kelly (Eds.), Why gesture?: How the hands function in speaking, thinking
and communicating (pp. 213–240). John Benjamins.
Holler, J., & Levinson, S. C. (2019). Multimodal language processing in human communication. Trends in Cogni-
Hostetter, A. (2011). When do gestures communicate? A meta-analysis. Psychological Bulletin, 137(2), 297–315.
Hostetter, A., & Alibali, M. W. (2008). Visible embodiment: Gestures as simulated action. Psychonomic Bulletin
& Review, 15(3), 495–514.
Kellerman, S. (1992). “I see what you mean”: The role of kinesic behaviour in listening and implications for
foreign and second language learning. Applied Linguistics, 13(3), 239–257.
Kelly, S. D. (2017). Exploring the boundaries of gesture-speech integration during language comprehension. In
R. Breckinridge Church, M. W. Alibali, & S. D. Kelly (Eds.), Why gesture?: How the hands function in speaking,
thinking and communicating (pp. 243–265). John Benjamins.
Kelly, S. D., Barr, D. J., Church, R. B., & Lynch, K. (1999). Ofering a hand to pragmatic understanding: The
role of speech and gesture in comprehension and memory. Journal of Memory and Language, 40(4), 577–592.
Kelly, S. D., Kravitz, C., & Hopkins, M. (2004). Neural correlates of bimodal speech and gesture comprehen-
sion. Brain and Language, 89(1), 253–260.
Kelly, S. D., McDevitt, T., & Esch, M. (2009). Brief training with co-speech gesture lends a hand to word learn-
ing in a foreign language. Language and Cognitive Processes, 24(2), 313–334.
Kelly, S. D., Özyürek, A., & Maris, E. (2010). Two sides of the same coin: Speech and gesture mutually interact
to enhance comprehension. Psychological Science, 21(2), 260–267.
Kendon, A. (1972). Some relationships between body motion and speech: An analysis of an example. In A. W.
Siegman & B. Pope (Eds.), Studies in dyadic communication (pp. 177–210). Pergamon.
Kendon, A. (1980). Gesticulation and speech: Two aspects of the process of utterance. In M. R. Key (Ed.), The
relationship of verbal and nonverbal communication (pp. 207–227). Mouton.
147
Marianne Gullberg
Kendon, A. (2004). Gesture. Visible action as utterance. Cambridge University Press.

Kita, S. (2009). Cross-cultural variation of speech-accompanying gesture: A review. Language and Cognitive Pro-
cesses, 24(2), 145–167.
Krahmer, E., & Swerts, M. (2007). The efects of visual beats on prosodic prominence: Acoustic analyses, audi-
tory perception and visual perception. Journal of Memory and Language, 57(3), 396–414.
Krauss, R. K., Chen, Y., & Gottesman, R. F. (2000). Lexical gestures and lexical access: A process model. In D.
McNeill (Ed.), Language and gesture (pp. 261–283). Cambridge University Press.
Kushch, O., Igualada, A., & Prieto, P. (2018). Prominence in speech and gesture favour second language novel
word learning. Language, Cognition and Neuroscience, 33(8), 992–1004.
Langton, S. R. H., O’Malley, C., & Bruce, V. (1996). Actions speak no louder than words: Symmetrical cross-
modal interference efects in the processing of verbal and gestural information. Journal of Experimental Psychol-
ogy: Human Perception and Performance, 22(6), 1357–1375.
Loehr, D. P. (2007). Aspects of rhythm in gesture and speech. Gesture, 7(2), 179–214.
Lücking, A., Bergman, K., Hahn, F., Kopp, S., & Rieser, H. (2013). Data-based analysis of speech and gesture:
The Bielefeld Speech and Gesture Alignment corpus (SaGA) and its applications. Journal on Multimodal User
Interfaces, 7(1), 5–18.
Macedonia, M., Müller, K., & Friederici, A. D. (2011). The impact of iconic gestures on foreign language word
learning and its neural substrate. Human Brain Mapping, 32(6), 982–998.
Macedonia, M., Repetto, C., Ischebeck, A., & Mueller, K. (2019). Depth of encoding through observed ges-
tures in foreign language word learning. Frontiers in Psychology, 10(33).
Mayberry, R. I., & Jaques, J. (2000). Gesture production during stuttered speech: Insights into the nature of
gesture-speech integration. In D. McNeill (Ed.), Language and gesture (pp. 199–214). Cambridge University
Press.
McCaferty, S. G. (1998). Nonverbal expression and L2 private speech. Applied Linguistics, 19(1), 73–96.
McNeill, D. (1992). Hand and mind. What gestures reveal about thought. University of Chicago Press.
McNeill, D. (2015). Why we gesture. Cambridge University Press.
Melinger, A., & Levelt, W. J. M. (2004). Gesture and the communicative intention of the speaker. Gesture, 4(2),
119–141.
Morett, L. M. (2014). When hands speak louder than words: The role of gesture in the communication, encod-
ing, and recall of words in a novel second language. The Modern Language Journal, 98(3), 834–853.
Müller, C., Cienki, A., Fricke, E., Ladewig, S., McNeill, D., & Tessendorf, S. (Eds.). (2013). Body—language—
communication: An international handbook on multimodality in human interaction. Mouton de Gruyter.
Nagpal, J., Nicoladis, E., & Marentette, P. (2011). Predicting individual diferences in L2 speakers’ gestures.
Nicoladis, E. (2007). The efect of bilingualism on the use of manual gestures. Applied Psycholinguistics, 28(3),
441–454.
Özyürek, A. (2014). Hearing and seeing meaning in speech and gesture: Insights from brain and behaviour.
Philosophical Transactions of the Royal Society B: Biological Sciences, 369(1651), 20130296.
Özyürek, A. (2017). Function and processing of gesture in the context of language. In R. B. Church, M. W.
Alibali, & S. D. Kelly (Eds.), Why gesture? How the hands function in speaking, thinking, and communicating (pp.
Özyürek, A., Kita, S., Allen, S., Furman, R., & Brown, A. (2005). How does linguistic framing of events
infuence co-speech gestures? Insights from crosslinguistic variations and similarities. Gesture, 5(1/2),
219–240.
Rogers, W. T. (1978). The contribution of kinesic illustrators toward the comprehension of verbal behavior
within utterances. Human Communication Research, 5(1), 54–62.
Rose, M. L. (2006). The utility of arm and hand gesture in the treatment of aphasia. Advances in Speech-Language
Pathology, 8(2), 92–109.
Slobin, D. I. (1996). From “thought and language” to “thinking for speaking”. In J. J. Gumperz & S. C. Levinson
(Eds.), Rethinking linguistic relativity (pp. 70–96). Cambridge University Press.
Stam, G. (2006). Thinking for Speaking about motion: L1 and L2 speech and gesture. International Review of
Applied Linguistics, 44(2), 143–169.
Stam, G., & Buescher, K. (2018). Gesture research. In A. Phakiti, P. De Costa, L. Plonsky, & S. Starfeld (Eds.),
Palgrave handbook of applied linguistics research methodology (pp. 793–809). Palgrave Macmillan.
Sueyoshi, A., & Hardison, D. M. (2005). The role of gestures and facial cues in second language listening com-
prehension. Language Learning, 55(4), 661–699.
Talmy, L. (1991). Paths to realization: A typology of event confation. In L. A. Sutton, C. Johnson, & R. Shields
(Eds.), Proceedings of the Berkeley Linguistics Society (Vol. 17, pp. 480–519). Berkeley Linguistics Society.
148
Tellier, M. (2008). The efect of gestures on second language memorisation by young children. Gesture, 8(2),
219–235.
Wittenburg, P., Brugman, H., Russel, A., Klassman, A., & Sloetjes, H. (2006). ELAN: A professional framework
for multimodality research. In Proceedings of the ffth international conference on language resources and evaluation
(LREC 2006) (pp. 1556–1559). Genoa.
Yap, D.-F., So, W.-C., Melvin Yap, J.-M., Tan, Y.-Q., & Teoh, R.-L. S. (2011). Iconic gestures prime words.
Cognitive Science, 35(1), 171–183.
149
13
CONDUCTING REACTION TIME
RESEARCH IN SECOND LANGUAGE
PSYCHOLINGUISTICS
Phillip Hamrick
Introduction
The goal of psycholinguistic research is to understand the mechanics of how language is learned, pro-
cessed, and produced. Perhaps the most widely employed method of investigating these questions is to use
chronometric data (also known as reaction time, response time, response latency, etc.), and the insights
gained from such reaction time (RT) studies have heavily infuenced psycholinguistics and related felds
(Baayen & Milin, 2010; see also Bernolet, 2023 [this volume]; Elgort & Warren, 2023 [this volume]).
The RTs that psycholinguists focus on are often quite small, with meaningful efects being found
on very short timescales (e.g., 10–50 milliseconds [ms]) and those efects often being smaller in sec-
ond language (L2) learners than in native speakers or monolinguals (e.g., Frenck-Mestre & Prince,
1997). In cases in which small RT efects are expected, it is paramount that our tools for detecting
such small efects are up to the job.
With modern technology, it would seem that detection of small RT efects should present no
problem for new computers running appropriate experimental software. This view is held, at least
tacitly, by any psycholinguist running RT experiments using such technology (and that appears to
be a good deal of the feld). Unfortunately, this trust in modern equipment is not necessarily well
placed. As we shall see, sometimes older hardware surpasses newer hardware in performance, and,
regardless, all hardware and software are subject to important technological limitations that research-
ers must be aware of as they design and conduct their chronometric research.
To that end, this paper aims to do the following:
1. Describe some technological factors that threaten chronometric research, which includes any
method involving a computer or computer peripherals (e.g., button box) as a means of collect-
ing timing data, including button-press data, oral response data, mouse- or eye-movement data,
event-related potentials, etc.;
2. Survey the benefts and drawbacks of some prominent hardware and software used in chrono-
metric research;
3. Propose solutions to these issues, including better reporting practices, fnding benchmark infor-
mation or producing it yourself, and more transparent handling of data preparation before
analysis.
Before doing so, I illustrate some general features of RT-based research in L2 psycholinguistics in
order to provide some points of reference for the concerns I will raise.
150 DOI: 10.4324/9781003018872-15

Reaction time research in second language psycholinguistics
Chronometry: This is the science of accurately measuring time. In psycholinguistics, it is the basis of
research methods that require synchronizing between a computer or time-recording device and
human language behavior.
Lag: (accuracy) Lag in response time or chronometric error is the delay between the intended time that
an event is presented or recorded and the time in which it is actually presented or recorded, usually in
ms. All software and hardware introduce lag.
Jitter (precision): The variability in the lag of presented or recorded events, often expressed as a range
or a standard deviation of the lag. Software and hardware vary in the amount of jitter they produce.
Tick: This is the amount of time it takes for a computer monitor or screen to “draw” and display a com-
plete screen. Most modern computer screens have a tick of approximately 16.67 ms. Importantly, the
visual information drawn in this time is not necessarily drawn all at once. Monitors and screens vary
in which information presented within a tick gets drawn frst.
Refresh rate: Tis is the number of ticks in a second, usually expressed in Hertz (Hz). For example,
a computer that takes 16.67 ms to produce a screen (one tick) has a refresh rate of 1,000 ms ÷
16.67 ms = 60 Hz.
Benchmarking: For chronometric research, benchmarking involves systematically measuring the lag and
latency of hardware and software to determine their overall accuracy and precision with setting timing
standards as a goal.
Chronometry in L2 psycholinguistics
Contemporary L2 researchers employ a variety of chronometric methods, such as lexical decision and
picture naming, along with more sophisticated techniques such as eye-tracking and electroencepha-
lography (EEG). These techniques all consider RT as crucial data. For example, lexical decision tasks
ask participants to view a string of letters or listen to a sequence of sounds and decide as quickly and
accurately as possible whether they form a real word. RTs for correct responses to real words are taken
to index lexical access. A diferent way to measure lexical access is via a picture naming task in which
participants are presented with pictures of objects and are asked to orally produce each object’s name
as quickly and accurately as possible. Here RTs are usually measured as the time from the onset of the
visual stimulus (i.e., the picture) to the onset of an orally produced target word. Thus, RT data from
two diferent tasks can give us converging insights on the nature of the mental lexicon (for more details
on these kinds of tasks, see Bernolet, 2023 [this volume]; Elgort & Warren, 2023 [this volume]).
Of course, chronometric methods (like any method) can only provide a window into the cogni-
tive and neural workings of language if we can trust their validity, including their timing. To illustrate
how timing errors might occur, consider a hypothetical chronometric study examining participants’
ability to detect whether an auditorily presented word matches a briefy presented printed word. Par-
ticipants are instructed to press a corresponding button (one for match and one for mismatch trials)
on a response device while brain-based response data are gathered via EEG. Figure 13.1 shows how
a given trial might play out as idealized by the experimenter compared against how things may play
out depending on the timing of the computer hardware and software.
Ideally, stimuli are presented precisely when intended and latencies in response to those stimuli
are, in general, timed against the onset (beginning) or ofset (end) of the stimuli. In reality, there are
timing errors in stimulus presentation and in the measurement of both overt behavioral responses
(button pressing) as well as with measures that do not require an overt response (EEG). If we recall
that many L2 psycholinguistic efect sizes in RT are quite small then the potential impact of these
151
Phillip Hamrick
Figure 13.1 Illustration of timing errors that normally occur (bottom) versus the ideal timing that most experi-
menters have in mind (top), inspired by Plant and Quinlan (2013).
timing errors becomes much more concerning (e.g., in one of the experiments in their seminal study,
Frenck-Mestre and Prince (1997) found semantic priming efects of 12 ms and 29 ms for L2 learners
and bilinguals, respectively).
In the next sections, I detail how basic characteristics of our experiment hardware and software
can add both lag and jitter to our RT data. Throughout, the reader should consider whether and to
what degree the issues raised constitute threats to the validity of their research.
Methods and paradigms: hardware

Each piece of hardware available to the modern psycholinguist comes with its own benefts and
drawbacks that researchers must weigh carefully. Here, I describe how hardware common to just
about all of psycholinguistic research (e.g., computer monitors, speakers, and response devices like
keyboards) is susceptible to timing error. Note that further hardware beyond my present focus, such
as eye-tracking and EEG (see, e.g., Jing et al., 2004, for timing issues in EEG), have their own timing
qualities that must be considered,1 and, at any rate, they almost always interface with the hardware
discussed here for data collection.
Visual presentation
The onset delay of a visual stimulus should be as near zero as possible, assuming the experimenter
wants the word to appear instantaneously. However, on standard computer monitors/screens nothing
appears instantaneously on the screen. Instead, computers “draw” the stimulus. The amount of time
it takes a monitor to draw a stimulus depends on the monitor’s drawing mechanisms and the com-
munication time between it and the computer. Despite the advances in liquid crystal display (LCD)
152
monitors, which represent the vast majority of the current market, refresh rates are generally faster
in older monitors like the cathode ray tube (CRT) monitors popular in computer systems from the
1980s and 1990s. A CRT monitor “draws” an image from the top left of a screen horizontally and
vertically, and upon completion in the bottom right corner of the screen returns to the top left and
begins again, taking approximately 10–12 ms in most CRTs, although some work as fast as every
4 ms (Garaizar, Vadillo et al., 2014).
While most current LCD monitors have refresh rates of around 60 Hz (16.67 ms/tick), they sufer
from other problems. For example, Plant and Turner (2009) report mean image onset lags for thin-
flm transistor (TFT) monitors (a type of faster, brighter, and more expensive LCD), ranging from
26.24 to 53.48 ms. Luckily, the jitter is quite low (SDs < 1 ms). Lagroix et al. (2012) found that esti-
mates of rise time (i.e., how long it takes for an image to “fade in” to view) are quite fast (e.g., 1–6 ms),
but still shy of the < 1 ms rise time found in most CRTs. Thus, when very small RT efects are the
focus of study, CRT monitors may provide important benefts over modern LCD screens (e.g., Neath
et al., 2011). This is also to say that if you are in a laboratory that still uses old CRT monitors, do not
think that you are necessarily working with inferior technology, at least for chronometric purposes.
Since timing errors may vary from monitor to monitor, researchers should keep in mind that data
collected with a single monitor will introduce a defnite image onset lag with (probably) minimal
jitter within that monitor. However, researchers who conduct studies with multiple monitors at
either single or multiple sites or via the Internet should bear in mind that there could be as much
as 25 ms of diference in image onset delays between monitors according to available benchmarked
data (e.g., Plant & Turner, 2009). Moreover, if any important research variable (e.g., experimental
condition, group, etc.) is confounded with the monitor used, then spurious results could follow. It
is recommended that research be carried out on the same computer and monitor if this technologi-
cal confound cannot be controlled (either by using multiple identical monitors or piloting to check
comparable lag and jitter rates across monitors; see Benchmarking and calibration).
Auditory presentation
Auditory presentation is critical for numerous psycholinguistic paradigms, including auditory lexi-
cal decision, self-paced listening, word monitoring, visual world eye-tracking, shadowing tasks, etc.
Unfortunately, computer sound cards, speakers, and headphone systems all introduce their own
additional lag and jitter. In fact, experiment-generation software manufacturers warn us that this is
the case. Consider Psychology Software Tools’ disclaimer for E-Prime:
E-Prime 3.0 reports millisecond accurate timing. This does not mean that E-Prime 3.0 is
capable of making hardware do things it cannot. For experiment paradigms that require
auditory stimuli, much care and concern should be considered with the hardware being
used. Sound cards can have good or poor startup latency. Startup latency is the time from
when E-Prime tells the sound card to play a sound to when the sound actually emits from
the sound card or speakers. Not all sound cards are created equal.
(Psychology Software Tools Admin, n.d.)
Indeed, Plant and Turner (2009) report several benchmarks of auditory onset times for a variety of
computer speakers that may be found in standard labs fnding quite a discrepancy in performance
across devices. The best performing speakers had a mean onset latency of 3.29 ms and a jitter of SD =
0.01 ms. In other words, these speakers introduced a reliable ~3.29 ms latency into the presentation
of an auditory stimulus. Compare this with the benchmark data for the lesser performing speakers
(M = 37.00 ms, SD = 1.31 ms), and it is clear that there is the potential for substantial variability in
audio latency across diferent speaker systems.
153
Phillip Hamrick
Sound cards in modern speakers vary similarly. For example, benchmark data from E-Prime
(Psychology Tools Software, n.d.), a popular experiment-generation software, show that some com-
binations of sound card, computer, and operating system can result in short lag and low jitter (e.g., i7
16 GB RAM, 64-bit running Windows 10 with E-Prime’s proprietary Chronos response box, M =
0.96 ms, SD = 0.30 ms) while others introduce more concerning latency errors and jitter (i7 8 GB
RAM 64-Bit running Windows 10 with a SoundBlaster X2 PCle and CoreAudio API, M = 52.00 ms,
SD = 38.78).
Most labs use a range of audio products, and it should therefore be expected that most labs
running an identical auditory experiment would have spurious latency diferences. At best, these
contribute noise to the data, lowering power; at worst they could lead to increased Type I/II error
rates, error in efect size estimates, and spurious replication failures. One solution to this is to use a
proprietary response device designed for timing auditory information, such as Psychology Software
Tools’ Chronos, which allows a computer-external clock to more accurately time audio onset, audio
responses, and button pressing more broadly.
Response devices
Chronometric studies often rely on participants making overt manual responses to stimuli, usually
pressing buttons on a computer keyboard, computer mouse, or a response box designed for research.
Common peripherals do often advertise low latencies for consumer purposes (e.g., gaming), but
these are not benchmarked with science in mind.
For many L2 psycholinguistic researchers, a keyboard is the response device of choice. They are
cheap and easy to use, and participants are familiar with their layout. Unfortunately, benchmark
data show a stark contrast in latency errors and jitter among various computer keyboard and mouse
devices (see Table 13.1). For example, standard keyboards introduce lags of 15–25 ms, with 1–3 ms
of jitter, while computer mouse latencies vary more substantially.
Industry-specifc proprietary peripherals such as Psychology Software Tools’ Chronos and the
Cedrus RB series of response boxes ofer more accurate and precise timing options, although
they come with a price tag ranging from many hundreds to a couple thousand dollars, and not all
researchers may be able to aford such a luxury. It is important to consider that the timing accuracy
of these products may difer as reported by the manufacturer or by independent benchmarking. For
example, Cedrus reports a 2 ms delay that is “jitter free” when using the RB-740 response box with
Table 13.1 A sample of benchmarked timing latencies for typical response devices used in psycholinguistic
research.
Device Mean Latency Lag SD
Belkin USB Keyboard1 18.30 ms 1.29 ms

Chic Intelligent USB Keyboard1 25.34 ms 3.68 ms
Acer USB Keyboard2 15.60 ms 2.58 ms
Interex Mouse2 15.57 ms 2.35 ms
Microsoft Optical USB PS/2 Mouse3 24.52 ms 3.14 ms
OEM Mouse3 66.30 ms 6.48 ms
Chronos (USB 3.0)2 0.30 ms 0.46 ms
1 Plant and Turner (2009)

2 Psychology Software Tools Admin (n.d.)
3 Plant et al. (2004)
154
Table 13.2 Response device options for L2 psycholinguistics research and their chronometric-relevant properties.
Device Pros Cons
Chronos (Psychology Software Tools) • High precision and accuracy1,2 • Expensive

• Times button presses as • Only operates with E-Prime
well as oral production with (at this time)
included microphone
• Includes photo sensor to
benchmark computer monitor
displays
Cedrus RB Response Pads • Extensive documentation • Expensive
from manufacturer • Poorer precision (~5 ms)2
• Come in a variety of layouts
• Operates with SuperLab, but
other developers have made it
compatible with their software
(e.g., PsychoPy)
Keyboards • Cheap • Variable in accuracy (see
• Multipurpose Table 13.1)
• Commonly used in L2 • Mac keyboards appear to have
psycholinguistic research less lag than Windows PC
keyboards2
• Increasingly wireless, rather
than being connected via
USB or serial port, which
increases latency issues
USB Game Pads • Cheap • Benchmarking data limited to
• Ergonomic, since they were manufacturer’s claims
designed to be used for long • Must be used in conjunction
periods of time with software that converts
• Gaming brands have quite button presses on the game
good accuracy reported by pad into keyboard-equivalent
manufacturers (e.g., Logitech, buttons or via coding in the
Gravis) experimental software
1 Psychology Software Tools (n.d.)

2 Freegard (2012)
SuperLab 6 (RB-740 Response Pad, 2020); however, this value difers from the independent bench-
marking of the previous model of the same device (RB-730), which had ~10 ms lag and ~5 ms of
jitter (Freegard, 2012). It is therefore unclear whether the discrepancy in timing is due to the diferent
model versions, the source of the benchmarking, or both. However, I know of no data indicating that
keyboard or mouse devices have accuracy or precision on par with specialized, proprietary response
devices.
To briefy survey some of these considerations, Table 13.2 provides a (necessarily incomplete)
overview of some common response devices for psycholinguistic research. It is worth noting that
the views expressed in this table (and Table 13.3) are somewhat subjective. For example, I regard
$1,000 as an expensive purchase, while others may not. Likewise, my experience with certain
technologies may lead to a more positive, negative, or detailed evaluation than I might give a
technology with which I have less experience. Researchers are encouraged to do their homework
on all the available information in these tables (and in the references in this paper and beyond)
to make a maximally informed decision before making their decisions about which software and
hardware to use.
155
Phillip Hamrick
Methods and paradigms: software

All experiment-generation software packages produce absolute timing lags (Bridges et al., 2020), and
some packages are better than others (for an overview, see Table 13.2). For example, independent
benchmark tests have shown that PsychoPy and Psychtoolbox have more precise timing than NBS
Presentation and E-Prime, although the diferences appear rather small (Bridges et al., 2020). I refer
the reader interested in more comprehensive coverage to the available literature on software timing
(e.g., Bridges et al., 2020; Garaizar & Vadillo, 2014; Garaizar et al., 2014; Plant, 2016).
As noted in the passage from E-Prime developers above, even when software can ofer millisecond
precision, it is only as true in application as the hardware it runs on. I will illustrate the point with
a simple example. Imagine we want to design a study in which, for example, we want to present a
word visually for 30 ms. Since common computer monitors have refresh rates of 16.67 ms, we cannot
do this with common modern computer screens no matter how precise the software we are using
because their 16.67 ms refresh rate does not evenly go into 30 ms. The closest that we can get will
be to show the stimulus for either 16.67 ms (1 frame) or 33.33 ms (2 frames). What is the solution?
Until relatively recently, only experiment generation software that allowed or forced researchers
to do some programming were typically able to be modifed to have timing based on each indi-
vidual refresh of the screen. If the software required using ms to specify stimulus durations, then the
researcher (should have) had to work out how to get the timing right based on the specifcations of
their hardware. More recently, experiment generation software has developed more user-friendly
GUIs that allow researchers to choose between millisecond timing and tick-based timing at their
discretion. Allowing experiments to be timed, especially those involving visual stimulus presentation,
in terms of ticks/refresh rate is an excellent option for researchers; however, this method is not com-
monly reported in the literature. A casual glance through L2 psycholinguistic research will quickly
show that timing in RT studies tends to be reported (and, we assume, conducted) as a function of
ms. Unless those experimenters took great care to make sure their visual stimulus presentation was
consistent with their computer monitor’s refresh rate, then the millisecond precision of their experi-
ment generation software is potentially in jeopardy.
That said, there are solutions and caveats. In particular, because DMDX and PsychoPy make use
of coding languages, fnding solutions for software/hardware issues is more transparent. On the other
hand, such solutions require that the researcher knows how to code, and this can place a skill burden
on researchers. Given how much time it takes to develop a good experiment, it may well be that
the researcher does not have the time or means to develop said coding skills (indeed, this would be
an ideal opportunity to fnd collaborators in related felds such as computer science). Other software
such as E-Prime allows some more user-friendly solutions to software/hardware issues via its GUI.
However, the software is not free, and may prove too costly for early-career researchers or those at
institutions with fewer resources.
Table 13.3 provides a brief survey of some of the most prominent available software packages for con-
ducting RT experiments. As with Table 13.2, this table is necessarily limited in scope, but I have prioritized
Table 13.3 Software options for L2 psycholinguistics research and their chronometric-relevant properties.
Software Pros Cons
E-Prime (Psychology Software Tools) • Efcient technical support • Expensive (~$1,000)

• Extensive documentation from • Advanced programming
manufacturer and benchmarking code is not a common
studies coding language
• Generally benchmarked as • Only operates on Windows
excellent, especially when used PC computers
with Chronos (Table 13.2)
156
Software Pros Cons
• Commonly used, increases

comparability
• Comes with software for
merging and fxing data fles
SuperLab (Cedrus) • Commonly used, increases • Coding options unavailable
comparability or limited
• User-friendly interface • Available benchmark data
• Comes with software for limited
merging data fles • Expensive if more than one
• Works with both Windows PC license is needed (~$2,000)
and Mac computers
• Cheaper student licenses available
DMDX • Free to use • Coding experiments has a
• Comes with some benchmarking somewhat steep learning
ability (TimeDX) curve
PsychoPy • Free to use • Data fles can only be
• Generally benchmarked as merged via coding or
excellent batching
• User-friendly interface and • Coding necessary for any
coding (Python) both available advanced experimental
within the same program designs
• Can run experiments both on a • Less widely used in the feld,
local computer as well as online lowering comparability
• Works with both Windows PC
and Mac computers and other
platforms
• Open-source code facilitates
open science
• Very precise1
OpenSesame • Free to use • Less widely used in the feld,
• User-friendly interface and lowering comparability
coding (Python) both available • Available benchmark data
within the same program limited
• Can run experiments both on a • Somewhat less precise than
local computer as well as online the others, especially for
• Works with both Windows PC auditory presentation1
and Mac computers and other
platforms
• Open-source code facilitates
open science
Psychophysics Toolbox • Benchmarked as excellent with • Requires coding skill
Windows PC and Linux1 • More advanced coding
• Free, but only if you have necessary to avoid timing
Matlab, which is expensive issues
• New standalone free version • Less widely used in the feld,
(Psychtoolbox) recently debuted lowering comparability
Presentation (NBS) • Benchmarked as excellent for • Not free, but range of
timing, but with difculty pricing options, some of
obtaining those results with which are inexpensive
advanced documentation1 • Only operates on Windows
• Widely used in research, but less PC computers
so in L2 psycholinguistics
• NBS team will program your
experiment for you for a fee
1 Bridges et al. (2020)

157
Phillip Hamrick
software that runs directly on lab computers that may have web-based compatibility over software that is
primarily web-based. For more on web-based software packages, I refer the reader to the timing megastudy
in Bridges et al. (2020), which includes prominent web-based software, such as jsPsych and Gorilla.
Web-based chronometric research

The new computing advances that have put the internet perpetually at our fngertips have also made
it much easier to collect vast quantities of data via web-based experiments with software such as
jsPsych, Gorilla, or PsychoJS (PsychoPy and OpenSesame now also ofer web-based functionality),
but web-based data collection introduces its own problems (e.g., HTML5 has a higher tendency to
have missing frames than non-browser software installed directly on the computer: Garaizar et al.,
2014). Although timing benchmarks for web-based studies are scarce, there are some useful data.
For example, de Leeuw and Motz (2016) compared RTs collected via the Psychophysics Toolbox
and JavaScript—a common programming language for conducting RT experiments on the web—all
within the same laboratory setting (i.e., to control for hardware diferences). The result was that
JavaScript-collected RTs were roughly 25 ms longer than those in the Psychophysics Toolbox, but
with no statistically signifcant diferences between platforms in jitter, suggesting a similar amount of
noise between the two platforms. On the other hand, Pronk et al. (2019) found that web-based stud-
ies using internet browsers introduce substantial lag (50–70 ms) and jitter (5–10 ms). Complicating
matters is the fact that benchmarking genuine web-based chronometric experiments (with partici-
pants responding from their personal devices) is rendered almost impossible because even if you fnd
timing errors, it is difcult to know whether they stem from, for example, delays in visual stimulus
presentation, detecting a button press response, etc. (Bridges et al., 2020).
Until more benchmarking data are available, the current evidence suggests that large RT efects
might still be readily attainable via web-based data collection, while studies of smaller RT efects
should, at least at this time, be restricted to laboratory settings. Despite these concerns, it is worth
noting that the larger sample sizes that internet-based research facilitates may alleviate much of these
concerns. That is, with a sufcient sample size, noise from web-based chronometric studies may be
ofset (Brand & Bradley, 2012).
Solutions and future directions

There are several steps a researcher can take to fx, or at least alleviate/address, the concerns raised in
this chapter. These need not be the only solutions, but they are good starting points.
Collect data on a single system

One solution is to collect data with a single system at one site (i.e., with the same computer, response
device, and software being used for every participant). This solution can be cost-efective since it
need only involve one system. If benchmarking information is available, then the researcher may also
be about to account for timing issues in data handling. A system with a lot of jitter will still lower
statistical power; however, any jitter or lag should be, more or less, similar across all experimental
variables, reducing the likelihood of confounds.
Accounting for lag and jitter before and after data collection
Before you even collect data, you can use any information you have on your system’s timing proper-
ties to set up your experiment. That is, program your experiment with the timing details of your
system in mind. For example, in studies with visual language stimuli, visual presentation durations
158
should be in multiples of the computer screen’s refresh rate, since a visual stimulus cannot be dis-
played for anything less than a tick. Programming the experiment in frames or ticks allows you to do
this more directly, but in software that does not allow for such a setup, calibrate your timing to com-
pensate. For example, the creators of E-Prime suggest setting stimulus durations to 10 ms below the
expected total duration of all the refresh cycles that one would like the stimulus to appear for. “Since
the visual duration is always rounded up to the next refresh, the display duration must be specifed
as some portion of one refresh below the target refresh duration” (Schneider et al., 2002, p. 99). If
you want a stimulus to appear on a common TFT screen for, say, three refresh cycles (e.g., 49.9 ms),
then program the software to show the stimulus for 39.9 ms. The visual duration will be rounded up
to appear for three refresh cycles, no more. Of course, diferent software may deal with timing issues
like this in diferent ways. It is highly recommended that you carefully consult the manufacturer’s
information to determine how to handle these timing issues with the software that you use.
Of course, part of programming the experiment correctly may require you to go beyond the
experiment programming options ofered by the convenient GUIs that many software packages ofer.
This could mean that to efectively implement your study with accurate and precise timing, you may
need to learn Python (as in PsychoPy), JavaScript (as in jsPsych), or some other programming lan-
guage (or, as noted earlier, fnd a collaborator with some computer science background). Obviously,
these solutions are not always readily available, and so researchers are encouraged to think about these
options before choosing which software to use to run their studies.
Benchmarking and calibration

The best possible solution for the technological hurdles mentioned here is to check and calibrate the
timing of your hardware and software, calibrate your system, and handle any remaining issues during
data cleaning. If benchmark data are already available for your software and response device (both of
which are somewhat likely), then you may choose to simply use the existing benchmark fgures to
guide your data collection and analysis. However, with such a wide variety of computers, processors,
operating systems, and keyboards used across research labs, it is much less likely that benchmarks will
already exist for each of them.
In all cases, one should consider the source of the benchmark data. In addition to the discrepancy
in timing benchmarks for Cedrus response pads mentioned earlier, there are benchmark discrepancies
between independent researchers and manufacturers elsewhere. For example, Orquin and Holmqvist
(2018) found spatial inaccuracy for certain eye-tracking devices to extend beyond the manufacturer’s
estimates. When in doubt, benchmarking and calibrating your own equipment is the gold standard;
however, in lieu of that expensive and resource-intensive solution, published independent bench-
marks should generally be treated as the next best option. Manufacturer benchmarks certainly have
the propensity to be biased in favor of that manufacturer’s product, but manufacturer benchmark
data are better than no benchmark data. When no other timing information is available outside the
manufacturer, it is still worth bearing in mind and reporting it.
If a researcher decides to benchmark their own hardware and software, they should note that it is
certainly doable, although the cost, complexity, and time involved varies. For example, you can check
visual presentation timing on your computer monitor using an oscilloscope which measures things
like refresh rate, dropped frames, and visual onset lag, by taking tens of thousands (or more) visual data
samples per second. Some cheaper options are available (e.g., BitScope). A similar tool (a photo sensor)
is available with Psychology Software Tools’ E-Prime and Chronos package, albeit at a more expensive
price. Perhaps the most robust system for assessing the timing issues that may arise in your lab is the
Black Box Toolkit (Plant et al., 2004); however, the system is substantially more expensive, currently
costing several thousand dollars on top of computer and software costs for running the experiment to
begin with (see Elgort & Warren, 2023 [this volume], for a similar point about eye-tracking).
159
Phillip Hamrick
Reporting practices and replicability

Considering all of the factors that can infuence the chronometric accuracy and precision of a study,
I recommend that, at a minimum, researchers include information about the kinds of technological
information in Table 13.4 used in their study. I also encourage readers to report this information as
it applies to other chronometric tools, such as eye-tracking (e.g., Godfroid, 2020; Godfroid & Hui,
2020) and EEG (Tatum et al., 2016). If more than one computer, response device, or type of software
is used, that also should be reported along with the precautions the researcher took to ensure that the
data were not afected substantially by using more than one system or device. If benchmark data are
available, they should be reported or referenced.
Beyond reporting these details, I also recommend that researchers look for chronometric irregu-
larities in their data (e.g., dropped frames or sound card startup delays) if they are available. Most
experiment generation software includes some information on these kinds of timing errors in the
data they output, but each varies, and it is recommended that researchers consult the manufacturer’s
literature or manual in depth before collecting data. Researchers should report what kinds of tim-
ing irregularities they screened for, what they found, and how they handled any issues that arose.
Given the often limited space in most research journals, it is recommended that researchers make this
information publicly available either as supplemental materials on the journal’s website or via an open
science repository (e.g., Open Science Framework database, etc.).
Importantly, reporting the aforementioned information adds an important dimension to under-
standing the replicability, generalizability, and validity of an experimental efect. For example, a suc-
cessful replication of a result with software and hardware that have diferent chronometric properties
than those used in the initial study indicates that the efect is unlikely to be due to idiosyncratic
efects of hardware and software chronometry. On the other hand, failure to replicate a result with
software or hardware that are similar or identical in their chronometric specifcations is a more com-
pelling failure to replicate than if the hardware or software were diferent. In any event, theoretically
interesting efects must be replicable, and timing errors may be one source of a failure to replicate.
Reporting practices like those suggested in this chapter may prove useful in explaining anomalous or
contradictory fndings, especially in future meta-analyses.
Table 13.4 Suggested details to report on technology used in chronometric research in L2 psycholinguistics
Technology Used What to Report
Computer hardware • Computer: make, model, processor, operating system

• Monitor: make, model, screen refresh rate in Hz or ticks
(e.g., 60 Hz refresh rate = 16 ms tick)
• Speakers and/or sound cards: make, model, any available
benchmark or manufacturer-reported data on accuracy (lag) and
precision (jitter)
Response devices • Keyboard: make, model, connection method (e.g., USB, serial port,
BlueTooth)
• Mouse: make, model, connection method (e.g., USB, serial port,
BlueTooth)
• Response device: make, model, connection method (e.g., USB,
serial port)
• Eye-tracking: make, model, set up (one computer, host and data
computer, etc.) connection method (e.g., USB, dedicated port)
Software • Software: make, version number
• Whether the experiment was programmed in ticks or ms
160
Other alleviating suggestions

In addition to following the guidelines in this section, a researcher with free software and a relatively
cheap computer can produce valid results, especially when combined with the conventional recom-
mendations for good science and proper conducting of chronometric research: (1) include more
trials and more participants (therefore more data observations); (2) engage in replication and open
science transparency; (3) avoid confounds in your design, including those between timing errors
and any experimental variables; (4) turn of all nonessential background computer processes that
might infuence the timing accuracy and precision of your study. Another option that researchers
should consider is to include appropriate random efects structures in multilevel modeling analyses
of their data.
Conclusion
This chapter might lead the reader to worry whether good chronometric research can be conducted
without access to expensive equipment or advanced coding skills. While it is true that expensive
equipment programmed correctly can give a researcher greater confdence in their chronometry, the
fact is that researchers may be able to conduct quality chronometric research with cheaper equip-
ment and free software provided that they do their due diligence to ensure that timing problems are
minimized as outlined in the preceding sections.
Despite the potentially concerning issues raised here, it is important keep our optimism up.
The reader should keep in mind that despite all of the concerns raised here, many psycholin-
guistic effects are easily replicated (e.g., word frequency effect, semantic priming, etc.). The
very existence of such replicable effects in the face of all the sources of timing error described
here indicates that the situation may not be terribly dire. If anything, it should provide us
with optimism. After all, a replicable RT effect in the face of all the timing errors presented
by our hardware and software suggests that many of the effect sizes in question may be bigger
than we think, and, therefore, may provide a better explanation of L2 psycholinguistics than
we even realize.
Can web-based chronometric experiments administered via internet browsers or some other interface
provide a viable alternative to laboratory-based experiments? If so, under what circumstances? How
small an efect can we reliably fnd via web-based RT experiments?
How can we meaningfully interpret RT efects for lower-profciency L2 learners, who, if they show RT
efects at all, will likely show smaller efects than advanced L2 learners or native speakers?
How can we separate or model (statistically) the variance in the data explained by individual diferences
in L2 speakers from variance explained by timing errors?
Note
1 For example, a standard eye-tracker includes a basic lag from the time when an image is captured by the sensor
to when valid gaze data can be time stamped (e.g., 25–35 ms for a Tobii X-50; these fgures vary by device;
Tobii Technology, 2010). See Godfroid (2020, section 9.1.3) for discussion of latency accuracy and precision
in the context of eye-tracking.
161
Phillip Hamrick
Further reading
Bridges, D., Pitiot, A., MacAskill, M. R., & Peirce, J. W. (2020). The timing mega-study: Comparing a range
of experiment generators, both lab-based and online. PeerJ, 8, e9414.
Plant, R. (2016). A reminder on millisecond timing accuracy and potential replication failure in computer-based
psychology experiments: An open letter. Behavior Research Methods, 48, 408–411.
These articles provide a general overview of the problem of timing errors.
Plant, R., Hammond, N., & Turner, G. (2004). Self-validating presentation and response timing in cognitive
paradigms: How and why? Behavior Research Methods, Instruments, & Computers, 36, 291–303.
Plant, R., & Turner, G. (2009). Millisecond precision psychological research in a world of commodity comput-
ers: New hardware, new problems? Behavior Research Methods, 41, 598–614.
These papers supply independent benchmarking data.
References
Baayen, R. H., & Milin, P (2010). Analyzing reaction times. International Journal of Psychological Research, 3,
12–28.
Bernolet. S. (2023). Studying second language production. In A. Godfroid & H. Hopp (Eds.), The Routledge
Brand, A., & Bradley, M. T. (2012). Assessing the efects of technical variance on the statistical outcomes of Web
experiments measuring response times. Social Science Computer Review, 30, 350–357.
Bridges, D., Pitiot, A., MacAskill, M. R., & Peirce, J. W. (2020, January 20). The timing mega-study: Compar-
ing a range of experiment generators, both lab-based and online. PeerJ, 8, e9414.
de Leeuw, J. R., & Motz, B. A. (2016). Psychophysics in a Web browser? Comparing response times collected
with JavaScript and Psychophysics Toolbox in a visual search task. Behavior Research Methods, 48, 1–12.
Elgort, I. & Warren, P. (2023). Studying second language comprehension. In A. Godfroid & H. Hopp (Eds.),
Freegard, G. (2012). Response pads: The design, development and results of a bespoke response box, and subsequent testing,
for comparison purposes, of various devices used for reaction time measurements in psychological experiments [Unpub-
lished manuscript], Swansea University, Swansea, Wales.
Frenck-Mestre, C., & Prince, P. (1997). Second language autonomy. Journal of Memory and Language, 37, 481–501.
Garaizar, P., & Vadillo, M. A. (2014). Accuracy and precision of visual stimulus timing in PsychoPy: No timing
errors in standard usage. PLoS ONE, 9, e112033.
Garaizar, P., Vadillo, M. A., & López-de-Ipiña, D. (2014). Presentation accuracy of the Web revisited: Animation
methods in the HTML5 era. PLoS ONE, 9, e109812.
Garaizar, P., Vadillo, M. A., López-de-Ipiña, D., & Matute, H. (2014). Measuring software timing errors in the
presentation of visual stimuli in cognitive neuroscience experiments. PLoS ONE, 9, 1–7.
Godfroid, A. (2020, February). Reporting practices in second language and bilingual eye-tracking research. Plenary talk
at the Eye-tracking Second Language Acquisition Research Symposium. University of Groningen, The
Netherlands.
Godfroid, A., & Hui, B. (2020). Five common pitfalls in eye-tracking research. Second Language Research, 36(3),
277–305.
Jing, H., Chojnowska, C., Heim, S., Thomas, J., & Benasich, A. A. (2004). Timing errors in auditory event-
related potentials. Journal of Neuroscience Methods, 138, 1–6.
Lagroix, H., Yanko, M., & Spalek, T. (2012). LCDs are better: Psychophysical and photometric estimates
of the temporal characteristics of CRT and LCD monitors. Attention, Perception, and Psychophysics, 74,
1033–1041.
Neath, I., Earle, A., Hallett, D., & Surprenant, A. M. (2011). Response time accuracy in Apple Macintosh
computers. Behavior Research Methods, 43, 353–362.
Orquin, J. L., & Holmqvist, K. (2018). Threats to the validity of eye-movement research in psychology. Behav-
ioral Research Methods, 50, 1645–1656.
Plant, R. (2016). A reminder on millisecond timing accuracy and potential replication failure in computer-based
psychology experiments: An open letter. Behavior Research Methods, 48, 408–411.
Plant, R., Hammond, N., & Turner, G. (2004). Self-validating presentation and response timing in cognitive
paradigms: How and why? Behavior Research Methods, Instruments, & Computers, 36, 291–303.
162
Plant, R., & Quinlan, P. (2013). Could millisecond timing errors in commonly used equipment be a cause of
replication failure in some neuroscience studies? Cognitive, Afective, and Behavioral Neuroscience, 13, 598–614.
Plant, R., & Turner, G. (2009). Millisecond precision psychological research in a world of commodity comput-
ers: New hardware, new problems? Behavior Research Methods, 41, 598–614.
Pronk, T., Wiers, R. W., Molenkamp, B., & Murre, J. (2020). Mental chronometry in the pocket? Timing accu-
racy of web applications on touchscreen and keyboard devices. Behavior Research Methods, 52, 1371–1382.
Psychology Software Tools. www.pstnet.com/eprimedevice.cfm
Psychology Software Tools Admin. (n.d.). TIMING: E-prime 3.0 timing data [24332]. https://support.pstnet.
com/hc/en-us/articles/360008833253-TIMING-E-Prime-3-0-Timing-Data-24332-
RB-740 Response Pad. (2020). https://cedrus.com/store/rb740.html
Schneider, W., Eschman, A., & Zuccolotto, A. (2002). E-Prime user’s guide. Psychology Software Tools.
Tatum, W. O., Olga, S., Ochoa, J. G., Clary, H. M., Cheek, J., Drislane, F., & Tsuchida, T. N. (2016). American
clinical neurophysiology society guideline 7: Guidelines for EEG reporting. Journal of Clinical Neurophysiology,
33(4), 328–332.
Tobii Technology. (2010). Timing guide for Tobii eye trackers and eye tracking software. www.tobiipro.com/
siteassets/tobii-pro/learn-and-support/design/eye-tracker-timing-performance/tobii-eye-tracking-timing.
pdf/?v=1.0
163
14
USING CORPORA IN RESEARCH
ON SECOND LANGUAGE
PSYCHOLINGUISTICS
Introduction
Second language (L2) psycholinguistics has long been associated with experimental method-
ologies. However, over the past thirty years corpus studies have increasingly attracted scholars’
attention as useful complementary or even alternative empirical approaches to address psycho-
linguistic questions in second language acquisition (SLA) in areas including processing, sentence
complexity, fuency, lexical diversity, vocabulary acquisition, cross-linguistic transfer, and lexical
knowledge in heritage language acquisition. Corpora are increasingly contributing to state-of-
the-art research methodologies in the feld although corpus applications emerged only relatively
recently in L2 psycholinguistics and therefore often lack the methodological sophistication of
existing experimental research (Gries, 2014b, p. 16). Across corpus linguistics (CL) and (L2)
psycholinguistics, L2 acquisition is operationalized diferently: While psycholinguistic research
often focuses on comprehension (rather than production) of relatively small samples of written
language and uses accuracy and reaction times as dependent variables, corpus research usually
utilizes written production data. Further, many corpus-based analyses focus on the conditional
probability of occurrence of a given form or meaning (as predicted by contextual information
from the corpus).
The contribution of CL to SLA is most noticeable within cognitive-linguistic frameworks
including constructionist, usage-based, and exemplar-based models of language acquisition and
use. In line with Lakof’s Cognitive Commitment, which assumes that language and linguistic
organization refect general principles of cognition (Lakof, 1991), these frameworks assume that
(i) language acquisition, representation, and processing are largely explicable with reference to
mechanisms of domain-general cognition, (ii) language use involves cognitive events that ulti-
mately shape the linguistic system (Kemmer & Barlow, 1999), and (iii) speakers’ knowledge of
lexical items correlates with and their uses in grammatical contexts (e.g., Langacker, 1987; Gold-
berg, 2006). Corpus data have gained recognition as “a source of relevant linguistic data and
consequently, quantitative and statistical tools now count as central methodologies” (Gries, 2014a,
p. 280). In this chapter, we use usage-based as an umbrella term for the previously listed cognitive-
linguistic frameworks.
Psycholinguistics and CL are intrinsically linked to the notion of frequency:1 Usage frequency
and repetition are central at all levels of language (Ellis, 2002) and are important for L2 learning
(Ellis, 2007). In psycholinguistics, token frequencies play an important role as control variables and
164 DOI: 10.4324/9781003018872-16

Using corpora in research on second language psycholinguistics
predictors in experimental studies, and frequency of use correlates with the entrenchment of an
expression as a unit (i.e., a linguistic structure with an established cognitive routine). As Ellis et al.
(2016, p. 45f ) put it, “The more times we experience something, the stronger our memory for it,
and the more fuently it is accessed” and
[c]onstructions that are frequent in the input are processed more readily than rare construc-
tions. Through experience, a learner’s perceptual system becomes tuned to expect construc-
tions according to their probability of occurrence in the input. . . . The same is true for the
strength of the mappings from form to interpretation.
(Ellis et al., 2016, pp. 46–47)
L2 acquisition and processing involve probabilistic/statistical knowledge (Ellis, 2002) and corpora
provide access to many kinds of frequencies of (co-)occurrence in learner language necessary to
understand how L2 knowledge is acquired and processed. Consequently, corpus-based psycholin-
guistic work often utilizes frequency as one predictor of acquisition (Gries, 2022) and usage-based
linguists assume that distributional characteristics of linguistic elements refect their functional
(e.g., semantic, pragmatic) characteristics. Thus, their distributional characteristics (i.e., their fre-
quencies of [co-]occurrence) refect the similarity of their functional characteristics (Gries, 2017,
p. 592). Of course,
[l]anguage learners do not consciously count language statistics, their stream of conscious-
ness concerns communication and understanding. The frequency tuning under consider-
ation here is computed by the learner’s system automatically during language usage. The
statistics are implicitly learned and implicitly stored.
(Ellis et al., 2016, pp. 64)
The usage-based literature posits that “[l]earning, memory and perception are all afected by fre-
quency, recency, and context of usage” (Ellis et al., 2016, p. 46). In the context of (S)LA, research
in the associative learning of cue-outcome contingencies (see Ellis, passim) has shown how domain-
general learning mechanisms such as entrenchment, productivity, recency, contingency, prototypicality,
Psycholinguistic notion Brief description Associated corpus method
Entrenchment Cognitive process by which a linguistic Token and type frequency of

pattern is established as a cognitive routine (co-)occurrence
Productivity Cognitive process by which a linguistic Type frequency of (co-)occurrence
pattern is extended to new cases
Recency Tendency to best remember Dispersion, concordancing,
information that is presented last structural/construction priming
Contingency Reliability of a linguistic form as a Co-occurrence (collocation,
predictor of a given interpretation colligation, collostruction)
Prototypicality Degree to which an expression is a central/ Frequency, co-occurrence,
and salience2 representative member of a category and contextual distinctiveness
stands out against other category members
Surprisal The degree to which a linguistic choice Frequency, co-occurrence
is unexpected, given its context
165
salience, and surprisal along with perceptual activity all play an important role in the (S)LA process.
While these notions have become central in usage-based research, corpus linguists over the past
decade have actively developed statistical approaches to operationalize these notions. Textbox 14.1
presents a list of these cognitive mechanisms and their associated corpus methods. In the following,
we discuss how these notions can be operationalized and captured quantitatively/statistically for
L2-psycholinguistic research. We will zoom in specifcally on Kim and Rah (2019), Ellis et al. (2016),
and Wulf and Gries (2019). Finally, we discuss some advantages and limitations of corpus approaches
to L2 psycholinguistics.
Entrenchment
Frequency and entrenchment are strongly correlated (Gries, 2014a) in that frequency of use is said
to promote entrenchment:
Every use of a structure has a positive impact on its degree of entrenchment, whereas
extended periods of disuse have a negative impact. With repeated use, a novel structure
becomes progressively entrenched, to the point of becoming a unit; moreover, units are
variably entrenched depending on the frequency of their occurrence.
(Langacker, 1987, p. 59)
Exploring the frequency of occurrence of linguistic elements in L2 corpora provides a way of under-
standing how language learners access, automatize, and process these elements (Gries & Ellis, 2015,
p. 4): As linguistic elements recur, speakers’ mental representations of linguistic systems are constantly
being updated (see Ellis, 2002, p. 147; Halliday, 2005, p. 67; Gries, 2022, p. 51).
Importantly, entrenchment triggers the acquisition of linguistic elements at diferent levels
of abstraction (i.e., more concrete vs. more abstract constructions), which is reminiscent of two
diferent types of frequencies, namely token frequencies (i.e., the number of times an element
is observed) vs. type frequencies (i.e., the number of diferent elements observed in a certain
position or slot such as the number of diferent verbs within a prepositional dative construction).
Thus, token frequency leads to the entrenchment of instances, whereas type frequency leads to
the formation and entrenchment of more abstract schemas. This distinction is key to understand-
ing, frst, the richness of exemplar memories and their associations, and second, more abstract
connectionist learning mechanisms. Accordingly, the relevance of token frequency grew more
and more (Ellis, 2002).
In L2 acquisition, token frequencies and entrenchment are similarly relevant: Firstly, token fre-
quency of linguistic elements in the input relates to age of acquisition (Casenhiser & Goldberg,
2005), to speed of lexical access (Schmid, 2000), to routinization, reduction (Aslin & Newport,
2012), and to category formation. For instance, exemplars with higher frequency are classifed more
accurately and as more typical (Ellis et al., 2016, p. 60f).
Token frequencies can be absolute or relative frequencies. Absolute frequencies refer to the number
of times an element is observed (often normalized to per-million-words counts to compare results
across diferently large corpora and inform context-free entrenchment (i.e., information about the
frequency of isolated linguistic elements independently of their context of occurrence). Relative
frequencies, however, provide information about contextual entrenchment (i.e., information about
the frequency/probability of elements given their linguistic or other context). Because usage-based
166
linguists assume context is relevant for all linguistic processes, they primarily focus on relative fre-
quencies. However, despite their perceived importance in psycholinguistics research, token frequen-
cies alone are often less important than is assumed (Gries, 2022, pp. 52–53) and studies such as
Adelman et al. (2006) and Baayen (2010) have begun to question the centrality of frequency in
general or of frequency-as-repetition in particular and raise the importance of supplementing it with
other predictors not discussed enough in psycholinguistics such as dispersion, association, and others
(Baayen, 2010; Gries, 2022).
Productivity
Traditionally, productivity has been associated with type frequency (Bybee & Hopper, 2001; Gold-
berg, 2006, p. 99), which refers to “the number of distinct lexical items that can be substituted in a
given slot in a construction, whether it is a word-level construction for infection or a syntactic con-
struction specifying the relation among words” (Ellis et al., 2016, p. 52). As Ellis et al. (2016, p. 53)
explain, “The more items in a certain position in a construction, the less likely the construction is
associated with a particular item, and the more likely it is that a general category is formed over
the items in that position.” Thus, type frequencies inform categorization in L2, i.e., how learners
build up constructional knowledge in the target language, in particular the connections between a
given construction and the words in its slot(s). In L2 psycholinguistics, two studies drawing on both
token and type frequencies, Godfroid and Uggen (2013) and Godfroid (2016), have discussed the
(non-)productivity of the strong verb paradigm in contemporary German. Godfroid (2016) studied
whether the observed learning was item- or system-based and justifed the inclusion of a generaliza-
tion posttest (to measure system-based learning) based on the notion that strong verbs form a fuzzy
grammatical category. In learner corpus research (LCR), large-scale investigations of L2 knowledge
have used collostructional analysis, which measures statistically the association between lexical items
and grammatical constructions and requires token frequencies as well as the complete set of elements
occurring in a construction’s slot(s). We return to this method below and discuss how this approach
can be integrated to L2 psycholinguistic studies.
Recency
Recency is the tendency to best remember information that is presented last. With corpora, recency
can be explored via the linguistic contexts of exemplars with concordancing, which allows linguists
to correlate frequencies of diferent linguistic choices with contextual information. Concordances
are displays of linguistic instances of a search word in a central column together with their preceding
and subsequent context (which can be defned in terms of numbers of words, numbers of characters,
intonation units, etc.). Figure 14.1 shows concordance lines for particles used in phrasal verb con-
structions in the International Corpus of Learner English.
Across corpus-linguistic methods, concordances provide the most context and can lead to fne-
grained analysis of many features on many dimensions (Gries, 2014a, p. 281). Their usefulness for L2
psycholinguistic research is that they provide all the context of a linguistic choice (to the extent it is
represented in the corpus [annotation]), so one can often determine what happened in the recent past
and how it might be correlated with the current investigated linguistic choice. Further, the cognitive
value of concordance lines lies in the notion that memory is context-dependent (that context being
of any nature, e.g., musical, linguistic); and information learned in a particular context is more readily
remembered when that context is reinstated. Context-dependent memory is sensitive to incidental
contextual information, it can recognize many diferent kinds of contextual similarities, and it can
infuence performance without awareness. Linguistically, context-based implicit learning has been
167
Figure 14.1 Concordance lines for particles used in phrasal verb constructions (in the International Corpus of
Learner English).
observed in areas such as homophone spelling (Smith et al., 1990), word-fragment completion (Ball
et al., 2010), and picture naming (Horton, 2007).
Recent LCR provides a good illustration of how precisely linguistic contexts can be explored with
concordancing, especially through annotating concordance for multiple linguistic predictors from
linguistic contexts and increasingly the inclusion of psycholinguistic predictors relevant to corpus-
based analyses. By applying a range of sophisticated (multifactorial/-variate) statistical techniques,
researchers can assess how linguistic and cognitive predictors (jointly) correlate with aspects of learn-
ers’ interlanguage. Many such studies model the probability of occurrence of a given form, which
contrasts with psycholinguist studies in which often the focus is on production/comprehension
accuracy and reaction times. Overall, combining comprehensive annotation and statistical analysis
has helped corpus linguists better understand how notions central to psycholinguistics contribute to
L2 acquisition with regards to word/sense entrenchment, the association/contingency of formal and
functional elements, and matters of categorization, amongst others (Gries, 2014a, p. 287). In section
3.3 of this chapter, Corpora as primary data, we present and discuss one such example, Wulf and
Gries (2019).
With their rich contextual information, concordances facilitate the exploration of recency efects,
which can be manifested through (i) (structural) priming (which, in statistical terms, would be mani-
fested as autocorrelation, a short-term efect) and (ii) dispersion (a kind of long-term recency, refer-
ring to the distribution of an element across texts, speakers, registers/genres, etc.). See for instance
McDonough and Trofmovich (2009) for key priming research in L2 psycholinguistics and see Gries
and Wulf (2005) for a corpus-based study of the syntactic priming of ditransitive and prepositional
dative constructions in L2.
As for (i), priming refers to the fact that an occurrence of x increases the probability of x recur-
ring beyond its (frequency-based) baseline. Priming occurs at all linguistic levels as well as non-
linguistic levels (e.g., conceptual representation) and can result from implicit learning and the pattern
extraction mechanisms assumed in usage-based frameworks (see, e.g., Rowland et al., 2012). While
language acquisition researchers can control for priming by experimental design, this is still rare,
and more comprehensive priming studies are mostly emerging in observational data (see Gries &
Kootstra, 2017). Thus, corpus-based research has not only enriched and validated our understanding
168
of priming, but has also been used for hypothesis generation (Gries & Kootstra, 2017) by providing
data often ecologically more valid than that from an experimental setting. However, corpus-based
studies on priming efects in L2 remain relatively rare and attention is needed to determine (i) what
these efects look like in L2 and (ii) how they can be best accounted for statistically.
As for dispersion, this notion relates to general learning processes (Ambridge et al., 2006) and has
been used most for lexis and response rates/reaction times in lexical decision tasks (Gries, 2019b) and
the (even) distribution of words (in constructional slots) across, say, a whole corpus. For example,
language users are more likely to experience constructions widely/evenly distributed in time/place.
When that happens, contextual dispersion indicates that a construction is broadly conventional-
ized and temporal dispersion shares out recency efects (Gries, 2019b, p. 114). Given recent results
(Baayen, 2010, Gries, 2019b, 2022), dispersion may well supersede frequency in its importance for
(L2) learning: (More) evenly distributed exemplars can be assumed to be (more broadly) conven-
tionalized and, therefore, to facilitate acquisition more. Generally, in psycholinguistics, dispersion is
a central factor to be accounted for because it afects every kind of frequency of (co)-occurrence in
a corpus.
In terms of computation, dispersion is best computed based on linguistically meaningful
corpus parts (e.g., fles/texts, sub-registers, registers/genres, language production modes); see
Gries (2008) or Egbert et al. (2020). Callies (2013) illustrates well how dispersion over fles/
speakers can account for individual-speaker variation. Disregarding dispersion in corpus analyses
can lead to generalizations over parts of the corpus that may or may not be valid (e.g., speech-
specifc patterns may be attributed to written language), which can undermine all conclusions
of an analysis.
Contingency
Associations (i.e., contingency) between linguistic forms and/or between them and their functions
also play an essential role in all aspects of language. Based on Siyanova-Chanturia and Pellicer-
Sanchez (2018) and Wray (2002), formulaic units are psychologically real and according to Durrant
(2014) and Durrant & Schmitt (2009), collocational knowledge is an important aspect of learner
language. For Ellis (2006, p. 7), “[l]anguage learning can be viewed as a statistical process requir-
ing the learner to acquire a set of likelihood-weighted associations between constructions and their
functional/semantic interpretations.” As speakers learn their L2, they acquire the ability to map forms
and functions reliably by keeping track of a wide range of co-occurrence information of both their
language comprehension and production (Gries & Ellis, 2015, p. 21). Contingency therefore drives
associative learning (Ellis, 2016, p. 62); for instance, collocational and phraseological knowledge is
central to the attainment of native-like fuency and native-like idiomaticity (Pawley & Syder, 1983).
To date, much corpus-based research on contingency has been conducted based on linguistic
co-occurrences such as collocations, colligations, and collostructions (i.e., lexico-grammatical co-
occurrences; see Gries & Durrant, 2020 for an overview). Specifcally, contingency, as manifested
by co-occurrence counts, helps to explore what-if relations, i.e., what happens if the context is like
this? Corpus-based contingency work assumes that (i) “human learning is . . . perfectly calibrated
with normative statistical measures of contingency” (Ellis, 2006, p. 7) and (ii) statistical associations
between linguistic elements found in corpus data refect the psychological associations in the minds
of language learners (Stefanowitsch, 2006).
Therefore, many association measures (AMs) have been developed including conditional prob-
ability, (logged) PFisher-Yates exact, t, z, odds ratio, MI but there is currently no consensus on how to best
measure contingency with regard to symmetry of association, type of metric, and frequency infor-
mation. However, “diferent AMs ofer diferent and complementary perspectives on collocation
169
learning” (Gries & Durrant, 2020; see Durrant (2014) on the relationship between L2 learners’
knowledge of English collocations and various measures of collocation frequency and association).
One approach widely adopted in SLA corpus-based research is collostructional analysis (CA; Gries &
Stefanowitsch, 2004), a family of approaches to quantify (i) degrees of attraction/repulsion of words
(typically verbs) to syntactically defned words in a construction (collexeme analysis), (ii) which
words are attracted/repelled by one of several constructions (distinctive collexeme analysis), and (iii)
identify (dis)preferred pairs in two slots of one construction (co-varying collexeme analysis). Findings
of such work inform many studies on issues such as item-based learning or generalization and the
question of which words are particularly attracted to particular constructions and, therefore, likely to
function as path-breaking words in constructional acquisition.
Prototypicality and salience

Concordancing and contingency are also useful to investigate the interrelated notions of proto-
typicality and salience. According to the infuential weighted-attribute approach, a prototype is an
abstract entity—not a concrete exemplar—which combines the most salient attributes of the cat-
egory, wherein (i) those attributes are those with a high cue validity for the category, and (ii) the
cue validity of an attribute A (e.g., fying) of object X (e.g., a sparrow) with regard to a category C
(e.g., birds) is the conditional probability of X being a member of category C given that X exhibits
A p(C|A) (see Taylor, 2011: Section 5.2; Ellis et al., 2016).
Prototypes, or more precisely, entities close to the abstract prototypes, exhibit a variety of
efects, many of which are measurable with corpora: They are acquired earlier, produced more
often (i.e., they are often more frequent and more associated to certain contexts), recognized faster,
invite generalizations more than more marginal category members, are perceptually more salient,
etc. (Taylor, 2011). But corpora can not only help identify specifc characteristics that are a part
of a prototype, but they can also determine to what degree these characteristics contribute to a
prototype. For example, Ellis et al. (2016) used corpora to study semantic prototypicality in L2
verb argument constructions (VACs) and build semantic networks based on verb type frequencies
as extracted from VAC distributions. Their data show how quantitative measurements of semantic
relations between verb types and VAC frames can be used to explore L2 speakers’ linguistic knowl-
edge based on co-occurrence patterns in corpora. Similarly, much of the work on collostructional
analysis identifes the verbs most strongly attracted to certain constructions because these verbs
refect the prototypical sense(s) of a construction (e.g., give and tell for the ditransitive construction;
see Gries & Stefanowitsch, 2004).
Surprisal
Surprisal is somewhat diferent from the other notions. On the one hand, it is a driving force of the
language learning process (Ellis et al., 2016). According to Jágrová et al. (2019, p. 244), “[i]ntuitively,
it can be thought of as measuring the information content conveyed by a linguistic unit [given its
(preceding) context] and it appears to scale the cognitive efort required to process this information.”
Thus, like most usage-based notions, surprisal implies a probabilistic approach to language. For
example, when a hearer hears a certain verb and, from that, expects (or predicts) a certain comple-
mentation pattern to follow, which then does not happen, the learning process is enhanced: “One
consequence is that, when prediction goes wrong, it is surprisal that maximally drives learning from
a single trial. Otherwise, the regularities of the usual course of our experiences add up little by little,
trial after trial, to drive our expectations” (Ellis, 2016, p. 344). Within the visual world eye-tracking
paradigm, this type of approach aligns with Jackson and Hopp’s (2020) exploration of whether pre-
diction failures during real-time processing drive language acquisition.
170
On the other hand, surprisal is also central as a moderator for some of the other notions discussed
earlier. For instance, Jaeger and Snider (2008) showed that surprisal can amplify priming efects:
Unexpected uses prime more than expected ones and, arguably, surprisal will also amplify salience
simply because an expected expression will “stick out more” as it is processed. However, corpus
studies involving surprisal in L2 acquisition research are extremely rare. One exception is Wulf et al.
(2018), a corpus analysis of optional that complementation in native and learner English. They use
surprisal—how surprising the transition is from a matrix clause to a complement clause—as a predic-
tor in a multifactorial analysis alongside twelve other linguistic factors and fnd that speakers smooth
over more surprising local transitions by using that while low-surprisal transitions (e.g., the as part of
the complement clause subject after I think) feature that much less often.
Example studies
Corpora can play diferent roles and serve diferent purposes in L2 psycholinguistic research: They
can (i) serve as methodological tools in setting up experimental studies; (ii) complement experiments
(i.e., when experimental and corpus approaches are triangulated); and (iii) serve as the main source
of empirical data. In this section, we illustrate each role by reviewing individual studies.
Corpora and experimental design

Amongst L2 experimentalists, native-language corpora have become widespread tools in the devel-
opment of experiments: They have been used to extract word frequencies in native language (L1)
to be used as predictors or control variables to assess learners’ performances (Gries & Ellis, 2015)
and they have allowed scholars to establish native-like baselines against which to contrast L2. This
approach has proved popular in processing (e.g., Spinner et al., 2017), morphology (e.g., Matusevych
et al., 2018), syntax (e.g., Hopp, 2017), collocational knowledge (e.g., Toomer & Elgort, 2019),
linguistic contexts and their efects on phonolexical processing (e.g., Chrabaszcz & Gor, 2014), and
constructional knowledge (e.g., Kim & Rah, 2019). In the case of Kim and Rah (2019), following
Johnson and Goldberg (2013), the authors used the Corpus of Contemporary American English
(COCA) to select verbs for an experiment designed to explore L2 learners’ sensitivity to construc-
tional information and learners’ efciency in integrating information from a verb and a construction
in real-time processing. Corpus data were integrated into the experiment by calculating lexical verb
frequencies from the COCA and by conducting a collostructional analysis to quantify how much
individual lexical verbs co-occur with the investigated constructions. This helped the authors to
choose experimental stimuli based on both frequencies and association strengths for lower-frequency
target verbs and their higher-frequency counterparts, which ultimately increased the generalizability
of the analysis. More theoretically, the authors showed that learners integrate argument roles between
a verb and a construction faster when focusing on constructions rather than lexical verbs, stressing the
importance of constructional information for L2 sentence processing.
Triangulating corpus-based and experimental approaches

Corpus data in L2 psycholinguistic research can involve triangulating corpus and experimental meth-
odologies complementarily. With corpora, L2 phenomena can be often explored at a (potentially)
much larger scale than experiments would allow and with greater ecological validity—with experi-
ments, much control can be exercised but potentially at the cost of ecological validity. Ellis et al.
(2016) adopted both corpus and experimental methodologies to explore the extent to which native
speakers’ knowledge of VACs difers from that of L2 learners and whether the type of learners’ L1 in
terms of verb semantics is a bias towards their knowledge of the L2. They frst focused on native data
171
and extracted and analyzed VACs from the British National Corpus to determine their contingen-
cies with verbs. Then, for each VAC, they identifed and quantifed core meanings and construction
semantic networks around the notions of prototypicality, semantic cohesion, and polysemy.
In a second step, they compared the VAC uses in the native data with those of German/Czech/
Spanish L2 learners, which they elicited experimentally with a survey. The corpus fndings estab-
lished that VACs promote learning in that (i) they are Zipfan (i.e., many words are very rare, very
few words are very frequent) in their verb type-token ratio constituency in usage, (ii) constructions
prefer certain verbs in them, and (iii) they are coherent in their semantics. For the experimental part,
native and L2 participants were administered an online VAC survey involving a free association task.
The verb responses collected for each VAC were lemmatized by verb type and ordered by verb token
frequencies. The authors then compared lists based on the learner responses with lists based on native
English responses, and focus was given to the potential efects of frequency, contingency, and seman-
tic prototypicality. Verb frequency in the VAC, VAC-verb contingency, and verb prototypicality co-
determined learners’ responses to VAC prompts, leading to the conclusion that “L2 VAC processing
involves rich associations, tuned by L2 verb type and token frequencies and their contingencies of
usage, which interface syntax, lexis and semantics.”
Corpora as primary data

Finally, corpora can be explored as primary sources of data. As mentioned earlier, frequency, recency,
and context afect SLA jointly and their mutual interaction afects how learners acquire their L2:
“The more times we experience conjunctions of features, the more they become associated in our
minds and the more these subsequently afect perception and categorization” (Ellis et al., 2016,
p. 46). LCR scholars are now exploring such interactions in corpora and how they contribute to
the process of L2 acquisition and use with multifactorial/multivariate approaches involving many
linguistic predictors derived from concordance lines (e.g., animacy, verb semantics, verb type, voice)
and cognitive predictors operationalizing many of the above notions.
For example, Wulf and Gries (2019) targeted the syntactic alternation between verb-particle-object
(VPO) vs. verb-object-particle (VOP) constructions across native English and over a dozen interlanguage
varieties with a multifactorial analysis of approximately 5,000 occurrences of the two constructions from
various L1 and L2 corpora. They explored how learners’ syntactic choices were infuenced by processing
demands, input efects, and L1 typology by analyzing the joint efects of 17 predictors, including, for
instance, the order of constituents in the verb particle construction (VPC), lengths of the particle and
the direct object noun phrase, the complexity of the direct object, and rhythmic alternation in the VPC.
They used the MuPDAR approach (Multifactorial Prediction and Deviation Analysis using Regression;
see Gries & Deshors, 2014), a two-step regression procedure that computes for every learner choice
what an L1 speaker would have chosen in the same context and then explores where and why the learner
choices difer from the L1 speaker choices. Among other things, Wulf and Gries found that, frstly,
learners overuse VPO constructions across nearly all contextual conditions. Secondly, particle stranding
in the VOP construction incurs a cognitive load, which impacts learners’ constructional choices more
strongly in speech than in writing, presumably due to the demands of online processing in speech.
Advantages and limitations of corpus data for L2 psycholinguistics
Advantages
Corpora have much to ofer to L2 psycholinguists: They ofer methodological options avoiding
potential problems associated with experimental designs, such as low ecological validity and input
misrepresentation.
172
Regarding the former, by nature, experimental studies are conducted under highly controlled
conditions based on carefully elicited or selected data. In contrast, as an authentic/spontaneous
and highly contextualized type of data, corpora can lead to studies with a higher level of eco-
logical validity because the data are not produced in, say, read-one-sentence-at-a-time kinds of
situations.
Regarding input distribution, the controlled (i.e., well balanced) design of experiments
often means that participants are exposed to unrepresentative distributions of investigated lin-
guistic elements (Gries, 2019a), which can be problematic given that “learning effects can be
observed even over a relatively small number of experimental stimuli” (Gries, 2019a, p. 235;
see also Baayen et al., 2018). Put differently, there can be within-experiment learning effects
given the unrepresentative input, but there can also be other within-experiment factors such as
fatigue or habituation that can distort the results of experiments especially with learners whose
probabilistic knowledge is not yet as robust as that of native speakers—corpora do not come
with these problems.
A fnal advantage is that corpora allow scholars to complement experimental L2 psycholin-
guistic research. For example, the study of how priming efects difer across diferent prime-target
distances in a setting not perturbed by unrepresentative experimental input can beneft much
from corpus studies, which are also well suited for exploratory investigations and hypothesis-
generating work.
Limitations
Despite, or in fact because of, their advantages, corpora also come with challenges: Ecological valid-
ity of data also means such data are often noisy and heterogeneous. For instance, given that conditions
of language production are not always strictly controlled, linguistic contexts may vary across the uses
of a given linguistic element across speakers. Further, data distribution can be problematic because
data are often unbalanced/Zipfan with frequencies of words that decrease as a power function of
their rank in the frequency table (Ellis et al., 2016), which can require very complex statistical analy-
ses. Also, corpus data can present challenges in terms of (i) collinearity between predictors and (ii)
the often necessary inclusion of many control variables to control statistically for what, with corpora,
cannot be controlled for by (experimental) design.
Finally, metadata regarding speakers are often lacking, making it hard to fully account for the
learning/sociodemographic background of the speakers in the corpus and how they may afect learn-
ers’ acquisition of the L2.

Much work remains to be done for corpus linguists involved in L2 psycholinguistic research; Textbox
14.2 summarizes the main direction for the research areas we’ve discussed in this chapter.
Regarding corpus compilation, various aspects of this process are important to address. Specif-
cally, we need (more) corpora that
• are bigger (to have more data points for proper statistical modeling);
• have richer metadata (e.g., degree of motivation, attitude towards the L2, cognitive variables,
etc., to know more about the speakers represented);
• are more varied—L1s, register, mode, etc. (to have more diverse data to generalize from);
• are longitudinal (to track development better than with cross-sectional studies);
• are not only accessible on a website (to compute statistics that web interfaces do not
provide).
173
Quantitative analyses of such corpora will bring a quality to both psycholinguistic and corpus-based
analyses that is as yet difcult to attain.
Regarding statistical approaches to corpora, we need more sophisticated conceptualizations of
the various forms in which frequencies in corpora can operationalize cognitive/psycholinguistic
notions. To return to an example from above, the feld needs to recognize that frequency must be
complemented with dispersion information (see again Adelman et al., 2006; Baayen, 2010, Gries,
2019b), given the evidence that frequency is a convenient, but imprecise, proxy for entrenchment:
Words of extremely similar frequencies can difer massively in their dispersion and their likelihood
of being known by learners. Thus, studies relying on frequency (as a predictor or as a control) alone
are likely to fail in properly capturing “entrenchment” or “exposure.” One can only wonder why
there are dozens of measures of association, lexical diversity, lexical dispersion, but an experimentally
supported variable such as dispersion is ignored.
Finally, much like surprisal, the notion of salience (and its role in SLA) remains to be operational-
ized in a valid corpus-based way (Gass et al., 2018). We are currently not aware of work incorporat-
ing salience in a systematic corpus-based/-driven fashion. It is conceivable that salience could in fact
be considered as a central component of surprisal and, thus, be operationalized the same way (see
Gries, 2019b, p. 65) but this awaits corpus-linguistic exploration.
It is our view that (methodological) innovation in L2 psycholinguistics is dependent upon close
collaboration between L2 psycholinguists and corpus linguists (see Rebuschat et al., 2017 for a spe-
cial issue on multidisciplinary research across experimental, computational, and corpus-based meth-
odological boundaries in [S]LA). Both types of researchers use corpora, but to varying degrees and
for diferent purposes: While most L2 psycholinguistics research published in fagship SLA journals
does not yet ofer analyses of corpora as primary data sources, learner corpus scholars do utilize cor-
pora primarily as main data sources but they are yet to conduct analyses that fully account for com-
prehensive ranges of psycholinguistic predictors. By working together, L2 psycholinguists and corpus
linguists will undoubtedly manage to shed more light on what it means to acquire a second language.
Corpus compilation: (Continue to) develop larger corpora with more varied data. Continue develop-
ment of longitudinal corpora.
Corpus metadata: Compile metadata databases on individuals, cognitive variables, aptitude, motivation,
etc.
Statistical approaches to corpora: (Continue to) develop corpus-based ways to handle noisy data.
Salience in L2: Follow the footsteps of Wulf et al. (2018) by developing studies that operationalize and
measure salience.
Notes
1 See Christiansen and Chater (2016) for a summary of the central role of frequency in psycholinguistic work.
Frequency efects are often observed in reaction times: Less frequent items incur higher reaction times com-
pared to more frequent items; note that this does not prove that frequency is the cause.
2 We include salience in Textbox 14.1 for a complete picture of the notions involved in L2 acquisition and
their operationalization in corpus studies. However, due to length constraints, salience is not extensively
discussed.
174
Further reading
Crossley, S., Kristopher, K., & Salsbury, T. (2016). A usage-based investigation of L2 lexical acquisition: The role
of input and output. The Modern Language Journal, 100(3), 702–715.
This corpus-based longitudinal study focuses on salience and the extent to which L2 learners are more likely
to produce lexical items in their L2 that are more salient in the L1 input.
Ellis, N. C., Römer, U., & O’Donnell, M. B. (2016). Usage-based approaches to language acquisition and processing:
Cognitive and corpus investigations of construction grammar. Language Learning Monograph.
This volume explores verb argument constructions in frst and second language acquisition, processing, and
use from a usage-based theoretical perspective. While reafrming the value of interdisciplinarity, the authors
show us the power of corpus data to better understand L2 acquisition and processing.
Gries, St. Th. (2019). Ten lectures on corpus-linguistic approaches: Applications for usage-based and psycholinguistic
research. Brill (Distinguished Lectures in Cognitive Linguistics series).
This volume connects psycholinguistics and CL by focusing on the operationalization and measurement of
cognitive notions such as frequency, dispersion, and context for quantitative analysis.
References
Adelman, J., Brown, G., & Quesada, J. (2006). Contextual diversity, nor word frequency, determines word-
naming and lexical decision times. Psychological Science, 17(9), 814–823.
Ambridge, B., Theakston, A., Lieven, E., & Tomasello, M. (2006). The distributed learning efect for children’s
acquisition of an abstract syntactic construction. Cognitive Development, 21(2), 174–193.
Aslin, R., & Newport, E. (2012). Statistical learning: From acquiring specifc items to forming general rules.
Current Directions in Psychological Science, 21, 170–176.
Baayen, H. (2010). Demythologizing the word frequency efect: A discriminative learning perspective. The
Mental Lexicon, 5(3), 436–461.
Baayen, H., van Rija, J., de Catc, C., & Wood, S. (2018). Autocorrelated errors in experimental data in the
language sciences: Some solutions ofered by Generalized Additive Mixed Models. In Mixed-efects regression
models in linguistics (pp. 49–69). Springer.
Ball, L., Shocker, J., & Miles, J. (2010). Odour-based context reinstatement efects with indirect measures of
memory: The curious case of rosemary. British Journal of Psychology, 101(4), 655–678.
Bybee, J., & Hopper, P. (Eds.). (2001). Frequency and the emergence of linguistic structure. John Benjamins.
Callies, M. (2013). Agentivity as a determinant of lexico-grammatical variation in L2 academic writing. Interna-
tional Journal of Corpus Linguistics, 18(3), 357–390.
Casenhiser, D., & Goldberg, A. (2005). Fast mapping between a phrasal form and meaning. Developmental Sci-
ence, 8, 500–508.
Chrabaszcz, A., & Gor, K. (2014). Context efects in the processing of phonolexical ambiguity in L2. Language
Learning, 415–455.
Christiansen, M., & Chater, N. (2016). Creating language: Integrating evolution, acquisition and processing. The MIT
Press.
Durrant, P. (2014). Corpus frequency and second language learners’ knowledge of collocations. International
Journal of Corpus Linguistics, 19(4), 443–477.
Durrant, P., & Schmitt, N. (2009). To what extent do native and non-native writers make use of collocations?
International Review of Applied Linguistics, 47(2), 157–177.
Egbert, J., Burch, B., & Biber, D. (2020). Lexical dispersion and corpus design. International Journal of Corpus
Linguistics, 25(1), 89–115.
Ellis, N. (2002). Frequency efects in language acquisition: A review with implications for theories of implicit
and explicit language acquisition. Studies in Second Language Acquisition, 24, 143–188.
Ellis, N. (2006). Language acquisition as rational contingency learning. Applied Linguistics, 27(1), 1–24.
Ellis, N. (2007). Language acquisition as rational contingency learning. Applied Linguistics, 27(1), 1–27.
Ellis, N. (2016). Salience, cognition, language complexity, and complex adaptive systems. Studies in Second Lan-
guage Acquisition, 38(2), 341–351.
Ellis, N., Römer, U., & O’Donnell, M. (2016). Usage-based approaches to language acquisition and processing: Cogni-
tive and corpus investigations of construction grammar. Language Learning Monograph Series. Wiley-Blackwell.
175
Gass, S., Spinner, P., & Behney, J. (Eds.). (2018). Salience in second language acquisition. Routledge.
Godfroid, A. (2016). The efects of implicit instruction on implicit and explicit knowledge development. Studies
Godfroid, A., & Uggen, S. (2013). Attention to irregular verbs by beginning learners of German—an eye move-
ment study. Studies in Second Language Acquisition, 35(2), 291–322.
Goldberg, A. (2006). Constructions at work: The nature of generalization in language. Oxford University Press.
Gries, St. Th. (2008). Dispersions and adjusted frequencies in corpora. International Journal of Corpus Linguistics,
13(4), 403–437.
Gries, St. Th. (2014a). Corpus and quantitative methods. In J. Taylor & J. Littlemore (Eds.), The bloomsbury
companion to cognitive linguistics (pp. 279–300). Bloomsbury.
Gries, St. Th. (2014b). Frequencies, probabilities, association measures in usage-/exemplar-based linguistics:
Some necessary clarifcations. In N. Gisborne & W. Hollmann (Eds.), Theory and data in cognitive linguistics
Gries, St. Th. (2017). Corpus approaches. In B. Dancygier (Ed.), Cambridge handbook of cognitive linguistics (pp.
590–606). Cambridge University Press.
Gries, St. Th. (2019a). Priming of syntactic alternations by learners of English: An analysis of sentence-completion
and collostructional results. In J. Egbert & P. Baker (Eds.), Using corpus methods to triangulate linguistic analysis
Gries, St. Th. (2019b). Ten lectures on corpus-linguistic approaches: Applications for usage-based and psycholinguistic
research. Distinguished Lectures in Cognitive Linguistics Series. Brill.
Gries, St. Th. (2022). On, or against?, (just) frequency. In H. Boas (Ed.), Directions for pedagogical Construction
Grammar: Learning and teaching (with) constructions (pp. 47–72). De Gruyter Mouton.
Gries, St. Th., & Deshors, S. (2014). Using regressions to explore deviations between interlanguage and native
language: Two suggestions. Corpora, 9(1), 109–136.
Gries, St. Th., & Durrant, P. (2020). Analyzing co-occurrence data. In M. Paquot & St. Th. Gries (Eds.), A
practical handbook of corpus linguistics (pp. 141–159). Springer.
Gries, St. Th., & Ellis, N. (2015). Statistical measures for usage-based linguistics. Language Learning, 65, 1–28.
Gries, St. Th., & Kootstra, G. (2017). Structural priming within and across languages: A corpus-based perspec-
tive. Bilingualism: Language and Cognition, 20(2), 235–250.
Gries, St. Th., & Stefanowitsch, A. (2004). Extending collostructional analysis: A corpus-based perspective on
“alternations”. International Journal of Corpus Linguistics, 9(1), 97–129.
Gries, St. Th., & Wulf, S. (2005). Do foreign language learners also have constructions? Evidence from priming,
sorting, and corpora. Annual Review of Cognitive Linguistics, 3, 182–200.
Halliday, M. (2005). Computational and quantitative studies. Continuum.
Hopp, H. (2017). Cross-linguistic lexical and syntactic co-activation in L2 sentence processing. Linguistic
Approaches to Bilingualism, 7, 96–130.
Horton, W. (2007). The infuence of partner-specifc memory associations on language production: Evidence
form picture-naming. Language and Cognitive Processes, 22(7), 1114–1139.
Jackson, C. N., & Hopp, H. (2020). Prediction error and implicit learning in L1 and L2 syntactic priming.
International Journal of Bilingualism, 24(5-6), 895–911.
Jaeger, F., & Snider, E. (2008). Implicit learning and syntactic persistence: Surprisal and cumulativity. In Proceed-
ings of the 29th annual Cognitive Science Society (pp. 1061–1066).
Jágrová, K., Avgustinova, T., Stenger, I., & Fischer, A. (2019). Language models, surprisal and fantasy in Slavic
intercomprehension. Computer Speech and Language, 53, 242–275.
Johnson, M., & Goldberg, A. (2013). Evidence for automatic accessing of constructional meaning: Jabberwocky
sentences prime associated verbs. Language and Cognitive Processes, 28, 1439–1452.
Kemmer, S., & Barlow, M. (1999). Introduction: A usage-based conception of language. In M. Barlow & S.
Kemmer (Eds.), Usage-based models of language (pp. 7–28). Center for the Study of Language and Information.
Kim, H., & Rah, Y. (2019). Constructional processing in a second language: The role of constructional knowl-
edge in verb-construction integration. Language Learning, 69, 1022–1056.
Lakof, G. (1991). Cognitive versus generative linguistics: How commitments infuence results. Language and
Communication, 11(1–2), 53–62.
Langacker, R. (1987). Foundations of cognitive grammar: Theoretical prerequisites (Vol. 1). Stanford University Press.
Matusevych, Y., Alishahi, A., & Backus, A. (2018). Quantifying cross-linguistic infuence with a computational
model: A study of case-marking comprehension. Linguistic Approaches to Bilingualism, 8, 561–605.
McDonough, K., & Trofmovich, P. (2009). Using priming methods in second language research. Routledge.
Pawley, A., & Syder, F. (1983). Two puzzles for linguistic theory: Nativelike selection and nativelike fuency. In
J. Richards & R. W. Schmidt (Eds.), Language and communication (pp. 191–225). Longman.
176
Rebuschat, P., Meurers, D., & McEnery, T. (Eds.). (2017). Language learning research at the intersection of
experimental, computational, and corpus-based approaches. Special issue of Language Learning, 67.
Rowland, C., Chang, F., Ambridge, B., Pine, J., & Lieven, E. (2012). The development of abstract syntax: Evi-
dence from structural priming and the lexical boost. Cognition, 125(1), 49–63.
Siyanova-Chanturia, A., & Pellicer-Sanchez, A. (Eds.). (2018). Understanding formulaic language: A second language
acquisition perspective. Routledge.
Schmid, H.-J. (2000). English abstract nouns as conceptual shells: From corpus to cognition. de Gruyter.
Smith, S., Heath, F., & Vela, E. (1990). Environmental context-dependent homophone spelling. The American
Journal of Psychology, 103(2), 229–242.
Spinner, P., Foote, R., & Upor, R. (2017). Gender and number processing in second language Swahili. Linguistic
Stefanowitsch, A. (2006). Distinctive collexeme analysis and diachrony: A comment. Corpus Linguistics and Lin-
guistic Theory, 2(2), 257–262.
Taylor, J. R. (2011). Prototype theory. In C. Maienborn, K. von Heusinger, & P. Portner (Eds.), Semantics
(pp. 643–664). De Gruyter Mouton.
Toomer, M., & Elgort, I. (2019). The development of implicit and explicit knowledge of collocations: A con-
ceptual replication and extension of Sonbul and Schmitt (2013). Language Learning, 69, 405–439.
Wray, A. (2002). Formulaic Language and the lexicon. Cambridge University Press.
Wulf, S., & Gries, St. Th. (2019). Particle placement in learner English: Measuring efects of context, frst lan-
guage, and individual variation. Language Learning, 69(4), 873–910.
Wulf, S., Gries, St. Th., & Lester, N. (2018). Optional that in complementation by German and Spanish learn-
ers. In A. Tyler, L. Huan, & H. Jan (Eds.), What is applied cognitive linguistics? Answers from current SLA research
(pp. 99–120). De Gruyter Mouton.
177
15
SYNTHESIS
Research methods in second language
psycholinguistics
Nan Jiang
Introduction
Psycholinguistic research explores mental representations and cognitive processes involved in lan-
guage use and acquisition. Because these representations and processes are not directly observable, we
can only make inferences about them based on observable language behavior. Real-life language use,
though, does not often provide the most efective basis for making such inferences because an indi-
vidual’s language behavior can be simultaneously afected by many individual, linguistic, contextual,
and sociocultural factors. Thus, psycholinguists usually observe language behavior under a highly
controlled environment that involves the adequate manipulation and rigorous control of variables.
Such observations then represent language processing data. The specifc forms of such data can vary
a great deal depending on the tasks and paradigms of a study. They may be specifc language output
(such as response words in a word association task), reaction times (as in naming pictures), accuracy
rates (as in a discrimination task in speech perception), rating scores (as in semantic rating study),
eye movement patterns (as used in the visual world paradigm), skin conductance or pupillary size
measures (as in studying emotion arousal in processing frst language (L1) and second language (L2)),
electrophysiological activities on the scalp (where event-related potentials are measured in examining
L2 semantic and morphosyntactic processing), or blood and oxygen activities of the brain cortex (as
in exploring cerebral involvement in L1 and L2 processing). When a pattern emerges from such data,
it becomes a psycholinguistic phenomenon which is often referred to as an efect, such as the transla-
tion priming efect or the interlingual neighborhood size efect. These phenomena provide the basis
for making inferences about mental representation and cognitive processes in L2 use and acquisition.
Given its emphasis on mental representation and processes, and the need to make valid inferences,
psycholinguistic research is known for its three methodological hallmarks. The frst is the use of a
rich repertoire of specifc experimental tasks and paradigms. Language use and acquisition is a very
complex process with many subprocesses (for instance, for sentence production, message formation,
lexical access, syntactic structure building, linguistic form assembly, articulation) at diferent levels of
the linguistic hierarchy (e.g., phonological, lexical, syntactic, pragmatic). A wide array of experimen-
tal tasks and paradigms are necessary for exploring diverse language processing phenomena. These
tasks are also more specifc than language use in real life, in the sense that they are quite limited in
what is required of the participants. For example, in a lexical decision task (LDT), all that is required
of a participant is to decide whether the stimulus is a word or not. In a naming task, participants
are asked to name a word or a picture as quickly as possible. Such specifc language tasks may allow
178 DOI: 10.4324/9781003018872-17

Synthesis
researchers to target a specifc cognitive process involved in language use. For example, picture nam-
ing can help explore concept activation, lexical decision is efective for examining lexical access, and
self-paced reading of sentences allows one to investigate the process of parsing, i.e., how a syntactic
structure is built step-by-step during reading comprehension. This does not imply, however, that
there is a one-to-one correspondence between a task and a cognitive process. In actuality, the map-
ping of behavioral tasks onto cognitive processes is complex.
The second hallmark is vigorous variable manipulation and control, which is the key to making
reasonable and valid inferences while interpreting and explaining the data. In studying whether L2
speakers are sensitive to a morphosyntactic violation such as plural errors, for example, researchers
construct a grammatical and an ungrammatical version of a sentence; these versions are identical
except for the plural error in the ungrammatical version to ensure other factors (i.e., stimulus-related
confounds) do not infuence participants’ responses. In comparing free recall accuracy in L1 and L2,
it is desirable to test the same group of bilinguals in their L1 and L2, that is, using a within-participant
design, so that individual diferences in factors such as working memory are controlled. While doing
so, it is also desirable to counterbalance the language order across the participants by asking half of
them to do L1 frst and the other L2 frst so that the results are not afected by language order (see
Chapter 2 of Jiang, 2012, for more detailed discussion and examples).
The third hallmark is the accuracy in stimulus display and behavior recording which ensures data
quality. This includes a millisecond (ms) accuracy in timing in both stimulus display and response
recording in studies in which the primary data are response latencies (see Hamrick, this volume). For
example, when the masked priming paradigm is adopted, the prime is often displayed for 60 ms or
shorter, since a longer display, e.g., 100 ms, would result in conscious perception of the prime by a
participant in spite of the presence of a mask. When eye movement is monitored or pupil size is mea-
sured, the eye-tracker should be calibrated correctly and frequently to ensure spatial or size accuracy.
Data collected under such circumstances often provide better information than real-life language
behavior for making valid inferences. For example, in a word association task in which bilinguals
are asked to provide the frst word coming to mind upon seeing a stimulus word, if they provided more
form-related responses (e.g., sheep–sheet) in their L2 than in their L1, researchers could reasonably infer
that form relatedness plays a more important role in the organization of the L2 lexicon than in that
of the L1 lexicon. Similarly, in a monolingual English LDT, if Spanish-English bilinguals’ responses to
English words are afected by how many orthographically similar words (or neighbors) they have in
Spanish, researchers can reasonably infer and conclude that both languages are activated and interact
with each other even in a monolingual task and thus lexical access is nonselective in bilinguals.
The chapters in this section illustrate the methodological complexities in conducting good
psycholinguistic L2 research and ofer helpful guidelines. Elgort and Warren (2023 [this volume])
organize their discussion around three data types in comprehension research: reaction times from
LDTs, priming, and self-paced reading; eye movements; and event-related brain potentials. For
each data type, they discuss specifc experimental tasks and paradigms, the specifc research topics
that these methods have been used to explore, and some key concepts and technical details related
to the methods. Bernolet (2023 [this volume]) ofers an overview of the tasks used in language
production research, with a focus on word and sentence production. Key methods are explained
with due attention to technical details, along with examples of published studies to illustrate the
kinds of research topics that these methods can be used to examine. Gullberg’s chapter (2023
[this volume]) deals with methods and paradigms that are particularly related to the study of the
multimodal nature of language processing and learning. With ample specifc examples to show
the methodological uniqueness of this research, e.g., how gesture production is analyzed in rela-
tion to oral speech production, Gullberg demonstrates that methodological choices and con-
siderations are often guided and shaped by the research questions under investigation. Hamrick
(2023 [this volume]) ofers a comprehensive treatment of a more technical aspect of psycholinguistic
179
Nan Jiang
L2 research: the accuracy of timing in stimulus display and response recording. He explains
potential threats to timing accuracy in three categories: hardware, software, and web-based data
collection, and, more importantly, provides practical suggestions for alleviating such threats.
Finally, Deshors and Gries (2023 [this volume]) ofer an insightful discussion of how corpus lin-
guistics can facilitate psycholinguistic L2 research both in the area of stimulus development and
in providing large-scale, ecologically valid production data.
In the remainder of this overview chapter, I hope to outline six overarching issues that should be
considered in psycholinguistic L2 research. Three of them have to do with the selection of tasks and
paradigms. The other three deal with variable control, minimizing the infuence of explicit knowl-
edge, and assessing L2 profciency.
Theoretical aspect of tasks

The experimental task or paradigm is at the very heart of research design in L2 psycholinguistics.
Under some circumstances, previous researchers may have paved the way in identifying an efective
task for a topic; for example, using lexical decision for studying word recognition, picture naming
for studying concept activation, and self-paced reading for studying parsing in sentence processing.
Under other circumstances, a task or paradigm has to be selected from multiple possible options. A
good example to illustrate the latter case is the study of whether multiple-word expressions such as
formulas are represented holistically. This topic has been approached with tasks such as eye-tracking
(Underwood et al., 2004), grammaticality judgment (Jiang & Nekrasova, 2007), self-paced read-
ing (Tremblay et al., 2011), and word monitoring (Jeong & Jiang, 2019), to name just a few. These
tasks provide multiple perspectives of the same phenomenon, and each has its own limitations and
advantages in approaching a research question; thus, converging evidence from multiple tasks is often
desirable.
Even though experimental tasks such as lexical decision and naming are intended to focus on a
particular cognitive process, and are thus more specifc than real-life language use, it is very rare that
only a single process is involved in their completion. For example, lexical decision may involve at
least three processes: input analysis (e.g., identifying visual features and letters), lexical access (locat-
ing the word in the mental lexicon and accessing lexical information), and response. To name a
picture, one has to recognize the picture, activate the concept, activate the lexical item associated
with the concept, and articulate the word. Thus, whether one is adopting a task conventionally used
for studying a topic or exploring an innovative use of a task, a researcher should have a clear under-
standing of the cognitive processes involved in the task and how such processes are afected by their
manipulation, or have “a theory of method” in Foss’s (1998) words. Good published studies will serve
as examples for fnding and presenting a theory of method, as I illustrate in the following example.
A theory of method is indispensable in linking the research design (the task and manipulations) of
a study to a specifc cognitive process and then to the study question under investigation. To use Jiang
and Nekrasova (2007) as an example, a grammaticality judgment task (GJT) was adopted to study
whether formulaic expressions are represented holistically or analytically (see Jiang & Nekrasova, for
more information). The critical stimuli were formulaic and nonformulaic expressions, e.g., at the
same time, at the same age, plus ungrammatical expressions (e.g., same the hope at) serving as fllers as
required by the task. The participants’ task was to determine whether these word sequences, which
were presented one by one, by themselves (not embedded in a sentence), and in a random order,
were grammatical or not. The GJT in this case had four components: a) recognize individual words,
b) perform a grammatical analysis to determine if a word sequence is grammatical, c) make a decision
about its grammaticality, and d) respond accordingly by pressing the YES or NO button. The under-
lying rationale of the method is that if a formula (e.g., at the same time) is represented holistically as a
single unit, it should be directly located in the lexicon and a positive decision can be made without
180
Synthesis
performing a grammatical analysis. In contrast, all four components are required in responding to
a regular phrase (e.g., at the same age), including syntactic analysis. This analysis leads to the predic-
tion that individuals should respond to a formula more quickly than to its nonformulaic counterpart
(assuming adequate control of other variables) because of the skipped grammatical analysis when
formulas are represented holistically. However, if formulas are not represented holistically and also
require grammatical analysis as nonformulaic expressions do, no RT diference should be present
between formulas and nonformulas (again assuming adequate control of other variables). Thus, by
examining the RT for the formulas and nonformulas in the GJT, an inference can be made regarding
the holistic representation of formulas. In other words, the adoption of the GJT in this study allows
the establishment of a relationship (through inferencing) between the RT data (faster RT for formu-
las than nonformulas) and the cognitive process afected (the skipping of syntactic analysis) and then
the conclusion (holistic representation of formulas).
In contrast, a lack of a proper understanding of the task may lead to faulty inferences. For example,
one may study the role of word shape in L2 visual word recognition by manipulating the appearance
of word stimulus. Thus, words are presented in alternating case (e.g., sTuDy) in one condition and in
regular, lower-case letters in another (e.g., study). A slower RT in an LDT for the alternating condi-
tion would seem to indicate that word shape does afect word recognition. The rationale underlying
this design may appear quite reasonable at frst glance. However, sublexical processing such as letter
recognition also afects performance in an LDT. East Asian learners of English as a second language
(ESL) often take more time in recognizing upper-case letters than lower-case letters (Jiang, 2021),
a phenomenon less known to some English native speakers. Thus, the long RT for case-alternating
words may arise from longer processing time needed for processing upper case letters rather than or
in addition to the unfamiliar shape of a word. In short, the inferences researchers make in interpret-
ing their data are valid only when there is a clear understanding of the cognitive processes involved
in a task and how such processes are afected by the experimental manipulations. By extension, a
researcher’s conclusion is convincing only when a reader can see a clear connection between the
data and the research questions under investigation. Thus, in writing an empirical study, it is almost
always desirable (but this part is often missing) to spell out in a clear way the rationale underlying
one’s research design. This includes a) an explanation of what cognitive processes are involved in a
task and how they are afected by the manipulation and b) what inferences are required in relating
the data to the purpose or the research questions of a study.
Technical aspect of tasks

In selecting a task, one also has to understand the technical aspect or “nuts and bolts” of the task.
All psycholinguistic tasks come with their own conventional ways of implementing the task and all
decisions about task implementation are made for a reason. For example, in an LDT, it is common
to include as many nonwords as words even though the former may not generate any useful data in
relation to the research question. This is done so that the task is not biased toward positive or negative
responses. The self-paced word-by-word reading task is often used to study whether L2 speakers have
developed native-like sensitivity to morphosyntactic violations. This sensitivity is assessed by examin-
ing if there is a delay in reading the critical region (where the error is) in ungrammatical sentences
as compared to grammatical sentences. Reading time has to be taken at the position following the
error as well, because the sensitivity may not materialize until reaching the word following the error,
a phenomenon referred to as the spill-over efect. In studies involving L2 speakers, it is desirable to
also include the reading time for one additional word since it may take them longer to show their
sensitivity (e.g., Jiang et al., 2011).
It is important to note that few tasks are uniform in their implementation. Instead, they often
have diferent versions that may tap diferent cognitive processes. For example, the naming task can
181
Nan Jiang
be done in a delayed and non-delayed version. The delayed version focuses more on the articulation
process, while multiple speech production processes such as conceptual activation are involved in the
non-delayed version. The word association task has multiple versions depending on input and output
modality (e.g., visual-written, cross-modal), whether responses are timed, and whether only one
response (the discrete version) or multiple responses (the continuous version) are required. There are
at least three diferent versions of the priming paradigm in terms of how stimuli are sequenced and
displayed (see Section 3.3.1 in Jiang, 2012, for more detail). The masked version taps early cognitive
processes in visual word recognition better, for example, than unmasked priming does. Knowing the
practical aspect of a task often means knowing how a particular version of a task should be imple-
mented and being able to link that back to theory.
It is always a good idea to pilot a study to determine if a task works with a particular topic, a par-
ticular set of test materials, and a particular participant group. Since many L2 psycholinguistic studies
involve the comparison of L1 and L2 speakers and take advantage of a psycholinguistic phenomenon
originally reported among L1 speakers, it is a good idea to test L1 speakers or L2 speakers in their L1
frst to see if the original fnding can be replicated with this task and your test materials. Piloting can
help discover many unforeseen issues. An example is the mask for masking prime words in a masked
priming study. In testing English and other Romance and Germanic languages, a set of English letters
or hash marks are often used as a mask, but such a mask does not mask a prime word well in an East
Asian language such as Korean, Japanese, or Chinese. A prime word is then often consciously visible
to a participant. A pilot study should help discover this problem. A solution in this case is to use a
graphic mask, such as a checkerboard graphic.
Task sensitivity
Tasks difer in their power or sensitivity in revealing a psycholinguistic efect or in responding to a
manipulation. If a group of Chinese ESL speakers are asked to identify an error in a sentence such as
John lost one of his key yesterday, they may be able to identify the missing plural with a high accuracy,
suggesting that they have developed native-like knowledge of English plural marking. However, if
one replaces the task with a self-paced reading task, in which reading time is compared between the
grammatical and ungrammatical versions of the same sentence, the ESL group may fail to show a
delay (as native speakers do) in reading the ungrammatical sentences (Jiang, 2004). In this example,
the online self-paced reading task in combination with RT data is more sensitive than the ofine
error identifcation task in helping reveal the diference in the integration of morphosyntactic knowl-
edge in L1 and L2 speakers.
Numerous studies have demonstrated the efect of task sensitivity. In masked translation priming
studies, L2 words were found to be less able to prime their L1 translations than the reverse in an LDT,
but they produced a reliable translation priming efect in semantic judgment (Grainger & Frenck-
Mestre, 1998) and episodic recognition (Jiang & Forster, 2001). Kim and Davis (2003) also demon-
strated that masked translation priming efects involving cognates, noncognates, and homophones
revealed themselves diferently in lexical decision, naming, and semantic categorization tasks. Simi-
larly, in comparing emotion arousal in processing L1 and L2, researchers have used tasks or measures
such as word association, semantic rating, free recall or recognition, Stroop color identifcation, skin
conductance, or pupillary response measures. Several studies demonstrated that these tasks varied in
diferentiating L1 and L2 performance. For example, Winskel (2013) compared Thai-English bilin-
guals in two tasks: emotionality rating and Stroop color identifcation. In the former, the participants
had to rate the unpleasantness of the target words, and in the latter, they were asked to press a button
based on the color of the target words, while ignoring the meaning of the words. The target words
were displayed in L1 and L2 and they were either negative or neutral words. The bilinguals produced
comparable performance in their L1 and L2 in the frst task, rating negative words as more unpleasant
182
Synthesis
in both languages. These results seemed to suggest no diference in emotion arousal in processing L1
and L2. However, in the Stroop task, color identifcation latencies showed a delay in responding to
negative words in L1 only, suggesting a stronger emotion involvement in L1 than in L2 processing.
Similar task- or measure-based diferences were also reported between word association and free
recall (El-Dakhs & Altarriba, 2019) and between the Stroop task and skin conductance response data
(Eilola & Havelka, 2011) in this line of research. Thus, in selecting a task, particularly when a null
efect is observed, or when no diference is observed between L1 and L2 speakers, one needs to ask
whether the task is sensitive enough for studying a phenomenon.
A related point is that RT data are most informative when obtained in a task that can be com-
pleted with ease and high accuracy. Deciding if a stimulus is a word or not (lexical decision), naming
a word or picture (naming), deciding if a stimulus refers to something man-made or natural (seman-
tic classifcation), deciding whether the input contains a particular sound or word (phoneme/word
monitoring), repeating a target word as soon as one hears it (shadowing), deciding if a stimulus is one
of those shown earlier (episodic recognition), and pressing a button after reading a word (self-paced
reading) are all good examples of such tasks. An easy and straightforward task allows the experimental
manipulation to produce the strongest possible efect on individuals’ performance, without contami-
nation from the complexity of the task itself. An easy task also helps minimize the loss of data due to
a high error rate, given the convention of analyzing RT data of correct responses only (see Hughes
et al., 2014, for data analysis methods that consider latency and accuracy simultaneously). Some other
tasks are less appropriate for generating useful RT data. For example, in a self-paced reading task in
which comprehension questions are used, the amount of time individuals take to answer compre-
hension questions is usually not very informative, since it is difcult to determine what afects the
latency in responding to comprehension questions. Similar examples include the recall task in which
an individual is asked to recall as many words as possible from a list presented previously, and the word
association task in which an individual is asked to provide the frst word coming to mind upon see-
ing a stimulus. Multiple processes and often less controllable factors may afect an individual’s RT in
such tasks. Under these circumstances, accuracy rates (in recall) or response type (in word association)
rather than response latencies are more informative dependent variables.
Variable control
Variable control is an intrinsic part of research design in almost any experimental research, but it is
particularly important in psycholinguistic L2 research, because this research relies heavily on depen-
dent variables such as RT and on inferences. A participant’s performance in an experimental task
is seldom driven by a single variable, and the manipulation of interest seldom afects just a single
aspect of the stimuli, the participant, the task, or the procedure. For example, when high- and low-
frequency words serve as stimuli to compare the size of the frequency efect in L1 and L2, the two
sets of words may also difer in length, concreteness, age of acquisition, or familiarity. They may
also difer in between-language properties such as cognate status, number of L1 translations, and L1
translation frequency. All these variables may afect the participants’ performance. When we compare
a high-profciency and a low-profciency group in an L2 processing task, these participants may
also difer in L1 background, the age and context of L2 acquisition, language aptitude, and current
language environment, which may also afect their performance. A primary task in research design,
often a challenging one, is to identify potential confounding variables and minimize the efect of such
variables through vigorous variable control.
One may classify potential confounding variables into three general categories: stimulus-related
variables, participant-related variables, and task- or procedure-related variables. All of them should
be considered in designing a study. For example, in addition to the lexical variables mentioned in
the preceding paragraph, one may want to consider if all stimulus words are familiar to prospective
183
Nan Jiang
participants, particularly when one is replicating an L1 study with the materials originally developed
for native speakers. One may also want to consider if L2 words in diferent conditions are matched
for their translatability in the participants’ L1. Furthermore, L2 speakers are generally a less homoge-
neous population compared to L1 speakers. They difer in many aspects, but four participant-related
variables are particularly important: L2 profciency, L1 background, age of onset, and L2 learning
context (e.g., classroom vs. naturalistic learners). L2 profciency and age of onset have been found
to afect processing outcome in many studies (Coughlin & Tremblay, 2013; Guillelmon & Grosjean,
2001; Ojima et al., 2005). The same is true of L1 background. L2 speakers with alphabetic and
logographic L1 backgrounds may difer in the extent to which they rely on orthographic and pho-
nological information in visual word recognition (e.g., Jiang & Pae, 2020; Wang et al., 2003), and the
morphological richness of L1 may afect the degree of decomposition in word recognition (Vainio
et al., 2014). In studying the issue of whether a bilingual’s two lexical systems can be selectively acti-
vated, it is important to consider bilinguals whose L1 and L2 either share or do not share the same
script (e.g., Moon & Jiang, 2012). In the area of procedure, numerous variables have to be consid-
ered, such as whether the target is displayed for a predetermined duration or remains on the screen
until a response is provided in an LDT, whether a 50 ms blank screen is inserted between a prime and
a target in a masked priming study, and whether a distractor word is presented simultaneously, prior
to, or following a target picture when the picture-word interference paradigm is adopted. Where the
same participants are tested in both languages or two tasks, it is usually desirable to counterbalance
the language and task order across participants. Successful variable control is essential to making valid
inferences and limiting alternative interpretations of the results (see Chapter 2 of Jiang, 2012, for
more information about variable control).
Dealing with explicit knowledge

A unique aspect of studying L2 learners is that many L2 speakers have obtained explicit knowledge
about the L2 through classroom instruction or self-study. Their L2 use may be driven or afected by
explicit knowledge to some extent, which sets them apart from L1 speakers, who may rely less on
explicit knowledge. In psycholinguistic research, however, if a researcher is interested in how linguistic
knowledge is represented and changes as an integral part of our mental representation that can be auto-
matically activated, an efort should be made to minimize the involvement of explicit knowledge, even
though the development and application of explicit knowledge in L2 acquisition is also a worthwhile
topic. The distinction between explicit and integrated knowledge is sometimes overlooked in extend-
ing L1-based methodological approaches to L2 speaker populations, which becomes problematic,
particularly in studying the representation and development of morphosyntactic knowledge in L2.
For example, both the grammaticality judgment task (GJT) (as a receptive task) and retell or pic-
ture description (as productive tasks) have been used for studying syntactic knowledge representation
and development in L1. However, these methods are less efective in studying the development of
integrated knowledge in L2 speakers because L2 learners are able to use their explicit knowledge
while completing such tasks, even when the task involves relatively spontaneous L2 production or is
performed under time pressure, as pointed out by DeKeyser (2003) and Jiang (2004). For example,
when studying L2 speakers’ sensitivity to morphosyntactic violations with electrophysiological data,
the GJT is often used. The use of the GJT, however, may function to draw the participants’ attention
to a morphosyntactic violation in the stimuli, thus enhancing their sensitivity to such violations (e.g.,
in comparison to a reading-for-comprehension task). This in turn may lead to the application and
refection of explicit knowledge in the ERPs, for instance in the N600 component, which can refect
the efects of both explicit and implicit knowledge (Morgan-Short et al., 2015).
L2 researchers have been long engaged in a quest to identify methods that can help study L2 mor-
phosyntactic development with minimum explicit knowledge involvement, from sentence matching
184
Synthesis
(e.g., Bley-Vroman & Masterson, 1989; Gass, 2001) to shadowing (Guillelmon & Grosjean, 2001).
Reading for comprehension under controlled conditions is now a widely used task. This task has three
versions depending on the type of data it generates: self-paced word-by-word reading with button-
pressing RT data (frst used in Jiang, 2004), eye-tracking with eye fxation data (frst used in Keating,
2009), and electrophysiological research with ERP data (Ojima et al., 2005). Other promising meth-
ods include word monitoring, as used in Granena (2013) and Suzuki and DeKeyser (2015) and the
visual world paradigm (Godfroid, 2020). These tasks are more successful than the GJT, for example, in
directing the participants’ attention to meaning or target words rather than to grammaticality.
Assessing L2 profciency
A long-standing challenge in psycholinguistic L2 research lies in the assessment of L2 profciency. L2
profciency has to be assessed in some studies because it is the focus of the research, for instance when
the goal is to determine whether the asymmetry in translation priming may be attributed to L2 pro-
fciency (e.g., Nakayama et al., 2016). In other studies, L2 profciency has to be assessed because it is
a potential confounding variable. Adequate L2 profciency assessment is also essential for comparing
and evaluating research fndings across studies.
A dilemma facing psycholinguistic L2 researchers is that processing data are usually collected from
a large number of participants, often one person at a time, and in relatively short test sessions, for
example, less than 30 minutes. It is not often practical to administer a comprehensive L2 profciency
test that can last for hours. As a result, researchers have explored other options. One frequently
adopted method is the use of a questionnaire. These questionnaires can be short surveys made
by individual researchers or much more detailed published versions, such as the Language History
Questionnaire by Li et al. (2014) or the Language Experience and Profciency Questionnaire (LEAP-
Q) by Marian et al. (2007) (see also Dunn & Tree, 2009 and Lim et al., 2008). These questionnaires
are used to obtain information about the participants’ language learning history (e.g., age of onset,
years of L2 learning, years of L2 immersion), their current use of languages (e.g., percentage of L1
and L2 use in diferent settings), and their self-reported profciency in listening, reading, speaking,
and writing in L1 and L2. The second method of assessment is the standardized L2 profciency tests,
particularly when existing scores are available from such tests. These tests often include TOEFL,
TOEIC, IELTS, the Michigan Test of English Language Profciency, and the Oxford Placement Test
(Allen, 1992). Other studies use a researcher-made cloze test to assess general L2 profciency. It is
also common to use a vocabulary test to assess L2 profciency, particularly when the topic is related
to lexical processing. Among the frequently used vocabulary tests are the Peabody Picture Vocabulary
Test-IV (Dunn & Dunn, 2007) or its adaptation in another language, LexTALE (Lemhöfer & Bro-
ersma, 2012) with its English, French, German, and Spanish versions, and the Productive Vocabulary
Levels Test (Laufer & Nation, 1999). In some studies, participants are classifed into diferent prof-
ciency groups based on their years at college (e.g., undergraduates versus graduates), years in a lan-
guage program (e.g., frst versus third year), or the language environment they live in (L1 versus L2).
It is clear that we face two problems. The frst is a lack of a profciency test that is validated to be
a good indication of overall L2 profciency and relatively easy and quick to administer. The second
is the inconsistency in assessment methods across studies. A coordinated efort of the whole feld is
needed to deal with these problems.
Conclusion
From a methodological perspective, researchers strive to obtain a set of data that allow one to make
reasonable and valid inferences about mental representations and processes involved in L2 use and
acquisition. This chapter outlines six overarching issues that have to be considered in order to achieve
185
Nan Jiang
this goal. New developments in data collection methods may raise new issues. For example, more
researchers are using Internet-based remote testing for data collection (see Witzel et al., 2013 for
an introduction to webDMDX, one of several programs that allow remote data collection, and Gor
et al., 2019 for an actual L2 processing study using the program), and the use of virtual reality for data
collection is likely to spread to L2 research soon (see Peeters, 2019 for an introduction and Heyseelar
et al., 2021 for an actual psycholinguistic study using this method). While these methods allow for
the considerable expansion of the participant pools, among their many advantages, they also raise
issues such as prescreening prospective participants. Whatever method is used, understanding the
nature of the task and rigorous variable control are at the center of a good research design.
References
Allen, D. (1992). The Oxford placement test. Oxford University Press.
Bernolet. S. (2023). Studying second language production. In A. Godfroid & H. Hopp (Eds.), The Routledge
Bley-Vroman, R., & Masterson, D. (1989). Reaction time as a supplement to grammaticality judgements in the
investigation of second language learners’ competence. University of Hawai’i Working Papers in ESL, 8, 207–237.
Coughlin, C. E., & Tremblay, A. (2013). Profciency and working memory based explanations for nonnative
speakers’ sensitivity to agreement in sentence processing. Applied Psycholinguistics, 34(3), 615–646.
DeKeyser, R. M. (2003). Implicit and explicit learning. In C. Doughty & M. H. Long (Eds.), The handbook of
second language acquisition (pp. 313–348). Blackwell.
Deshors, S. & Gries, St. Th. (2023). Using corpora in research on second language psycholinguistics. In
Dunn, L., & Dunn, D. (2007). Peabody picture vocabulary test (4th ed., PPVT-4). Pearson.
Dunn, L., & Tree, J. (2009). A quick, gradient bilingual dominance scale. Bilingualism: Language and Cognition,
12, 273–289.
Eilola, T. M., & Havelka, J. (2011). Behavioral and physiological responses to the emotional and taboo Stroop
tasks in native and non-native speakers of English. International Journal of Bilingualism, 15, 353–369.
El-Dakhs, D. A. S., & Altarriba, J. (2019). How do emotion word type and valence infuence language process-
ing? The case of Arabic—English bilinguals. Journal of Psycholinguistic Research, 48, 1063–1085.
Foss, D. J. (1998). Two strands of scholarship on language comprehension: Phoneme monitoring and discourse
context. Journal of Psycholinguistic Research, 27, 191–201.
Gass, S. M. (2001). Sentence matching: A re-examination. Second Language Research, 17, 421–441.
guide. Routledge.
Gor, K., Chrabaszcz, A., & Cook, S. (2019). A case for agreement: Processing of case infection by early and late
learners. Linguistic Approaches to Bilingualism, 9, 6–41.
Grainger, J., & Frenck-Mestre, C. (1998). Masked priming by translation equivalents in profcient bilinguals.
Language and Cognitive Processes, 13, 601–623.
Granena, G. (2013). Individual diferences in sequence learning ability and second language acquisition in early
childhood and adulthood. Language Learning, 63, 665–703.
Guillelmon, D., & Grosjean, F. (2001). The gender marking efect in spoken word recognition: The case of
bilinguals. Memory and Cognition, 29, 503–511.
Gullberg, M. (2023). Studying multimodal language processing: the integration of speech and gestures. In
Routledge.
Heyseelar, E., Peeters, D., & Hagoort, P. (2021). Do we predict upcoming speech content in naturalistic envi-
ronments? Language, Cognition and Neuroscience, 36, 440–461.
Hughes, M. M., Linck, J. A., Bowles, A. R., Koeth, J. T., & Bunting, M. F. (2014). Alternatives to switch-cost
scoring in the task-switching paradigm: Their reliability and increased validity. Behavior Research Methods, 46,
702–721.
186
Synthesis
Jeong, H., & Jiang, N. (2019). Representation and processing of lexical bundles: Evidence from word monitor-
ing. System, 80, 188–198.
Jiang, N. (2004). Morphological insensitivity in second language processing. Applied Psycholinguistics, 25,
603–634.
Jiang, N. (2021). Examining L1 infuence in L2 word recognition: A case for case. Journal of Second Language
Studies, 4, 1–18.
Jiang, N., & Forster, K. I. (2001). Cross-language priming asymmetries in lexical decision and episodic recogni-
tion. Journal of Memory and Language, 44, 32–51.
Jiang, N., & Nekrasova, T. (2007). The processing of formulaic sequences in a second language. The Modern
Language Journal, 91, 433–445.
Jiang, N., Novokshanova, E., Masuda, K., & Wang, X. (2011). Morphological congruency and the acquisition
of L2 morphemes. Language Learning, 61, 940–967.
Jiang, N., & Pae, H. (2020). The role of phonology in L2 visual word recognition. In P. Booth & J. Clenton
(Eds.), First language infuences on multilingual lexicons. Routledge.
Keating, G. D. (2009). Sensitivity to violations of gender agreement in native and nonnative Spanish: An eye-
movement investigation. Language Learning, 59, 503–535.
Kim, J., & Davis, C. (2003). Task efects in masked cross-script translation and phonological priming. Journal of
Laufer, B., & Nation, P. (1999). A vocabulary-size test of controlled productive ability. Language Testing, 16,
33–51.
Lemhöfer, K., & Broersma, M. (2012). Introducing LexTALE: A quick and valid lexical test for advanced learn-
ers of English. Behavior Research Methods, 44, 325–343.
Li, P., Zhang, F., Tsai, E., & Puls, B. (2014). Language history questionnaire (LHQ 2.0): A new dynamic web-
based research tool. Bilingualism: Language and Cognition, 17, 673–680.
Lim, V. P., Liow, S. J. R., Lincoln, M., Chan, Y. H., & Onslow, M. (2008). Determining language dominance
in English-Mandarin bilinguals: Development of a self-report classifcation tool for clinical use. Applied Psy-
cholinguistics, 29, 389–412.
Marian, V., Blumenfeld, H. K., & Kaushanskaya, M. (2007). The language experience and profciency question-
naire (LEAP-Q): Assessing language profles in bilinguals and multilinguals. Journal of Speech, Language, and
Hearing Research, 50, 940–967.
Moon, J., & Jiang, N. (2012). Non-selective lexical access in diferent-script bilinguals. Bilingualism: Language
and Cognition, 15, 173–180.
research to issues in explicit and implicit second language acquisition. In P. Rebuschat (Ed.), Implicit and
explicit learning of languages (pp. 349–384). Benjamins.
Nakayama, M., Ida, K., & Lupker, S. J. (2016). Cross-script L2-L1 noncognate translation priming in lexical
decision depends on L2 profciency: Evidence from Japanese—English bilinguals. Bilingualism: Language and
Cognition, 19, 1001–1022.
Ojima, S., Nakata, H., & Kakigi, R. (2005). An ERP study of second language learning after childhood: Efects
of profciency. Journal of Cognitive Neuroscience, 17, 1212–1228.
Peeters, D. (2019). Virtual reality: A game-changing method for the language sciences. Psychonomic Bulletin and
Review, 26, 894–900.
Suzuki, Y., & DeKeyser, R. (2015). Comparing elicited imitation and word monitoring as measures of implicit
knowledge. Language Learning, 65, 860–895.
Tremblay, A., Derwing, B., Libben, G., & Westbury, C. (2011). Processing advantages of lexical bundles: Evi-
dence from self-paced reading and sentence recall tasks. Language Learning, 61, 569–613.
Underwood, G., Schmitt, N., & Galpin, A. (2004). The eyes have it: An eye-movement study into the pro-
cessing of formulaic sequences. In N. Schmitt (Ed.), Formulaic sequences: Acquisition, processing, and use (pp.
153–172). Benjamins.
Vainio, S., Pajunen, A., & Hyönä, J. (2014). L1 and L2 word recognition in fnnish. Studies in Second Language
Acquisition, 36, 133–162.
Wang, M., Koda, K., & Perfetti, C. A. (2003). Alphabetic and nonalphabetic L1 efects in English word identi-
fcation: A comparison of Korean and Chinese English L2 learners. Cognition, 87, 129–149.
Winskel, H. (2013). The emotional Stroop task and emotionality rating of negative and neutral words in late
Thai—English bilinguals. International Journal of Psychology, 48, 1090–1098.
Witzel, J., Cornelius, S., Witzel, N., Forster, K. I., & Forster, J. C. (2013). Testing the viability of webDMDX
for masked priming experiments. The Mental Lexicon, 8, 421–449.
187
SECTION III
Theoretical perspectives on second

language psycholinguistics
16
WORD AND MULTI-WORD
PROCESSING
Memory-based and linguistic approaches
Kira Gor
Introduction
Recent years have seen a surge in interest in the properties of the nonnative mental lexicon and
the storage and retrieval of nonnative words (see Textbox 16.1). This chapter will survey diferent
research areas in word recognition and juxtapose two approaches to word recognition and encoding
in bilinguals, focusing on the difculties second langauge (L2) learners experience with the encoding
of words in L2. We will evaluate two theoretical claims about L2 lexical processing—memory-based
and linguistic representations-based—against the existing empirical evidence. The memory-based
episodic L2 hypothesis postulates that L2 speakers rely on episodic rather than long-term semantic/
lexical memory in lexical processing of both novel and familiar L2 words (Jiang & Forster, 2001).
This pervasive L2 reliance on episodic memory stands in contrast with L1 consolidation of recently
formed episodic lexical representations into lexical/semantic representations for long-term storage.
Conversely, the fuzzy lexical representations hypothesis centers on the ambiguity in the phonological
and semantic encoding, or fuzziness of lexical representations for less familiar L2 words at the level
of form, meaning, and form-meaning mappings (Bordag et al., 2022a; Gor Cook, 2020; Gor et al.,
The mental lexicon is the mental storage of words where their linguistic properties are encoded. Words
are organized in the mental lexicon in networks based on form, meaning, and morphological struc-
ture (see Jarema & Libben, 2007).
Phonological encoding is a process of categorization of sound sequences as phonemes of a particular lan-
guage based on the phonetic cues found in the segment as well as the phonetic context.
Phonolexical representations are (sublexical) phonemic sequences representing the sound forms of words
in the mental lexicon.
Lexical representations encode forms, meanings, form-meaning mappings and linguistic properties of
words, such as their morphological structure.
Phonological neighbors are words that difer from each other in one phoneme that is substituted, added,
or deleted.
DOI: 10.4324/9781003018872-19 191

Kira Gor
2021). Unfaithful encoding of the lexical representations may lead to inefcient word recognition
resulting in slowness, uncertainty, and errors in the retrieval of incorrect lexical items from the mental
lexicon. As with the other aspects of language processing, L2 speakers show a developmental trajec-
tory in lexical processing with more robust lexical encoding in high-profciency speakers. I will sug-
gest a possible synthesis of these two approaches and spell out its implications for L2 acquisition. In
addition, the chapter will survey the processing of morphologically complex words and the features
of multi-word processing in L2.

L2 speakers know fewer words than native speakers (e.g., Pellicer-Sánchez, 2019), and they are
often uncertain about the exact form or meaning of words, which leads to lexical confusion. While
the reasons for smaller L2 mental lexicons are rather transparent—lower exposure to L2 input (see
Conklin & Thul, 2023 [this volume])—the exact nature of L2-specifc word encoding, storage,
and retrieval is still insufciently understood. The two main approaches to L2 lexical processing are
memory-based (Jiang & Forster, 2001; Qiao & Forster, 2017; Witzel & Forster, 2012) or based on
diferences in linguistic representations (Cook & Gor, 2015; Gor & Cook, 2020).
The memory-based episodic L2 hypothesis

The memory-based approach is exemplifed by two strands. According to the episodic L2 hypoth-
esis (Jiang & Forster, 2001; Witzel & Forster, 2012), all L2 words are represented in episodic
memory, and not in lexical (i.e., long-term) memory, while most L1 words are represented in
lexical memory, with only a few that have recently appeared in the input represented in episodic
memory. Episodic memory stores richly detailed representations of personal events with the infor-
mation about where and when the event took place, or the source and the context of the episodic
representation (Francis et al., 2019). In turn, once a rich representation of the word is linguisti-
cally encoded in long-term memory, many nonlinguistic aspects of the episodic representation,
such as the voice of the speaker or the font in which the word was printed are lost unless they are
emotionally or otherwise important for the individual. According to the complementary learning
systems hypothesis (see Lindsay & Gaskell, 2010), there are two interdependent learning systems
in the brain converting episodic memory representations to long-term memory representations:
the fast-learning system for episodic memory representations located in the hippocampus and the
slow-learning system storing long-term (lexical) memory representations located in the neocortex.
The hippocampus consolidates episodic representations of words and converts them into lexical
representations to be integrated into lexical networks in long-term lexical memory. These two
types of representations—episodic ones that are rich, “raw,” and short-lived, and lexical/semantic
ones that are compactly linguistically encoded to be stored long-term—may not be as categorically
diferent in L2 as they generally are in L1.
In its strong version, the episodic L2 hypothesis postulates that L1 and L2 words are stored in
diferent memory systems, and therefore, they cannot activate each other cross-linguistically (Jiang
& Forster, 2001). The evidence regarding the potential episodic nature of L2 lexical representations
mainly comes from masked translation priming studies (Jiang & Forster, 2001; Witzel & Forster,
2012; yet see Lee et al., 2018). In the L2 prime–L1 target condition in a lexical decision task with
a prime in L2 English and the target—its recently studied translation equivalent—in L1 Chinese,
Chinese speakers of English did not show any priming efects. In contrast, they showed a priming
efect for the words they had studied before in an episodic recognition task, in which participants
frst study a list of words (in their case, Chinese L1 words), and they respond to each word whether
it is new or old (trained). One could argue that it is difcult to conceive of all L2 words as stored in
192
Word and multi-word processing
episodic memory, which is not intended for long-term lexical storage. Nor is it plausible that L1 and
L2 words stored in diferent memory systems cannot activate each other, since in some tasks, such as
semantic categorization, they can (see Jiang & Forster, 2001).
Yet, more evidence in support of L2 speakers’ reliance on recently activated episodic representa-
tions of L2 words comes from studies using recognition memory tasks (Francis & Strobach, 2013). In
these tasks, participants study word lists in their L1 or L2 and are later tested on whether they rec-
ognize the studied words. Several observations point to a stronger reliance of bilinguals on episodic
memory, especially, when processing L2 words. They are more accurate at recognizing studied words
than monolinguals, and more accurate in their less profcient L2 than in their L1.
Additionally, L2 speakers are less prone to developing false memories for semantic lures added to
the test list (Bialystok et al., 2020) in the L2 compared to the L1. For example, when the studied list
includes “thread,” “pin,” “eye,” and “sewing,” bilinguals more often incorrectly “remember” the criti-
cal lure “needle,” belonging to the same semantic feld but absent from the studied list, in their L1 than
in their L2. Taken together, these observations indicate that bilinguals are quite efcient at retrieving
recently formed episodic memories, especially in L2. Two possible factors that may contribute to this
strong reliance on episodic representations are (1) greater episodic distinctiveness of L2 words due to
their subjective lower frequency, i.e., a stronger novelty efect, especially, for low-frequency words
(Francis & Strobach, 2013), and (2) weaker associative links in L2 semantic memory—L1 words are
more integrated with the words from the same semantic feld and are more easily confounded.
Another memory-based account of the L1-L2 diferences in lexical storage and retrieval makes
use of a phenomenon called the prime lexicality efect (see Qiao & Forster, 2017). In a masked
orthographic form priming experiment, a nonword prime that difers in one letter from a real target
word (e.g., prime: banara, target: BANANA) induces facilitation. Conversely, a real-word prime
will induce either no priming or inhibition (as in the prime-target pair contract-CONTRAST). This
efect was exploited in novel word learning experiments to demonstrate that newly learned L1 words
showed a prime lexicality efect, i.e., they behaved like real words, while L2 words did not (Qiao &
Forster, 2017). Such evidence prompted the authors to hypothesize that L2 words are stored in a
diferent memory system compared to L1 words.
The fuzzy representations hypothesis

The absence of the prime lexicality efect in L2 word learning also fts with the idea of the represen-
tations-based approach claiming that L2 learners rely on the same memory systems as monolinguals,
yet they have fuzzy lexical representations of less familiar L2 words that do not get engaged in lexical
competition and do not show the inhibition associated with it. According to this approach (Gor &
Cook, 2020; Gor et al., 2021; see also Bordag et al., Bordag et al., 2022a, 2022b.), L2 speakers deal
with multiple encoding problems in L2, starting with phonological categorization difculties. Also,
the exact form and meaning of novel L2 words may not be retained because they occur in the input
insufciently often. As a result, lexical representations of novel and less familiar L2 words encoun-
tered in the past are fuzzy. According to this approach, it is the quality of linguistic encoding of words
that sets less familiar L2 words apart from L1 words. Note, however, that the mechanisms underlying
the encoding and retrieval of fuzzy lexical representations in L2 are not unique to nonnative process-
ing; rather, it is the pervasiveness of fuzzy representations in L2 that is unique.
In the following section, I review several strands of research that provide support for L2 fuzzy
lexical representations and argue that the fuzziness in diferent aspects of lexical encoding, from pho-
nology to morphological structure, is what makes lexical processing in lower-profciency L2 speakers
more efortful and less efcient than in L1 speakers. Specifcally, I explore the extent to which an
approach in terms of fuzzy lexical representations can capture the data that have been used to argue
for the episodic memory approach.
193
Kira Gor
Perceptual problems with diffcult L2 phonological contrasts

lead to problems with phonolexical encoding
Research on spoken word recognition is based on the understanding that words are stored in the
mental lexicon, and the process of word recognition includes several stages: activation of lexical
entries that are more or less compatible with the auditory input, their competition for selection,
and the selection of the best-matching candidate (Marslen-Wilson, 1987). Nonnative listeners have
to deal with an additional issue in L2 word recognition: They may experience lexical confusion
when processing words that difer in a confusable phoneme in L2. We will refer to the phonologi-
cal encoding of words as phonolexical encoding, in keeping with the idea that words are encoded
in the mental lexicon as sequences of phonemes (see, e.g., Hickok, 2001). For instance, L1 Japanese
learners of English experience difculties with the diferentiation of the /r/-/l/ contrast (Yamada,
1995). Their low sensitivity to the contrast leads to problems with phonolexical encoding, and as
a result, they confuse words “light” and “right.” Such confusion may lead to spurious activation of
irrelevant words in the mental lexicon, thereby increasing lexical competition in the L2. An increase
in lexical competition, in turn, slows down the retrieval of the target lexical entry and interferes
with speech processing. These efects are measurable in phonological priming (see Bernolet, this
volume). When a prime inhibits the selection of a target, this indicates that the listener is dealing
with increased competition, while facilitation indicates that the competition was light (see Gor,
2018). For instance, increases in lexical competition were explored in Dutch speakers of English in
phonological priming experiments with minimal word pairs (e.g., “fesh” and “fash”) or words with
overlapping onsets (e.g., “dafodil” and “defnite”), which led to inhibition in English monolinguals,
indicating increased lexical competition (Broersma & Cutler, 2008; Broersma, 2012). Conversely,
lexical competition was reduced in L2, thereby lending support to the idea that spurious activation
of irrelevant competitors difering from the target word by a problematic L2 phoneme plays a limited
role in L2 word recognition. Rather, weak competition points to fuzzy phonolexical representations
of L2 words.
Priming experiments test lexical competition in an indirect way, namely by assuming that the prime
is still activated when the target is presented, and phonological neighbors are activated and compete
for selection. In contrast, eye tracking using the visual world paradigm targets lexical competition in a
more direct way by exploring how lexical competition is afected when L2 listeners have low sensitiv-
ity to a contrast that diferentiates the initial segments of the target and competitor word. In one study,
L1 Dutch speakers of L2 English and L1 English speakers listened to pairs of words such as “panda”
and “pencil” while looking at corresponding pictures in the visual display (Weber & Cutler, 2004). For
native speakers of English, the words start to diverge at the frst vowel—/æ/ and /ɛ/; yet, for native
speakers of Dutch, these two vowels may sound similar given that Dutch does not have the phoneme
/æ/. Because the L2 listeners kept looking to both the panda and the pencil when hearing “panda”
for longer than L1 listeners, the study demonstrated that lexical competition in the L2 is resolved later
compared to the L1 because of a sound segment presenting problems for phonemic categorization.
One of the frst studies to address the consequences of phonological confusion for word rec-
ognition looked at pairs of Catalan words and nonwords, which difered in the phonemes that are
contrasted in Catalan, but not Spanish: /e/-/ɛ/, /o/-/ɔ/, and /s/-/z/ (Pallier et al., 2001). The par-
ticipants were Catalan-Spanish adult bilinguals with either Catalan or Spanish as their frst language.
While L1 Catalan speakers did not respond more quickly when they heard mes /mɛs/ “month” after
hearing més /mes/ “more,” L1 Spanish speakers did (although see Barrios et al., 2016).
Other studies also provide evidence in support of unfaithful or fuzzy lexical representations of
nonnative words with perceptually confusable L2 phonemes (Darcy et al., 2012; Darcy & Thomas,
194
2019). For instance, Darcy and colleagues (2012) observed an asymmetry when comparing the prim-
ing efects in repetition priming with the pattern of error rates in a test of phonetic discrimination for
French high /y/-/u/ as in puce /pys/ (“fea”) and pouce /pus/ (“thumb”) and mid /œ/-/ɔ/ as in seul
/sœl/ (“alone”) and sol /sɔl/ (“foor”) front and back rounded vowel contrasts in intermediate and
advanced learners of French with L1 American English. Intermediate learners were more accurate in
the phonetic discrimination of /y/-/u/ than in the priming experiment involving words diferentiated
by this contrast, while advanced learners showed higher sensitivity to both contrasts in the priming
than in the phonetic categorization test. This asymmetry indicates that successful phonolexical encod-
ing is not purely a result of phonetic categorization. In addition, lexical frequency—a proxy of word
familiarity—may contribute to the efciency of phonological encoding, with more frequent words
more accurately encoded (see also Conklin & Thul, 2023 [this volume]). The source of problems with
phonolexical encoding in the L2 may be located in the L1 (Darcy & Thomas, 2019).
To summarize, there is evidence in support of fuzzy phonolexical representations for L2 words
with difcult phonological contrasts. Phonological confusion may be the source of lexical confusion
in L2 when the words are diferentiated by a difcult contrast that resists proper encoding.
Problems with L2 phonolexical encoding transcend diffcult

phonological contrasts
Learning the phonological form of new words in the L2 is a challenge for the learner beyond deal-
ing with particular problematic L2 contrasts, since the phonological categorization of L2 segments
based on L2-specifc phonetic cues is difcult across the board. As a result, words that sound some-
what similar to each other and do not occur in the input frequently may have “approximative” fuzzy
phonolexical representations that are easily confusable. The idea that less familiar L2 words have
fuzzy representations of form and form-meaning mappings was prompted by the observation that
L2 speakers of Russian showed a diferent pattern of phonological priming efects for low-frequency
Russian primes and targets compared to L1 speakers (Gor & Cook, 2020). While both L1 and L2
speakers responded more slowly to high-frequency targets that overlapped with the primes in three
initial phonemes, as in vrak–VRACH (enemy–doctor, /vrak/–/vraʃ/), which is a sign of dealing with
lexical competition, L2 learners were faster to access the targets in the low-frequency prime-target
pairs kabyla–KABAK (mare–pub, /kabɨla/–/kabak/). Facilitation in phonological priming points
to sublexical processing and signals that lexical competition is weak (see Gor, 2018). Another study
directly evaluated L2 prime and target familiarity in a separate test of lexical knowledge and observed
facilitation for low-familiarity (“recognizable”) L2 primes and targets in a phonological priming
experiment involving L2 Russian words and L1 speakers of English (Cook & Gor, 2015).
The hypothesis that fuzzy phonolexical representations can lead to fuzzy form-meaning map-
pings was tested in a pseudo-semantic priming experiment with L2 Russian primes and targets
in two critical conditions: semantically related, as in cow–MILK (/karova/–/malako/), and with
the prime semantically unrelated to the target, but related to a phonologically similar word, as in
cow–HAMMER (/karova/–/malatok/), with “hammer” being phonologically similar to “milk” in
Russian (Cook et al., 2016). As expected, the prime “cow” activated the semantic representation of
“milk,” but it did not elicit any priming efects for “hammer” in L1 speakers. Conversely, when L2
speakers who expect “milk” after hearing “cow” hear “hammer” as a target instead, they should tem-
porarily confuse “hammer” with the expected “milk.” This prediction was confrmed: L2 speakers
showed inhibition when they were induced to retrieve the wrong lexical entry.
The studies reviewed above make a claim for fuzzy phonolexical representations of less familiar
L2 words and fuzzy form-to-meaning mappings in L2 words based on auditory experiments. One
can argue that in these experiments, phonological perception rather than the stored representations
could cause the observed difculties with L2 word recognition. A line of research that does not rely
195
Kira Gor
on auditory input can tease these possibilities apart (Ota et al., 2009). The experiments test semantic
associations between orthographically represented words and ask L2 speakers of English to rate the
semantic relatedness of English words, such as “key” and “lock.” When the participants’ L1 was Japa-
nese, they were likely to respond that “key” and “rock” are also semantically related. This can only
happen when problems with the /r/-/l/ contrast percolate to the lexical encoding of these words in
the mental lexicon.
Overall, the reviewed studies suggest that the phonological form of L2 words is not robustly
encoded for less familiar L2 words. This is especially true for the words with problematic phono-
logical contrasts. Moreover, confusion may occur for L2 words that difer in more than just one
phoneme, as an analysis of the role of the Levenshtein distance, a measure of phonological distance
between words, has demonstrated in L2 lexical confusion (see Cook et al., 2016). At the same time,
this form-centered approach that is concerned with the encoding of the word forms and form-
meaning mappings treats word meanings as fxed, and does not inquire about how those meanings
came to be established in the L2. However, the development of semantic representations of L2 words
is a complex process, and it has recently attracted researchers’ attention. The next section will focus
on the semantic representations of L2 words.
Fuzzy semantic representations of newly learned and less familiar L2 words

For incidental vocabulary learning, for instance while reading or watching television, the number of
exposures is the strongest predictor of success in learning meanings of new words (Godfroid et al.,
2018). Several incidental and intentional vocabulary learning experiments with advanced L2 learners
of German investigated how L2 learners acquire the semantic properties of novel L2 words (Bordag
et al., 2015; Bordag et al., 2017; Bordag et al., 2018). The fndings for novel L2 words documented
a type of engagement with the L2 semantic network that was diferent from the one established for
familiar well-known L2 words. Facilitation is usually observed in semantic priming experiments
when the prime and the target are semantically related, as in the pair nurse–DOCTOR; yet, newly
incidentally learned words used as primes produced inhibition for semantically related familiar tar-
gets (Bordag et al., 2015). In a follow-up study that used intentional learning of novel words from
defnitions, whether semantic priming led to facilitation or inhibition depended on how well the
participants acquired their meaning (Bordag et al., 2017). Well-acquired words showed facilitation,
while inhibition was observed for the words for which only the form was acquired. The third study
manipulated the type of meaning assigned to the novel words: existent as opposed to fctitious and
non-existent in L1 (Bordag et al., 2018). As predicted, only novel word primes with non-existent
meanings produced semantic inhibition—apparently, novel meanings are not easily encoded into the
existing semantic space. Bordag and colleagues attributed inhibition in priming to the weak repre-
sentations of novel words. However, it is not obvious how weak primes can inhibit stronger familiar
targets. It is possible that newly acquired words serving as primes have fuzzy semantic representations,
which generates uncertainty about the exact boundaries of their meaning and their semantic associa-
tions with the targets and other competitors.
Summary
What emerges from the review is that L2 lexical processing is characterized by several linguistic and
memory-related properties. Linguistic encoding, including form, meaning, and form-meaning map-
pings, is less efcient in L2. As we have discussed, L2 speakers deal with multiple encoding problems
in L2, beginning with phonological categorization difculties, and also with lower retention of
the exact form and meaning of novel L2 words, both of which lead to fuzzier lexical representa-
tions of novel and less familiar L2 words. As a consequence, L2 speakers may tend to rely more on
196
recently formed episodic representations as a compensatory mechanism—since the linguistic encod-

ing of L2 words is not robust, they make use of “raw” acoustic-phonetic representations of form and
“patchy” representations of meaning. Such an overreliance on episodic information by bilinguals
cannot be entirely strategic, since it is observed both under intentional and incidental encoding con-
ditions (Francis et al., 2019). Note that an L2 episodic memory advantage is systematically observed
in episodic memory tasks with recently encountered words, whether familiar or novel. As men-
tioned already, recently encountered familiar words in a less dominant L2 are more salient, leading
to enhanced episodic memories. Even when L2 speakers encode words in long-term memory, these
long-term lexical representations may not always be properly encoded, or be fuzzy, and therefore
also show the properties of episodic representations—for example, they are not as actively engaged
in lexical competition as L1 lexical representations are. The fact that the representations of less famil-
iar L2 words are not properly consolidated (or fuzzy) makes them behave as if they were episodic
representations under certain experimental conditions. Future research will establish whether the
fuzzy representations approach can capture all data that have been used to argue for the L2 episodic
memory approach or whether both are needed as complementary approaches.
Current research in the linguistic processing of words and multi-word units
Processing of morphologically complex words

Another key issue in L2 lexical processing is how L2 learners process morphologically complex
words and whether they engage in morphological decomposition in the L2. Research has studied
polymorphemic words involving infection, as in “jump-ed,” derivation, as in “sing-er,” and com-
pounding, as in “honey-moon.”
According to one point of view, L2 speakers do not decompose morphologically complex words
and store and access them as whole words. This view is expressed in the Shallow Structure Hypoth-
esis (Clahsen & Felser, 2006, 2018) in its application to morphological processing in the L2 (Clahsen
et al., 2010). Evidence in support of the absence of morphological decomposition in the L2 comes
primarily from masked priming experiments with infected and derived word forms that prime their
base forms in English and German native speakers (e.g. Clahsen et al., 2010). L2 speakers showed
no priming for regularly infected English verbs and German sufxed participles, while they showed
partial facilitation for English derived words such as “neat-ness” or “valid-ity.” Similar efects were
reported for typologically distant languages, such as Turkish (Kirkici & Clahsen, 2013) and Hebrew
(Farhy et al., 2018). However, other studies reported L2 facilitation for infected words under masked
priming conditions (Coughlin & Tremblay, 2015; Foote, 2017; Foote et al., 2020; Voga et al., 2014).
Furthermore, L2 speakers show more robust masked priming efects for derivation (Ciaccio & Clah-
sen, 2020; Diependaele et al.,, 2011; Jacob et al., 2018) and compounding (Li et al., 2017) than for
infection. One of the possible reasons for infected word forms to show less robust masked priming
in the L2 than derived words is that familiar derived words are represented in the L2 mental lexicon,
while not all infected word forms of low-frequency words with rich infectional paradigms are, and
accordingly, infected word forms require greater processing costs to be morphologically analyzed
and recognized.
At the same time, several masked morphological priming studies have reported that L2 participants
showed facilitation in an orthographic control condition with morphologically and semantically
unrelated primes and targets sharing the same number of graphemes as the ones in the morphologi-
cal condition (Diependaele et al., 2011; Feldman et al., 2010; Heyer & Clahsen, 2015). Thus, both
“scanner” and “scandal” produced facilitation for “scan” in the L2 (Heyer & Clahsen, 2015). This
fnding may potentially undermine the role of morphological facilitation in the L2 and point to
form-based priming instead. Indeed, it is conceivable that orthographic form facilitation, which is
197
Kira Gor
akin to facilitation in phonological priming discussed earlier, is interpreted as morphological facilita-

tion. However, frstly, orthographic facilitation is not always observed in orthographic controls for
morphologically related primes and targets (see Jacob et al., 2018); and, secondly, it still does not
explain the diferences between the L2 priming patterns for infection vs. derivation or for regular
vs. irregular infection. Therefore, while the role of form priming in the L2 should be recognized, it
does not cancel the morphological priming efects reported in numerous studies.
One reason for the absence of morphological priming in the L2 could be due to the experimen-
tal task of masked priming—a short exposure to an infected masked prime may be insufcient for
lower-profciency L2 speakers to morphologically analyze it given that they are generally slower in
processing L2. Indeed, a number of studies using overt priming, auditory, visual, or cross-modal, have
shown robust morphological priming efects in L2 speakers (e.g., for L2 English: Basnight-Brown
et al., 2007; for L2 Russian: Gor & Jackson, 2013; for L2 Arabic: Freynik et al., 2017). Based on
these fndings, one could hypothesize that L2 speakers rely on native-like mechanisms in processing
morphologically complex words, but the masked priming conditions put them at a disadvantage.
Yet, recent research suggests that, while L2 speakers may decompose infected words rather than
storing them as whole words, they do not fully process the morphosyntactic information carried
by the infections, such as case marking in nouns. For instance, a study of L2 processing of Russian
case-infected nouns using an auditory lexical decision task demonstrated how native-like process-
ing costs for nouns infected in the genitive case (“of the dog/s”) compared to the nominative case,
or the citation form (“dog”) emerged only in high-profciency L2 speakers (Gor et al., 2017). In
another study, both early and late L2 learners of Russian were at chance level in rejecting nonwords
with real stems and real infections combined in an illegal way (e.g., *beautifuls; Gor et al., 2018).
Since L2 learners accepted such nonwords as real words, they must have successfully decomposed
the infected nonwords and accessed their existing stems; however, they failed to check the whole
word for the compatibility of the stem and infection. The fnding that they ignored the incongruent
infection may explain why L2 decomposition into stem and infection does not guarantee efcient
morphosyntactic processing of infected words in L2. The study also documented a developmental
trajectory in that the rate of errors in accepting such nonwords decreased at higher profciency levels.
Overall, research on morphological processing in the L2 indicates that while L2 speakers are
sensitive to morphological structure of L2 words, they are less efcient than L1 speakers at checking
the information carried by the grammatical morphemes in real time—a skill that they can develop
at higher profciency levels.
Multiword units processing

Formulaic sequences have two key properties; namely, they occur more often than freely co-occur-
ring word sequences and they possess certain semantic cohesiveness. Native speakers access formulaic
sequences faster than free sequences composed of words with similar frequency, an efect termed
as idiom superiority (e.g., Carrol & Conklin, 2014, see also Conklin & Thul, 2023 [this volume]).
To account for this fnding, the dual route model of representation of idioms assumes that they are
represented and processed both as sequences of constituents and as noncompositional units of mean-
ing (Wray, 2002). For L2 learners, research addresses the issue of whether both routes, the sequential
compositional, and direct access of the idiom meaning, are available in the L2, and the general con-
sensus is that L2 speakers rely predominantly on the sequential compositional route and compute the
literal meaning frst (Carrol & Conklin, 2017; Siyanova-Chanturia et al., 2011).
Idioms are often characterized by a noncompositional, i.e., semantically opaque meaning, for
example, “to beat around the bush” when referring to an evasive way of expressing oneself. This sec-
tion will focus on the linguistic aspects of L2 processing of idiomatic phrases, which are mainly defned
by diferent degrees of noncompositional meaning. Idiomaticity in formulaic expressions presents
198
a learning challenge for L2 speakers. Several studies, such as eye-tracking during reading, explored
how L2 speakers handle nonnative idiomatic expressions that are formulaic, i.e., whose meaning is
nondecomposable (Carrol & Conklin, 2014, 2017, Conklin & Schmitt, 2008; Siyanova-Chanturia
et al., 2011). L1 and L2 speakers of English read sentences with formulaic sequences in which the
context supported their idiomatic interpretation (e.g., “take the bull by the horns” meaning “attack a
problem”) or their literal interpretation (“wrestle an animal”) faster than control phrases (Conklin &
Schmitt, 2008). Conversely, an eye-tracking study found that L2 speakers did not show a processing
advantage for idioms compared to novel phrases in a biasing story context, in which the fgurative
meaning was processed more slowly than the literal meaning (Siyanova-Chanturia et al., 2011).
Idiomatic phrases usually belong to a particular language and their fgurative meaning gets lost
in translation into another language. To explore whether L1 idiomatic phases could be transferred
to L2, priming and eye-tracking reading studies tested L1 Chinese learners of L2 English (Carrol &
Conklin, 2014, 2017). In the eye-tracking study, L1 and L2 participants read two-sentence Eng-
lish context sentences containing English idioms, literal translations of Chinese idioms, and control
phrases with the last word replaced by another semantically congruent word (e.g., “spill the beans/
chips” in English or “draw a snake and add feet/hair,” which means “to ruin something by adding
unnecessary detail” in Chinese). L1 Chinese participants were able to recognize the Chinese idiom
in its literal English translation, yet they processed the fgurative meaning more slowly than the literal
meaning, while English L1 participants processed fgurative meanings of the English idioms faster
than their literal meanings. According to the authors, these fndings suggest that, most likely, Chi-
nese speakers recognize L1 idioms in their English translations by processing them sequentially rather
than directly recognizing their meanings. Importantly, this pattern of results ultimately suggests that
conceptual idiomatic meaning is language-specifc. In summary, while L2 speakers show sensitivity
to idiomatic sequences, especially at high profciency levels, they appear to rely more on sequential
combinatorial processing of L2 idioms rather than direct recognition and access of their idiomatic
meaning (Carrol & Conklin, 2017).

This chapter has reviewed the studies exploring diferent aspects of L2 lexical representations and points
to a general property of L2 lexical encoding: fuzziness in the representations of word form, meaning,
and form-meaning mappings for novel or low-familiar words. Further research will determine whether
idiomatic multiword expressions have fuzzy representations in the L2 mental lexicon, which is likely
given that L2 speakers often favor the literal meaning of idioms. L2 speakers’ uncertainty about the
congruence of stems and infections indicates that fuzziness may also involve combinatorial aspects of
lexical storage and access in representing the morphological structure of L2 words and processing the
grammatical information carried by infections. Furthermore, both phonological and visual masked
priming indicate that L2 speakers are more (and possibly, longer) engaged in the sublexical level of word
form processing than L1 speakers are. Fuzzy lexical representations are associated with unfaithful pho-
nological/orthographic, semantic, and grammatical encoding of lexical items in the L2 mental lexicon.
They contribute to slower and error-prone lexical access and morphosyntactic processing in the L2.
These observations lead to further questions about the complex nature of L2 lexical encoding.
Firstly, it remains to be seen to what extent fuzziness characterizes lexical representations of low-
frequency L1 words, as suggested by the Lexical Quality Hypothesis (Perfetti, 2007). Secondly, fuzzy
lexical representations in the L2 may contribute to problems with building syntactic structures, and
more research is needed to understand the L2 processing issues arising at the interface of the lexicon
and sentence processing, as identifed by the Lexical Bottleneck Hypothesis (Hopp, 2018). Finally, L2
speakers are exposed to continuous speech rather than isolated words, and, accordingly, research into
the role that sentence context plays in shaping L2 lexical representations is welcome.
199
Kira Gor
What are the diferences between L1 and L2 lexical representations? Is fuzziness in forms, meanings, and
form-meaning mappings present for less familiar words both in L1 and L2?
What is the role of L2 lexical representations in sentence processing?
How does sentence context help to disambiguate fuzzy L2 lexical representations?
What are the contributions of nonnative phonology and new semantic concepts, as they interact with
diferent memory types, to the learning of new words and their integration into lexical networks?
How do L2 speakers process infected words beyond morphological decomposition and stem recognition:
Do they access the grammatical information carried by infected words?
What are the triggers for the activation of idiomatic meaning and form in the production of idiomatic
multiword sequences in L2?
Further reading
Gor, K., Cook, S., Bordag, D., Chrabaszcz, A., & Opitz, A. (2021). Fuzzy lexical representations in adult second
language speakers. Frontiers in Psychology, 12, 732030.
The article lays out the fuzzy lexical representations hypothesis and reviews empirical evidence in its support.
Bordag, D., Gor, K., & Opitz, A. (2022a). Ontogenesis Model of the L2 lexical representation. Bilingualism:
The Ontogenesis Model captures the developmental trajectory of L2 lexical representations, from a fuzzy
lexical representation to the representational optimum.
second language: An eye-movement study. Studies in Second Language Acquisition, 40(2), 341–366.
The study addresses L2 difculties with online retrieval of meaning in adult L2 vocabulary learning from
reading.
Chrabaszcz, A., & Gor, K. (2017). Quantifying contextual efects in second language processing of phonolexi-
cally ambiguous and unambiguous words. Applied Psycholinguistics, 38(4), 909–942.
The article investigates the role of sentence context-induced expectations in the recognition of words with
fuzzy phonolexical representations.
References
Barrios, S., Jiang, N., & Idsardi, W. J. (2016). Similarity in L2 phonology: Evidence from L1 Spanish late-learners’
perception and lexical representation of English vowel contrasts. Second Language Research, 32(3), 367–395.
Basnight-Brown, D. M., Chen, L., Hua, S., Kostic, A., & Feldman, L. B. (2007). Monolingual and bilingual
recognition of regular and irregular English verbs: Sensitivity to form similarity varies with frst language
experience. Journal of Memory and Language, 57, 65–80.
Bialystok, E., Dey, A., Sullivan, M. D., & Sommers, M. S. (2020). Using the DRM paradigm to assess language
processing in monolinguals and bilinguals. Memory & Cognition, 1–14.
Bordag, D., Gor, K., & Opitz, A. (2022a). Ontogenesis Model of the L2 lexical representation. Bilingualism:
Bordag, D., Gor, K., & Opitz, A. (2022b). Refning key concepts of the Ontogenesis Model of the L2 lexical
representation. Bilingualism: Language and Cognition, 25(2), 236–241.
Bordag, D., Kirschenbaum, A., Rogahn, M., Opitz, A., & Tschirner, E. (2017). Semantic representation of
newly learned L2 words and their integration in the L2 lexicon. Studies in Second Language Acquisition, 39(1),
197–212.
200
Bordag, D., Kirschenbaum, A., Tschirner, E., & Opitz, A. (2015). Incidental acquisition of new words during
reading in L2: Inference of meaning and its integration in the L2 mental lexicon. Bilingualism: Language and
Cognition, 18(3), 372–390.
Bordag, D., Opitz, A., Rogahn, M., & Tschirner, E. (2018). The strength of meaning. The Mental Lexicon,
13(1), 74–104.
Broersma, M. (2012). Increased lexical activation and reduced competition in second-language listening. Lan-
guage and Cognitive Processes, 27, 1205–1224.
Broersma, M., & Cutler, A. (2008). Phantom word activation in L2. System, 36(1), 22–34.
Carrol, G., & Conklin, K. (2014). Getting your wires crossed: Evidence for fast processing of L1 idioms in an
L2. Bilingualism: Language and Cognition, 17(4), 784–797.
Carrol, G., & Conklin, K. (2017). Cross language lexical priming extends to formulaic units: Evidence from
eye-tracking suggests that this idea “has legs”. Bilingualism: Language and Cognition, 20(2), 299–317.
Chrabaszcz, A., & Gor, K. (2017). Quantifying contextual efects in second language processing of phonolexi-
cally ambiguous and unambiguous words. Applied Psycholinguistics, 38(4), 909–942.
Ciaccio, L. A., & Clahsen, H. (2020). Variability and consistency in frst and second language processing: A
masked morphological priming study on prefxation and sufxation. Language Learning, 70(1), 103–136.
3–42.
Clahsen, H., & Felser, C. (2018). Some notes on the shallow structure hypothesis. Studies in Second Language
Acquisition, 40(3), 693–706.
Clahsen, H., Felser, C., Neubauer, K., Sato, M., & Silva, R. (2010). Morphological structure in native and non-
native language processing. Language Learning, 60(1), 21–43.
Conklin, K., & Schmitt, N. (2008). Formulaic sequences: Are they processed more quickly than nonformulaic
language by native and nonnative speakers? Applied Linguistics, 29(1), 72–89.
Hopp (Eds.). The Routledge handbook of second language acquisition and psycholinguistics (pp. 203–215). Routledge.
Cook, S. V., & Gor, K. (2015). Lexical access in L2: Representational defcit or processing constraint? The Mental
Lexicon, 10(2), 247–270.
Cook, S. V., Pandža, N. B., Lancaster, A. K., & Gor, K. (2016). Fuzzy nonnative phonolexical representations
lead to fuzzy form-to-meaning mappings. Frontiers in Psychology, 7, 1345.
Coughlin, C. E., & Tremblay, A. (2015). Morphological decomposition in native and non-native French speak-
ers. Bilingualism: Language and Cognition, 18(3), 524–542.
Darcy, I., Dekydtspotter, L., Sprouse, R. A., Glover, J., Kaden, C., McGuire, M., & Scott, J. H. (2012). Direct
mapping of acoustics to phonology: On the lexical encoding of front rounded vowels in L1 English—L2
French acquisition. Second Language Research, 28(1), 5–40.
Darcy, I., & Thomas, T. (2019). When blue is a disyllabic word: Perceptual epenthesis in the mental lexicon of
second language learners. Bilingualism: Language and Cognition, 22(5), 1141–1159.
Diependaele, K., Duñabeitia, J. A., Morris, J., & Keuleers, E. (2011). Fast morphological efects in frst and
second language word recognition. Journal of Memory and Language, 64(4), 344–358.
second language: An eye-movement study. Studies in Second Language Acquisition, 40(2), 341–366.
Farhy, Y., Veríssimo, J., & Clahsen, H. (2018). Do late bilinguals access pure morphology during word recogni-
tion? A masked-priming study on Hebrew as a second language. Bilingualism: Language and Cognition, 21,
945–951.
Feldman, L. B., Kostić, A., Basnight-Brown, D. M., Filipović Đurđević, D., & Pastizzo, M. J. (2010). Mor-
phological facilitation for regular and irregular verb formations in native and little evidence for two distinct
mechanisms. Bilingualism: Language and Cognition, 13, 119–135.
Foote, R. (2017). The storage and processing of morphologically complex words in L2 Spanish. Studies in Second
Language Acquisition, 39(4), 735–767.
Foote, R., Qasem, M., & Trentman, E. (2020). Morphological decomposition in L2 Arabic: A masked priming
study. Journal of Psycholinguistic Research, 49(2), 291–317.
Francis, W. S., & Strobach, E. N. (2013). The bilingual L2 advantage in recognition memory. Psychonomic Bul-
letin & Review, 20(6), 1296–1303.
Francis, W. S., Strobach, E. N., Penalver, R. M., Martínez, M., Gurrola, B. V., & Soltero, A. (2019). Word—
context associations in episodic memory are learned at the conceptual level: Word frequency, bilingual pro-
fciency, and bilingual status efects on source memory. Journal of Experimental Psychology: Learning, Memory,
and Cognition, 45(10), 1852–1871.
201
Kira Gor
Freynik, S., Gor, K., & O’Rourke, P. (2017). L2 processing of Arabic derivational morphology. The Mental
Lexicon, 12(1), 21–50.
Godfroid, A., Ahn, J., Choi, I., Ballard, L., Cui, Y., Johnston, S., Lee, S., Sarkar, A., & Hyung-Jo Yoon, H.-J.
(2018). Incidental vocabulary learning in a natural reading context: An eye-tracking study. Bilingualism: Lan-
guage and Cognition, 21(3), 563.
Gor, K. (2018). Phonological priming and the role of phonology in nonnative word recognition. Bilingualism:
Language and Cognition, 21(3), 437.
Gor, K., Chrabaszcz, A., & Cook, S. (2017). Processing of native and nonnative infected words: Beyond afx
stripping. Journal of Memory and Language, 93, 315–332.
Gor, K., Chrabaszcz, A., & Cook, S. (2018). Early and late learners decompose infected nouns, but can they tell
which ones are infected correctly? Journal of Second Language Studies, 1(1), 106–140.
Gor, K., & Cook, S. V. (2020). A mare in a pub? Nonnative facilitation in phonological priming. Second Language
Research, 36(1), 123–140.
Gor, K., Cook, S., Bordag, D., Chrabaszcz, A., & Opitz, A. (2021). Fuzzy lexical representations in adult second
language speakers. Frontiers in Psychology, 12, 732030.
Gor, K., & Jackson, S. (2013). Morphological decomposition and lexical access in a native and second language:
A nesting doll efect. Language and Cognitive Processes, 28(7), 1065–1091.
Heyer, V., & Clahsen, H. (2015). Late bilinguals see a scan in scanner AND in scandal: Dissecting formal overlap
from morphological priming in the processing of derived nouns. Bilingualism: Language and Cognition, 18,
543–550.
Hickok, G. (2001). Functional anatomy of speech perception and speech production: Psycholinguistic implica-
tions. Journal of Psycholinguistic Research, 30(3), 225–235.
Jacob, G., Heyer, V., & Veríssimo, J. (2018). Aiming at the same target: A masked priming study directly com-
paring derivation and infection in the second language. International Journal of Bilingualism, 22(6), 619–637.
Jarema, G., & Libben, G. (2007). The mental lexicon: Core perspectives. Emerald Group Publishing Limited.
Jiang, N., & Forster, K. I. (2001). Cross-language priming asymmetries in lexical decision and episodic recogni-
tion. Journal of Memory and Language, 44(1), 32–51.
Kirkici, B., & Clahsen, H. (2013). Infection and derivation in native and non-native language processing:
Masked priming experiments on Turkish. Bilingualism: Language and Cognition, 16, 776–794.
Lee, Y., Jang, E., & Choi, W. (2018). L2-L1 translation priming efects in a lexical decision task: Evidence from
low profcient Korean-English bilinguals. Frontiers in Psychology, 9, 267.
Li, M., Jiang, N., & Gor, K. (2017). L1 and L2 processing of compound words: Evidence from masked priming
experiments in English. Bilingualism: Language and Cognition, 20(3), 384–402.
Lindsay, S., & Gaskell, M. G. (2010). A complementary systems account of word learning in L1 and L2. Language
Learning, 60, 45–63.
Marslen-Wilson, W. D. (1987). Functional parallelism in spoken word-recognition. Cognition, 25(1–2), 71–102.
Ota, M., Hartsuiker, R. J., & Haywood, S. L. (2009). The key to the rock: Near-homophony in nonnative visual
word recognition. Cognition, 111, 263–269.
Pallier, C., Colomé, A., & Sebastián-Gallés, N. (2001). The infuence of native-language phonology on lexical
access: Exemplar-based versus abstract lexical entries. Psychological Science, 12(6), 445–449.
Pellicer-Sánchez, A. (2019). Examining second language vocabulary growth: Replications of Schmitt (1998) and
Webb & Chang (2012). Language Teaching, 52(4), 512–523.
Perfetti, C. (2007). Reading ability: Lexical quality to comprehension. Scientifc Studies of Reading, 11(4), 357–383.
Qiao, X., & Forster, K. I. (2017). Is the L2 lexicon diferent from the L1 lexicon? Evidence from novel word
lexicalization. Cognition, 158, 147–152.
Siyanova-Chanturia, A., Conklin, K., & Schmitt, N. (2011). Adding more fuel to the fre: An eye-tracking study
of idiom processing by native and nonnative speakers. Second Language Research, 27, 251–272.
Voga, M., Anastassiadis-Symeonidis, A., & Giraudo, H. (2014). Does morphology play a role in L2 processing?:
Two masked priming experiments with Greek speakers of ESL. Lingvisticae Investigationes, 37(2), 338–352.
Weber, A., & Cutler, A. (2004). Lexical competition in non-native spoken-word recognition. Journal of Memory
and Language, 50, 1–25.
Witzel, N. O., & Forster, K. I. (2012). How L2 words are stored: The episodic L2 hypothesis. Journal of Experi-
mental Psychology: Learning, Memory, and Cognition, 38(6), 1608–1621.
Wray, A. (2002). Formulaic language in computer-supported communication: Theory meets reality. Language
Awareness, 11(2), 114–131.
Yamada, R. (1995). Age and acquisition of second language speech sounds: Perception of American English /ɹ/
and /l/ by native speakers of Japanese. In W. Strange (Ed.), Speech perception and linguistic experience: Issues in
cross-language research (pp. 305–320). York Press.
202
17
WORD AND MULTIWORD
PROCESSING
Cognitive approaches
Introduction
As language speakers, we have a “store” of words and multiword sequences that we produce and rec-
ognize. This store can be thought of as our mental dictionary, which we refer to as the lexicon and
the items within it as lexical entries (for a discussion of approaches to word recognition and encoding
in bilinguals see Gor, 2023 [this volume]). Researchers who study the lexicon are interested in what
information lexical entries contain, how the entries are organized, and how comprehenders access
entries in their lexicon. It is the last question that is the focus of the current chapter, in which we will
look at arguably the most signifcant determinant of word and multiword recognition and processing
in a frst language (L1) and a second language (L2): word frequency.
Recognizing single words and multiword sequences involves a complex set of processes that
are infuenced by a wide range of variables. Frequency, or how often comprehenders encounter a
word or sequence, plays a central role in the entrenchment and processing of lexical entries. Words
and sequences that are encountered frequently are recognized and responded to more quickly than
those that are encountered less often, while those that are rarely encountered are responded to
more slowly yet (e.g., “cat” vs. “rat” vs. “gnat”). Frequency efects have been found in tasks that
present words on a computer screen and measure decision/response times (RTs). For example, in a
word naming task, participants say frequent words aloud more quickly than infrequent ones, and in
a lexical decision task (deciding whether a string of letters is a word or not), frequent words elicit
faster “yes” responses than less frequent ones (e.g., Forster & Chambers, 1973; Hudson & Bergman,
1985). In eye-tracking, there are fewer and shorter eye fxations on high-frequency words (e.g.,
Rayner & Raney, 1996). Word frequency generally explains 30% to 40% of the variance in L1 word
recognition tasks (Brysbaert et al., 2016) and more than any other variable (e.g., Balota et al., 2004;
Whaley, 1978). Word frequency efects have also been demonstrated in the L2 (e.g., Diependaele et
al., 2013; Whitford et al., 2016). Similarly to single words, there is faster recognition and processing
for more frequent multiword sequences in the L1 (e.g., Siyanova-Chanturia et al., 2011). Evidence
of a frequency advantage for multiword sequences in the L2 is somewhat mixed (for discussion see
Conklin, 2019).
On top of frequency, a range of other factors are known to infuence single-word processing,
albeit to a lesser extent: word length; age of acquisition (AoA); and neighborhood size (both number
and frequency), as well as the regularity of neighbors’ grapheme-to-phoneme correspondence (for
an overview see Harley, Chapter 6, 2014). The impact of such factors has been well documented in
DOI: 10.4324/9781003018872-20 203

the L1, and some work has been done exploring these factors the L2 (for a discussion see Dijkstra &
van Heuven, 2018). Another important factor in word processing is a word’s predictability (McDon-
ald & Shillcock, 2003): More predictable words have decreased processing times, and, in reading
tasks, highly predictable words are often skipped altogether (Ehrlich & Rayner, 1981; McDonald &
Shillcock, 2003). Predictability has important implications for the processing of multiword items. For
example, the word abject predicts a very limited set of upcoming words (e.g., poverty, failure, misery,
horror, terror, apology), with some of the combinations being multiword items (e.g., the collocation
abject poverty). Crucially, the predictability of these upcoming words will (largely) be predicated on
language exposure—in other words the frequency of their co-occurrence. For a word to become
predictable, it will need to have occurred in the same and/or a very similar context a sufcient num-
ber of times.
In addition, a considerable amount of research has looked at how cross-language overlap infu-
ences single- and multi-word recognition and processing in bilinguals, even when the social and
linguistic context calls for the use of only one language (for reviews see Conklin & Carrol, 2018;
Dijkstra & van Heuven, 2018). This has primarily been investigated using words that share lexical
properties across languages: cognates overlapping in semantics and phonology, as well as orthography
for languages that share a script, as in tableEnglish, tableFrench, and テーブル Japanese (‘teburu’);1 interlin-
gual homographs overlapping in orthography and somewhat in phonology, but not semantics, as
in painEnglish–“hurt,” painFrench–“bread”; and interlingual homophones overlapping in phonology but
Lexicon is the mental “store” of words and other linguistic expressions (e.g., idioms) that is thought to
have information about their form (phonological and orthographic), meaning, and morphological
and syntactic properties.
Lexical entries are the items that are stored in long-term memory with information about the sound and
meaning of words and frequent multiword strings (e.g., idioms “kick the bucket,” phrasal verbs “put of,”
etc.). Entries also include information about the morphological and syntactic properties of the items.
Multiword sequences are sequences of words that are (1) recurrent in language, occurring more
frequently than comparable novel phrases; and (2) (generally) processed more quickly than
matched novel phrases. Multiword sequences encompass a broad range of phenomena that fullfl
a number of communicative functions, such as binomials bride and groom; collocations abject pov-
erty; idioms kick the bucket; lexical bundles you know what; and phrasal verbs pick up. The terms
multiword units/sequences, formulaic language, and formulaic units/sequences are often used
interchangeably.
Interactive activation (IA) model is a computational model for word recognition. Each word has an acti-
vation level and is recognized when its activation level passes the activation threshold. A word’s activa-
tion level is infuenced by the activation levels of other words including neighbors.
Neighbors, orthographic and phonological, are words that difer from one another by only one letter
or sound respectively. For example, the word “hint” has the neighbors “hilt,” “hind,” “hunt,” “mint,”
“tint,” “fint,” “lint,” “pint,” etc.
Resting level activation (RLA) is the activation level of a word in the absence of any input and interaction.
Activation threshold is the critical value that the activation level of a word has to exceed for it to be
recognized.
204
Word and multiword processing
not orthography or semantics, as in /kle/ for clayEnglish–“stif, sticky fne-grained earth that can be
molded,” clefFrench–“key.” Multiword items can also overlap across languages in various ways, as dem-
onstrated by the following English and French examples: meaning and form (e.g., throw money out the
window/jeter l’argent par les fenêtres, literally in both languages “to throw money out of the windows”
and fguratively “to waste money”); and meaning but not form (e.g., feel blue/avoir le cafard, literally
in French “to have the cockroach” and fguratively in both languages “to feel sad”).
As a whole, research on cognates and interlingual homographs and homophones demonstrates
that word recognition is language-independent, meaning that all lexical representations that are
similar to, that is have form overlap with, the input are activated (for review, see Conklin & Car-
rol, 2018; Dijkstra & van Heuven, 2018; van Hell & Tanner, 2012; and for a meta-analytic review
of cognates Lauro & Schwartz, 2017). Activation is mediated by the amount of phonological,
orthographic, and semantic overlap between the words in the two languages (e.g., for discus-
sion see Dijkstra & van Heuven, 2018; and for examples in diferent script languages, see Allen
& Conklin, 2013, 2017). It can further be infuenced by the biasing strength of the sentence
context (Whitford et al., 2016). Investigations of cross-language overlap in multiword items have
primarily focused on idioms and collocations. Overall, research points to a cross-language efect,
such that L1-L2 overlap benefts multiword processing (Conklin & Carrol, 2018; yet see Beck &
Weber, 2016).
Frequency and single-word recognition

Of all of the variables that infuence word recognition, word frequency is by far the most powerful
predictor of RTs (Whaley, 1978). Indeed, it has been said that frequency of occurrence should be
given “its rightful place as a fundamental shaper of a lexical system always dynamically responsive to
experience” (Monsell, 1991, p. 150). The impact of word frequency on RTs is generally called a “fre-
quency efect.” However, it is important to note that what word frequency refers to varies. In some
studies, word frequencies are those extracted from a corpus (with a possible distinction between word
form and lemma, see Deshors & Gries, 2023 [this volume]), while, in other investigations, word
frequency refers to the number of occurrences of that particular word in an experiment. Despite
these diferences, frequency efects likely arise because each encounter with an item strengthens the
memory traces that make up its lexical representation (Murray & Forster, 2004), which, in turn,
raises its resting level activation (RLA).
It is important to note that there is a limit to how fast the processing system can get; it reaches
what we call an asymptote. This means that once a word has received a certain amount of expo-
sure, recognition cannot get any faster even with additional exposure to that item. Consequently,
additional exposure afects high- and low-frequency words diferently. We can understand this
better in terms of the words in Figure 17.1A: “gnat” that is low frequency, “rat” that is higher
frequency, and “cat” that is even higher frequency. We can see that, at occurrence x, “gnat” has
a lower RLA than “rat” and “cat,” and in turn “rat” has a lower RLA than “cat.” As is illustrated
in the fgure, words with lower RLA are farther away from the activation threshold. Note that an
additional occurrence (x+1) raises the RLA of “gnat” much more than that of “rat” or “cat,” and
that of “rat” more than that of “cat.” Further, we see that eight exposures yield much less change
for the RLA of “cat” than they do for “gnat.” Thus, at each occurrence “gnat” has a much big-
ger shift toward the activation threshold/asymptote, although the shift is less with each additional
occurrence. In contrast, “cat” moves less with each additional occurrence, since it already starts
near the threshold/asymptote.
205
Figure 17.1 Panel A. Efect of additional occurrences on low frequency “gnat,” higher frequency “rat,” and
even higher frequency “cat,” demonstrating that exposure to a lexical item raises its RLA in proportion to its
distance from the activation threshold/asymptote. Panel B. Efect of additional occurrences on monolingual
high- and low-frequency words and bilingual L1 and L2, high- and low-frequency words, demonstrating that
increased exposure in bilinguals has a greater efect on RLA.
Note: Curves labeled mono-high and mono-low refer to high- and low-frequency words for monolinguals,
respectively, while L1-high and L2-high denote high-frequency words in a bilingual’s L1 and L2, respectively,
and L1-low and L2-low denote low-frequency words in the L1 and L2.
It is important to consider how frequency efects are impacted by speaking more than one language.
Learning, knowing, and using more than one language means that language use is by necessity
divided. In other words, second language speakers generally use both of their languages (L2 and L1)
less frequently than monolinguals use their one language. This gives rise to reduced lexical entrench-
ment, which has been accounted for in diferent ways. One view holds that the connections between
an item’s orthographic form, phonological form, and its meaning are strengthened as a function
of exposure. When there is less use, this results in weaker ties between an item’s form (phonology,
orthography) and concepts (e.g., weaker links hypothesis: Gollan et al., 2008). An alternative view
is that using more than one language leads to reduced RLA of L2 words and potentially L1 words as
well (e.g., bilingual interactive activation plus model, BIA+: Dijkstra & van Heuven, 2002).
On the latter account, knowing an L2 reduces lexical entrenchment (i.e., the strength of word
and multiword representations in memory)—possibly even in the L1—which in turn leads to more
of an efect of frequency (i.e., a greater increase in RLA with additional exposures). This can be
understood in terms of Figure 17.1B. We see that, in monolinguals, high-frequency words start very
close to the asymptote. In unbalanced bilinguals, while L1 high-frequency words also start close to
the asymptote, they are not as close as the high-frequency monolingual ones. Therefore, more expo-
sure should have more of an impact on their RLA. L2 high-frequency words are further still from
206
the asymptote, as are L1 and L2 low-frequency words in turn. In unbalanced bilinguals, this means
that increased exposures should have a greater impact on particular words: L2 low-frequency words
> L1 low-frequency words > L2 high-frequency words > L1 high-frequency words. Further, more
exposures should raise RLA more in bilinguals than monolinguals and, thus, we would expect more
of an efect of increased frequency in bilinguals.
Indeed, research has demonstrated that there is a greater efect of frequency for bilinguals in
their L2 than L1. In a lexical decision task with unbalanced Dutch-English bilinguals, Duyck et al.
(2008) found a frequency efect that was almost twice as large in the L2 as in the L1. Using eye-
tracking during paragraph reading with unbalanced English-French and French-English bilinguals,
Whitford and Titone (2012) found larger frequency efects in gaze duration and total reading time
in the L2 than in the L1. Further, the efect was modulated by the current level of exposure, such
that the greater the L2 exposure prior to the study, the smaller the frequency efects. It is important
to note that Figure 17.1B predicts smaller frequency efects in monolinguals than for bilinguals
in their L1 (i.e., less change in RLA with increased exposures for monolinguals). However, this
has not always been found to be the case. An investigation by Cop et al. (2015) demonstrated
that frequency efects are similar for monolinguals and unbalanced bilinguals in their L1 and that
larger frequency efects in bilinguals are driven by a disproportionate lowering of the RLA of low-
frequency words.
A consequence of higher RLA is faster processing. More precisely, single- and multiword items
that are more frequent are processed more quickly because they are closer to the activation threshold.
This has been found for single words in the L1 and L2 across a range of tasks (for a discussion see
Brysbaert et al., 2018; Cop et al., 2015; Diependaele et al., 2013; Monaghan et al., 2017).
Frequency and multiword units

At the outset of this chapter, it was suggested that multiword sequences demonstrate frequency efects
akin to those of single-word items. This means each and every occurrence of a multiword item in the
L1 and L2 contributes to its entrenchment in memory, with more encounters leading to faster pro-
cessing. Research has demonstrated that, in the L1, children and adults are sensitive to the frequency
of multiword sequences, with more frequent multiword items being processed more quickly than less
frequent ones (for discussion, see Conklin, 2019; Siyanova-Chanturia & Sidtis, 2018).
When studying the processing of multiword sequences, the frequency of the sequences themselves
is generally not what is under investigation. Instead, the processing of multiword sequences is com-
pared to that of matched “novel” language. In other words, frequently co-occurring word sequences
are compared to word sequences that rarely or never co-occur (e.g., idiom spill the beans vs. matched
spill the chips). In the L1, a processing advantage is generally observed for multiword sequences relative
to matched controls, but this advantage is not always found in the L2, in particular for idioms (e.g.,
Conklin & Schmitt, 2008; Siyanova-Chanturia et al., 2011). Idioms may be problematic for learn-
ers (both L1 children and L2 children and adults) because computing the meaning of the individual
words does not generally yield the fgurative meaning (e.g., spill + the + beans ≠ “reveal a secret”).
If exposure strengthens the links between an item’s form and meaning, it may simply be that learn-
ers have not encountered idioms a sufcient number of times for a strong connection to develop
between the form (spill the beans) and meaning (“reveal a secret”). In fact, amongst the various types
of multiword items, idioms tend to be fairly low frequency.
Since the processing advantage for multiword sequences is not always observed in the L2, it has
been hypothesized that processing is more compositional (i.e., word by word) in the L2, which
makes the properties of the individual words and their meanings more salient than phrase-level
frequency and meaning (Cieślicka et al., 2014). Thus, L2 speakers encountering kick the bucket
would be more likely to arrive at the meaning “strike a pail with the foot” than “die.” In contrast,
207
a usage-based approach, in which the linguistic system is shaped by experience with language
(e.g., Bybee, 2013; Tomasello, 2003), would say that, if a processing advantage is not evident for
multiword sequences, it is simply that learners have not encountered the sequences a sufcient
number of times for them to become entrenched in memory and for a processing advantage to
emerge. A set of studies by Carrol and Conklin (2014, 2017) explored this question using lexical
decision and eye-tracking. English monolinguals and Chinese-L1–English-L2 speakers read Eng-
lish idioms and controls (such as spill the beans/chips) and translated Chinese idioms and controls
(like draw a snake and add feet/hair). The latter did not exist in English and were translated into
sequences that had a frequency of zero in English, but which were well known in Chinese. In
line with previous research, Carrol and Conklin found a processing advantage for English idi-
oms for monolinguals, but not for L2 speakers. Crucially, the Chinese speakers had a processing
advantage for Chinese idioms translated into English compared to their controls, demonstrating
that frequent multiword sequences that are entrenched in memory do show the typical processing
advantage in the L2.
Research on more compositional multiword sequences (i.e., sequences where the meaning is the
sum of its parts such as salt + and + pepper = “salt and pepper”) generally demonstrates a processing
advantage relative to matched controls (e.g., “pepper and salt”) as well as a sensitivity to frequency in
the L2, such that more frequent sequences are processed more quickly than less frequent ones (for an
overview see Conklin, 2019; Siyanova-Chanturia & Sidtis, 2018). Thus, much like with single word
items, every occurrence of a multiword item in both the L1 and L2 contributes to its entrenchment
memory, with more encounters leading to faster processing.

In order to better understand and explain word recognition, as well as the role of frequency in
it, computational models have been developed. Such models contain variables that are thought to
characterize the word recognition system as well as rules that describe how these variables interact
and evolve over time. These rules refect our beliefs about the dynamic processes that unfold during
word recognition. Simulations typically explore how model variables change over time as model
parameters are adjusted. Through the interplay of model parameters, variables, and proposed rules,
computational simulations can provide us with insights into the cognitive processes that underlie
word recognition in the L1 and L2 (Norris, 2013).
A main goal of these models is to quantitatively explain and/or predict measured RTs (i.e., the
time needed to make a response in a task and/or read a word in eye-tracking) and their distribu-
tions. It is important to note that we expect a distribution of RTs because of frequency. For example,
given a specifc word such as “cat,” we can measure RTs of that word across numerous participants.
Due to individual diferences—including diferent degrees of exposure to the word (i.e., the word
frequency)—RTs will difer among participants and hence have a particular distribution. Alterna-
tively, we could measure RTs for several words, such as “cat” vs. “rat” vs. “gnat,” from single partici-
pants. We would expect a distribution of RTs due to their diferent frequencies.
While computational models difer in their mathematical complexity and conceptual foun-
dations, they all share a common key principle: RTs are determined by the time that it takes a
word (or multiword sequence) to reach a predefned activation threshold. In one of the main
classes of models, each word has an activation level, and, at heart, these models describe how
a word’s activation level changes over time. The details of the mechanisms that are responsible
for changing activation levels vary greatly between diferent models. Nevertheless, all models
require a link between word frequency and a component of the model, such as its RLA. We will
illustrate this by focusing on interactive activation (IA) models pioneered by McClelland and
Rumelhart (McClelland & Rumelhart, 1981; Rumelhart & McClelland, 1982). In particular,
208
we will discuss the representation of word frequency in Multilink, a bilingual IA model (Dijkstra
et al., 2018).
In an interactive activation model (McClelland & Rumelhart, 1981), the RLA of a word is
thought to be determined by its frequency, while in distributed connectionist approaches (e.g.,
Plaut, 1997; Seidenberg & McClelland, 1989), the strength of the connections between process-
ing units (which are linked to, for example, orthography or phonology) is a function of word fre-
quency. Multilink (Dijkstra et al., 2018) is a recently developed IA model that accounts for bilingual
word recognition, production, and translation. It is based on the bilingual IA (BIA) model and its
successor, the BIA+ model (Dijkstra & van Heuven, 1998, Dijkstra & van Heuven, 2002). In IA
models, each word has its own activation level and a specifc word is recognized if its activation
level crosses a pre-defned threshold more quickly than any other word (or letter string in lexi-
cal decision tasks). The time that it takes to reach the threshold is the RT. Crucially, interactions
between words can substantially impact RTs. These interactions, which can be either excitatory or
inhibitory, occur at various levels including phonology, orthography, and semantics. In the absence
of any input or interaction, activation levels are assumed to decay to a baseline value, which cor-
responds to the RLA. Importantly, each word has its own RLA, and words typically difer in their
RLAs. In Multilink, the BIA+, and the BIA models, word frequency is linked to the RLA. This
notion was already put forward by McClelland and Rumelhart (1981) who remarked that high-
frequency words should have higher RLAs than low-frequency words. Thus, words with higher
RLAs (i.e., high-frequency words) have shorter RTs compared to those with lower RLAs (i.e.,
low-frequency words). While the relationship between RT and RLA appears intuitive, a hallmark
of computational models is their ability to make quantitative predictions and explain observed
data. As a consequence, we need to translate the verbal relationships between RT, RLA, and word
frequency into formal links.
When thinking about how to relate RLA to RT, we should start by keeping in mind that IA
models track the temporal evolution of activation levels, which means that they require an initial
value for a word’s activation level. At the beginning of an experimental trial, each word is at
rest (it has not received any input yet), meaning that a word’s activity level equals its RLA. If we
assume that the activation level for each word grows at a constant rate and is independent of the
activation level of any other word (i.e., neglecting the aforementioned interactions from phonol-
ogy, orthography, and semantics), RTs are proportional to the distance between the RLA and the
activation threshold. This is equivalent to saying that RTs and RLA are linearly related and estab-
lishes a formal relationship between RLA and RT. Moreover, the fnding that high-frequency
words are recognized more quickly than low-frequency words entails that, for a fxed activation
threshold, the RLA for high-frequency words is higher than that for low-frequency words. It is
important to note that this relationship between word frequency and RLA is a prediction of the
model, which is consistent with the intuition put forward by McClelland and Rumelhart (1981).
Figure 17.2 illustrates the connection between RLA and RT. Each solid straight line represents
the time evolution of the activation level of a word starting from its RLA (open upright triangle)
until it reaches the activation threshold (open upside-down triangle). In addition, we see the
distributions for RT and RLA, respectively, and how a certain RLA is mapped onto an RT value
(solid dots and dashed lines). Because the relationship between RLA and RT is linear, it entails
that the RLA distribution has the same shape as the RT distribution, only stretched, reversed,
and moved along the y-axis.
Figure 17.2 illustrates how a given RLA yields a corresponding RT. To explain frequency
efects in RTs, we also need to relate word frequency to RLA. So how would we do this? Consider
the solid black line in Figure 17.2. It maps a high RLA to a short RT. Since high-frequency words
have short RTs, we can also interpret the black line as relating a high RLA to a high-frequency
word. Analogously, the light grey line relates a low RLA to a long RT. Given that long RTs are
209
Figure 17.2 Relationship between RLA and RT distribution. Each solid straight line represents the time evolu-
tion of the activation level for a word starting from a specifc RLA (open upright triangle) until the activation
threshold is reached (open upside-down triangle).
associated with low frequency words, this connection can be understood as linking a low-fre-
quency word with a low RLA. The important point here is that because we know the quantitative
relation between RLA and RT, we can use it to establish a quantitative relationship between word
frequency and RLA.
In Figure 17.2, we assume that the relationship between RLA and RT is linear. Crucially, it can
be shown on conceptual grounds that RLA and RTs should be linearly related in Multilink. Hence,
we can use the argument presented earlier to quantitatively link word frequency and RLA in Multil-
ink. It is important to emphasize that not only do Multilink simulations result in a linear relationship
between RLA and RT, but also that this should be true a priori.

As we have seen, frequency plays a prominent role in single-word and multiword sequence pro-
cessing. This is in line with usage-based approaches to language acquisition and processing, in
which exposure (i.e., frequency) leads to entrenchment in memory, with increased exposure
yielding stronger entrenchment (e.g., Bybee, 2013; Tomasello, 2003). It is important to point
out that frequency measures from corpora are often viewed as an index of linguistic experience
(i.e., exposure). Thus, if word y (or multiword sequence y) is frequent in a large, representative
corpus, it is assumed that speakers have encountered it more than word z (or sequence z) that is
less frequent in the corpus. Intuitively we know that corpora do not provide a metric of all of a
210
speaker’s wide and varied experience with language. As Gernsbacher (1984) notes, corpora only
provide an approximation of actual exposure, which may be particularly problematic for low-
frequency words. A study by Gardner et al. (1987) demonstrated that engineering students were
faster at making a lexical decision to low-frequency engineering words while nurses were faster
for low-frequency medical words. Thus, corpus frequency does not necessarily align with actual
experience with a language, even in the L1. Crucially, to draw convincing conclusions about the
relationship between input and processing, we need to have a more accurate picture of the input,
in particular in the L2. As we will see, recent work has been attempting to do this, and future work
should build on these initial fndings.
One way to relate input to processing is to establish what input learners actually receive. While
this is extremely challenging, in some contexts it may be possible—for example when learners
get very little L2 exposure outside the classroom and textbooks. Across several studies, North-
brook and Conklin (2018, 2019, 2021) looked at just such a situation. Firstly, Northbrook and
Conklin (2018) created a corpus of Japanese Ministry of Education-mandated junior high school
English textbooks. They compared the most frequent textbook lexical bundles (a type of multi-
word sequence) to those in a native speaker reference corpus. They found that, although lexical
bundles were very frequent in the textbooks and were similar to those in a native-speaker corpus
at shorter lengths (three-word lexical bundles), they deviated considerably at longer ones (four,
fve, and six words). Northbrook and Conklin (2019) found that junior high school students had
a processing advantage for frequent three-word lexical bundles from their textbooks (do you play)
but not matched ones that were frequent in a native speaker corpus (do you hear) that did not
occur in their textbooks. Notably, the processing advantage showed a frequency efect, such that
more frequent multiword sequences in the textbook were processed more quickly than less fre-
quent ones. Thus, young EFL learners were not only sensitive to whether or not items occurred
in their texts, but also to the frequency of occurrence of a given item in them. Northbrook and
Conklin (2021) demonstrate that Japanese learners of English found the textbook multiword
sequences more natural than matched native-English sequences many years after junior high
school and their perception of naturalness varied with the frequency of the items in the texts.
Thus, initial input and its frequency appears to have long-lasting efects, which is a fnding that
needs further exploration.
Another way to explore the relationship between input and processing is to provide the input
in an experimental context (for some examples see, Godfroid et al., 2018; Pellicer-Sánchez,
2017). Such studies have presented L2 speakers with existing multiword items; however, even
with pre-tests to assess knowledge, we cannot be certain that the items had never been encoun-
tered before. Alternatively, researchers have used nonwords; however, attention may unduly be
drawn to these items, since they are unknown “words.” A recent study by Conklin and Carrol
(2021) addressed these concerns by monitoring monolingual English readers’ eye movements
while reading short stories containing existing binomials (a type of multiword sequence, like time
and money) and novel ones (wires and pipes). In line with the wider literature, they found a pro-
cessing advantage for existing multiword sequences. At the beginning of the story there was no
processing advantage for the novel multiword sequences. However, after encountering them four
or fve times, an advantage emerged. This happened during the course of reading a story, demon-
strating that the frequency of occurrence of multiword items allows them to become entrenched
in memory and to elicit a processing advantage. However, there are many open questions that
remain to be explored (see Textbox 17.2), such as how long-lasting/durable the processing advan-
tage is; if exposure is spread across multiple texts and/or days, how many occurrences are needed
for a processing advantage to emerge; and how the difculty of the text impacts the emergence
of a processing advantage.
211
Beyond frequency for single words in the L2: A wide range of factors are known to infuence L1 word
processing. Do we see efects in the L2 akin to those in the L1 for length, predictability, taboo and
emotion-laden words, etc.?
Frequency in multiword sequences: Are frequency efects for multiword items in the L2 similar to those
for single-word items in the L1, and is this the same across multiword sequence types (e.g., idioms,
collocations, binomials, lexical bundles, etc.)?
RLA beyond word frequency: Currently, RLA is most commonly computed from word frequencies
alone. How does the value of RLA change if other factors are taken into account such as neigh-
bors, AoA, etc.?
Interaction strength: In IA models, nodes representing orthographic, phonological, and semantic features
interact with each other. How do we determine the strengths of these connections in a principled manner?
Varied populations, languages, and contexts:2 Research on word and multiword processing has primarily
focused on adult L2 learners in university contexts. We need far more research on a greater range
of populations like young learners, and learners in classroom vs. immersion contexts. We need to
investigate a wider range of languages by expanding the focus beyond English and other European
languages. We need to understand processing in more natural contexts and in production instead of
simply in comprehension.
Notes
1 In linguistic terms, Japanese words like table/テーブル’ (“teburu”) are in fact “loanwords” because they have
been borrowed into the language. However, it is the overlap in form and meaning that underpins the process-
ing advantage for cognates and not their linguistic origins. Because loanwords share form and meaning, they
are treated as cognates in the processing literature.
2 There have been some recent studies on more varied populations; for example, López and Vaid (2018)
investigated informal translators or “language brokers.” There has also been some recent work on multi-
word sequence production (Siyanova-Chanturia & Janssen, 2018). However, far more of this type of work is
needed.
Further reading
Conklin, K. (2020). Processing single- and multi-word items. In S. Webb (Ed.), Routledge handbook of vocabulary
studies. Routledge.
This chapter discusses single- and multiword processing, with a focus on the infuence of word frequency
and linguistic overlap.
Conklin, K., & Carrol, G. (2018). First language infuence on the processing of formulaic language in a second
language. In A. Siyanova-Chanturia & A. Pellicer-Sánchez (Eds.), Understanding formulaic language: A second
language acquisition perspective. Routledge.
This chapter provides a detailed overview of the infuence of cross-language overlap (congruency) on the
processing of multiword items, in particular on idioms, collocations, and translated formulaic language.
Dijkstra, T., & van Heuven, W. J. B. (2018). Visual word recognition in multilinguals. In S.-A. Rueschemeyer &
M. G. Gaskell (Eds.), The Oxford handbook of psycholinguistics (2nd ed.). Oxford University Press.
This chapter discusses single-word visual processing in multilinguals. It has a strong focus on cognate and
interlingual homograph representation and processing.
212
References
Allen, D., & Conklin, K. (2013). Cross-linguistic similarity and task demands in Japanese-English bilingual
processing. PLoS ONE, 8(8), 1–14.
Allen, D., & Conklin, K. (2017). Naturalistic reading in the L2 and the impact of word frequency and cross-
linguistic similarity. Journal of the Ochanomizu University English Society, 7, 41–57.
Balota, D. A., Cortese, M. J., Sergent-Marshall, S. D., Spieler, D. H., & Yap, M. J. (2004). Visual word recogni-
tion of single-syllable words. Journal of Experimental Psychology: General, 133(2), 283–316.
Beck, S. D., & Weber, A. (2016). Bilingual and monolingual idiom processing is cut from the same cloth: The
role of the L1 in literal and fgurative meaning activation. Frontiers in Psychology, 7, 1350.
Brysbaert, M., Mandera, P., & Keuleers, E. (2018). The word frequency efect in word processing: An updated
review. Current Directions in Psychological Science, 27(1), 45–50.
Brysbaert, M., Stevens, M., Mandera, P., & Keuleers, E. (2016). The impact of word prevalence on lexical deci-
sion times: Evidence from the Dutch Lexicon Project 2. Journal of Experimental Psychology: Human Perception
and Performance, 42(3), 441–458.
Bybee, J. (2013). Usage-based theory and exemplar representation. In T. Hofman & G. Trousdale (Eds.), The
Oxford handbook of construction grammar (pp. 49–69). Oxford University Press.
Carrol, G., & Conklin, K. (2014). Getting your wires crossed: Evidence for fast processing of L1 idioms in an
L2. Bilingualism: Language and Cognition, 17(4), 784–797.
Carrol, G., & Conklin, K. (2017). Cross language priming extends to formulaic units: Evidence from eye-
tracking suggests that this idea “has legs”. Bilingualism: Language and Cognition, 20(2), 299–317.
Cieślicka, A. B., Heredia, R. R., & Olivares, M. (2014). It’s all in the eyes: How language dominance, salience,
and context afect eye movements during idiomatic language processing. In M. Pawlak & L. Aronin (Eds.),
Essential topics in applied linguistics and multilingualism (pp. 21–41). Springer International Publishing.
Conklin, K. (2019). Processing single- and multi-word items. In S. Webb (Ed.), Routledge handbook of vocabulary
studies (pp. 174–188). Routledge.
Conklin, K., & Carrol, G. (2018). First language infuence on the processing of formulaic language in a second
language. In A. Siyanova-Chanturia & A. Pellicer-Sánchez (Eds.), Understanding formulaic language: A second
language acquisition perspective (pp. 62–67). Routledge.
Conklin, K., & Carrol, G. (2021). Words go together like “bread and butter”: The rapid, automatic acquisition
of lexical patterns. Applied Linguistics, 42(3), 492–513.
Conklin, K., & Schmitt, N. (2008). Formulaic sequences: Are they processed more quickly than nonformulaic
language by native and nonnative speakers? Applied Linguistics, 29(1), 72–89.
Cop, U., Keuleers, E., Drieghe, D., & Duyck, W. (2015). Frequency efects in monolingual and bilingual natural
reading. Psychonomic Bulletin & Review, 22(5), 1216–1234.
Deshors, S., & Gries, S. (2023). Using corpora in research on L2 psycholinguistics. In A. Godfroid & H. Hopp
(Eds.), The Routledge handbook of second language acquisition and psycholinguistics. Routledge.
Diependaele, K., Lemhöfer, K., & Brysbaert, M. (2013). The word frequency efect in frst and second lan-
guage word recognition: A lexical entrenchment account. Quarterly Journal of Experimental Psychology, 66(5),
843–863.
Dijkstra, T., & van Heuven, W. J. B. (1998). The BIA model and bilingual word recognition. In J. Grainger & A. Jacobs
(Ed.), Localist connectionist approaches to human cognition (pp. 189–225). Lawrence Erlbaum Associates Publishers.
Dijkstra, T., & van Heuven, W. J. B. (2002). The architecture of the bilingual word recognition system: From
identifcation to decision. Bilingualism: Language and Cognition, 5(3), 175–197.
Dijkstra, T., & van Heuven, W. J. B. (2018). Visual word recognition in multilinguals. In S.-A. Rueschemeyer &
M. G. Gaskell (Eds.), The Oxford handbook of psycholinguistics (2nd ed., pp. 118–143). Oxford University Press.
Dijkstra, T., Wahl, A., Buytenhuijs, F., Van Halem, N., Al-Jibouri, Z., De Korte, M., & Rekké, S. (2018). Mul-
tilink: A computational model for bilingual word recognition and word translation. Bilingualism: Language
Duyck, W., Vanderelst, D., Desmet, T., & Hartsuiker, R. J. (2008). The frequency efect in second-language
visual word recognition. Psychonomic Bulletin & Review, 15(4), 850–855.
Ehrlich, S. F., & Rayner, K. (1981). Contextual efects on word perception and eye movements during reading.
Journal of Verbal Learning and Verbal Behavior, 20(6), 641–655.
Forster, K. I., & Chambers, S. M. (1973). Lexical access and naming time. Journal of Verbal Learning and Verbal
Behavior, 12(6), 627–635.
Gardner, M. K., Rothkopf, E. Z., Lapan, R., & Laferty, T. (1987). The word frequency efect in lexical deci-
sion: Finding a frequency-based component. Memory & Cognition, 15(1), 24–28.
Gernsbacher, M. A. (1984). Resolving 20 years of inconsistent interactions between lexical familiarity and
orthography, concreteness, and polysemy. Journal of Experimental Psychology: General, 113(2), 256–281.
213
Godfroid, A., Ahn, J., Choi, I., Ballard, L., Cui, Y., Johnston, S., . . . & Yoon, H. J. (2018). Incidental vocabu-
lary learning in a natural reading context: An eye-tracking study. Bilingualism: Language and Cognition, 21(3),
563–584.
Gollan, T. H., Montoya, R. I., Cera, C., & Sandoval, T. C. (2008). More use almost always means a smaller
frequency efect: Aging, bilingualism, and the weaker links hypothesis. Journal of Memory and Language, 58(3),
787–814.
Gor, K. (2023). Word and multiword processing: memory-based and linguistic approaches. In A. Godfroid & H.
Harley, T. A. (2014). The psychology of language: From data to theory (4th ed.). Psychology Press.
Hudson, P. T., & Bergman, M. W. (1985). Lexical knowledge in word recognition: Word length and word fre-
quency in naming and lexical decision tasks. Journal of Memory and Language, 24(1), 46–58.
Lauro, J., & Schwartz, A. I. (2017). Bilingual non-selective lexical access in sentence contexts: A meta-analytic
review. Journal of Memory and Language, 92, 217–233.
López, B., & Vaid, J. (2018). Fácil or A piece of cake: Does variability in bilingual language brokering experience
afect idiom comprehension? Bilingualism: Language and Cognition, 21(2), 340–354.
McClelland, J. L., & Rumelhart, D. E. (1981). An interactive activation model of context efects in letter percep-
tion: I. An account of basic fndings. Psychological Review, 88(5), 375–407.
McDonald, S. A., & Shillcock, R. C. (2003). Eye movements reveal the on-line computation of lexical prob-
abilities during reading. Psychological Science, 14(6), 648–652.
Monaghan, P., Chang, Y. N., Welbourne, S., & Brysbaert, M. (2017). Exploring the relations between word
frequency, language exposure, and bilingualism in a computational model of reading. Journal of Memory and
Language, 93, 1–21.
Monsell, S. (1991). The nature and locus of word frequency efects in reading. In D. Besner & G. W. Humphreys
(Eds.), Basic processes in reading: Visual word recognition (pp. 148–197). Erlbaum.
Murray, W. S., & Forster, K. I. (2004). Serial mechanisms in lexical access: The rank hypothesis. Psychological
Review, 111(3), 721–756.
Norris, D. (2013). Models of visual word recognition. Trends in Cognitive Sciences, 17(10), 517–524.
Northbrook, J., & Conklin, K. (2018). “What are you talking about?”—An analysis of lexical bundles in Japa-
nese junior high school textbooks. International Journal of Corpus Linguistics, 23(3), 311–334.
Northbrook, J., & Conklin, K. (2019). Is what you put in what you get out?—Textbook-derived lexical bundle
processing in beginner English learners. Applied Linguistics, 40(5), 816–833.
Northbrook, J., & Conklin, K. (2021). “That sounds about right”–An examination of lexical bundle naturalness
intuitions in native and non-native English speakers. Vocabulary Learning and Instruction, 10(1), 49–63.
Pellicer-Sánchez, A. (2017). Learning L2 collocations incidentally from reading. Language Teaching Research,
21(3), 381–402.
Plaut, D. C. (1997). Structure and function in the lexical system: Insights from distributed models of word read-
ing and lexical decision. Language and Cognitive Processes, 12(5–6), 765–806.
Rayner, K., & Raney, G. E. (1996). Eye movement control in reading and visual search: Efects of word fre-
quency. Psychonomic Bulletin & Review, 3(2), 245–248.
Rumelhart, D. E., & McClelland, J. L. (1982). An interactive activation model of context efects in letter per-
ception: Part 2. The contextual enhancement efect and some tests and extensions of the model. Psychological
Review, 89(1), 60–94.
Seidenberg, M. S., & McClelland, J. L. (1989). A distributed, developmental model of word recognition and
naming. Psychological Review, 96(4), 523–568.
Siyanova-Chanturia, A., Conklin, K., & Schmitt, N. (2011). Adding more fuel to the fre: An eye-tracking study
of idiom processing by native and non-native speakers. Second Language Research, 27(2), 251–272.
Siyanova-Chanturia, A., Conklin, K., & van Heuven, W. J. (2011). Seeing a phrase “time and again” matters:
The role of phrasal frequency in the processing of multiword sequences. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 37(3), 776–784.
Siyanova-Chanturia, A., & Janssen, N. (2018). Production of familiar phrases: Frequency efects in native speak-
ers and second language learners. Journal of Experimental Psychology: Learning, Memory, and Cognition, 44(12),
2009–2018.
Siyanova-Chanturia, A., & Sidtis, D. V. L. (2018). What online processing tells us about formulaic language. In
Understanding formulaic language: A second language acquisition perspective (pp. 38–61). Routledge.
Tomasello, M. (2003). Constructing a language: A usage-based theory of language acquisition. Harvard University Press.
van Hell, J. G., & Tanner, D. (2012). Second language profciency and cross-language lexical activation. Language
Learning, 62(S2), 148–171.
214
Whaley, C. P. (1978). Word-nonword classifcation time. Journal of Verbal Learning and Verbal Behavior, 17(2),
143–154.
Whitford, V., Pivneva, I., & Titone, D. (2016). Eye movement methods to investigate bilingual reading. In R. R.
Heredia, J. Altarriba, & A. B. Cieślicka (Eds.), Methods in bilingual reading comprehension research (pp. 183–211).
Springer Science+ Business Media LLC.
Whitford, V., & Titone, D. (2012). Second-language experience modulates frst- and second-language word
frequency efects: Evidence from eye movement measures of natural paragraph reading. Psychonomic Bulletin
and Review, 19(1), 73–80.
215
18
SENTENCE PROCESSING IN
A SECOND LANGUAGE
Linguistic approaches
Holger Hopp
Introduction
Research on second language (L2) sentence processing investigates how L2 users build grammati-
cal structure and sentence interpretation incrementally as they encounter sentences word by word
in real-time parsing. Language users rapidly integrate diferent types of information in parsing. For
one thing, they make use of universal as well as language-specifc linguistic information at the gram-
matical, lexical-semantic, and discourse level. For another, they employ processing routines that
refect the architecture of the language processor and are constrained by cognitive resources. Finally,
language users refer to probabilistic information gathered from the experience of using the language,
allowing users to adapt their processing to patterns encountered in the input and to predict the most
likely continuation of an unfolding sentence (for overview, van Gompel, 2013).
From a linguistic point of view, the main question in research on L2 sentence processing is how
L2 users access or apply grammatical knowledge and grammar-based processing routines to assign the
input string a grammatical representation. This chapter traces the main themes in linguistic research
on L2 sentence processing by reviewing how L2 learners resolve syntactic ambiguities, how they apply
syntactic constraints in processing, and how they use morphosyntactic information in parsing (see
Textbox 18.1). I then outline three approaches to L2 sentence processing and present future directions.
Parsing: Parsing is the real-time assignment of grammatical structure and meaning during sentence comprehension.
Incrementality: Sentence processing is incremental in that comprehenders immediately analyze and inter-
pret the incoming input, assigning it a preliminary parse.
Garden-path sentences: These are temporarily ambiguous sentences that initially suggest a likely analysis
which then needs to be reanalyzed later on, so that processing difculty ensues.
Filler-gap dependencies: These are sentences, such as wh-questions or relative clauses, containing con-
stituents like wh-words (fllers) in positions diferent from where they are interpreted (gap), e.g. “Who
did Jim see ___ last night?”
Referential dependencies: These are sentences containing refexives or pronouns that depend in interpre-
tation on other elements in the sentence, e.g., “Joe saw the mani cut himselfi.”
216 DOI: 10.4324/9781003018872-21

Sentence processing in a second language
The study of L2 sentence processing is situated within the larger concerns of SLA research on ultimate
attainment and diferences between native speakers (NS) and non-native speakers (NNS). Most stud-
ies pose questions about the “native-likeness” of grammatical processing by comparing NS to NNS
sentence processing. Over the years, the research interests have shifted and developed. In the 1990s,
L2 sentence processing research emerged mainly in the context of formal approaches to SLA, which
were couched in generative frameworks of universal grammar (UG; e.g., White, 2014). Originally,
the rationale for using time-sensitive measures of sentence comprehension was to supplement ofine,
i.e., untimed, data obtained in grammaticality judgments. Reaction-time data were thought to be less
tainted by metalinguistic refection and thus could provide an additional window into learners’ implicit
linguistic knowledge (e.g., Jufs & Harrington, 1996). As a consequence, studies tested how L2 learners
processed ungrammaticalities or syntactic constraints on possible interpretations in the context of claims
about age efects that may limit learners’ “access” to Universal Grammar after a purported critical period
(Lenneberg, 1967). The focus on age efects on grammatical knowledge gave way to an increased
interest in potential diferences in processing routines between NS and NNS in the early 2000s. Either
explicitly or tacitly, most researchers in L2 sentence processing assume formal parsing models proposed
for NS processing, such as Frazier’s garden-path model (Frazier, 1987), Gibson’s dependency locality
theory (Gibson, 1998) or Pritchett’s generalized theta attachment theory (Pritchett, 1992). Studies
then investigated whether L2 learners make use of the same processing routines as proposed for native
speakers in these models and particularly investigated whether L1-L2 diferences afected L2 parsing. In
their infuential Shallow Structure Hypothesis, Clahsen and Felser (2006) posited that adult L2 speakers
underuse grammatical information and over-rely on surface heuristics, lexical-semantic, and discourse
information instead. As a consequence, a multitude of studies sought to assess how adult L2 learners
apply grammatical knowledge versus other information types in sentence processing.
Currently, researchers are establishing closer links between monolingual models of processing
about cue interference in parsing (e.g., Cunnings, 2017), the role of grammatical prediction in sen-
tence processing (e.g., Kaan, 2014; Grüter et al., 2017), and individual diferences in the use of gram-
matical information (e.g., Hopp, 2015a). Other than in cognitive approaches to L2 processing (see
Sagarra, 2023 [this volume]), the primary aim of linguistic approaches to sentence processing is not
to model the contributions of individual learner factors to L2 sentence processing. Rather, studies
control for individual diferences in cognitive processing to gauge whether evidence of grammatical
signatures emerges in L2 sentence processing when L2 users are best matched to native speakers.
Resolving syntactic ambiguities: garden-paths and reanalysis

Garden-path sentences have long been a staple in the experimental diet of researchers in sentence pro-
cessing. They allow researchers to investigate (a) whether L2 learners process sentences incrementally
using the same processing strategies as native speakers, and (b) whether L2 learners can revise an initial
misanalysis by reanalyzing the parse once they encounter disambiguating information. The sentences
in (1) illustrate common types of temporarily ambiguous sentences used in research on sentence pro-
cessing. In (1), the ambiguous region is in italics, and the disambiguating information is in bold print.
(1) a. While the band played the song pleased all the customers. (object-subject ambiguity)
b. The inspector warned the boss would destroy very many lives. (object-subject ambiguity)
c. The celebrity interviewed by the media was angry. (reduced relative clause ambiguity)
d. The dean liked the secretary of the professor who hurt himself. (relative clause attachment
ambiguity)
217
Holger Hopp
According to the garden-path model of parsing, the parser employs economy principles in assign-
ing grammatical structure to the input (Frazier, 1987). For instance, the parsing principle of minimal
attachment holds that the parser posits the fewest grammatical nodes compatible with an analysis, and
the principle of late closure states that the parser accommodates incoming elements into the phrase
currently being processed. In conjunction, these principles can account for garden-path efects for
the sentences in (1). For instance, an incoming noun phrase after a verb in (1a) and (1b) is prefer-
entially integrated as its object rather than a subject of a new clause as per late closure. When readers
then encounter disambiguating information that goes against the parsing principles, they slow down,
indicating that they need to reanalyze the parse.
For object-subject ambiguities as in (1a), adult L2 learners, like natives, indeed slow down on
the main clause verb (pleased) in ambiguous sentences compared to non-ambiguous sentences with
intransitive verbs or a comma separating the preposed adjunct clause and the main clause (When
the band played, the song . . .). These efects hold irrespective of whether the learners’ L1 is verb-
fnal and thus does not allow for postverbal objects (Jufs & Harrington, 1996); yet, L1-L2 difer-
ences in the subcategorization properties of a verb can modulate the size of the garden-path efect
(Frenck-Mestre & Pynte, 1997). Like natives, L2 readers also access verb bias information refect-
ing the usage frequency of verbs and their complements. L1 Spanish readers of L2 English showed
reading-time diferences in garden-path efects depending on whether the verb in sentences like
(1b) occurred more often with direct objects (e.g., warned) or had a bias towards sentential comple-
ments (e.g., admitted; Dussias & Cramer-Scaltz, 2008). Subsequent studies found verb bias efects
also for high-profciency L2 learners whose L1s are verb-fnal, so that verb bias could never guide
incremental processing in their L1s and verb biases could not have been transferred (Lee et al.,
2013; Şafak & Hopp, 2021). Taken together, garden-path studies illustrate that, across L1s, L2
learners process sentences incrementally and employ parsing principles and lexical-thematic infor-
mation in processing.
Even so, L2 readers appear to difer from native speakers in how easily they can recover the target
interpretation once they encounter disambiguating information. For instance, Roberts and Felser
(2011) manipulated the plausibility of the postverbal NP as an object in sentences as in (1a, song
vs. beer) and (1b, boss vs. crimes). For native speakers, an implausible NP-as-object (beer/crimes) eased
reanalysis compared to plausible NPs (song/boss). For L2 readers, plausibility diferences did not afect
reanalysis in preposed adjunct clauses as in (1a), suggesting that reanalysis remains incomplete in
complex garden-path sentences (though see Hopp, 2015a). Further evidence for greater reanalysis
efort in L2 readers comes from studies on semantic persistence. In an eye-tracking study by Jacob
and Felser (2016), native English speakers and German-speaking L2 learners of English read sen-
tences as in (2).
(2) While the gentleman was eating(,) the burgers were still being reheated in the microwave.
Both native English and L1 German L2 readers of English had longer reading times for the disambig-
uating region (were still) in the syntactic disambiguating region in the ambiguous sentences without
a comma, with the L2 readers showing lower reading time diferences, indicating weaker reanalysis.
Further on, both groups also demonstrated slowdowns in the semantic disambiguation region being
reheated. Apparently, the syntactic disambiguation was not sufcient to erase the initial analysis of the
burgers as an object, leading to semantic persistence efects. Moreover, in subsequent comprehen-
sion questions which directly targeted the direct-object misinterpretation (the gentleman was eating
the burgers), both L1 and L2 participants accepted this interpretation, with the L2 group showing
even higher acceptance. This pattern suggests greater semantic persistence in L2 processing in that
initial misanalyses linger more, and L2 learners often fail to complete reanalysis (see also Pozzan &
Trueswell, 2016; Şafak & Hopp, 2021).
218
Relative clause attachment ambiguities as in (1d) have been studied predominantly to assess the
degree to which L1 diferences in parsing afect L2 sentence processing. Across languages, attachment
preferences for relative clauses vary in that native speakers of some languages prefer to interpret rela-
tive clauses as referring to the frst noun of a complex noun phrase (the secretary in (1d); high attach-
ment; e.g., German, Spanish, Greek), whereas other languages prefer low attachment (the professor in
(1d); e.g., English, Italian, Brazilian Portuguese). Gibson et al. (1996) propose that these diferences in
attachment preferences refect cross-linguistic diferences in the relative weighting of two structure-
based parsing principles, recency and predicate proximity. Similar to the principle of late closure, recency
favors attaching an incoming element to the most recent grammatical node, while predicate proximity
guides the parser to attach a node as closely to the head of a predicate as possible. Several studies
reported that advanced L2 learners did not demonstrate any attachment preferences for sentences in
(1d) in reading times in the absence of lexical-thematic or discourse biases for disambiguation, even
when the L1 preference matched the L2 preference (Felser et al., 2003; Pan & Felser, 2011; Papado-
poulou & Clahsen, 2003). Clahsen and Felser (2006) interpreted these fndings as evidence against
L1 efects in L2 processing and suggested that L2 learners generally do not refer to structure-based
parsing principles. However, later research reporting L1-based or native-like preferences (Hopp,
2014; Witzel et al., 2012) suggests that L2 learners (a) are sensitive to grammatical structure in parsing
ambiguous sentences at higher levels of processing automaticity and (b) that L1 parsing preferences
inform L2 processing but can be overcome as L2 learners achieve higher profciency.
In sum, like native processing, L2 sentence processing is characterized by incremental analysis,
the use of parsing principles, and grammatical structure. L2 learners readily employ lexical-thematic
and discourse information in the processing of garden-path sentences and temporary ambiguities, yet
they appear to experience greater difculty efecting and completing syntactic reanalysis than native
speakers.
Using syntactic constraints in processing
Filler-gap dependencies
In questions or relative clauses, a wh-phrase, such as who or which boy, does not always appear next to
the predicate, and needs to be related to its subcategorizer (e.g., the verb) for interpretation by virtue
of a fller-gap dependency. For instance, in an English object wh-question such as Which boy did the girl
meet?, the fller, the wh-phrase which boy, needs to be related to the subcategorizing verb meet for inter-
pretation in that comprehenders posit a gap for the fller following the verb (see Textbox 18.1). More-
over, wh-phrases can be displaced across clause boundaries in so-called long-distance dependencies.
Crucially, there appear to be grammatical constraints on the formation of long-distance dependencies
(e.g., Ross, 1967), so that wh-phrases cannot be extracted across any phrase boundary. Studies on the
L2 processing of fller-gap dependencies examine how L2 learners relate wh- phrases to subcategorizers
and whether they obey grammatical constraints on wh-extraction. In processing wh-questions, native
speakers attempt to link the fller (the wh-phrase) to its lexical subcategorizer as soon as possible by pos-
tulating a gap during sentence processing. In line with such an active fller strategy (Clifton & Frazier,
1989), readers show slowdowns at the potentially frst gap site following the verb roast in (3) when they
realize the potential gap site is flled by the NP the potatoes (e.g., Boland et al., 1995).
(3) The chef decided which carrots he would roast the potatoes with ___.
L2 learners also pursue an “active” gap-flling strategy (Dallas & Kaan, 2008; Williams et al., 2001).
For instance, Omaki and Schulz (2011) tested whether L2 learners would show plausibility difer-
ences indicative of active gap flling in sentences as in (4).
219
Holger Hopp
(4) The book/city that the author wrote (___) regularly about ___ was named for an explorer.
Both the nouns in (4), the book and the city, are plausible fllers following the preposition about, yet only
the book constitutes a plausible fller at the earlier gap site following the verb wrote. Self-paced reading
data for English natives and an L1 Spanish–L2 English group demonstrated plausibility diferences in
reading times at the earlier gap site, which suggests that L2 learners pursue an active gap-flling strategy
(see also Kim, Baek et al., 2015). However, these studies leave open whether L2 learners simply posit
a gap following the frst possible lexical subcategorizer they encounter or whether they are sensitive to
syntactic constraints on the licensing of gaps.
To address this issue, several studies investigated whether L2 learners also posit gaps within so-
called grammatical islands from which no fller can be extracted, for instance relative clauses (5).
(5) The book/city that the author [who wrote *(___) regularly] saw ___ was named for an explorer.
In Omaki and Schulz (2011), L1 and L2 readers demonstrated plausibility diferences in reading times
at the potential gap site in the non-island sentences in (4), yet not in the matched sentences contain-
ing a relative clause island in (5). Other studies using diferent methods (Felser et al., 2012) replicated
such diferential efects, which suggests that L2 learners obey syntactic island constraints (though see
Kim et al., 2015, for L1 efects).
Another line of research considers whether L2 learners posit structurally required gaps. According
to current versions of syntactic theory, cross-clausal wh-dependencies project intermediate gaps at
clause boundaries (Chomsky, 1995). Hence, the sentence in (6a) with a clausal complement contains
an intermediate gap preceding the complementizer that, while a matched sentence with a nominal
complement does not project an intermediate gap (6b).
(6) a. The nurse who the doctor argued [___ that the rude patient had angered ___] is refusing to
work late. (extraction across CP)
b. The nurse who the doctor’s [argument about the rude patient had angered __] is refusing to
work late. (extraction across NP)
Marinis et al. (2005) tested L1 Chinese, L1 German, L1 Greek, and L1 Japanese advanced L2
learners and native English controls in a self-paced reading task on the sentences in (6) as well
as matched control sentences that did not contain any extraction. Native speakers read the verb
before the fnal gap (angered) more quickly in the extraction-across-CP than in the extraction-
across-NP condition, while there were no RT diferences in the matched non-extraction con-
ditions. Apparently, fller integration is easier for native speakers when they could previously
reactivate the fller at the intermediate gap site in (6a). In contrast, none of the L2 groups demon-
strated reading diferences between (6a) and (6b), which led Marinis and colleagues to conclude
that L2 learners do not project abstract grammatical structure during sentence processing (see also
Felser & Roberts, 2007). However, highly profcient learners who had been immersed in an L2
context for longer periods of time demonstrate native-like processing of intermediate gaps in the
same type of task (Pliatsikas & Marinis, 2013), and even non-immersed L2 learners have been
found to posit gaps in studies using more fne-grained pupillometry measures during eye-tracking
(Fernandez et al., 2018).
In sum, studies on fller-gap dependencies underscore that L2 learners process the L2 incremen-
tally using similar parsing principles as NS. Further, grammatical structure constrains L2 syntac-
tic processing in that advanced L2 learners posit gaps only when and where they are syntactically
licensed.
220
Referential processing
On top of fller-gap dependencies, researchers examine the resolution of referential dependencies
within a sentence, which are subject to grammatical binding constraints. According to binding the-
ory (e.g., Chomsky, 1981), refexive and reciprocal pronouns (e.g., himself, each other) must be bound
locally, i.e., typically within a clause, by a coreferential antecedent (Principle A). In an eye-tracking
study, Felser and Cunnings (2012) manipulated coreference options in terms of stereotypical gender
as in (7), such that a refexive would have either only one likely antecedent (frefghter) or two possible
antecedents (he, frefghter), of which only the local antecedent (frefghter) was grammatically accessible
as per binding theory.
(7) John/Jane looked on as the building burnt down. He/She realized that the frefghter had
injured himself/herself on some sharp pieces of glass. Luckily nobody else was seriously injured.
Native English readers showed delays in reading refexive pronouns with non-stereotypical local
antecedents (frefghter—herself), yet reading times did not difer depending on the gender of the
non-local distractor (He/She). During sentence comprehension, native speakers thus obey locality
restrictions on binding and do not consider other discourse-accessible, yet non-local antecedents (see
also Sturt, 2003). In contrast to the native speakers, L1 German learners of English had longer ini-
tial reading times on the refexives when the non-local distractor mismatched in gender (she). They
retrieved the correct grammatical antecedent (frefghter) only later, as shown in longer re-reading
times on the refexive pronoun for the non-stereotypical local antecedent. While the application of
binding principles for refexive pronouns (Principle A) appears to be delayed during L2 processing,
several studies suggest that L2 readers do not violate Principle B, which prohibits personal pronouns
from being bound locally. For instance, Patterson et al. (2014) studied how L1 German learners of
English resolved the pronoun him in sentences as in (8).
(8) a. Jane remembered that John had taught him a new song on the guitar.
b. John remembered that Jane had taught him a new song on the guitar.
Both native and L2 readers slowed down after the pronoun when the grammatically accessible ante-
cedent (the main clause subject) mismatched the pronoun in gender (8a). However, gender mis-
matches between the pronoun and the grammatically inaccessible embedded clause subject (8b) did
not lead to delays in either group. Unlike native speakers, however, L2 readers also did not show any
efects of gender mismatches between the pronoun and the embedded clause subject when both the
main clause and the embedded clause subject were grammatically accessible antecedents in sentences
with spatial prepositions as in (9).
(9) Andy noticed Ryan place a chair next to him at the front of the hall.
In conjunction, this fndings that L2 readers prefer non-local antecedents for pronouns sug-
gests they prioritize discourse-prominent antecedents, such as main clause subjects, irrespective
of whether they are grammatically accessible (see also Cunnings, 2017; Felser, 2019). While
preferring discourse-salient subjects as antecedents leads to non-target processing for refexives
(Principle A), it occasions the proliferation of seeming Principle B efects that rule out local
antecedents for pronouns. However, such diferences between NS and NNS in referential pro-
cessing do not generalize across all binding contexts. Studies on cataphoric pronouns, in which
the pronoun precedes its coreferential noun, suggest that native and non-native readers behave
221
Holger Hopp
similarly in establishing coreference and applying binding principles online (Drummer & Felser,
2018). In such forward-looking referential dependencies, both native and non-native speakers
initially allow for coreference between pronouns and grammatically inaccessible nouns and only
apply binding constraints later.
By contrasting discourse and syntactic constraints on coreference, studies on referential L2
processing add insights into the weighting of grammatical vs. other constraints. Compared to
native speakers, L2 learners experience protracted difculty integrating discourse information with
syntactic binding principles in that they assign greater priority to discourse cues and sometimes
apply syntactic binding constraints later than native speakers (see also Kim, Montrul et al., 2015;
Roberts et al., 2008).
Using morphosyntactic information in processing

Another line of work in L2 sentence processing considers how L2 learners recruit morphosyn-
tactic knowledge in the real-time processing of agreement relations. Here, the main question of
interest is whether diferences between the learners’ L1 and the L2 in the availability or realiza-
tion of morphosyntactic features impact L2 processing. These studies relate to formal theories
about the transfer and accessibility of syntactic features in adult SLA (e.g., Lardiere, 2009).
According to representational defcit models, the L1 features become fxed due to a critical
period, so that adult L2 learners are constrained by the L1 inventory of features in acquiring
L2 morphosyntax (e.g., Hawkins & Chan, 1997). In this context, studies on morphosyntactic
agreement in L2 processing probe whether L2 learners can come to process morphosyntactic
features in a native-like way even when those features are not or are diferently realized in
their L1s. Research in this line of work often employs ungrammatical sentences and measures
reading-time delays or electrophysiological correlates of anomaly detection in order to test
whether L2 learners are sensitive to morphosyntactic agreement violations. In design, these
studies typically compare morphosyntactic features unique to the L2, such as grammatical gen-
der, to features that are shared between L1 and L2, like number (e.g. Sabourin & Stowe, 2008),
or they compare morphosyntactic features like case to non-grammatical cues such as animacy
(e.g. Jackson & Roberts, 2010). This section cannot even begin to do justice to the vast litera-
ture on the processing of morphosyntactic agreement violations (for a recent review, see Jiang,
2018: Chapter 7); not least because much research carried out within the frameworks of cue-
based and emergentist approaches to SLA also addresses agreement processing (see Sagarra, 2023
[this volume]).
From a linguistic perspective, studies on agreement processing are particularly relevant insofar
as they test asymmetries during L2 sentence processing that bear refexes of the underlying lin-
guistic representation of these features. In L2 production, it has been observed that L2 learners do
not make random agreement errors (Prévost & White, 2000). L2 learners are more likely to omit
infection or use an underspecifed default form (e.g., “He sing”) than to produce a fully specifed
form, such as a verb marked for third-person singular, with a mismatching subject that clashes in
person or number features (e.g., “You sings”). Such asymmetries suggest that interlanguage gram-
mars represent linguistic defaults and morphosyntactic markedness hierarchies (McCarthy, 2012) as
proposed in linguistic frameworks such as distributed morphology (Embick & Noyer, 2007). Sev-
eral L2 processing studies considered whether L2 learners demonstrate greater sensitivity to feature
clashes than to default errors in the processing of agreement violations. In speeded grammaticality
judgments, intermediate and advanced L1 English learners of Spanish have lower accuracy rates
for default errors involving singulars than feature clashes with plurals for number agreement, yet
show the reverse pattern of over-accepting feminine feature clashes for masculine nouns with
222
gender agreement (Lopez-Prego & Gabriele, 2014). In contrast, native Spanish speakers performed
better on feature clash than default errors for both number and gender agreement when process-
ing the stimuli at increased presentation speeds. Using an ERP paradigm, Alemán Bañón et al.
(2017) examined whether diferences in person markedness would afect the processing of subject-
verb agreement violations in L1 English advanced learners of Spanish. According to theoretical
accounts (Harley & Ritter, 2002), frst-person singular denotes a marked speaker role, while third-
person singular constitutes the unmarked default person. Participants read sentences with frst-
person pronominal or third-person subjects paired with third-person or frst-person marked verbs,
respectively, in Spanish. While native Spanish speakers had larger processing costs (i.e., P600s) for
agreement violations with marked frst-person subjects paired with third-person verbs, the L2
learners evinced stronger efects for marked frst-person verbs paired with unmarked third-person
subjects. Taken together, these studies suggest that grammatical diferences in markedness afect L2
processing, albeit not always in native-like ways.
Other studies consider the status of diferent morphosyntactic features in sentence processing. For
instance, Hopp (2006) probed how L2 learners employed case and number marking in the revision
of non-canonical object-frst sentences in German. According to Fodor and Inoue (2000), case and
number marking difer in their informativity for reanalysis, with case marking carrying information
on how to repair the parse, while number marking simply signals that the initial subject-frst analysis
was erroneous. At near-native profciency levels, L1 English, L1 Dutch, and L1 Russian learners of
German completed reanalysis more accurately and more quickly for sentences that marked object-
frst orders by case than by number in self-paced reading and speeded grammaticality judgment
experiments. These fndings suggest that L2 learners, like native speakers, employ morphosyntactic
features diferentially in sentence processing
In sum, studies reporting that L2 learners make grammatical distinctions in L2 processing
provide some evidence that L1 and L2 processing implicate similar mechanisms and underlying
representations, even when L2 users are not native-like in absolute performance levels or reaction
times.
Linguistic approaches to L2 sentence processing

The Shallow Structure Hypothesis (SSH) by Clahsen and Felser (2006) constitutes the central
model in L2 sentence processing and has stimulated most research on L2 sentence processing
since the mid-2000s. Assuming a critical period for grammar, the original version of the SSH
states that adult L2 learners cannot recruit abstract syntactic information during real-time process-
ing. Instead, they allegedly over-rely on non-grammatical information, such as lexical-thematic,
semantic, and discourse information or surface heuristics, e.g., word order patterns. In processing
terms, the SSH takes inspiration from dual-pathway models of monolingual sentence processing
like the good-enough approach by Ferreira and colleagues (Ferreira & Patson, 2007). According
to good-enough processing, speakers compute surface mappings of input strings to interpretation
on top of using a syntactic route to sentence interpretation as per structure-based models of pars-
ing. The SSH claims that adult L2 learners are “largely restricted” to the shallow route (Clahsen
& Felser, 2006, p. 34). In a revision of the SSH, Clahsen and Felser (2018) accommodate recent
evidence of native-like L2 grammatical processing of long-distance dependencies and sensitivity to
structure-based constraints in referential processing. They argue that shallow processing is a matter
of degree among L2 learners and may also play out in delays in the timing or a reduced weight-
ing of grammatical versus non-grammatical information. While the revised SSH can thus fexibly
accommodate evidence both of native-like and non-native processing, it raises questions as to how
this approach could be falsifed.
223
Holger Hopp
Cunnings (2017) presents an alternative model to the SSH that is grounded in cue-based
approaches to parsing (e.g., McElree, 2000). According to these approaches, comprehenders
encode and retrieve diferent cues in a capacity-restricted working memory during parsing.
With a view to L2 processing, Cunnings claims (a) that L2 learners weight discourse cues more
strongly than grammatical cues and (b) that L2 learners experience greater retrieval interference
than native speakers when they need to reconstruct grammatical information from memory
in parsing. This account captures a large array of processing data on L1-L2 diferences; yet it
remains unclear why exactly L2 learners should difer from NS in the two ways suggested by
Cunnings.
Finally, a number of approaches hold that many diferences between NS and NNS parsing do
not point to fundamental diferences between the two, yet follow from the increasing cognitive
demands of processing a non-native language (e.g. Hopp, 2010; McDonald, 2006). Among them,
linguistically based approaches posit additional strains on L2 processing at linguistic interfaces
connecting diferent linguistic domains. For instance, the Interface Hypothesis (Sorace, 2011)
identifes the external interface between grammar and discourse as a locus of processing dif-
fculty in bilinguals. In consequence, L2 learners have trouble mapping discourse information
onto syntax and demonstrate protracted diferences in referential processing. Another approach
focusing on the lexicon-syntax interface, the Lexical Bottleneck Hypothesis (Hopp, 2018), posits
that slower lexical access in L2 learners impedes or delays the construction of syntactic struc-
ture and dependencies in real-time processing. Neither hypothesis amounts to a comprehensive
model of NS-NNS diferences; instead each identifes NS-NNS diferences that follow from the
architecture and the mechanisms of the grammar and the bilingual language processing system.
Whereas these approaches capture many fndings in which L2 learners show delays or incomplete
parsing compared to natives, they cannot easily account for observations that L2 learners some-
times demonstrate greater sensitivity to discourse cues than natives (Pan & Felser, 2011) or prefer
computationally more costly non-local over local antecedents in establishing coreference (Felser
& Cunnings, 2012).

This review suggests that there do not seem to be fundamental diferences in the processing strate-
gies, constraints, and the information native and L2 speakers deploy in sentence processing. Since
L2 learners have been shown to recruit both grammatical and non-grammatical information in pro-
cessing, the initial question of WHAT information types are predominantly used in L2 processing
is gradually being replaced by the more nuanced question of WHEN diferent types of information
are recruited during L2 processing. Recent work has started to explore the timing of grammatical
constraints (Felser, 2019). Time-sensitive methods such as ERP and eye-tracking allow for disentan-
gling possible diferences in timing or temporal staging in the application of grammatical and other
constraints in L2 sentence processing.
Besides refning experimental designs and methods, the feld is slowly moving beyond the question of
whether adult L2 sentence processing can be native-like with reference to adult native speakers (for dis-
cussion, Hopp, 2022). An increasing number of studies makes comparisons across bilingual populations.
Research on sentence processing in other bilingual populations, e.g., bilingual children and child L2 learners,
suggests that the amount of cross-linguistic interactions and the degree of target-like parsing depend on the
age of acquisition and amounts of exposure (e.g., Lemmerth & Hopp, 2019). Further, fndings of increased
reanalysis costs in L2 processing (section “Resolving syntactic ambiguities: garden-paths and reanalysis”)
resemble results from child L1 sentence processing (Trueswell et al., 1999), suggesting that some apparent
diferences between native and non-native sentence processing may be characteristics of the develop-
ing parser. Other studies compare L2 learners with L1 speakers who may have undergone L1 attrition
224
after long-term exposure to an L2 (see Schmid et al., 2023 [this volume]). Among L1 speakers, processing
preferences for relative clause attachment may switch after lengthy L2 exposure (Dussias & Sagarra, 2007).
In contrast, the use of L1 grammatical information remains resilient in L1 parsing even after long-term
immersion in an L2 (Grüter & Hopp, 2021, though see Kasparian & Steinhauer, 2017). Comparative
bilingual processing research along these lines can elucidate whether lower sensitivity to grammatical infor-
mation or greater cue interference is specifc to late L2 learners or generalizes across bilingual populations
(see Textbox 18.2).
Finally, L2 sentence processing research is beginning to link up more with research on L2 acqui-
sition. Recent accounts stress the nexus between processing and learning by way of predictive pro-
cessing (Phillips & Ehrenhofer, 2015). Making active predictions in sentence processing allows the
language learner to test their interlanguage grammars against the incoming input. The input then
provides immediate feedback as to whether the learner’s grammar matches the target language. In
consequence, experiencing prediction errors allows the learner to update their predictions and thus to
gradually converge on the target language grammar. Many studies report that adult L2 learners tend to
engage less in prediction than native speakers (Grüter et al., 2017; see Kaan, 2014). Moreover, when
their predictions turn out to be wrong, adult L2 learners often continue making erroneous predictions
(Hopp, 2015b) or suspend making predictions altogether (Hopp, 2016). Lower degrees of prediction
among L2 learners may thus impede learning and provide processing obstacles in L2 acquisition.
To test if and how language users learn from sentence processing, current research on syntactic adapta-
tion investigates how input changes in the course of an experiment lead to diferent processing (Fine &
Jaeger, 2013). By tracking processing changes over time, these studies examine whether L2 learners adapt
their parsing preferences (Kaan et al., 2019) or whether L2 learners come to integrate morphosyntactic
information in sentence processing (Hopp, 2020) simply from reading sentences containing the target
structure for comprehension. Processing studies ofering repeated exposure to the target grammar in
reading experiments or in syntactic priming (see Jackson, 2023 [this volume]) can be used to test implicit
learning of grammatical detail by virtue of sentence processing. In these ways, research on L2 sentence
processing can connect to theories of L2 development and L2 instruction. These recent developments
illustrate that L2 sentence processing research increasingly engages with core topics in SLA while at the
same time making the case that sentence processing is key to understanding L2 acquisition.
What is the relation between the parser and the grammar, especially in L2 processing when parsing can
be informed by more than one grammar?
Why are L1 efects much more limited in L2 sentence processing than in lexical and phonological pro-
cessing or in ofine SLA studies on grammar?
How does sentence processing vary across diferent bilingual populations?
How does processing feed into learning, i.e., how do learners process to learn, given that learners also
need to learn (how) to process?
Further reading
Jiang, N. (2018). Second language processing: an introduction. Routledge.
This book provides an accessible summary of second language processing and contains two chapters on
sentence processing. The wider focus helps readers to contextualize sentence processing in the larger feld of
bilingual language processing.
225
Holger Hopp
Jufs, A., & Rodriguez, G. A. (2014). Second language sentence processing. Routledge.
This monograph presents the issues in L2 sentence processing research and traces its history from a linguistic
perspective.
Van Gompel, R. P. G. (Ed.). (2013). Sentence processing. Psychology Press.
An edited volume on sentence processing in monolinguals, children, and bilinguals that presents all major
approaches and themes in sentence processing research.
References
Alemán Bañón, J., Miller, D., & Rothman, J. (2017). Morphological variability in second language learners: An
examination of electrophysiological and production data. Journal of Experimental Psychology: Learning, Memory,
and Cognition, 43, 1509.
Boland, J. E., Tanenhaus, M. K., Garnsey, S. M., & Carlson, G. N. (1995). Verb argument structure in parsing
and interpretation: Evidence from wh-questions. Journal of Memory and Language, 34(6), 774–806.
Chomsky, N. (1981). Lectures on government and binding. Foris.
Chomsky, N. (1995). The minimalist program. MIT Press.
Clahsen, H., & Felser, C. (2018). Some notes on the Shallow Structure Hypothesis. Studies in Second Language
Clifton, C., & Frazier, L. (1989). Comprehending sentences with long-distance dependencies. In Linguistic struc-
ture in language processing (pp. 273–317). Springer.
Cunnings, I. (2017). Parsing and working memory in bilingual sentence processing. Bilingualism: Language and
Cognition, 20, 659–678.
Dallas, A., & Kaan, E. (2008). Second language processing of fller-gap dependencies in late learners. Language
and Linguistics Compass, 2, 372–388.
Drummer, J.-D., & Felser, C. (2018). Cataphoric pronoun resolution in native and non-native sentence com-
prehension. Journal of Memory and Language, 101, 97–113.
Dussias, P., & Cramer-Scaltz, T. (2008). Spanish—English L2 speakers’ use of subcategorization bias infor-
mation in the resolution of temporary ambiguity during second language reading. Acta Psychologica, 128,
501–513.
Dussias, P., & Sagarra, N. (2007). The efect of exposure on syntactic parsing in Spanish-English bilinguals.
Bilingualism: Language and Cognition, 10, 101–116.
Embick, D., & Noyer, R. (2007). Distributed Morphology and the syntax/morphology interface. In The Oxford
handbook of linguistic interfaces (pp. 289–324). Oxford University Press.
Felser, C. (2019). Structure-sensitive constraints in non-native sentence processing. Journal of the European Second
Language Association, 3(1).
Felser, C., & Cunnings, I. (2012). Processing refexives in English as a second language: The role of structural
and discourse-level constraints. Applied Psycholinguistics, 33, 571–603.
Felser, C., Cunnings, I., Batterham, C., & Clahsen, H. (2012). The timing of island efects in nonnative sentence
Felser, C., & Roberts, L. (2007). Processing wh-dependencies in a second language: A cross-modal priming
study. Second Language Research, 23, 9–36.
Felser, C., Roberts, L., Marinis, T., & Gross, R. (2003). The processing of ambiguous sentences by frst and
second language learners of English. Applied Psycholinguistics, 24, 453–489.
Fernandez, L., Höhle, B., Brock, J., & Nickels, L. (2018). Investigating auditory processing of syntactic gaps with
L2 speakers using pupillometry. Second Language Research, 34, 201–227.
Ferreira, F., & Patson, N. (2007). The good enough approach to language comprehension. Language and Lin-
guistics Compass, 1, 71–83.
Fine, A. B., & Jaeger, T. F. (2013). Evidence for implicit learning in syntactic comprehension. Cognitive Science,
37(3), 578–591.
Fodor, J. D., & Inoue, A. (2000). Syntactic features in reanalysis: Positive and negative symptoms. Journal of
Psycholinguistic Research, 29, 25–36.
Frazier, L. (1987). Sentence processing: A tutorial review. In M. Coltheart (Ed.), Attention and performance 12:
The psychology of reading (pp. 559–586). Lawrence Erlbaum Associates.
226
Frenck-Mestre, C., & Pynte, J. (1997). Syntactic ambiguity resolution while reading in second and native lan-
guages. Quarterly Journal of Experimental Psychology, 50A, 119–148.
Gibson, E. (1998). Linguistic complexity: Locality of syntactic dependencies. Cognition, 68, 1–76.
Gibson, E., Pearlmutter, N., Canseco-Gonzalez, E., & Hickok, G. (1996). Recency preferences in the human
sentence processing mechanism. Cognition, 59, 23–59.
Grüter, T., & Hopp, H. (2021). How permeable are native and non-native syntactic processing to crosslinguistic
infuence? Journal of Memory and Language, 121, 104281.
Grüter, T., Rohde, H., & Schafer, A. J. (2017). Coreference and discourse coherence in L2: The roles of gram-
matical aspect and referential form. Linguistic Approaches to Bilingualism, 7, 199–229.
Harley, H., & Ritter, E. (2002). Person and number in pronouns: A feature-geometric analysis. Language, 78,
482–526.
Hawkins, R., & Chan, C. Y. H. (1997). The partial availability of Universal Grammar in second language acqui-
sition: The “failed functional features hypothesis”. Second Language Research, 13, 187–226.
Hopp, H. (2006). Syntactic features and reanalysis in near-native processing. Second Language Research, 22,
369–397.
Hopp, H. (2010). Ultimate attainment in L2 infectional morphology: Performance similarities between non-
native and native speakers. Lingua, 120, 901–931.
Hopp, H. (2014). Working memory efects in the L2 processing of ambiguous relative clauses. Language Acquisi-
tion, 21, 250–278.
Hopp, H. (2015a). Individual diferences in the second language processing of object-subject ambiguities.
Applied Psycholinguistics, 36, 129–173.
Hopp, H. (2015b). Semantics and morphosyntax in L2 predictive sentence processing. International Review of
Hopp, H. (2016). Learning (not) to predict: Grammatical gender processing in adult L2 acquisition. Second
Language Research, 32, 277–307.
Hopp, H. (2018). The bilingual mental lexicon in L2 sentence processing. Second Language Research, 17,
5–27.
Hopp, H. (2020). Morphosyntactic adaptation in adult L2 processing: Exposure and the processing of case and
tense violations. Applied Psycholinguistics, 41, 627–656.
Hopp, H. (2022). L2 sentence processing. Annual Review of Linguistics, 8, 235–256.
(Eds.). The Routledge handbook of second language acquisition and psycholinguistics (pp. 38–49). Routledge.
Jackson, C. N., & Roberts, L. (2010). Animacy afects the processing of subject—object ambiguities in the sec-
ond language: Evidence from self-paced reading with German second language learners of Dutch. Applied
Psycholinguistics, 31, 671–691.
Jacob, G., & Felser, C. (2016). Reanalysis and semantic persistence in native and non-native garden-path recov-
ery. The Quarterly Journal of Experimental Psychology, 69, 907–925.
Jufs, A., & Harrington, M. (1996). Garden path sentences and error data in second language processing research.
Kaan, E. (2014). Predictive sentence processing in L2 and L2. What is diferent? Linguistic Approaches to Bilingual-
ism, 4, 257–282.
Kaan, E., Futch, C., Fuertes, R. F., Mujcinovic, S., & de la Fuente, E. Á. (2019). Adaptation to syntactic struc-
tures in native and nonnative sentence comprehension. Applied Psycholinguistics, 40, 3–27.
Kasparian, K., Vespignani, F., & Steinhauer, K. (2017). First language attrition induces changes in online mor-
phosyntactic processing and re-analysis: An ERP study of number agreement in complex Italian sentences.
Cognitive Science, 41, 1760–1803.
Kim, E., Baek, S., & Tremblay, A. (2015). The role of island constraints in second language sentence processing.
Kim, E., Montrul, S., & Yoon, J. (2015). The on-line processing of binding principles in second language acqui-
sition: Evidence from eye tracking. Applied Psycholinguistics, 36, 1317–1374.
Lardiere, D. (2009). Some thoughts on the contrastive analysis of features in second language acquisition. Second
Lee, E. K., Lu, D. H. Y., & Garnsey, S. M. (2013). L1 word order and sensitivity to verb bias in L2 processing.
Lemmerth, N., & Hopp, H. (2019). Gender processing in simultaneous and successive bilingual children: Efects
of lexical and syntactic cross-linguistic infuence. Language Acquisition, 26, 21–45.
Lenneberg, E. (1967). Biological foundations of language. Wiley.
227
Holger Hopp
Lopez-Prego, B., & Gabriele, A. (2014). Examining the nature of morphological variability in native and non-
native Spanish. Linguistic Approaches to Bilingualism, 4, 192–221.
Marinis, T., Roberts, L., Felser, C., & Clahsen, H. (2005). Gaps in second language sentence processing. Studies
in Second Language Acquisition, 27, 53–78.
McCarthy, C. (2012). Modeling morphological variation and development: Person and number in L2 Spanish.
Linguistic Approaches to Bilingualism, 2, 25–53.
McDonald, J. (2006). Beyond the critical period: Processing-based explanations for poor grammaticality judge-
ment performance by late second language learners. Journal of Memory and Language, 55, 381–401.
McElree, B. (2000). Sentence comprehension is mediated by content-addressable memory structures. Journal of
Psycholinguistic Research, 29, 111–123.
Omaki, A., & Schulz, B. (2011). Filler-gap dependencies and island constraints in second-language sentence
Pan, H., & Felser, C. (2011). Referential context efects in L2 ambiguity resolution: Evidence from self-paced
reading. Lingua, 121, 221–236.
Papadopoulou, D., & Clahsen, H. (2003). Parsing strategies in L1 and L2 sentence processing: A study of relative
clause attachment in Greek. Studies in Second Language Acquisition, 25, 501–528.
Patterson, C., Trompelt, H., & Felser, C. (2014). The online application of binding condition B in native and
non-native pronoun resolution. Frontiers in Psychology, 5, 147.
Phillips, C., & Ehrenhofer, L. (2015). The role of language processing in language acquisition. Linguistic
Pliatsikas, C., & Marinis, T. (2013). Processing empty categories in a second language: When naturalistic expo-
sure flls the (intermediate) gap. Bilingualism: Language and Cognition, 16, 167–182.
Pozzan, L., & Trueswell, J. (2016). Second language processing and revision of garden-path sentences: A visual-
world study. Bilingualism: Language and Cognition, 19, 636–643.
Prévost, P., & White, L. (2000). Missing surface infection or impairment in second language acquisition? Evi-
dence from tense and agreement. Second Language Research, 16, 103–133.
Pritchett, B. L. (1992). Grammatical competence and parsing performance. University of Chicago Press.
Roberts, L., & Felser, C. (2011). Plausibility and recovery from garden-paths in second-language sentence pro-
cessing. Applied Psycholinguistics, 32, 299–331.
Roberts, L., Gullberg, M., & Indefrey, P. (2008). Online pronoun resolution in L2 discourse: L1 infuence and
general learner efects. Studies in Second Language Acquisition, 30, 333–357.
Ross, J. (1967). Constraints on variables in syntax [Unpublished doctoral dissertation], Massachusetts Institute of
Technology, USA.
Sabourin, L., & Stowe, L. A. (2008). Second language processing: When are frst and second languages processed
similarly? Second Language Research, 24, 397–430.
Şafak, D. F., & Hopp, H. (2021). Verb bias and semantic persistence efects in L2 ambiguity resolution. Second
Sagarra, N. (2023). Sentence processing: cognitive approaches to second language morphosyntactic and mor-
phological processing. In A. Godfroid & H. Hopp (Eds.). The Routledge handbook of second language acquisition
and psycholinguistics (pp. 229–241). Routledge.
Schmid, M.S., Soto, C., & Heimann, J. (2023). The psycholinguistics of frst language attrition. In A. God-
froid & H. Hopp (Eds.). The Routledge handbook of second language acquisition and psycholinguistics (pp. 61–72).
Routledge.
1, 1–33.
Sturt, P. (2003). The time-course of the application of binding constraints in reference resolution. Journal of
Trueswell, J. C., Sekerina, I., Hill, N. M., & Logrip, M. L. (1999). The kindergarten-path efect: Studying on-
line sentence processing in young children. Cognition, 73, 89–134.
van Gompel, R. P. (2013). Sentence processing. Psychology Press.
White, L. (2014). Linguistic theory, universal grammar, and second language acquisition. In B. VanPatten &
J. Williams (Eds.), Theories in second language acquisition: An introduction (pp. 34–53). Routledge.
Williams, J., Möbius, P., & Kim, C. (2001). Native and non-native processing of English wh-questions: Parsing
strategies and plausibility constraints. Applied Psycholinguistics, 22, 509–540.
Witzel, J., Witzel, N., & Nicol, J. (2012). Deeper than shallow: Evidence for structure-based parsing biases in
second-language sentence processing. Applied Psycholinguistics, 33, 419–456.
228
19
SENTENCE PROCESSING
Cognitive approaches to second language
morphosyntactic and morphological processing
Nuria Sagarra
Introduction
The human brain receives massive amounts of information every minute, and working memory
(WM) limits how much data can be computed in real time (see Li, 2023 [this volume]). The brain
selects what is processed following probabilistic associations based on perceptual salience, redundancy,
frequency, and consistency. In language, probabilistic associations are fundamental for processing
(Ellis, 2006; MacWhinney, 2012; Romberg & Safran, 2010) and prediction (Kuperberg & Jaeger,
2016). According to usage-based models, language is a learned skill like any other skill developed and
honed through experience, and frst language (L1) and second language (L2) processing depend on
the same cognitive mechanisms ranging over probabilistic associations.
Currently, cognitive approaches to L2 morphosyntactic and morphological processing investigate
how physical, psychological, and experiential factors shape adults’ ability to compute new and enduring
L2 probabilistic patterns under the time demands typical of real-time processing, by means of online
methods such as self-paced reading, eye-tracking, and event-related potentials (ERPs) (see Pliatsikas &
Marinis, 2020, for an overview of the benefts of online methods in SLA research). Ofine methods
typical of earlier usage-based studies focus on the product rather than the process. This is problematic
because they measure responses after the entire sentence has been processed. Also, ofine methods
provide learners with time to think about the answer. This is troublesome because they can easily assess
conscious explicit knowledge.
This chapter reviews current key topics in cognitive research focusing on morphosyntactic
processing, i.e., morphological associations between words, like subject-verb number agreement
and determiner-noun gender agreement, and morphological processing, i.e., morphologi-
cal associations within words, like stress-sufx and tone-sufx tense associations in Spanish and
Swedish verbs. Morphosyntactic associations are explicitly taught and widely studied, whereas
morphological associations are implicitly learned and have barely been investigated until recently.
The emphasis will be on salience, contingency, and learned-attention explanations to L2 mor-
phosyntactic and morphological processing. The chapter begins with a summary of critical issues
in frequency-based models, followed by a section laying the groundwork for empirical stud-
ies examining these issues. The chapter ends with an examination of current trends and future
directions.
DOI: 10.4324/9781003018872-22 229

Nuria Sagarra
Associative learning: The process of acquiring new information via the formation of connections between
two stimuli that happen together.
Cue availability: The number of times a cue is available over the times it is needed.
Cue reliability: The number of times a cue is correct over the total number of occurrences.
Cue validity: The product of cue reliability and cue availability.
Learned attention: Attention shifts due to language experience.
Blocking: Inattention to further associations due to a prior association, caused by learned attention.
Salience: The ability of a stimulus to stand out from the rest of the input and attract attention.
Working memory: The activation and maintenance of short-lived information while learners perform
complex and time-consuming cognitive tasks (Barrouillet & Camos, 2007).
In the 1970s, studies with monolingual children (e.g., Brown, 1973) and adult L2 learners (e.g., Gass,
1979) revealed that L1 and L2 morphology is acquired systematically, but L2 morphology often follows
a diferent order and shows persistent variability in L2 learners compared to children. In the 1980s and
1990s, the competition model sparked a large number of studies examining cue competition related
to word order, case marking, agreement, intonation, and verb-based expectations (see MacWhinney,
2012, for a review). In 1986, Rumelhart and McClelland laid out the tenets of connectionist models,
a class of computational, neural-network models based on learning through graded, malleable, and
associative experiences. The early 2000s witnessed the consolidation of associative learning (Kruschke
& Blair, 2000), which claims that all species learn through unconscious probabilistic associations that
facilitate successful decision-making, an essential precursor of survival and performance. In 2006, Nick
Ellis applied the associative learning principles to language, in a model conceiving SLA as construction-
based involving the acquisition of form-meaning constructions based on prior experience, rational in
its refection of prior L1 usage, exemplar-driven, emergent in that few and simple learning mechanisms
conspire to form complex, dynamic, and adaptive knowledge networks, and dialectic (CREED). In
this framework, probabilistic associations depend on physical characteristics (cue salience and redun-
dancy), psychological characteristics (cue availability and reliability), and experiential characteristics,
such as cue entrenchment, caused by frequency, regularity, and acquisition order—frst learned cues
block later ones. Physical and psychological characteristics determine why some stimuli are attended
to more than others, whereas experiential characteristics defne cue attention based on an individual’s
language history. Finally, WM is associated with attention to less salient cues.
Physical characteristics
Perceptual salience refers to the ability of a stimulus to stand out from the rest and to attract attention
by virtue of physical characteristics (Ellis, 2019; Styles, 2006). Perceptual salience infuences all areas
of L1 and L2 acquisition (see Dube et al., 2019, for a recent review), and it plays a crucial role in L2
morphosyntactic and morphological processing.
In L2 morphosyntactic processing, attention increases with shorter rather than longer distance
between the constituents (e.g., Foote, 2011; Biondo et al., 2018), sufxes preceding rather than
following lexical cues conveying the same meaning (VanPatten et al., 2012), and salient lexical
cues rather than non-salient morphological ones. This lexical bias could also be a result of lexical
230
Sentence processing
cues often preceding morphological cues, although Ellis and Sagarra (2010a, 2011) reported that
learners’ reliance on equally salient lexical and morphological cues was unrelated to cue posi-
tion. Lexical bias could also be due to universal L2 processing strategies (VanPatten, 1996), even
though L1 efects have been reported (Sagarra & Ellis, 2013). Finally, lexical bias could emerge
from lexical cues being physically long and oftentimes stressed, (Ellis, 2006) versus morphologi-
cal cues being short, fused with stem verbs (bound), and often unstressed (Cutler & Carter, 1987)
(compare yesterday with walk-ed). Regarding L2 morphological processing, attention increases
with free rather than bound morphology (e.g., Behney et al., 2018; VanPatten, 1996), stressed
rather than unstressed morphemes (DeKeyser et al., 2018), long rather than short sufxes (Simoens
et al., 2018), and words at the beginning and at the end of sentences rather than in the middle
(Barcroft & VanPatten, 1997).
Psychological characteristics
Strong probabilistic associations involve a high number of examples (high availability) and few or
no exceptions (high reliability). Consider grammatical gender agreement in Spanish. Hispano-
phones can use lexical and syntactic probabilistic associations to compute gender agreement. The
lexical probabilistic associations link specifc noun endings to gender classes in gender transpar-
ent nouns: “-o” designates masculine gender (carro “carMASC.SG”) and “-a” feminine gender (mesa
“tableFEM.SG”); in fact, 66% of masculine singular Spanish nouns and 70% of feminine singular
Spanish nouns show these sufxes (medium availability). Of these nouns, the association is true
99.8% (masculine) and 96.3% (feminine) of the times (high reliability), but occurs only in the
input 62% (masculine) and 70% (feminine), leading to low reliability (Harris, 1991). In contrast,
opaque nouns have endings not associated with a particular gender class (reloj “watchMASC.SG”) and
require knowledge of a noun’s gender or, when available, the article-gender class syntactic associa-
tion. Syntactic associations connect specifc article endings to gender classes: Articles ending in a
consonant (el/un “the/aMASC”) denote that the noun will be masculine and those ending in “-a”
(la/una “the/aFEM”) denote that it will be feminine. This association is highly frequent (high avail-
ability), but there are nearly 600 exceptions of masculine nouns ending in “-a” (díaMASC “dayMASC”;
Harris, 1991) (low reliability).
Cognitive models claim that access to a noun’s gender depends on the reliability of noun
endings in marking gender, and that agreement markers are used when noun endings are gen-
der-ambiguous. Although monolingual adults focus on reliable cues, monolingual children and
adult L2 learners initially prefer available cues, and gradually settle on reliable cues as language
profciency increases (MacWhinney, 2012). However, the low perceptual salience and low reli-
ability of morphological cues afect L1 and L2 acquisition alike and cannot explain diferences
in child-adult acquisition. Instead, cognitive models attribute L2 learners’ difculty in acquiring
new L2 morphology to experiential diferences in the learning history which afect the allocation
of attention.
Experiential characteristics
Associative learning theory (e.g., Kamin, 1969; Kruschke & Blair, 2000) assumes that learning that
a cue is associated with an outcome (e.g., a bell predicting food) blocks the subsequent learning of
other cues paired with the same outcome (e.g., a light predicting food). Blocking is a robust and
ubiquitous phenomenon across species and is an efect of learned attention. Learned attention is the
attention shifting to certain cues as a result of language experience and depends on language experi-
ence with the L1 (transfer) and the L2 (profciency). The accumulation of L1 and L2 experiences
forms the processes and attentional biases that constitute the L2 system.
231
Nuria Sagarra
L1 transfer
Transfer plays a paramount role in cognitive models, which claim that properties similar in the L1
and the L2 depend on the same cognitive and neural mechanisms (e.g., Bates & MacWhinney, 1982;
Ellis, 2006). However, most studies investigating L1 transfer efects on L2 morphosyntactic and mor-
phological processing lack groups with diferent L1s and are unable to link their fndings conclusively
to cross-linguistic diferences. The few L2 morphosyntactic processing studies comparing learners
of varying profciencies whose L1 has or lacks a linguistic phenomenon indicate that beginning
learners opt for the least efortful interpretation independently of their L1, unless they have a high
WM span (Sagarra, 2008). Apparently, the scarcity of cognitive resources makes beginning learners
initially select salient cues (e.g., adverbs, quantifers) before non-salient cues (e.g., sufxes; VanPat-
ten, 1996). At intermediate and advanced levels, transfer efects appear and WM efects dissipate,
and at very advanced and near-native levels transfer vanishes because of extensive L2 knowledge (see
Sagarra, 2020, for a review). This proposal is supported by studies employing diferent techniques
(e.g., self-paced reading, eye-tracking, ERPs), structures (e.g., adverb-verb tense agreement, subject-
verb number agreement, noun-adjective gender/number agreement), populations (e.g., adult early
and late bilinguals), L1s (e.g., Chinese, English, Latin, Spanish, Romanian, Russian), and L2s (e.g.,
Spanish, English). Regarding transfer and L2 morphological processing, Gosselke Berthelsen and
colleagues (2020) found that beginning learners with a tonal L1 are sensitive earlier to tone-sufx
associations during spoken word recognition than non-tonal learners.
L2 profciency
Profciency infuences L2 morphosyntactic and morphological processing in the verbal and nominal
domain. For L2 morphosyntactic processing in verbal agreement, behavioral and neurocognitive data
indicate that learners are sensitive to subject-verb number agreement violations at higher profciency
(Rodríguez & Reglero, 2015; Yao & Chen, 2017), but delayed sensitivity (Rossi et al., 2006) or
no sensitivity (Sagarra, 2014) at lower profciency. Also, higher profciency goes hand in hand with
increased attention to morphological cues and decreased attention to lexical cues (Armstrong et al.,
2018; Sagarra, 2020; Sagarra & Ellis, 2013). In the nominal domain, learners are sensitive to gender
and number agreement violations at higher but not lower profciency levels (e.g., Keating, 2010;
Morgan-Short et al., 2010; Sagarra & Herschensohn, 2010). Similarly, higher profciency learners
consistently use gender and number sufxes to anticipate an upcoming noun (Dussias et al., 2013;
Hopp, 2013), whereas lower profciency learners show variability (Hopp, 2016; Lew-Williams &
Fernald, 2010). Furthermore, during morphosyntactic processing, beginning and intermediate L2
learners activate distinct neural representations (Dehaene et al., 1997), whereas monolinguals and
advanced learners activate similar brain areas (Vingerhoets et al., 2003). Profciency also afects L2
morphological processing. In the verbal domain, advanced, but not beginning, learners make stress-
sufx present-past verb associations (Sagarra & Casillas, 2018). In the nominal domain, non-tonal
advanced learners make tone-sufx singular-plural noun associations but those with longer L2 expe-
rience show more native-like processing (Schremm et al., 2016).
In sum, L1 transfer and L2 profciency infuence L2 morphosyntactic and L2 morphological pro-
cessing and depend on each other in that L1 transfer is more pronounced at lower profciency levels.
Profciency is also associated with WM efects (see Mitchell et al., 2015, for a review).
Cognitive characteristics
Two pieces of evidence suggest that L2 morphosyntactic processing may be cognitively demanding.
Firstly, Dutch monolinguals with low WM spans made more agreement errors than those with high
232
Sentence processing
WM spans, but WM did not afect their production of quantifers (Hartsuiker & Barkhuysen, 2006).
These results indicate that morphological cues are cognitively more taxing than lexical cues, even for
monolinguals. Second, miniature Latin learners attended more to lexical than morphological cues
when linguistic complexity was increased, independently of their L1 (Sagarra & Ellis, 2013). Across
domains, many studies show that high WM span correlates with more native-like L2 morphosyntac-
tic processing across profciency levels, but a few studies show no WM efects in advanced learners
(see Sagarra, 2020, for a review). This discrepancy may be due to the advanced learners of the stud-
ies showing no WM efects being more advanced than those revealing WM efects. In addition to
profciency, WM is associated with L1 transfer, because WM efects are stronger in learners whose
L1 difers more from the L2 (Sagarra, 2020). In sum, attention to infectional morphology may be
cognitively taxing, yet WM is not a stand-alone factor, since it clearly interacts with a person’s L1
and L2 history.

In this section, I review studies providing evidence about the role of physical, psychological, expe-
riential, and WM characteristics on the development of L2 morphosyntactic and morphological
associations during real-time processing. Importantly, not all the studies ascribe to cognitive models,
yet their results lend themselves to interpretation within this framework.
Tense
Time is fundamental to human cognition and action. L2 learners initially mark temporal refer-
ence with adverbs, prepositional phrases, serialization, and calendric reference, with verbal sufxes
appearing only slowly thereafter, if at all. Associative learning phenomena of learned attention and
blocking have been widely investigated with adverb-verb tense associations. For instance, Ellis and
Sagarra (2010a) explored learned attention efects in two experiments with a subset of Latin. In
phase 1 (training), English speakers received adverb training (seeing hodie “today” and heri “yesterday”
36 times), verb training (seeing cogito “I think” and cogitavi “I thought” 36 times), or no training.
In phase 2 (learning), all groups saw congruent adverb-verb/verb-adverb combinations in the past,
present, and future tenses (e.g., seeing heri cogitavi), determined the tense, and received feedback
(e.g., if they chose past, they saw Correct!, but if they chose present or future, they saw Incorrect. Heri
cogitavi is past). In phase 3 (a perception test), all groups saw Latin adverbs, verbs, and congruent/
incongruent adverb-verb/verb-adverb combinations and provided a rating on a scale of 1 (past) to
5 (future) (e.g., hodie cogito—3). The phase 3 data with incongruent trials showed that adverb train-
ing participants relied on adverbs 97% of the times, verb pretraining participants 12%, and controls
60% (see Ellis et al., 2014, for an eye-tracking study, partially replicating these fndings). In phase 4
(a production/translation test), everybody saw English adverbs, verbs, and congruent/incongruent
adverb-verb/verb-adverb combinations (e.g., hodie cogito) and typed the Latin translation (e.g., today
I think). Experiment 1 explored short-term learned attention efects. The phase 4 data revealed that
adverb training participants overused adverbs, while verb training participants were more accurate on
verb translations, and controls were unbiased. Experiment 2 revealed that a new group of Chinese
controls relied more on adverbs and produced more target-like adverbs and less target-like verbs
than the English controls of experiment 1, arguably because Chinese lacks tense morphology. These
experiments comprised lexical and morphological cues that were equally salient, available, and reli-
able, and they can therefore disentangle learned attention from salience efects. However, lexical cues
are often more salient, available, and reliable than morphological cues. This cue imbalance could
explain adults’ prolonged difculty acquiring L2 infectional morphology but cannot account for
why monolingual children are not afected or why the English controls were not lexically biased.
233
Nuria Sagarra
Several explanations may be possible. Firstly, children attend more to morphological than lexical cues
because they do not fully understand the meaning of temporal adverbs, and they lack experience
showing that lexical cues are more salient, available, and reliable than morphological cues. Secondly,
the English controls may have been adverb-biased due to the stimuli’s oversimplicity. Indeed, Ellis
and Sagarra’s (2011) replication of their 2010a study adding second- and third-person singular condi-
tions revealed that all learners became lexical. These studies suggest that L2 entrenchment acquired
via repeated exposure to L2 cues is stronger than the blocking originating from years of experience
with the L1.
The studies described here can speak to associative learning of miniature but not complete lan-
guages. In Sagarra’s (2008) study, beginning English-Spanish learners read L2 sentences containing
adverb-verb/verb-adverb congruencies/incongruencies and were unable to process sufxes unless
they had high WM span. To determine whether this null efect was due to the cognitive demands
typical of non-cumulative moving-window tasks, Ellis and Sagarra (2010b) replicated the study using
eye-tracking. They found that Spanish monolinguals looked longer at verbs in incongruent than in
congruent sentences, but English monolinguals and intermediate English-Spanish learners looked
longer at adverbs in incongruent than in congruent sentences, and the latter relied more on adverbs
than English monolinguals. To discover whether these fndings were due to the learners’ L1 or to
insufcient L2 experience, LaBrozzi (2009) compared the intermediate group with a new intermedi-
ate group that had immersion experience. Both groups were sensitive to adverb-verb/verb-adverb
incongruencies, but immersed learners relied more on morphological cues than non-immersed
learners, and only high-WM immersed learners looked longer at verbs in adverb-verb incongruen-
cies than adverb-verb congruencies. Finally, to ascertain whether the adverb bias observed in Ellis
and Sagarra’s (2010b) intermediate English-Spanish learners was a result of a default or an L1-based
strategy, Sagarra and Ellis (2013) replicated the study on Latin with intermediate and advanced
English-Spanish and Romanian-Spanish learners. Romanian and Spanish are morphologically richer
than English. In this study, all groups were sensitive to tense incongruencies, but Romanian-Spanish
learners and Romanian monolinguals looked longer at verbs than the rest.
Regarding L2 morphological processing of tense, advanced, but not beginning, L2 learners
use prosodic cues that are absent or used diferently in their L1 to predict verb tense during sen-
tence processing. For instance, a visual world paradigm task and a gating task showed that Spanish
monolinguals and advanced, but not beginning, Anglophones use stress to predict sufxes in L2
Spanish regular verbs (stressed frst-syllable present: lim- pia “cleansPRES”; unstressed frst-syllable
past: lim-pió “cleanedPAST”; Sagarra & Casillas, 2018). To determine whether the advanced learn-
ers were superior to the beginners due to higher L2 profciency or to increased anticipatory expe-
rience, Lozano-Argüelles et al. (2019) compared advanced learners with and without interpreting
experience on the assumption that interpreters have more anticipatory experience. The results
revealed that monolinguals anticipate earlier than interpreters and non-interpreters but interpret-
ers anticipate at the fastest rate, and interpreters anticipate earlier than non-interpreters in all
conditions except in unstressed, CV conditions. Furthermore, an ERP verb identifcation task
revealed that Swedish monolinguals and advanced non-tone L1 learners of Swedish use Swedish
tone-sufx associations for tense marking (low tone present: häll-er “pourPRES”; high tone past:
häll-de “pourPAST”; Schremm et al., 2016). Crucially, mismatch-match diferences are bigger in
monolinguals than learners, and learners with more than less language experience. Finally, WM
did not afect the use of stress-sufx associations in Spanish monolinguals or in advanced or
beginning English-Spanish learners in an eye-tracking task. Yet, higher WM increased accuracy
in a gating task in all groups.
Taken together, these studies on tense demonstrate that language experience (L1 transfer, L2 pro-
fciency, immersion), cognitive complexity, and WM afect L2 morphosyntactic and morphological
processing.
234
Sentence processing
Number
Adverb-verb tense congruencies do not constitute morphosyntactic agreement relations as such and
can only speak to morphological processing accompanied, and often overshadowed, by more salient
lexical competitors. Subject-verb number agreement allows researchers to investigate the role of
perceptual salience, language experience, and WM on L2 morphosyntactic processing in the absence
of lexical cues.
As for perceptual salience, intermediate and advanced learners attend more to salient verbs (long
or stressed sufxes) than non-salient verbs (Sagarra, 2020). In addition, non-advanced learners are
sensitive to verb-subject person-number violations, in which the verb is not redundant because it
appears frst, but not to subject-verb person-number violations, in which the verb is redundant
because it appears after the adverb (VanPatten et al., 2012)
Regarding L1 transfer efects, advanced Chinese-English learners show no sensitivity (Jiang,
2004) or non-native-like brain responses (Armstrong et al., 2018) to subject-verb number violations
in English, and intermediate English-French learners are insensitive to similar violations in French
(Reichle et al., 2013). This pattern may arise because Chinese lacks number morphology and English
only has it in third-person singular of the present tense. However, these studies lack L2 learners of a
morphologically rich L1 and cannot determine whether their fndings are due to L1 transfer. Studies
comparing learners of morphologically null/poor and rich L1s suggest transfer efects at intermediate
and advanced levels, but not at beginning, very advanced, and near-native levels. For instance, both
intermediate and advanced English-Spanish and Romanian-Spanish learners and Spanish monolin-
guals are sensitive to subject-verb number violations, but the L1 English learners regress more to
subjects (Sagarra, 2020). This may be due to the lower degree of morphological richness in English.
In contrast, beginning learners of the same L1-L2 pairs are insensitive to agreement violations, while
very advanced and near-native English, Dutch, and Russian learners of German are sensitive (Hopp,
2010). These fndings underscore that L1 transfer efects vary with L2 profciency.
Concerning L2 profciency efects in the processing of subject-verb number violations, some
studies show sensitivity in intermediate learners (eye-tracking task: Sagarra, 2020; ERP task: Rossi
et al., 2006), advanced learners (eye-tracking task: Sagarra, 2020; Yao & Chen, 2017; self-paced
reading task: Rodríguez & Reglero, 2015; ERP task: Rossi et al., 2006; speeded grammaticality
judgments: Hopp, 2013), and near-native learners (self-paced reading task: Hopp, 2010; Jegerski,
2016). In contrast, other studies reveal no sensitivity in beginning learners (eye-tracking task: Sagarra,
2014), intermediate learners (self-paced reading task: Rodríguez & Reglero, 2015; VanPatten et al.,
2012; Yao & Chen, 2017; ERP task: Reichle et al., 2013), and advanced learners (self-paced reading
task: Yao & Chen, 2017; Jegerski, 2016; ERP task: Chen et al., 2007). Overall, these studies show
sensitivity in near-natives and no sensitivity in beginners. The inconsistent fndings in intermediates
and advanced may be related to the amalgamated infuence of L1 transfer (L1 Chinese learners are
always insensitive), task demands (simpler tasks with cumulative displays like eye-tracking steadily
produce sensitivity efects), cognitive load (Rodríguez & Reglero and VanPatten et al. mixed person
and number agreement violations), and cognitive resources (WM).
As for WM efects, some studies show sensitivity to subject-verb agreement violations in eye-
tracking, yet not self-paced reading (Yao & Chen, 2017), increased sensitivity in intermediate
immersed high-WM than immersed low-WM and non-immersed high- and low-WM learners
(Sagarra & LaBrozzi, 2018), and earlier sensitivity (sensitivity during frst vs. second pass reading) in
intermediate and advanced high- than low-WM learners (Sagarra, 2020).
Studies exploring nominal L2 morphological associations provide further evidence of language
experience efects. ERP data indicated that Swedish speakers, but not beginning German-Swedish
learners, employ Swedish tone-sufx noun associations (low-tone singular: bil-en “carSING”; high-
tone plural: bil-ar “carPL”; Gosselke Berthelsen et al., 2018).
235
Nuria Sagarra
To discover whether beginners’ lack of sensitivity resulted from the lack of tones in their L1, tonal
and non-tonal learners of Swedish learned artifcial word tones related to gender and number suf-
fxes (Gosselke et al., 2020). ERP data revealed delayed processing of tone-sufx associations in both
groups, but early processing only in the tonal group. Notably, training increased accuracy and short-
ened RTs (Schremm et al., 2017), but training did not produce a P600 efect and did not shorten
RTs for invalid trials, suggesting non-native predictive processing persisted even after training (Hed
et al., 2019).
In sum, L2 morphosyntactic and morphological processing and prediction of number depend on
language and anticipatory experience and beneft from input-based training.
Gender
The studies on tense and number speak to verbal, but not nominal, L2 morphosyntactic associations
like gender agreement. Regarding L2 profciency efects, lower-profciency learners are insensitive
to gender agreement violations (e.g., L2 English: Weber-Fox & Neville, 1996; L2 French: Osterhout
et al., 2008; L2 German: Hahne & Friederici, 2001; L2 Spanish: Tokowicz & MacWhinney, 2005)
or process them as semantic anomalies (L2 Spanish: Morgan-Short et al., 2010). Also, higher- but
not lower-profciency Anglophones are sensitive to gender agreement violations across materials and
methods (L2 Spanish: self-paced reading: Sagarra & Herschensohn, 2010; eye-tracking: Keating,
2010; ERPs: Gabriele et al., 2013). Furthermore, length of L2 contact predicted changes in neural
response to noun-adjective gender disagreement within non-immersed, but not immersed, learners
(Faretta-Stutenberg & Morgan-Short, 2018).
Studies on predictive processing ofer further support for language experience efects during L2
gender agreement processing. For L1 transfer, beginning English-Dutch learners are more accurate
and faster, and fxate earlier at nouns preceded by an article that is similar between the two lan-
guages (dit–deze correspond to “this–these”) compared to articles that are diferent (het–de merged
into “theSG/PL”; Liburd, 2014). As for L2 profciency, advanced learners use an article’s gender to
anticipate a noun before hearing it (e.g., Dussias et al., 2013; Grüter et al., 2012; Hopp, 2013), but
they are slower than natives, and intermediates are unable to make article-noun gender predictions
(Lew-Williams & Fernald, 2010). The difculty processing gender agreement and the overreliance
on lexical and syntactic distributional regularities typical of low-profciency L2 learners of gender-
less L1s seem to be related to weak and unstable sufx-gender associations. Two pieces of evidence
support this claim. Firstly, some studies found a contingency between speed of lexical access and
predictive syntactic gender agreement (Hopp, 2013). Secondly, L2 learners have greater difculty
predicting syntactic gender agreement with familiar nouns than novel nouns learned through immer-
sion involving typical cue-sufx associations linking a specifc sufx to a specifc gender (Grüter
et al., 2012).
The previous paragraph covered gender agreement, that is L2 syntactic gender associations. Evi-
dence of L2 lexical sufx-to-gender associations can also be found in studies showing that L2 learners
are faster and more accurate at assigning gender to transparent than to opaque nouns (L2 German:
Bordag et al., 2006; L2 Russian: Taraban & Kempe, 1999; L2 Spanish: Cafarra et al., 2017). Impor-
tantly, higher use of Spanish on a daily basis increases sensitivity to gender agreement violations with
opaque, but not transparent, nouns in early bilinguals (Cafarra et al., 2017). These fndings are in
line with cognitive models, which posit that lower experience produces probabilistic associations,
and higher experience generates abstract lexical representations.
Finally, studies show WM efects on L2 gender agreement processing in lower- (Faretta-
Stutenberg, & Morgan-Short, 2018), but not higher-profciency learners (Sagarra & Herschensohn,
2010), unless gender agreement is cognitively taxing when presented across non-adjacent constitu-
ents (Keating, 2010).
236
Sentence processing

This chapter reviewed cognitive approaches to L2 morphosyntactic processing investigating physi-
cal, psychological, experiential, and cognitive factors underlying adult bilinguals’ difculties when
processing infectional morphology in their non-dominant language. Yet, numerous studies have
important limitations or have failed to address important questions (see Textbox 19.2).
Firstly, many studies either mix salient and non-salient morphemes, lack a group whose L1 is
similar to the L2, focus on a single profciency level, do not use WM tests, or employ cognitively
taxing tasks or tasks with explicit judgments that can afect fndings and performance during real-
time comprehension. More studies that test salient and non-salient morphemes separately, compare
L2 learners with an L2 similar to and diferent from their L1, include a wide range of profciency
levels, and add a WM test are needed.
Secondly, most studies compare L2 learners to monolinguals and cannot unravel processing
patterns typical of adult L2ers from those representative of other types of bilinguals. Online and
ofine studies comparing early and late bilinguals repeatedly reveal that neither group behaves
like monolinguals (e.g., Rodríguez & Reglero, 2015), suggesting that L2 learners’ non-nativeness
is partially due to the simple fact of juggling multiple languages in a single mind. Surprisingly,
studies examining early and late bilinguals’ morphological and morphosyntactic processing are
rare. The feld of SLA is in desperate need of early bilingual data to fully understand L2 process-
ing patterns.
Thirdly, cognitive approaches attribute monolingual young children’s preference for morpho-
logical over lexical cues to insufcient language and world experience. However, this explanation
is based on a few ofine studies with English monolingual children (e.g., Valian, 2006). We need
monolingual and bilingual child studies comparing lexical and morphological cues encoding the
same meaning longitudinally in diferent languages. Such studies should employ online methodolo-
gies with lexical and morphological cues that are equally salient, available, and reliable, as well as
extensive naturalistic corpora.
Fourthly, L2 learners have persistent difculty processing morphosyntactic agreement. Recent
studies show that lower profciency learners have weaker lexical representations (Grüter et al., 2012)
and over-rely on distributional regularities (see Cafarra et al., 2017), and that gender lexical rep-
resentations increase with profciency (Sagarra, 2020). However, additional studies are necessary to
determine which cues L2 learners employ to compute agreement in real time.
Fifthly, the last years have witnessed an explosion of L2 studies exploring predictive processing.
Predictive processing depends on cue-outcome associations gradually acquired through frequency
and consistency. Monolingual children and adults use a myriad of cues to predict, but L2 studies are
scarce and inconclusive and those with early bilinguals are virtually non-existent (yet see Lemmerth
& Hopp, 2019). L2 predictive processing studies are crucial to identify what causes L2 morphosyn-
tactic and morphological difculties.
Finally, cognitive accounts relate L1-L2 diferences to insufcient input exposure and low percep-
tual salience. The missing opportunities to attend to morphological cues and wrong/missing predic-
tions can hinder L2 learning. Some approaches to L2 instruction suggest solving this problem by
training learners via grammar instruction (Cintrón-Valentín & Ellis, 2016), input enhancement (Ellis
& Sagarra, 2010b), and repeated exposure to distributional regularities (Schremm et al., 2017) and
infectional morphology (Ellis & Sagarra, 2010b). L2 training studies along these lines can provide a
more fne-grained guide to help learners attend to infectional morphology.
Cognitive approaches to L2 sentence processing have opened a fascinating window into bilinguals’
mental processes. However, language cognition is not just “in the head” (Ellis, 2019). An integrated
science of language that amalgamates diferent language disciplines will provide a more holistic view
of SLA.
237
Nuria Sagarra
Which processing patterns are unique to L2 speakers and which are common to bilinguals?
Why do children favor morphological over lexical cues?
Do adults use abstract lexical representations or distributional associations during L2 morphosyntactic
processing? Do they transfer the cues used in their L1?
What cues do monolingual and bilingual children and adults use to make morphosyntactic and morpho-
lexical predictions during L2 sentence processing?
Can we train learners to make faster and better predictions?
Further reading
Gass, S. M., Spinner, P., & Behney, J. (2018). Salience in second language. Routledge.
This edited volume on salience in second language acquisition contains several chapters on perceptual salience.
MacWhinney, B. (2018). A unifed model of frst and second language learning. In M. Hickmann, E. Veneziano,
& H. Jisa (Eds.), Sources of variation in frst language acquisition: Languages, contexts, and learners (pp. 287–312).
John Benjamins.
This chapter provides a recent overview of the Unifed Competition Model.
Wulf, S., & Ellis, N. C. (2018). Usage-based approaches to second language acquisition. In D. Miller, F. Bayram,
J. Rothman, & L. Serratrice (Eds.), Bilingual cognition and language: The state of the science across its subfelds
This chapter reviews usage-based models to second language acquisition, focusing on associative-learning
explanations of L2 morphology acquisition.
References
Armstrong, A., Bulkes, N., & Tanner, D. (2018). Quantifcational cues modulate the processing of English
subject-verb agreement by native Chinese speakers: An ERP study. Studies in Second Language Acquisition,
20, 731–754.
Barcroft, J., & VanPatten, B. (1997). Acoustic salience of grammatical forms: The efect of location, stress, and
boundedness on Spanish L2 input processing. In W. Glass & A. Teresa Pérez-Leroux (Eds.), Contemporary
perspectives on the acquisition of Spanish (Vol. 2, pp. 109–121). Cascadilla Press.
Barrouillet, P., & Camos, V. (2007). The time-based resource-sharing model of working memory. In N. Osaka,
R. H. Logie, R. Logie, & M. D’Esposito (Eds.), The cognitive neuroscience of working memory (pp. 59–80).
Oxford University Press.
Bates, E., & MacWhinney, B. (1982). Functionalist approaches to grammar. In E. Wanner & L. Gleitmann
(Eds.), Language acquisition: The state of the art (pp. 173–218). Cambridge University Press.
Behney, J., Spinner, P., Gass, S. M., & Valmori, L. (2018). The L2 acquisition of Italian tense. The role of salience.
In S. M. Gass, P. Spinner, & J. Behney (Eds.), Salience in second language acquisition (pp. 21–40). Routledge.
Biondo, N., Vespignani, F., Rizzi, L., & Mancini, S. (2018). Widening agreement processing: A matter of time,
features and distance. Language, Cognition and Neuroscience, 33(7), 890–911.
Bordag, D., Opitz, A., & Pechmann, T. (2006). Gender processing in frst and second languages. The role of
noun termination. Journal of Experimental Psychology: Learning, Memory, and Cognition, 32, 1090–1101.
Brown, R. (1973). A frst language: The early stages. Harvard University Press.
Cafarra, S., Barber, H., Molinaro, N., & Carreiras, M. (2017). When the end matters: Infuence of gender cues
during agreement computation in bilinguals. Language, Cognition and Neuroscience, 32(9), 1069–1085.
Chen, L., Shu, H. U. A., Liu, Y., Zhao, J., & Li, P. (2007). ERP signatures of subject—verb agreement in L2
learning. Bilingualism: Language and Cognition, 10(2), 161–174.
Cintrón-Valentín, M. C., & Ellis, N. C. (2016). Salience in second language acquisition: Physical form, learner
attention, and instructional focus. Frontiers in Psychology, 7, 1284.
238
Sentence processing
Cutler, A., & Carter, D. M. (1987). The predominance of strong initial syllables in the English vocabulary. Com-
puter Speech and Language, 2, 133–142.
Dehaene, S., Dupoux, E., Mehler, J., Cohen, L., Paulesu, E., Perani, D., & Le Bihan, D. (1997). Anatomical
variability in the cortical representation of frst and second language. Neuroreport, 8(17), 3809–3815.
DeKeyser, R. M., Alf-Shabtay, I., Ravid, D., & Shi, M. (2018). The role of salience in the acquisition of He-
brew as a second language: Interaction with age of acquisition. In S. M. Gass, P. Spinner, & J. Behney (Eds.),
Salience in second language acquisition (pp. 131–146). Routledge.
Dube, S., Kung, C., Brock, J., & Demuth, K. (2019). Perceptual salience and the processing of subject-verb
agreement in 9–11-year old English-speaking children: Evidenced from ERPs. Language Acquisition, 26(1),
73–96.
Dussias, P., Valdés Krof, J., Guzzardo Tamargo, R., & Gerfen, C. (2013). When gender and looking go hand
and hand: Grammatical gender processing L2 Spanish. Studies in Second Language Acquisition, 35, 353–387.
Ellis, N. C. (2006). The associative-cognitive CREED. In B. VanPatten & J. Williams (Eds.), Theories in second
language acquisition: An introduction (pp. 77–95). Earlbaum.
Ellis, N. C. (2019). Essentials of a theory of language cognition. The Modern Language Journal, 103(1), 39–60.
Ellis, N. C., Hafeez, K., Martin, K. I., Chen, L., Boland, J., & Sagarra, N. (2014). Learned attention in adult
language acquisition: Overt then covert. Applied Psycholinguistics, 35(3), 547–579.
Ellis, N. C., & Sagarra, N. (2010a). The bounds of adult language acquisition: Blocking and learned attention.
Studies in Second Language Acquisition, 32(4), 1–28.
Ellis, N. C., & Sagarra, N. (2010b). Learned attention efects in L2 temporal reference: The frst hour and the
next eight semesters. Language Learning, 60(2), 85–108.
Ellis, N. C., & Sagarra, N. (2011). Learned attention in adult language acquisition: A replication and generaliza-
tion study and meta-analysis. Studies in Second Language Acquisition, 33, 589–624.
Faretta-Stutenberg, M., & Morgan-Short, K. (2018). The interplay of individual diferences and context of
learning in behavioral and neurocognitive second language development. Second Language Research, 34(1),
67–101.
Foote, R. (2011). Integrated knowledge of agreement in early and late English-Spanish bilinguals. Applied Psy-
cholinguistics, 32, 187–220.
Gabriele, A., Fiorentino, R., & Alemán Bañón, J. (2013). Examining second language development using event-
related potentials: A cross-sectional study on the processing of gender and number agreement. Linguistic
Gass, S. (1979). Language transfer and universal grammatical relations. Language Learning, 29, 327–344.
Gosselke Berthelsen, S., Horne, M., Brännström, K. J., Shtyrov, Y., & Roll, M. (2018). Neural processing of
morphosyntactic tonal cues in second-language learners. Journal of Neurolinguistics, 45, 60–78.
Gosselke Berthelsen, S., Horne, M., Shryrov, Y., & Roll, M. (2020). Diferent neural mechanisms for rapid
acquisition of words with grammatical tone in learners from tonal and non-tonal backgrounds: ERP evi-
dence. Brain Research, 1729(146614), 1–15.
Grüter, T., Lew-Williams, C., & Fernald, A. (2012). Grammatical gender in L2: A production or real-time
processing problem? Second Language Research, 28, 191–215.
Hahne, A., & Friederici, A. D. (2001). Processing a second language: Late learners’ comprehension mechanisms
as revealed by event-related brain potentials. Bilingualism: Language and Cognition, 4, 123–141.
Harris, J. (1991). The exponence of gender in Spanish. Linguistic Inquiry, 22, 27–62.
Hartsuiker, R. J., & Barkhuysen, P. N. (2006). Language production and working memory: The case of subject-
verb agreement. Language and Cognitive Processes, 21, 181–204.
Hed, A., Schremm, A., Horne, M., & Roll, M. (2019). Neural correlates of second language acquisition of
tone-grammar associations. The Mental Lexicon, 14(1), 98–123.
Hopp, H. (2010). Ultimate attainment in L2 infection: Performance similarities between non-native and native
speakers. Lingua, 120(4), 901–931.
Hopp, H. (2013). Grammatical gender in adult L2 acquisition: Relations between lexical and syntactic variabil-
ity. Second Language Research, 29(1), 33–56.
Hopp, H. (2016). The timing of lexical and syntactic processes in second language sentence comprehension.
Applied Psycholinguistics, 37(5), 1253.
Jegerski, J. (2016). Number attraction efects in near-native Spanish sentence comprehension. Studies in Second
Jiang, N. (2004). Morphological insensitivity in second language processing. Applied Psycholinguistics, 25,
603–634.
Kamin, L. J. (1969). Predictability, surprise, attention, and conditioning. In B. A. Campbell & R. M. Church
(Eds.), Punishment and aversive behavior (pp. 276–296). Appleton-Century-Crofts.
239
Nuria Sagarra
Keating, G. D. (2010). The efects of linear distance and working memory on the processing of gender agree-
ment in Spanish. Research in Second Language Processing and Parsing, 53, 113–134.
Kruschke, J. K., & Blair, N. J. (2000). Blocking and backward blocking involve learned inattention. Psychonomic
Bulletin and Review, 7, 636–645.
Kuperberg, G. R., & Jaeger, T. F. (2016). What do we mean by prediction in language comprehension? Lan-
guage, Cognition and Neuroscience, 31, 32–59.
LaBrozzi, R. (2009). Processing lexical and morphological cues in a study abroad context [Unpublished doctoral
dissertation], Pennsylvania State University.
Lemmerth, N., & Hopp, H. (2019). Gender processing in simultaneous and successive bilingual children: Cross-
linguistic lexical and syntactic infuences. Language Acquisition, 26(1), 21–45.
Lew-Williams, C., & Fernald, A. (2010). Real-time processing of gender-marked articles by native and non-
native Spanish speakers. Journal of Memory and Language, 63, 447–464.
Li, S. (2023). Working memory and second language learning: a critical and synthetic review. In A. Godfroid
& H. Hopp (Eds.). The Routledge handbook of second language acquisition and psycholinguistics (pp. 348–360).
Routledge.
Liburd, T. (2014). Investigating prediction in L2 morpho-syntax: A visual world study [Unpublished doctoral disserta-
tion], University of Pittsburgh.
Lozano-Argüelles, C., Sagarra, N., & Casillas, J. V. (2019). Slowly but surely: Interpreting facilitates L2 morpho-
logical anticipation based on suprasegmental and segmental information. Bilingualism: Language and Cognition,
23(4), 752–762.
MacWhinney, B. (2012). The logic of the unifed model. In S. Gass & A. Mackey (Eds.), Handbook of second
Mitchell, A. E., Jarvis, S., O’Malley, M., & Konstantinova, I. (2015). Working memory measures and L2 prof-
ciency. In Z. Wen, M. Borges Mota, & A. McNeill (Eds.), Working memory in second language acquisition and
processing (pp. 270–284). Routledge.
Morgan-Short, K., Sanz, C., Steinhauer, K., & Ullman, M. T. (2010). Second language acquisition of gender
agreement in explicit and implicit training conditions: An event-related potential study. Language Learning,
60(1), 154–193.
Osterhout, L., Poliakov, A., Inoue, K., McLaughlin, J., Valentine, G., Pitkanen, I., Frenck-Mestre, C., & Her-
schensohn, J. (2008). Second-language learning and changes in the brain. Journal of Neurolinguistics, 21(6), 509–521.
Pliatsikas, C., & Marinis, T. (2020). Online psycholinguistic methods in second language acquisition research. In
C. A. Chapelle (Ed.), The concise encyclopedia of applied linguistics. Wiley. ISBN 9781119147367.
Reichle, R. V., Tremblay, A., & Coughlin, C. A. (2013). Working-memory capacity efects in the processing of
non-adjacent subject-verb agreement: An event-related brain potentials study. In E. Voss et al. (Eds.), Selected
proceedings of the 2011 second language research forum (pp. 54–69). Cascadilla Proceedings Project.
Rodríguez, E., & Reglero, L. (2015). Heritage and L2 processing of person and number features: Evidence from
Spanish subject-verb agreement. EuroAmerican Journal of Applied Linguists and Languages, 2(2), 11–30.
Romberg, A., & Safran, J. (2010). Statistical learning and language acquisition. Wiley Interdisciplinary Reviews.
Cognitive Science, 1(6), 906–914.
Rossi, S., Gugler, M. F., Friederici, A. D., & Hahne, A. (2006). The impact of profciency on syntactic second-
language processing of German and Italian: Evidence from event-related potentials. Journal of Cognitive Neu-
roscience, 18(12), 2030–2048.
Rumelhart, D. E., & McClelland, J. L. (1986). Parallel distributed processing: Explorations in the microstructure of
cognition. MIT Press.
Sagarra, N. (2008). Working memory and L2 processing of redundant grammatical forms. In Z. Han (Ed.),
Understanding second language process (pp. 133–147). Multilingual Matters.
Sagarra, N. (2014). Absence of morphological transfer in beginners. In Z. Han & R. Rast (Eds.), First exposure
to a second language: Learners’ initial input processing (pp. 139–170). Oxford University Press.
Sagarra, N. (2020). When more is better: Higher L1-L2 similarity, L2 profciency, and working memory facili-
tate L2 morphosyntactic processing. In M. Leeser, W. Wong, & G. Keating (Eds.), Research on second language
processing and processing instruction. Studies in Honor of Bill VanPatten. John Benjamins.
Sagarra, N., & Casillas, J. V. (2018). The use of stress in L1 and L2 lexical access. Special issue on Second Lan-
guage Processing (Ed. Jiang). Journal of Second Language Studies, 1(1), 31–60.
Sagarra, N., & Ellis, N. (2013). From seeing adverbs to seeing morphology. Language experience and adult
acquisition of L2 tense. Special issue “Eye tracking and SLA” (Eds. A. Godfroid, S. Gass, & P. Winke). Studies
Sagarra, N., & Herschensohn, J. (2010). The role of profciency and working memory in gender and number
agreement processing in L1 and L2 Spanish. Lingua, 20, 2022–2039.
240
Sentence processing
Sagarra, N., & LaBrozzi, R. (2018). Benefts of study abroad on L2 morphosyntactic processing, and of working
memory on L2 semantic processing. In C. Sanz & A. Morales (Eds.), The Routledge handbook of study abroad
research and practice (pp. 149–164). Routledge.
Schremm, A., Hed, A., Horne, M., & Roll, M. (2017). Training predictive L2 processing with a digital game:
Prototype promotes acquisition of anticipatory use of tone-sufx associations. Computers and Education, 114,
206–221.
Schremm, A., Söderström, P., Horne, M., & Roll, M. (2016). Implicit acquisition of tone-sufx connections in
L2 learners of Swedish. The Mental Lexicon, 11(1), 55–75.
Simoens, H., Housen, A., & De Cuypere, L. (2018). The efect of perceptual salience on processing L2 in-
fectional morphology. In S. M. Gass, P. Spinner, & J. Behney (Eds.), Salience in second language acquisition
Styles, E. A. (2006). The psychology of attention (2nd ed.). Psychology Press.
Taraban, R., & Kempe, V. (1999). Gender processing in native and nonnative Russian speakers. Applied Linguis-
tics, 20, 119–148.
Tokowicz, N., & MacWhinney, B. (2005). Implicit and explicit measures of sensitivity to violations in second
language grammar: An event-related potential investigation. Studies in Second Language Acquisition, 27(2),
173–204.
Valian, V. (2006). Young children’s understanding of present and past tense. Language Learning and Development,
4, 251–276.
VanPatten, B. (1996). Input processing and grammar instruction. Ablex Publishing Corporation.
VanPatten, B., Keating, G., & Leeser, M. (2012). Missing verbal infections as a representational problem: Evi-
dence from self-paced reading. Linguistic Approaches to Bilingualism, 2(2), 109–140.
Vingerhoets, G., Van Borsel, J., Tesink, C., Van den Noort, M., Dehblaere, K., Seurinck, R., Vandemaele, P., &
Achten, E. (2003). Multilingualism: An fMRI study. NeuroImage, 20, 2181–2196.
Weber-Fox, C. M., & Neville, H. J. (1996). Maturational constraints on functional specializations for lan-
guage processing: ERP and behavioral evidence in bilingual speakers. Journal of Cognitive Neuroscience, 8(3),
231–256.
Yao, P., & Chen, B. (2017). Cross-linguistic diferences afect late Chinese-English learners on-line processing
of English tense and aspect. International Journal of Bilingualism, 21(3), 268–290.
241
20
DISCOURSE PROCESSING
Saveria Colonna
Introduction
Discourse occurs as soon as at least two utterances are produced in a row in speech or in writing.
Its processing requires the comprehension of individual utterances, as well as their linking, to form
a coherent representation of the discourse as a whole. The coherent mental representation that the
addressee constructs during discourse processing is known as a mental model or situation model. The
construction of such a representation is based on the addressee’s ability to generate inferences guided
by multiple sources of information ranging from linguistic expressions, which serve as cohesive devices
(see Textbox 20.1), to world knowledge. Cohesive devices can be taken as processing instructions
to the addressee that facilitate the linking of the utterances of a discourse. However, these links of
continuity are not always explicitly marked. The addressee must then infer them on the basis of his
or her world knowledge.
Discourse processing involves a complex interplay between comprehension of linguistic
input, prior discourse information, and world knowledge. In a second language (L2), this inte-
gration of information from multiple sources can be slower and more difficult compared to the
first language (L1) due to limited processing resources in, e.g., working memory capacity, and
(possibly) due to competing constraints between the L1 and the L2. In comparison to sentence
processing, discourse processing has received little attention in the second language acquisi-
tion (SLA) literature. Furthermore, while most of these studies examine L2 discourse produc-
tion under (more or less) controlled production conditions, relatively few studies examine
real-time discourse processing using experimental psycholinguistic methods. In this chapter,
I first provide an overview of the role of linguistics devices in the construction of a coherent
representation of discourse in an L2. I examine cohesive devices like anaphoric pronouns and
connectives that encode referential and relational coherence respectively, as well as linguistic
devices marking information structure, like word order. I then consider competing hypotheses
to explain the differences in discourse processing between L1 and L2. Finally, I present recent
experimental studies dealing with the ability of L2 learners to make the inferences involved in
discourse processing.
242 DOI: 10.4324/9781003018872-23

Discourse processing
Mental model: A mental model (also termed a situation model) is a mental representation of the discourse
content, i.e., the events, actions, and persons described in the discourse (see, among others, Graesser
et al., 1994; Johnson-Laird, 1983; Kintsch, 1988; Van Dijk & Kintsch, 1983; Zwaan & Radvansky,
1998). This mental representation is constructed during discourse processing from the situation explic-
itly mentioned in utterances and/or inferred from world knowledge. It is a mental representation of the
described state of afairs in the discourse, rather than only a mental representation of the discourse itself.
Inferences in discourse: An inference is information that is not expressed explicitly in the discourse, but
mentally constructed by the addressees on the basis of their knowledge (for review, see Noordman &
Vonk, 2015). Inferencing plays a crucial role in reference resolution and in the interpretation of dis-
course structure and coherence relations. Inferences can either be logical or pragmatic inferences. For
instance, the deduction that Peter is older than John follows logically from Peter is older than Mary and
Mary is older than John. Inferences that do not follow logically from the propositions of the discourse
are pragmatic inferences. An example is the inference that the millionaire was murdered from The mil-
lionaire died on a dark and stormy night. The killer left no clues for the police to trace (Just & Carpenter, 1978).
While addressees are able to make logical inferences in comprehension, the construction of discourse
coherence is mainly based on pragmatic inferences (Graesser & Kreuz, 1993). In this chapter, when
we use the term “inference,” it is to designate a pragmatic inference.
Cohesive devices: Discourse processing is guided by certain linguistic expressions, known as cohesive
devices (Halliday & Hasan, 1976). These are markers that convey interpretive instructions facilitating
access to the links between the diferent utterances of a discourse. Cohesive devices are explicit lexical
and grammatical clues, among which are markers of relational coherence and markers of referential
coherence (e.g., Asher & Lascarides, 2003; Reinhart, 1981; Sanders et al., 1992). Relational cohesive
devices such as connectives (e.g., because, but, and) explain the relations connecting the discourse units,
e.g., cause-consequence relations. Referential cohesive devices such as pronouns and demonstratives
allow one to relate various linguistic mentions to the same mental referent.
Information structure: Information structure (or information packaging) refers to how speakers organize the
information in their utterances in order to adapt to the knowledge of the other discourse participants.
The main dimensions of information structure are givenness (maintained vs. new information), about-
ness (topic vs. comment), and foregrounding (e.g., contrastive topic, focus; for review, see Krifka, 2008).

Most studies on L2 discourse production show that learners’ discourse usually difers from native
speaker production, even at very advanced levels of L2 profciency (see for example the collected vol-
umes, Benazzo et al., 2012; Dimroth & Starren, 2003; Dimroth & Lambert, 2008; Hendriks, 2005).
These studies underline that, even when learners master the use of grammatical structures in the L2,
their productions maintain non-native uses of discourse functions, in the expression of motion events,
the use of temporal and aspectual markers, and the use of referring expressions. Since the L2 learner
typically does not receive feedback or negative evidence on infelicitous discourse aspects of their
utterances, such non-native uses are difcult for the learner to correct. It remains to be ascertained
whether the diferences between native and non-native productions refect conceptual transfer from
the L1 (Jarvis & Pavlenko, 2008), or whether they constitute a learner-specifc variety which cannot
be directly traced to either the L1 or L2 (for overview, see Pavlenko, 2011). Similarly, SLA studies
243
Saveria Colonna
on comprehension have shown that even highly profcient L2 speakers do not behave in a target-like
manner during discourse processing. To better understand the causes of these diferences between L1
and L2, studies have predominantly examined L2 learners’ interpretations of linguistic devices used
to link utterances together in coherent discourse. The underlying assumption in these studies is that
cross-linguistic variation in linguistic devices may cause learning or processing problems.
Referential discourse coherence in SLA

As already stated, discourse is not a sequence of unrelated utterances, because these need to be
linked together in a coherent mental representation to form a discourse. One of the ways in which
a discourse coheres is through referential coherence, that is, the establishment of referential links
across utterances. The sentences of a discourse are connected in that they make repeated reference to
the same entities (e.g., persons, places, or objects). Anaphoric devices such as pronouns are used to
signal referential coherence in discourse. For instance, to understand the short discourse in (1), the
addressee has to interpret the pronoun he and the boy as indicating the same referent.
(1) The boy is sad. He lost his balloon.
Research on SLA has identifed anaphoric pronouns as a particularly difcult domain to acquire
(Sorace & Filiaci, 2006; Sorace, 2011). Roberts et al. (2008) examined pronoun resolution in Ger-
man and Turkish advanced learners of L2 Dutch in short discourses as in (2). In a comprehension
questionnaire, learners and native speakers had to make an ofine choice regarding the preceding
referent, i.e., the antecedent, of the ambiguous pronoun hij “he.”
(2) Peter en Hans zitten in het kantoor. Terwijl Peter aan het werk is, eet hij een boterham.
“Peter and Hans are in the ofce. While Peter is working, he is eating a sandwich.”
While Turkish is a pro-drop language, German and Dutch are both non-pro-drop languages; that is,
they do not allow null subjects. In pro-drop languages, null pronouns typically refer to the antecedent
in the subject and topic position (as do overt pronouns in non-pro-drop languages), while overt pro-
nouns prefer an antecedent in a less salient position (Carminati, 2002). In interpretation, L1 German
learners of Dutch showed the same referent preference as Dutch native speakers, both interpreting the
pronoun hij as referring to the sentence-internal subject Peter, whereas L1 Turkish learners chose this
interpretation in just over half of the cases, and instead preferred the sentence-external referent Hans
for the remainder. These ofine results suggest that the Turkish learners transferred the interpretation
preference of the Turkish overt pronoun to the Dutch overt pronoun. The authors conclude that
pronoun interpretation might pose an acquisition problem when pronominal systems difer in the L1
and L2. In addition, in an eye-tracking experiment during reading, Roberts et al. (2008) observed
that the learner groups patterned alike and both performed diferently from Dutch native speakers.
Longer reading times were found for the Turkish and German L2 learners on the ambiguous pronoun
and the following verb. These results suggest that, when having to establish co-reference online, L2
learners have more difculties integrating syntactic and discourse-pragmatic information than native
speakers, irrespective of L1 diferences. The overall results show that, while the L1 infuences ofine
interpretation preferences, there is no similar infuence during online processing.
In line with the idea that learners encounter difculties integrating syntactic and discourse information,
several L2 studies have focused on the learners’ acquisition of discourse-related conditions determining the
use of overt subjects in pro-drop languages. Sorace and Filiaci (2006) studied the interpretation of overt and
null pronouns in near-native speakers of Italian (pro-drop) whose L1 was English (non-pro-drop). Native
speakers and near-native L2 speakers were presented with sentences like (3) in a picture verifcation task.
244
(3) La mamma dà un bacio alla fglia, mentre lei/Ø si mette il cappotto.

“The mother kisses her daughter, while she/Ø is wearing her coat.”
The null or overt pronoun in the subordinate clause could variously refer to the subject or to the
object of the main clause. The results revealed that near-natives preferred the subject as antecedent
of null pronouns to the same degree as natives, but they interpreted overt pronouns as being co-
referential with the subject antecedent more often than the natives. Hence, near-natives had acquired
a target-like interpretation only for null pronouns. In line with this fnding, Belletti et al. (2007)
reported an overuse of overt pronominal subjects in production for L1 English near-native speakers
of Italian. These studies showed that, while near-natives use and interpret null subjects in a native-like
way, this is not the case for overt pronominal subjects. Sorace and colleagues interpret these results
as difculties among L2 learners to acquire the discourse conditions determining overt pronominal
realization at the interface between syntax and discourse. In contrast, the null pronouns realization
rules present less of a problem, since they are purely syntactically determined (Sorace, 2011).
Relational discourse coherence in SLA

A further aspect of discourse coherence is relational coherence. Discourse segments are connected
by establishing coherence relations between them, such as causal, contrastive, or concessive rela-
tions. These conceptual relations can be (optionally) made explicit by linguistics markers such as
connectives (Halliday & Hasan, 1976; Mann & Thompson, 1988; Sanders et al., 1992). In example
(1) shown earlier, the addressee can infer that there is a causal relation between being sad and los-
ing the balloon. The causal relation can also be explicitly marked by the conjunction because as in
The boy is sad because he lost his balloon and instructs the addressee to make the causal inference. In
native speakers, a facilitating role of connectives in real-time text comprehension has been reported
in several studies. During the reading process, the presence of a connective linking clauses together
decreases reading time on the words following the connective, facilitating the integration of upcom-
ing words. However, reading time increases on the last words of the sentence, which may refect the
causal inference process triggered by the connective (Cozijn et al., 2011). In addition, the presence
of a connective improves accuracy in comprehension questions after reading and in recall (for review,
Degand & Sanders, 2002). For L2 learners, studies on the production of connectives point to misuse
as well as overuse and underuse compared to native speakers, even in advanced learners (Zuferey &
Gygax, 2017). Few studies examined the role of connectives in L2 discourse comprehension. Degand
and Sanders (2002) explored the role of causal markers such as because in reading comprehension in
L2. They observed that both L1 Dutch speakers and L1 French learners of Dutch beneftted from
the presence of the causal marker, as shown by their improved ability to answer questions about
text content when connectives were used. Zuferey et al. (2015) investigated learners’ sensitivity
to non-native-like semantic uses of connectives in English that corresponded to licit uses of similar
connectives in their L1 (French and Dutch). The authors used ofine grammaticality judgments
and eye-tracking during reading of discourses containing non-native uses of connectives typically
produced by L1 French and L1 Dutch learners of English. Example (4) illustrates a conditional sen-
tence involving the incorrect use of when instead of if. Example (5) illustrates a contrastive sentence
with an incorrect use of if instead of while. While advanced learners of English were able to detect
non-native uses of connectives during online reading, they failed to detect incorrect uses when they
corresponded to licit uses in their L1.
(4) The kids don’t look tired today. When they don’t sleep now, we can go out for a walk.
(5) People had a very diferent reaction to the president’s speech. If in New York most people agreed
with him, in Texas people were appalled.
245
Saveria Colonna
Zuferey and Gygax (2017) further explored learners’ mastery of connectives by testing L1 German
speakers’ comprehension of the French connective en efet (lit. “in fact”), which encodes either a causal
relation or a confrmation relation depending on context, as illustrated in (6) and (7) respectively. In
German, these relations map to diferent connectives, namely in der Tat (indeed) for confrmation
relations and denn (because) or nämlich (namely) for causal relations. The authors used an acceptability
rating task and a self-paced reading experiment to evaluate the interpretation of en efet in L1 and L2
speakers of French. Experimental items for both relations were presented in two versions, one con-
taining the connective, and one without a connective, leaving the intended relation implicit.
(6) Susanne ne fait manifestement pas attention à ses afaires. En efet, elle/Elle a oublié son portefeuille dans
le bus.
“Suzanne is obviously rather careless with her belongings. CONNECTIVE, she/She forgot her
purse in the bus.”
(7) Susanne avait l’impression qu’il lui manquait quelque chose. Et en efet, elle/Elle a oublié son portefeuille
dans le bus.
“Susanne felt that she had lost something. And CONNECTIVE, she/She forgot her purse in
the bus.”
Reading times showed that L1 German learners of L2 French did not react to the lack of the con-
nective in confrmation relations (7) as French native speakers did. Both ofine and online results
suggested that learners do not master the complex form-function mapping of en efet. Learners are
not aware that confrmation relations must be explicitly marked to maintain coherence in French.
Overall, the studies on relational coherence in L2 highlight that cross-linguistic diferences in the
marking of discourse relations may lead to difculties in mastering discourse structuring, even at
advanced stages of L2 learning.
Information structure in SLA

L2 learners need to deal with another linguistic phenomenon which contributes to the overall
coherence and topic continuity of a discourse, namely information structure (IS). The expression
of IS distinctions, such as topic and focus, varies across languages, and L2 learners need to acquire
the syntactic and prosodic means of information selection and organization in discourse in the L2.
Hopp (2009) examined the acquisition of word order variation in German scrambling, i.e., object-
subject reordering in the middle feld, by learners of various L1s: English (no scrambling), Russian
(scrambling with IS efects similar to German), and Dutch (scrambling with a slightly diferent func-
tion). In German, objects are fronted as given information in discourse contexts in which the sub-
ject is new information. In a self-paced reading experiment, sentences with scrambled word order
as in (8a) or canonical word order as in (8b) were presented in felicitous or infelicitous contexts. A
context such as In the factory, the worker was distracted by someone last Monday. Who distracted the worker?
was felicitous for the scrambling word order in (8a), because it designated the subject as new infor-
mation. In contrast, a context such as In the factory, the apprentice distracted someone last Monday. Who
did the apprentice distract? was more appropriate for the canonical subject-object word order in (8b).
(8) a. Ich glaube, dass den Arbeiter am Montag der Lehrling abgelenkt hat.
I think that the.ACC worker on Monday the.NOM apprentice distracted has.
“I think that the APPRENTICE has distracted the worker on Monday.”
(8) b. Ich glaube, dass der Lehrling am Montag den Arbeiter abgelenkt hat.
I think that the.NOM apprentice on Monday the.ACC worker distracted has.
“I think that the apprentice has distracted the WORKER on Monday.”
246
Self-paced reading times showed that L1 Russian and L1 English learners performed like Ger-
man natives, i.e., they were sensitive to the IS constraints on German scrambling. When scram-
bled sentences were presented in felicitous contexts, reading times were shorter. Only the L1
Dutch learners treated German scrambled sentences in felicitous and infelicitous contexts alike,
probably due to L1 transfer. Interestingly, the results for English learners showed that IS-related
word order optionality was acquirable even when the L1 does not encode similar IS-to-syntax
mappings. Hopp et al. (2020) asked L1 English learners of L2 German and L1 German learners
of L2 English to judge the naturalness of different types of non-canonical fronted constructions
across different discourse contexts in both languages. The authors observed that L2 learn-
ers were as sensitive to the IS constraints on fronting constructions as native speakers. The
results of both studies (Hopp, 2009; Hopp et al., 2020) suggest that learners, with the exception
of the L1 Dutch learners of German, have acquired the discourse-appropriate uses of non-
canonical word orders in L2. Smeets (2019) investigated whether L1 English and L1 German
learners of L2 Dutch are sensitive to IS constraints on object movement to the prefield. Felicity
judgments showed that both groups of learners have acquired the syntax-to-discourse mappings
for object fronting to the prefield, although the relevant syntactic structure was available in the
L1 of only one group (German). In line with Hopp (2009), these results suggest that native-
like acquisition is possible, even when the syntactic construction acquired is not available in
the L1. In contrast, different syntax-to-discourse mappings in the L1 may interfere with the IS
acquisition in the L2.

As we have seen, discourse processing requires the simultaneous processing of various types of infor-
mation such as syntax, semantics, and pragmatics. In this section, we present several hypotheses that
have been proposed to explain the difculties encountered by L2 learners in the integration of infor-
mation from multiple sources, including discourse information, and, more generally, to account for
the processing diferences between L1 and L2.
The Interface Hypothesis

Sorace proposed the Interface Hypothesis (IH), according to which non-native-like acquisition is
due to difculty at the linguistics interfaces, specifcally in the integration of information from syn-
tactic and non-syntactic sources during processing (Sorace & Filiaci, 2006; Sorace, 2011; Sorace &
Serratrice, 2009). The IH suggests that L2 learners encounter difculty in integrating syntactic
and discourse-level information, i.e., when the choice of a syntactic option is guided by discourse
constraints. Here, L2 speakers must learn to link syntactic constructions with the appropriate dis-
course features, and acquiring this mapping could be problematic. In a visual world eye-tracking
experiment, Wilson (2009) observed that L2 learners difered from native German speakers in
their interpretation of the German demonstrative pronoun der, a so-called d-pronoun. In German,
the division of labor between personal pronouns and d-pronouns is similar to that between overt
and null pronouns in pro-drop languages. D-pronouns prefer less salient referents than personal
pronouns. According to the IH, the conditions on the use of d-pronouns in German would be
difcult to acquire, because they are located at the syntax-discourse interface. As predicted by the
IH, Wilson (2009) found that L1 English learners of German acquired the syntactic dependencies
of personal pronouns, preferring subject antecedent, but showed difculties with the discourse
dependency of d-pronouns. The IH assumes that L2 learners’ difculties at the syntax-discourse
interface are due to a limitation in processing resources or a resource allocation defcit rather than
to a representational defcit.
247
Saveria Colonna
The RAGE hypothesis

Grüter and colleagues proposed that non-native speakers have a Reduced Ability to Generate Expec-
tations during language processing, the so-called RAGE hypothesis (Grüter et al., 2014, 2017). The
idea behind the RAGE hypothesis is that, although non-native speakers have the relevant L2 knowl-
edge, they cannot use this knowledge to predict upcoming information. Various studies have shown
that the ability of non-native speakers to predict upcoming information at the lexical and syntactic
level is reduced when compared to native speakers (for review, see Kaan, 2014). Near-native learn-
ers are nevertheless capable of predictive processing of grammatical gender (Hopp, 2013), just as
bilinguals can predict upcoming words when close constructions exist in both L1 and L2 (Foucart
et al., 2014). Little research has looked at the role of expectations in L2 processing at the discourse
level. Grüter et al. (2017) examined the ability of non-native speakers to predict upcoming referential
choices by using discourse information such as event structure. L1 Korean and L1 Japanese learners
of L2 English were asked to provide story continuations after a one-sentence context as in (9). The
grammatical aspect of verbs expressing transfer or possession in the context sentence was manipu-
lated; it was either a perfective or imperfective (progressive) verb. The perfective aspect indicates an
event as completed, favoring continuation with the referent associated with the end state of the trans-
fer event (i.e., the goal). The imperfective aspect describes an event as ongoing, favoring continua-
tion with the referent associated with the start state of the event (i.e., the source). The continuation
began with a pronoun, she in (9), which had two potential referents, the source argument in subject
position of the context sentence (Emily), or the goal argument in the prepositional object position of
the context sentence (Melissa).
(9) Emily brought/was bringing a drink to Melissa. She . . .
While English native speakers tended to continue with the subject pronoun referring to the source
referent in the imperfective condition more than in the perfective condition, the event structure had
little infuence in L2 learners’ referent choices. These results suggest that, compared to native speak-
ers, L2 learners did not use verb aspect to generate an anticipatory bias for the source or goal refer-
ent. In other words, learners showed a reduced ability to make expectations about how a discourse
was likely to continue. As an underlying explanation, the authors proposed that fnite processing
resources of the L2 speaker are exhausted by information integration, and, as a result, few or no
resources remain for anticipatory processing, such as prediction at the discourse level.
The Shallow Structure Hypothesis

While many studies look at the difculties encountered by L2 learners during discourse processing,
others predict that L2 learners encounter no specifc difculties in acquiring discourse constraints.
With respect to L2 production, the learner varieties approach assumes that the acquisition of syntactic
knowledge in an L2 may be particularly difcult, and that learners, at least in the early stages of acqui-
sition, would therefore rely on semantic and discourse-level factors to a higher degree than native
speakers (Dimroth et al., 2003; Klein & Perdue, 1997). In line with the idea that learners give more
weight to discourse-level cues than L1 speakers, Clahsen and Felser (2006a, 2006b, 2018) proposed
the Shallow Structure Hypothesis (SSH) to account for diferences in L1 and L2 real-time processing,
even in highly profcient L2 speakers. According to the SSH, L2 learners underuse syntactic informa-
tion during processing, relying more on lexical-semantic and discourse information. The underuse of
syntactic information during parsing is supposed to be a result of a representational defcit in that L2
speakers are unable to store syntactic representations in a form accessible for parsing. Therefore, shal-
low parsing is assumed to predominate in L2 processing. Consequently, in comprehension, learners
248
may over-rely on semantic and discourse-level cues compared to syntactic cues. In accordance with
the prediction that learners are more sensitive to discourse-level cues than native speakers, Pan and
colleagues reported that learners of L2 English are reliably more infuenced by the referential context
than native speakers in the online resolution of syntactic ambiguities (Pan & Felser, 2011; Pan et al.,
2015). Further evidence comes from studies on the interpretation of refexive and personal pronouns
in L2, showing that learners were more sensitive to discourse salience than native speakers (Felser
et al., 2009; Felser & Cunnings, 2012; Patterson et al., 2014; Schimke & Colonna, 2016).

In this section, we report recent experimental psycholinguistic studies examining the ability of L2
learners to make inferences involved in discourse processing, therefore challenging the theoretical
hypotheses presented above.
Destruel and Donaldson (2017) examined whether learners of L2 French generate the same
pragmatic inferences as native speakers of French when interpreting cleft constructions. In an ofine
questionnaire, participants were presented with target sentences as in (10) in three diferent condi-
tions: a c’est-cleft (10a), a canonical structure (10b), or an exclusive seul/seulement (10c) construction.
They were asked to choose between three continuations that corrected the exhaustive interpretation
of the target sentence (10a-c) (i.e., that nobody besides the farmer brushed the horse) with various
degrees of strength (11a-c).
(10) a. C’est le fermier qui a brossé le cheval.

“It’s the farmer who brushed the horse.”
b. Le fermier a brossé le cheval.
“The farmer brushed the horse.”
c. Seul le fermier a brossé le cheval.
“Only the farmer brushed the horse.”
(11) a. Non, la cavalière aussi a brossé le cheval.
“No, the rider also brushed the horse.”
b. Oui, mais la cavalière aussi a brossé le cheval.
“Yes, but the rider also brushed the horse.”
c. Oui, et la cavalière aussi a brossé le cheval.
“Yes, and the rider also brushed the horse.”
Native speakers of French preferred to use (11a) to follow the construction with an explicit exclusive
particle (10c), while choosing (11b) and (11c) for the other two constructions. The continuation
(11b) was chosen more often than (11c) for the cleft construction (10a), and conversely for the
unmarked construction in (10b), the continuation (11c) was chosen more often than (11b). These
results suggest that the exhaustive interpretation of the cleft construction is pragmatically inferred
rather than semantically encoded, as is the case with the exclusive particle. The most advanced learn-
ers showed the same choice of continuations as the natives, whereas the beginners did not seem to
make any distinction between the unmarked construction (10b) and the cleft construction (10a),
showing no preference for (11b) or (11c) as continuation in either case. Yet, learners of all profciency
levels overwhelmingly preferred (11a) as continuation for constructions with the exclusive particle
(10c), just like the native speakers. So, while the exhaustive interpretation of the exclusive particle
was also observed in the least advanced learners, the interpretation of the cleft depends on the level
of profciency of the learners: Only the advanced learners exhibited the native-like inferential pat-
tern of exhaustivity associated with cleft constructions. Since exhaustivity in c’est-cleft sentences is
not lexically encoded, its interpretation involves the integration of syntactic and discourse-pragmatic
249
Saveria Colonna
information, and the fact that the high profciency L2 learners performed on a native-like level chal-
lenges the IH. While this study used ofine measures, Reichle and Birdsong (2014) examined the
processing of focus-marking cleft constructions in L2 French in an event-related potential (ERP)
study. Participants were presented with a picture of a household object on a table followed by a
question and response relating to the contents of the picture. The question in (12) requires marteau
“hammer” to be focused in a cleft construction as in (13a), whereas (13b) is an infelicitous answer
because on the table is marked as focal, despite being given as the topic in question.
(12) C’est quoi qu’on voit sur la table?

“WHAT do we see on the table?”
(13) a. C’est un marteau qu’on voit sur la table.
“It’s a HAMMER that we see on the table.”
b. #C’est sur la table qu’on voit un marteau.
#“It’s a hammer that we see on the TABLE.”
The authors examined whether the same ERP efects measured in focus structure processing in
native speakers would be seen in L2 learners of French. While the low-profciency L2 learners
showed a diferent response, the higher-profciency L2 French learners exhibited brain signatures
comparable to those of native speakers, suggesting that it is possible to attain native-like processing.
Contrary to the IH, both these studies (Destruel & Donaldson, 2017; Reichle & Birdsong, 2014)
suggest that learners’ difculties at the syntax-discourse interface can be overcome at sufciently high
levels of L2 attainment.
In an eye-tracking visual world experiment, Contemori and Dussias (2019) examined the ability
to predict an upcoming pronoun referent based on the implicit causality bias of the verb of English
native speakers and Spanish-English bilinguals. They looked at the interpretation of the ambiguous
pronoun he in sentences such as (14), wherein the main clause contained either a subject-bias verb
(apologize) or an object-bias verb (believe).
(14) Kevin apologized to/believed Dave because he . . .
Eye-tracking data showed that bilinguals were not slower than monolinguals in processing the infor-
mation associated with the implicit causality of the verb. A sentence continuation task confrmed that
monolinguals and bilinguals alike used implicit causality information associated with the verb. Mono-
linguals and bilinguals produced more continuations in which the pronoun referred unambiguously
to the subject of the matrix clause in the subject-bias verb than in the object-bias verb condition. The
results suggest that highly profcient and early exposed bilinguals can develop similar on-line and of-
line predictions as monolingual speakers. These fndings challenge the RAGE hypothesis mentioned
earlier (Grüter et al., 2017), which hypothesizes weaker predictive abilities in the L2 compared to the
L1. While the learners tested in Grüter et al. (2017) were late L2 learners at intermediate profciency
levels, the participants in Contemori and Dussias (2019) were highly profcient bilinguals with child-
hood exposure to the L2, i.e., partly heritage speakers (see Montrul, 2023 [this volume]). Hence, the
diference in learner profciency and language background may explain why the bilinguals demon-
strated the same predictive abilities as natives, whereas the less profcient learners failed to do so.
Finally, in an ERP study, Foucart et al. (2016) examined whether L2 speakers can integrate mul-
tiple sources of information in discourse comprehension. English native speakers were compared to
L1 Spanish speakers who were highly profcient in L2 English during a reading task of a text whose
interpretation required simple or complex inferences. The texts presented involved three levels of
causality linking the fnal sentence: highly causally related as in (15a), intermediately related as in
(15b), or causally unrelated to its context as in (15c). In a discourse such as (15a), the last sentence
250
logically followed the frst two, and required a simple inference based on world knowledge. In (15b),
a more complex inference was required to link the second sentence to the third. Finally, in (15c), the
last sentence was not a logical continuation of the frst two, and information from both the discourse
context and world knowledge had to be processed to make sense of it.
(15) a. Jill had very fair skin. She forgot to put sunscreen on. She had sunburn on Monday.
(15) b. Jill had very fair skin. She usually remembered to wear sunscreen. She had sunburn on
Monday.
(15) c. Jill’s skin always tanned well. She always put on sunscreen. She had sunburn on Monday.
The results for native speakers replicated a gradual N400-like efect previously observed in L1 speak-
ers (Kuperberg et al., 2011), namely, a larger negativity in the causally unrelated condition (15c)
than in the highly related condition (15a), with the intermediately related condition (15b) falling
in between. Similarly to native speakers, L2 learners exhibited an N400 efect slightly larger in the
intermediate condition (15b) than in (15a), but this efect was longer lasting for learners. These
results suggest that highly profcient L2 speakers can process diferent sources of information to
make inferences online, even though it requires extra processing in the L2 compared to the L1.
These fndings support the position that the mechanisms involved in L1 and L2 are similar, and, as a
consequence, that comprehension diferences in L1 and L2 processing originate from computational
restrictions rather than a representational defcit.

SLA studies have paid relatively little attention to discourse processing compared to grammatical
aspects of sentence processing. Most of the hypotheses presented in this chapter have been proposed
to account for sentence processing, although as seen, they can be extended to discourse processing.
Furthermore, the results from studies on discourse processing in L2 appear to be conficting. On the
one hand, it has been shown that learners are sensitive to discourse constraints from the beginning of
L2 acquisition. On the other hand, it has been claimed that the native-like organization of discourse
is very hard if not impossible to achieve even by very advanced learners. This apparent paradox is
mainly due to the fact that these studies do not refer to the same discourse constraints. Studies fnd-
ing early and robust use of discourse information are concerned with the impact of very general and
potentially universal discourse constraints like the “topic frst” constraint, whereas the latter deal with
language-specifc preferences for discourse organization that interact closely with grammatical prop-
erties of the L2 (for discussion, see Dimroth & Narasimhan, 2012). Finally, the data on discourse pro-
cessing in L2 come from various experimental contexts/tasks (production tasks with varying degrees
of control, ofine and online comprehension tasks) and a wide variety of populations (diferent levels
of learner profciency, diferent L1 and L2 pairs), which makes across-study comparisons difcult. For
instance, we have seen that apparently conficting results concerning the learners’ ability to predict
upcoming pronoun referents during online discourse processing may be, at least partly, due to the
diference in the profciency or the ages of onset of the learners tested. As a consequence, in order
to ascertain whether L2 learners can indeed process discourse information like native speakers, what
the role of the L1 is, and how the level of L2 profciency afects discourse processing, experimental
studies on discourse acquisition must consider a wider range of factors. Among these factors are the
experimental task (production, comprehension, metalinguistic judgments), the degree and length of
exposure to the L2, the level of profciency in the L2, the individual learner’s characteristics, and the
typological diferences between L1 and L2. A larger number of psycholinguistic studies on discourse
acquisition will provide a more comprehensive picture of diferences and similarities between L1 and
L2 processing and inform current debates in SLA.
251
Saveria Colonna
How can we explain diferences in L2 discourse processing across diferent types of discourse constraints?
Are there systematic diferences in L2 discourse comprehension between ofine and online tasks?
How can ERP research provide additional insights into L2 discourse processing?
How does discourse processing in the L2 difer depending on age of acquisition and language dominance?
Further reading
Howard, M., & Leclercq, P. (Eds.). (2017). Tense-aspect-modality in a second language: Contemporary perspectives
(Studies in Bilingualism). Benjamins.
This volume brings together a set of contributions on the expression of modality in L2, the efect of frst
language transfer and aspectuo-temporal marking at a wider discourse level.
Roberts, L., González Alonso, J., Pliatsikas, C., & Rothman, J. (2018). Evidence from neurolinguistic meth-
odologies: Can it actually inform linguistic/language acquisition theories and translate to evidence-based
applications? Second Language Research, 34(1), 125–143.
This paper presents the current state of neurolinguistic data in the feld of L2 acquisition and processing
and assesses how the neurophysiological methods might inform us with respect to linguistic and language
acquisition theories.
Van den Bosch, L. J., Segers, E., & Verhoeven, L. (2018). Online processing of causal relations in beginning frst
and second language readers. Learning and Individual Diferences, 61, 59–67.
While the studies reviewed in this chapter focused on adult L2 learners, this paper explores the online pro-
cessing of causal relations during text comprehension in child L2 learners.
References
Asher, N., & Lascarides, A. (2003). Logics of conversation. Cambridge University Press.
Belletti, A., Bennati, E., & Sorace, A. (2007). Theoretical and developmental issues in the syntax of subjects:
Evidence from near-native Italian. Natural Language & Linguistic Theory, 25(4), 657–689.
Benazzo, S., Flecken, M., & Soroli, E. (Eds.). (2012). Typological perspectives on second language acquisition:
“Thinking for speaking” in L2. Special Issue of Language, Interaction & Acquisition, 3(2).
Carminati, M. N. (2002). The processing of Italian subject pronouns [Ph.D. dissertation], University of Massachu-
setts, Amherst.
Clahsen, H., & Felser, C. (2006a). How native-like is non-native language processing? Trends in Cognitive Sciences,
10, 564–570.
Clahsen, H., & Felser, C. (2006b). Grammatical processing in language learners. Applied Psycholinguistics, 27(1),
3–42.
Clahsen, H., & Felser, C. (2018). Some notes on the Shallow Structure Hypothesis. Studies in Second Language
Acquisition, 40(3), 693–706.
Contemori, C., & Dussias, P. E. (2019). Prediction at the discourse level in Spanish—English bilinguals : An
eye-tracking study. Frontiers in Psychology, 10, 6.
Cozijn, R., Noordman, L., & Vonk, W. (2011). Propositional integration and world-knowledge inference :
Processes in understanding because sentences. Discourse Processes, 48(7), 475500.
Degand, L., & Sanders, T. (2002). The impact of relational markers on expository text comprehension in L1 and
L2. Reading and Writing, 15(7/8), 739–757.
Destruel, E., & Donaldson, B. (2017). Second language acquisition of pragmatic inferences: Evidence from the
French c’est-cleft. Applied Psycholinguistics, 38(3), 703–732.
Dimroth, C., Gretsch, P., Jordens, P., Perdue, C., & Starren, M. (2003). How to express fniteness in Germanic
languages: A stage-model for frst and second language acquisition. In C. Dimroth & M. Starren (Eds.),
Linguistic structure and the dynamics of acquisition (pp. 65–94). Benjamins.
252
Dimroth, C., & Lambert, M. (Eds.). (2008). La structure informationnelle chez les apprenants L2. Special Issue of
Acquisition et interaction en langue étrangère, 26.
Dimroth, C., & Narasimhan, B. (2012). The acquisition of information structure. In M. Krifka & R. Musan
(Eds.), The expression of information structure. Mouton de Gruyter.
Dimroth, C., & Starren, M. (Eds.). (2003). Information structure and the dynamic of language acquisition. Benjamins.
Felser, C., & Cunnings, I. (2012). Processing refexives in a second language: The timing of structural and
discourse-level information. Applied Psycholinguistics, 33(3), 571–603.
Felser, C., Sato, M., & Bertenshaw, N. (2009). The on-line application of binding principle A in English as a
second language. Bilingualism: Language and Cognition, 12(4), 485–502.
Foucart, A., Martin, C., Moreno, E., & Costa, A. (2014). Can bilinguals see it coming? Word anticipa-
tion in L2 sentence reading. Journal of Experimental Psychology: Learning, Memory and Cognition, 40(5),
1461–1469.
Foucart, A., Romero-Rivas, C., Gort, B. L., & Costa, A. (2016). Discourse comprehension in L2: Making sense
of what is not explicitly said. Brain and Language, 163, 32–41.
Graesser, A. C., & Kreuz, R. J. (1993). A theory of inference generation during text comprehension. Discourse
Processes, 16(12), 145–160.
Graesser, A. C., Singer, M., & Trabasso, T. (1994). Constructing inferences during narrative text comprehen-
sion. Psychological Review, 101(3), 371–395.
Grüter, T., Rohde, H., & Schafer, A. J. (2014). The role of discourse-level expectations in non-native speakers’
referential choices. In Proc. 38th BUCLD (pp. 179–191). Cascadilla.
Grüter, T., Rohde, H., & Schafer, A. J. (2017). Coreference and discourse coherence in L2 : The roles of gram-
matical aspect and referential form. Linguistic Approaches to Bilingualism, 7(2), 199–229.
Halliday, M. A. K., & Hasan, R. (1976). Cohesion in English. Longman.
Hendriks, H. (Ed.). (2005). The structure of learner varieties. Mouton de Gruyter.
Hopp, H. (2009). The syntax-discourse interface in near-native L2 acquisition: Of-line and on-line perfor-
mance. Bilingualism: Language and Cognition, 12, 463–483.
Hopp, H. (2013). Grammatical gender in adult L2 acquisition : Relations between lexical and syntactic vari-
ability. Second Language Research, 29(1), 33–56.
Hopp, H., Bail, J., & Jackson, C. N. (2020). Frequency at the syntax—discourse interface: A bidirectional study
on fronting options in L1/L2 German and L1/L2 English. Second Language Research, 36(1), 65–96.
Jarvis, S., & Pavlenko, A. (2008). Crosslinguistic infuence in language and cognition. Routledge.
Johnson-Laird, P. N. (1983). Mental models : Towards a cognitive science of language, inference, and consciousness. Har-
vard University Press.
Just, M. A., & Carpenter, P. A. (1978). Inference processes during reading: Refections from eye fxations.
In J. W. Senders, D. F. Fisher, & R. A. Monty (Eds.), Eye movements and the higher psychological functions
(pp. 157–174). Erlbaum.
Kaan, E. (2014). Predictive sentence processing in L2 and L1 : What is diferent? Linguistic Approaches to Bilingual-
ism, 4(2), 257–282.
Kintsch, W. (1988). The role of knowledge in discourse comprehension: A construction-integration model.
Psychological Review, 95, 163–182.
Klein, W., & Perdue, C. (1997). The basic variety (or: Couldn’t natural languages be much simpler?). Second
Krifka, M. (2008). Basic notions of information structure. Acta Linguistica Hungarica, 55, 243–276.
Kuperberg, G. R., Paczynski, M., & Ditman, T. (2011). Establishing causal coherence across sentences: An ERP
study. Journal of Cognitive Neuroscience, 23(5), 1230–1246.
Mann, W. C., & Thompson, S. A. (1988). Rhetorical structure theory: Toward a functional theory of text
organization. Text, 8(3), 243–281.
Montrul, S. (2023). The psycholinguistics of heritage language acquisition. In A. Godfroid & H. Hopp (Eds.).
Noordman, L., & Vonk, W. (2015). Inferences in discourse, psychology of. In International encyclopedia of the social
& behavioral sciences (pp. 37–44). Elsevier.
Pan, H. Y., & Felser, C. (2011). Referential context efects in L2 ambiguity resolution: Evidence from self-paced
reading. Lingua, 121(2), 221–236.
Pan, H.-Y., Schimke, S., & Felser, C. (2015). Referential context efects in non-native relative clause ambiguity
resolution. International Journal of Bilingualism, 19(3), 298–313.
Patterson, C., Trompelt, H., & Felser, C. (2014). The online application of binding condition B in native and
non-native pronoun resolution. Frontiers in Psychology, 5.
Pavlenko, A. (2011). Thinking and speaking in two languages. MIT Press.
253
Saveria Colonna
Reichle, R., & Birdsong, D. (2014). Processing focus structure in L1 and L2 French: L2 profciency efects on
ERPs. Studies in Second Language Acquisition, 36(3), 535–564.
Reinhart, T. (1981). Pragmatics and linguistics : An analysis of sentence topics. Philosophica, 27, 53–94.
Roberts, L., Gullberg, M., & Indefrey, P. (2008). On-line pronoun resolution in L2 discourse : L1 Infuence and
general learner efects. Studies in Second Language Acquisition, 30(3), 333–357.
Sanders, T., Spooren, W., & Noordman, L. (1992). Toward a taxonomy of coherence relations. Discourse Pro-
cesses, 15(1), 135.
Schimke, S., & Colonna, S. (2016). Native and non-native interpretation of pronominal forms: Evidence from
French and Turkish. Studies in Second Language Acquisition, 38(1), 131–162.
Smeets, L. (2019). The acquisition of object movement in Dutch: L1 transfer and near-native grammars at the
syntax—discourse interface. Second Language Research, 35(4), 479–504.
1(1), 133.
Sorace, A., & Filiaci, F. (2006). Anaphora resolution in near-native speakers of Italian. Second Language Research,
22(3), 339–368.
Sorace, A., & Serratrice, L. (2009). Internal and external interfaces in bilingual language development : Beyond
structural overlap. International Journal of Bilingualism, 13(2), 195210.
Van Dijk, T. A., & Kintsch, W. (1983). Strategies of discourse comprehension. Academic Press.
Wilson, F. (2009). Processing at the syntax-discourse interface in second language acquisition [Unpublished Ph.D. thesis],
University of Edinburgh.
Zuferey, S., & Gygax, P. M. (2017). Processing connectives with a complex form-function mapping in L2 : The
case of French “En Efet”. Frontiers in Psychology, 8.
Zuferey, S., Mak, W., Degand, L., & Sanders, T. (2015). Advanced learners’ comprehension of discourse con-
nectives : The role of L1 transfer across on-line and of-line tasks. Second Language Research, 31(3), 389–411.
Zwaan, R. A., & Radvansky, G. A. (1998). Situation models in language comprehension and memory. Journal
of Experimental Psychology: Learning, Memory, and Cognition, 21, 162–185.
254
21
CODE-SWITCHING
Janet G. van Hell
Introduction
When bilinguals speak with each other, they regularly use words and phrases from their two lan-
guages and integrate these diferent language codes within a single utterance, often in a fexible and
seemingly efortless way. This phenomenon is called code-switching and is unique to bilingual lan-
guage processing. Code-switching overtly demonstrates one of the key fndings in research on second
language acquisition and bilingualism, namely that a bilingual’s two languages are continuously active
to some extent (for reviews, see Kroll et al., 2006; Van Hell & Tanner, 2012). Code-switching also
illustrates that bilingual speakers can fexibly use and integrate both of their languages and that bilin-
gual listeners need to continuously regulate and adjust the level of activation of their two languages
to enable their comprehension of the code-switched utterance as it unfolds over time. The study of
code-switching, therefore, is an ideal testbed for studying how bilingual speakers navigate their two
language systems during speech planning and how bilingual comprehenders incrementally integrate
new information, potentially coming from diferent languages, with the preceding context in order
to build up a message-level interpretation (e.g., Beatty-Martinez et al., 2020; Van Hell et al., 2015).
Why do bilingual speakers code-switch? If you ask this to a speaker who just code-switched, they
may respond that the code-switch occurred unintentionally and just popped up without their real-
izing it (Grosjean, 2010; Heredia & Altarriba, 2001). Other speakers, particularly those who are less
profcient in their second language, may say they did not know the particular word or phrase in the
language they were using, so they resorted to their other language. Although there are still many gaps
in our current knowledge on why bilinguals code-switch, the currently available evidence indicates
that code-switching is rarely driven by a lack of lexical knowledge in the other language in fuent
bilinguals. Furthermore, while a bilingual may feel that a switch to the other language simply “popped
up,” research has shown that intra-sentential code-switching is not haphazard, but governed by linguis-
tic and socio-contextual principles. Research has also shown that switching into the other language is
often associated with measurable cognitive efort, both in speaking and in comprehension, reminiscent
of the well-document cognitive costs associated with task-switching (Kiesel et al., 2010).
This chapter focuses on intra-sentential code-switching, i.e., the alternative use of two languages
within a clause or sentence (for studies examining inter-sentential code-switching, see, for example,
Gullifer et al., 2013; Ibáñez et al., 2010; Pérez & Duñabeitia, 2019). The language into which a con-
stituent from the other language has been inserted is often referred to as the base language or matrix
language (cf., Myers-Scotton, 1993), since this language determines the morphosyntactic frame of
DOI: 10.4324/9781003018872-24 255

Janet G. van Hell
the sentence. The language from which the constituent is taken is referred to as the embedded lan-
guage. Intra-sentential code-switches can be characterized structurally as alternations, insertions, or
congruent lexicalizations (Deuchar et al., 2007; Muysken, 2000). Alternations are code-switches that
involve a complete switch of a sentence constituent (phrase, clause, etc.) into the other language (1).
(1) Je telephone à Chantal, he, meestal voor commiskes te doen en eten.

“I call Chantal, hm, mostly to go shopping and eat.”
(Trefers-Daller, 1994, p. 213)
Insertions are characterized by the insertion of one grammatical constituent, typically consisting of
one word or a few words, from one language into the sentence structure of the other language (2).
(2) Yo anduve in a state of shock por dos días.

“I walked in a state of shock for two days.”
(Pfaf, 1979, p. 219)
In congruent lexicalizations both languages are used to fll the grammatical slots, which typically
takes place when the syntactic structure is shared (structurally congruent) between two languages (3).
(3) Dy lei dea yné hoeke, verroerde gjin vin.

“He lay dead in the corner, did not stir at all.”
(Wolf, work in progress, as cited in Muysken, 2000, p. 138)
Code-switching is a discourse phenomenon, since it comprises the production and the comprehension of
code-switched utterances. As we will see, most theoretical perspectives pertain to the production of code-
switched utterances, but empirical studies have examined both the production and the comprehension of
code-switched utterances. Furthermore, code-switching occurs more frequently in spoken than in writ-
ten discourse. Although code-switching can occur during writing, existing corpora of code-switched dis-
course are all based on spoken utterances, usually as they emerged in natural discourse settings. However,
a substantial part of the psycholinguistic and neurocognitive evidence on the comprehension of code-
switching is based on bilinguals’ reading of code-switched utterances using, for example, eye-tracking (e.g.,
Guzzardo Tamargo et al., 2016) or electrophysiological methodology (see Van Hell et al., 2018).
Some lab-based psycholinguistic studies focus on externally induced switches, i.e., non-
spontaneous language switches whereby bilinguals switch languages as prompted by an external cue
(e.g., a color cue or a fag denoting a particular response language); whether switches are externally
induced or internally generated potentially impacts behavioral patterns associated with the produc-
tion of code-switched speech (see, e.g., Gollan & Ferreira, 2009). Studies of language corpora, on
the other hand, typically refect internally generated switches in natural discourse. Both externally
induced switches and internally generated switches enable researchers to examine relevant questions
related to language switching (see Gullberg et al., 2009).
Intra-sentential code-switching: The alternative use of two languages within a clause or sentence.
Externally induced switches: Non-spontaneous language switches whereby bilinguals switch languages as
prompted by an external cue.
Internally induced switches: Spontaneous language switches occurring during natural discourse.
256
Code-switching
Historical perspectives and critical questions in the cross-disciplinary

study of code-switching
Code-switching emerges when two languages come into contact (Torres Cacoullos & Travis, 2018).
For several decades, researchers have studied structural linguistic or sociolinguistic aspects of code-
switching, often on the basis of linguistic corpora (for overviews, see Bullock & Toribio, 2009;
Guzzardo Tamargo et al., 2016; Isurin et al., 2009; Stell & Yakpo, 2015). Structural linguistic
approaches to code-switching seek to identify universal grammatical constraints on code-switching.
Several structural accounts have been particularly infuential in the study of code-switching, includ-
ing Poplack’s (1980) model of the equivalence constraint and the free morpheme constraint, Myers-
Scotton’s matrix language framework (e.g., Myers-Scotton, 1993), and the minimalist program
approach proposed by MacSwan (2000). Focusing on lexical constraints, the triggering hypothesis,
originally proposed by Clyne (1967, 2003), posits that code-switches are more likely to occur in
the presence of language-ambiguous words, such as cognates, i.e., words that share semantics, pho-
nology, and orthography across languages, such as Dutch-English appel–apple or Spanish-English
piano–piano.
Sociolinguistic perspectives on code-switching examine the social, cultural, and communicative
motivations driving code-switched speech, often studied through observations of code-switched dis-
course in naturalistic settings (Stell & Yakpo, 2015). Building on seminal work by Weinreich (1953)
and Blom and Gumperz (1972), researchers have studied how code-switching patterns vary with, for
example, language-ideological settings and language privileges (e.g., Muysken, 2000), the extent to
which code-switching is socially acceptable within a community (e.g., Torres Cacoullos & Travis,
2018) or characteristics of the interlocutors, such as language background (e.g., Bentahila & Davies,
1992) or language attitudes (e.g., Badiola et al., 2018).
Linguistic and sociolinguistic perspectives have traditionally been tested via analyses of corpora
of code-switched speech or observations of natural discourse. More recently, these perspectives have
been examined in studies that used psycholinguistic and neurocognitive experimental research tech-
niques to elicit code-switched speech (e.g., Fairchild & Van Hell, 2017; Kootstra et al., 2010; Lipski,
2019) or examine the online processing of code-switched sentences (e.g., Dussias, 2001; Vaughan-
Evans et al., 2020; Suurmeijer et al., 2020; for reviews, see Beatty-Martínez et al., 2018; Valdés Krof
et al., 2018; Van Hell et al., 2018). These studies typically merge linguistic or sociolinguistic perspec-
tives on code-switching with cognitive processing models of speech production or comprehension
and manipulate linguistic materials and experimental conditions to test specifc predictions on the
production or comprehension of code-switched sentences. This approach will be illustrated by dis-
cussing psycholinguistic research that tested specifc linguistic and sociolinguistic perspectives: the
triggering hypothesis and the impact of interlocutor characteristics (see Sections “Triggering hypoth-
esis” and “Sociolinguistic perspectives: the impact of interlocutor characteristics”, respectively).
Another focus in psycholinguistic research on intra-sentential code-switching focuses on (1)
cognitive eforts and (2) cognitive control mechanisms associated with switching. This perspective
originates from a long tradition in the broader feld of cognitive psychology on switching costs and
cognitive control (for a review, see Bobb & Wodniecka, 2013), and applies these cognitive mecha-
nisms to the production and comprehension of code-switched sentences. As discussed in more detail
in Section “Factors that modulate switching costs”, psycholinguistic and neurocognitive research on
intra-sentential code-switching has adapted the notion of switching costs to examine the extent to
which more naturalistic intra-sentential switching incurs processing cost and which factors mitigate
these costs, including switching direction and bilinguals’ daily language use and code-switching habits.
A second line of psycholinguistic research on code-switching that builds on decades of research
in cognitive psychology focuses on cognitive control mechanisms that are associated with the pro-
duction of code-switched speech. In their control process model (CPM) of code-switched speech,
257
Janet G. van Hell
Green and Wei (2014; see also Green, 2018) relate cognitive control processes to code-switching; this
model will be further discussed in Section “Control process model and intra-sentential code-switching”.
Theoretical perspectives and empirical studies

I will discuss four diferent theoretical approaches to intra-sentential code-switching and outline the
empirical work in these domains by highlighting illustrative studies. Two positions originate in the
linguistic literature (triggering theory, socio-pragmatics of interlocutor characteristics) and two in the
psychological literature (modulators of switching costs and the control process model).
Triggering hypothesis
The triggering hypothesis was proposed by Clyne (e.g., 1967, 2003) based on corpus data on Ger-
man, Dutch, Hungarian, Italian, Spanish, Croatian, and Vietnamese immigrants in Australia. Clyne
noted that code-switches tend to occur when a sentence contains one or more cognates. He pro-
posed that cognates, as well as other language-ambiguous words like proper nouns and homophones
(words that share phonology, but not meaning across languages), can function as triggers that facili-
tate a switch to the other language. For example, the Croatian-English cognate “tennis” triggered
a switch to English in “Imam puno zadaca I sutra mi igramo tennis . . . that’s about all” ([“I have a
lot of assignments and tomorrow we are playing tennis . . . that’s about all”]; Clyne, 2003, p. 164).
Broersma and De Bot (2006) integrated Clyne’s original triggering hypothesis with psycholinguistic
models of bilingual speech production in their adjusted triggering hypothesis. Assuming that trig-
gering occurs at the lemma level, this adjusted triggering hypothesis states that the selection of the
lemma of a trigger word (e.g., a cognate) co-activates words in the currently used language A, but
also words in the speaker’s other language B. This increased activation of language B makes it more
likely that the speaker will subsequently select lemmas from language B as the sentence progresses.
The notion that cognates co-activate words in two languages and boost activation of the non-target
language is in line with a wealth of studies on bilingual lexical processing reporting that cognates are
typically produced and recognized more quickly and more accurately than noncognates (for a review,
see Van Hell & Tanner, 2012).
The idea that code-switches can be triggered by language-ambiguous words, in particular cog-
nates, has been tested in various corpus analyses (Broersma, 2009; Broersma & De Bot, 2006; Bro-
ersma et al., 2020; Broersma et al., 2009; Fricke & Kootstra, 2016; Soto et al., 2018) and experimental
studies (Broersma, 2011; Bultena et al., 2015a, 2015b; Gullifer & Titone, 2019; Kootstra et al., 2012,
2020; Li & Gollan, 2018). In the frst systematic test of the triggering hypothesis, Broersma and De
Bot (2006) analyzed conversations of three Moroccan Arabic-Dutch bilinguals and coded code-
switches and trigger words (cognates and proper nouns). They found that code-switches occurred
more frequently directly after a trigger word than could be expected by chance, especially if the
trigger and code-switch appeared in the same clause. These fndings were paralleled in subsequent
corpus studies analyzing conversational speech of Dutch immigrants in New Zealand and Australia
(e.g., Broersma, 2009), Russian immigrants in the USA (Broersma et al., 2009), and Welsh-English
bilingual speakers (Broersma et al., 2020). In the latter study, Broersma et al. (2020) analyzed a large-
scale corpus of over 50 natural conversations of approximately 100 Welsh-English bilingual speakers.
Not only did they observe more code-switches in clauses containing cognates than in clauses without
a cognate, they also found that a higher number of cognates within a clause increased the likelihood
that speakers switched at some point later in the utterance. Furthermore, speakers who used more
cognates code-switched more frequently throughout the conversation, but a speaker’s production
of cognates did not increase the likelihood that their discourse partner code-switched when they
started to talk in turn. This indicates that hearing (rather than producing) a lexical trigger does
258
Code-switching
not afect the listener’s likelihood to code-switch when it was their turn to speak. Together, these
corpus studies provide compelling quantitative evidence that trigger words co-occur with code-
switches in a speaker’s utterance, and that these efects can be observed across a variety of diferent
language combinations.
A second line of research testing the predictions of the triggering hypothesis sought to provide
more insight into the processing mechanisms underlying triggered code-switching, using experi-
mental techniques that enable specifc manipulations of critical variables (Broersma, 2011; Bultena
et al., 2015a, 2015b; Gullifer & Titone, 2019; Kootstra et al., 2012, 2020; Li & Gollan, 2018). Even
though these techniques generally come with less natural and ecologically valid discourse situations,
they allow for a more systematic control of potentially confounding or modulating factors. The
evidence provides mixed support for the lexical triggering hypothesis, particularly in the context of
intra-sentential code-switching, with studies using picture naming paradigms for nouns mostly sup-
porting the triggering hypothesis (Broersma, 2011; Li & Gollan, 2018).
Using the structural priming technique (see Bernolet, 2023 [this volume]), Kootstra et al. (2012)
frst asked Dutch-English bilinguals to repeat a code-switched sentence that started in Dutch and
switched into English, and then describe a target picture using a sentence that switched from Dutch
to English. The object noun in the prime sentence (e.g., “bal” in “De jongen gooit een bal to the
butcher” [“The boy throws a ball to the butcher”]) was either a cognate (here: bal—ball) or a non-
cognate, and was either repeated in the subsequent target picture (e.g., a picture of a boy throwing a
ball to a diver) or not. Analyses of the target sentences produced to describe the pictures showed that
the location of the switch in the target sentences was more likely to be similar to the switch location
in the prime sentence when the object noun was a cognate than a noncognate, and when the same
noun was repeated across prime and target sentences. These cognate-related switching efects were
more pronounced in the group of highly profcient bilinguals than in lower-profciency bilinguals.
This study suggests that cognates enhance the likelihood that the switch location in a spoken sentence
aligns with the switch location of a previously produced utterance, indicating that cognate nouns can
afect code-switching patterns.
Mimicking a dialogue situation using confederate-scripted priming, Kootstra et al. (2020) exam-
ined the impact of diferent types of lexical triggers on the production of code-switched sentences.
Dutch-English bilingual participants and confederates took turns in describing a pictured event to
each other after which their discourse partner had to select the correct picture from their screen.
Each picture contained an actor, an action, and a patient (e.g., a picture of a hunter putting a rose
on a chair). The patient was either a cognate (e.g., roos–rose), an interlingual homophone (rok [skirt]–
rock), or a noncognate (e.g., fets–bike). The confederate was scripted to code-switch from Dutch to
English directly after the critical trigger word (cognate or homophone) or the non-trigger noncog-
nate in half of the picture descriptions, or produce the entire sentence in Dutch in the remaining
half. After the participant had selected the picture just described from two pictures on their screen,
the participant was presented with a pictured event they had to describe. The patient objects in the
participants’ pictures (e.g., a woman putting a baby on a chair) were either cognates, homophones,
or noncognates. Participants were free to use English or Dutch and were not forced to code-switch.
The critical question was whether participants would code-switch more frequently when their pic-
ture contained a trigger word, and whether their code-switching behavior would be afected by that
of the confederate. Participants code-switched more frequently when pictures contained a cognate
rather than a noncognate, but only when their discourse partner had code-switched in the preceding
trial. Homophones did not enhance the likelihood of switching.
Cognate triggering efects in sentences, if they occur at all, may be restricted to noun cognates
and do not emerge in verb cognates (Bultena et al., 2015a, 2015b). Using a shadowing task, Bultena
et al. (2015a) tested whether verb cognates would modulate the production of code-switches pre-
sented downstream in the sentence. Dutch-English bilinguals listened to sentences that started in
259
Janet G. van Hell
Dutch and switched into English, or vice versa; the code-switch was preceded by a cognate or
noncognate verb. Upon the start of the sentence, participants were asked to repeat (“shadow”) what
they heard as quickly and as accurately as possible. Shadowing latencies (delay between word onset
in the original recording and the participant’s reproduction of the word) showed that producing verb
cognates or noncognates did not afect the later production of the code-switch, in either switching
direction and in both syntactic structures tested (word orders shared [SVO] or not shared [XVSO]
across Dutch and English).
Bultena et al.’s (2015a) fndings were paralleled in a self-paced reading study in Dutch-English
bilinguals (Bultena et al., 2015b): Reading a verb cognate did not facilitate the reading of the sub-
sequent code-switched word. Testing French-English bilinguals, Gullifer and Titone (2019) also
observed that the reading of switched nouns with higher or lower form overlap that appeared early
in the sentence did not afect the reading of words downstream in the sentence.
In sum, quantitative analyses of corpus studies suggest that lexical triggers co-occur with code-
switches in diferent types of bilingual speakers, in line with Clyne’s (1967, 2003) original trigger
hypothesis and the adjusted triggering hypotheses proposed by Broersma and De Bot (2006). How-
ever, the psycholinguistic studies using controlled lab-based experimental techniques indicate that
lexical triggering in and of itself is not a fundamental cognitive mechanism that elicits code-switched
speech in all circumstances or afects the processing of code-switched sentences. Rather, in a speak-
ing context in which bilinguals were free to code-switch (or not), cognates or false friends only
triggered a code-switch when the speaker’s discourse partner had just code-switched (Kootstra et al.,
2020), suggesting that lexical triggering is restricted to discourse situations in which code-switching
occurs frequently. Actually, this discourse situation best mimics the contextual situation in which
linguistic corpora are collected, suggesting that quantitative analyses of code-switches in corpora may
refect code-switching behavior in specifc discourse situations.
Sociolinguistic perspectives: the impact of interlocutor characteristics

Recently, psycholinguistic studies have adopted sociolinguistic perspectives on code-switching and
use experimental research techniques to examine how social factors afect the comprehension and
production of intra-sentential code-switches. Here, I will highlight psycholinguistic and neuro-
cognitive studies that examined how interlocutor characteristics infuence code-switching patterns
(Blanco-Elorrieta & Pylkkänen, 2017; Grosjean, 1997; Kaan et al., 2020; Kapiley & Mishra, 2018).
To examine how language mode afected bilinguals’ code-switching behavior, Grosjean (1997)
had 15 French-English bilinguals, all French native speakers who had immigrated to the United
States, retell stories to three French interlocutors who also lived in the United States. Half the sto-
ries were French-only, and described typical French situations accompanied by cartoons of French
scenes. The other half of the stories were also in French, but contained English code-switches and
described typical American situations. Prior to the story retelling task, participants were told that the
interlocutors difered in their fuency in English and their attitude towards code-switching (ranging
from being a French purist to having positive attitudes towards code-switching). Diferent code-
switching patterns emerged during the retelling of the stories: The French “purist” interlocutor
and French-only stories triggered the least amount of code-switches in the participants’ retellings,
whereas retelling stories to the interlocutor with the most positive attitude towards code-switching
elicited the most code-switches.
More recently, Blanco-Elorrieta and Pylkkänen (2017) examined neural responses associated with
the production and comprehension of cued language switches in Arabic-English bilinguals, using
magneto-encephalography. The cues were either static portraits of individuals who had been intro-
duced as bilinguals or monolinguals (social cue), or were color cues indicating a particular language
(color cue). In the cued picture naming task, the bilingual portrait cue indicated that participants
260
Code-switching
could freely choose the output language, and the monolingual portrait cue indicated they had to
name in the language of the monolingual. The color cue indicated which particular language to use.
In the comprehension task, bilinguals were presented with one of the cues, followed by an audi-
tory stimulus. They then had to judge whether this auditory stimulus matched the picture that was
presented next. The fnding most relevant to the present discussion was behavioral switching costs
and measurable neural efects in the executive function network were smaller, or even absent, in the
socially cued conditions than in the condition that used color as a switch cue.
Awareness of interlocutor identity also infuenced code-switching patterns in Telugu-English
bilinguals (Kapiley & Mishra, 2018). Participants were introduced to cartoon images of interlocutors
who were either more or less profcient in their second language (L2) English. In a picture nam-
ing task, speakers switched into L2 English less often in the presence of low-profciency cartoons
as opposed to high-profciency cartoons. This suggests that bilingual speakers are sensitive to the
identity of their interlocutor and their communicative needs, which infuences how they plan their
language use and switching behavior.
Studying intra-sentential code-switching rather than isolated item-switching (Blanco-Elorrieta &
Pylkkänen, 2017; Kapiley & Mishra, 2018), Kaan et al. (2020) had Spanish-English bilinguals read
sentences with or without a code-switch in the presence of an individual who they knew was a
Spanish-English bilingual speaker or an English monolingual speaker. The rationale here is that code-
switching is not socially allowed in the presence of a monolingual, which may then afect the degree
to which the bilingual participants expect a switch or co-activate their Spanish knowledge. Brain
activity was recorded during reading using the event-related potential technique. Switches elicited
an early fronto-central positivity, which was attenuated in the presence of the bilingual speaker dur-
ing the frst part of the study. In addition, the switch-related late positivity was smaller in the pres-
ence of a bilingual than a monolingual, but only for participants who were sensitive to the other’s
language knowledge as measured in their ofine judgments. Kaan et al. (2020) take these results to
imply that, when initially joined by a bilingual individual, bilingual readers expected and activated
both languages and more easily accommodated intra-sentential code-switches in the presence of a
bilingual speaker relative to a monolingual speaker, suggesting that sentence revision and updating
is easier when code-switching is socially permitted. This study provides important insights into the
prominent role of social factors, like speaker identity, on the comprehension and production of intra-
sentential code-switches: Even the identity of silent bystanders can afect bilinguals’ quiet reading of
code-switched sentences.
The fndings of the studies discussed in this section support the idea that code-switching patterns
are afected by speaker identity and perceived language knowledge of the individual (real or imag-
ined) present in the code-switching context, in line with sociolinguistic perspectives on code-switch-
ing. Moreover, the corollary that language control associated with the production or processing of
code-switched utterances can be modulated by interlocutor characteristics aligns with the control
process model posited by Green and Wei (2014; Green, 2018), which will be discussed in more detail
in Section “Control process model and intra-sentential code-switching”.
Factors that modulate switching costs

Psycholinguistic and neurocognitive research on cognitive eforts associated with intra-sentential
code-switching is largely based on the psycholinguistic literature on processing costs that reports lon-
ger response times associated with switching languages when producing or processing single items,
such as pictures, numbers, or words (Bobb & Wodniecka, 2013). Even though bilinguals often report
that producing or listening to code-switched speech in natural discourse is efortless, analyses of
natural discourse in linguistic corpora confrm that code-switching comes with measurable changes
in speech variables. In their analysis of spontaneous code-switches in the Bangor Miami corpus
261
Janet G. van Hell
(Deuchar et al., 2014), Fricke et al. (2016) found that bilingual speakers slowed their speech rate in
anticipation of a language switch and adapted VOTs of words close to an upcoming switch (see also
Balukas & Koops, 2015; but see Johns & Steuck, 2021).
Recent psycholinguistic and neurocognitive studies on intra-sentential code-switching examined
which factors mitigate cognitive eforts associated with the production or comprehension of code-
switched utterances. One modulating factor is switching direction (Bultena et al., 2015a, 2015b;
Gullifer & Titone, 2019; Suurmeijer et al., 2020). Specifcally, switching from the dominant lan-
guage or L1 into the weaker language of L2 has been found to be cognitively more efortful than
switching in the opposite direction, as evidenced in reduced shadowing and reading times (Bultena
et al., 2015a; b). Electrophysiological research using ERP and time-frequency analyses also indicates
that neurocognitive processes involved in reading (Litcofsky & Van Hell, 2017) and listening to (Fer-
nandez et al., 2019) code-switched sentences vary by switching direction, as evidenced by diferent
neural signatures associated with the two switching directions (see Van Hell et al., 2018, for more
details). More specifcally, processing sentences that begin in the dominant language and halfway
switch into the weaker language engage lexical processes as well as sentence-level reanalysis to inte-
grate the weaker language into the sentence structure the comprehender is incrementally building. In
contrast, processing sentences that switch from the weaker to the dominant language engages lexical-
semantic integration accompanied by inhibition processes: As the sentence starts to unfold in their
weaker language, comprehenders must inhibit their dominant language, and this inhibition must be
released upon hearing a switch into the dominant language.
Bilinguals’ daily code-switching practices and language experience also afect cognitive eforts
associated with intra-sentential code-switching (Adamou & Shen, 2019; Beatty-Martínez & Dussias,
2017; Kheder & Kaan, 2016; Valdés Krof et al., 2018; Valdés Krof et al., 2020). For example,
presenting written code-switched sentences while recording participants’ ERPs, Beatty-Martínez
and Dussias (2017) observed an early frontal positivity in Spanish-English bilinguals recruited from
an environment in which code-switching was rather uncommon (Spain), whereas Spanish-English
bilinguals who frequently code-switched (recruited in the USA) did not show this ERP signature.
Testing fuent Romani-Turkish bilinguals recruited from a community in which code-switching is
highly common, Adamou and Shen (2019) also observed no switching costs for sentences that con-
tained Turkish verbs that are frequently used in spontaneous conversations.
In sum, rather than considering switching costs as a given in bilinguals’ switching between lan-
guages, an increasing number of studies on inter-sentential code-switching seek to understand which
factors mitigate cognitive eforts associated with code-switching. These studies often relate their
fndings to cognitive models of language processing and experienced-based or usage-based accounts,
positing a strong link between cognitive representations and frequency of use of particular construc-
tions, and argue that language processing is shaped by expectations based on recent and long-term
exposure to the language (e.g., Backus, 2015).
Control process model and intra-sentential code-switching

Relating code-switching to cognitive control, Green and Wei (2014) proposed the control process
model to account for code-switching in production (see also, Green, 2018). In a nutshell, this model
posits that language task schemas that govern cognitive control processes cooperate in some contexts
but compete in others, depending on specifcs of the interactional context (single language, dual lan-
guage, and dense code-switching contexts). Competitive control operates in a single or dual language
context when only one language is selected, and items from the language not in use are blocked from
entering the language planning system. Coordinate control operates in a code-switching context,
involving insertion, alternation, and congruent lexicalization (Muysken, 2000) in one of two ways:
coupled control or open control. Under coupled control, control passes back and forth such that the
262
Code-switching
most appropriate item enters the language planning system. Alternation and insertion are realized by
coupled control: The matrix language turns over control to the other language to allow the intended
alternation or insertion after which control is returned. Open control realizes congruent lexicaliza-
tion: The control processes have no top-down control or biasing input and allow items from either
language to enter the planned utterance on an opportunistic basis, whichever is most appropriate at
a given time. A further prediction of the model is that under open control, language switching costs
are minimized, whereas coupled control can confer a switching cost.
So far, most studies testing the control process model have examined how bilinguals’ language
experience and code-switching practices modulate performance on executive function tasks, such
as Flanker (Adler et al., 2019; Hofweber et al., 2020), verbal Stroop and non-verbal global—local
(Lai & O’Brien, 2020), or task-switching (Hartanto & Yang, 2016) tasks (for discussion, see Poarch,
2023 [this volume]). An emergent psycholinguistic literature on the processing of code-switched
sentences focuses on the model’s assumption that language switching costs are related to diferent
types of cognitive control (Beatty-Martínez & Dussias, 2017; Broersma et al., 2020). For example,
Beatty-Martínez and Dussias (2017) interpret the early positivity present in non-habitual code-
switchers but absent in habitual code-switchers as an attentional shift from a more competitive to a
more cooperative state of bilingual language control. Further research is needed to test the model’s
predictions related to code-switched sentence processing.

A growing body of research on code-switched sentence processing examines the cognitive and
neural mechanisms that underlie the comprehension and production of code-switched utterances,
using quantitative corpus analyses, laboratory-based experimental-behavioral, and neurocognitive
techniques that seek to maximize experimental rigor while maintaining the integrity of natural
code-switched sentences common in bilingual discourse. These studies have provided a wealth of
insights, but there are still many topics that require further experimentation or questions that as of
yet remained largely unexplored (see Textbox 21.2).
One of the topics requiring further experimentation pertains to the role of bilinguals’ code-
switching practices and experiences on the production and comprehension of code-switched sen-
tences. Green and Wei’s (2014) control process model that relates variability in bilingual experience
to diferent types of language control is an excellent starting point to examine these factors in the
context of the production and comprehension of code-switched sentences.
Furthermore, much of our current insights into the cognitive and neural mechanism of the com-
prehension of code-switched sentences is based on reading rather than listening to code-switched
utterances. Because code-switching occurs more frequently in spoken than in written discourse,
more knowledge on the comprehension of code-switched speech is needed. Studying code-switched
speech also opens up a new line of research into the role of speaker identity on code-switched sen-
tence processing. For example, many bilinguals who code-switch have a perceivable accent in one or
in both of their languages, and accented speech may cue or facilitate the listeners’ comprehension of
code-switched speech (cf. Fernandez & Van Hell, submitted). Relatedly, studying speech compre-
hension using the visual world paradigm, Fricke et al. (2016) found that bilingual listeners can exploit
low-level phonetic cues to anticipate that a code-switch is coming. Do listeners exploit these cues
to predict an upcoming switch and how do these cues afect their comprehension of code-switched
speech downstream? How does the listener’s knowledge of a speaker’s code-switching practices and
language experiences modulate these mechanisms?
In addition, theoretical models on the cognitive and neural basis of language processing need to
be tested in experimental studies, but also via computational or neural network modeling. Some frst
eforts have been made to test some specifc theories of code-switched sentence processing using
263
Janet G. van Hell
computational modeling (Cohen et al., 2021; Tsoukala et al., 2021), but clearly more work is needed
in this area as well as in the neural network domain.
As of yet, most psycholinguistic work on the comprehension and production of code-switched
sentences studied Indo-European languages, recruiting bilinguals in the USA and Europe. We know
little about the cognitive and neural mechanisms of code-switching in bilingual/multilingual com-
munities in other regions of the world, for example, the African continent, Oceania, or Southeast
Asia. We also know very little about code-switching patterns in trilingual speakers.
Finally, code-switching in bimodal bilinguals, users of a signed language and a spoken language,
is particularly interesting, because in their code-switching behavior (termed code-blending) propo-
sitional content can be expressed simultaneously (saying “cat” and signing “cat” at the same time),
rather than linearly as in spoken languages (Emmorey et al., 2008; Emmorey et al., 2020). Exploiting
the simultaneity ofered by the spoken and manual modalities, these code-blends provide unique
insights into theoretical models of sentence production and comprehension.
How do bilinguals’ code-switching practices and experiences infuence the production and comprehen-
sion of code-switched sentences?
How does the listener’s knowledge of a speaker’s code-switching practices and language experiences
modulate cognitive and neural processes associated with the comprehension of code-switched speech?
Code-switching across the lifespan: What are the developmental dynamics of the comprehension and
production of code-switched speech across the lifespan?
Acknowledgements
The writing of this chapter was supported by NSF Grants BCS-1349110 and OISE-1545900.
Further reading
Special Issue in Frontiers in Psychology, March, 2021: Behavioral and neurophysiological approaches to codeswitch-
ing and language switching. Editors: J. Trefers-Daller, E. Ruigendijk, and J. Hofweber.
This issue contains 19 articles addressing (1) the relationship between code-switching/language-switching
and cognitive control; (2) linguistic processing of code-switches/language-switches; (3) neural and elec-
trophysiological correlates of switching; and (4) linguistic and orthographic analyses of code-switches/
language-switches.
Special Issue in Linguistic Approaches to Bilingualism, 2018, Issue 1: Methodologies for intra-sentential code-
switching research. Editors: A. Munarriz-Ibarrola, M. C. Parafta Couto, and E. Vanden Wyngaerd.
This issue contains six articles describing psycholinguistic methodologies used in intra-sentential code-
switching research.
References
Adamou, E., & Shen, X. R. (2019). There are no language switching costs when codeswitching is frequent.
International Journal of Bilingualism, 23, 53–70.
Adler, R. M., Valdés Krof, J. R., & Novick, J. M. (2019). Does integrating a code-switch during comprehen-
sion engage cognitive control? Journal of Experimental Psychology: Learning, Memory, and Cognition, 46(4),
741–759.
264
Code-switching
Backus, A. (2015). A usage-based approach to code-switching: The need for reconciling structure and function.
In G. Stell & K. Yakno (Eds.), Code-switching between structural and sociolinguistic perspectives (pp. 19–38). De
Gruyter.
Badiola, L., Delgado, R., Sande, A., & Stefanich, S. (2018). Code-switching attitudes and their efects on
acceptability judgment tasks. Linguistic Approaches to Bilingualism, 8, 5–24.
Balukas, C., & Koops, C. (2015). Spanish-English bilingual voice-onset time in spontaneous code-switching.
International Journal of Bilingualism, 19, 423–443.
Beatty-Martínez, A. L., & Dussias, P. E. (2017). Bilingual experience shapes language processing: Evidence from
codeswitching. Journal of Memory and Language, 95, 173–189.
Beatty-Martínez, A. L., Navarro-Torres, C. A., & Dussias, P. E. (2020). Codeswitching: A bilingual toolkit for
opportunistic speech planning. Frontiers in Psychology, 11, 1699.
Beatty-Martínez, A. L., Valdés Krof, J. R., & Dussias, P. E. (2018). From the feld to the lab: A converging
methods approach to the study of code-switching. Languages, 3(2), 19.
Bentahila, A., & Davies, E. (1992). Code-switching and language dominance. Advances in Psychology, 83,
443–468.
Bernolet. S. (2023). Studying second language production. In A. Godfroid & H. Hopp (Eds.). The Routledge
Blanco-Elorrieta, E., & Pylkkänen, L. (2017). Bilingual language switching in the laboratory versus in the wild:
The spatiotemporal dynamics of adaptive language control. Journal of Neuroscience, 37, 9022–9036.
Blom, J. P., & Gumperz, J. (1972). Social meaning in linguistic structures: Codeswitching in northern Norway. In
J. Gumperz & D. Hymes (Eds.), Directions in sociolinguistics: The ethnography of communication (pp. 407–434). Holt.
Bobb, S. C., & Wodniecka, Z. (2013). Language switching in picture naming: What asymmetric switch costs
(do not) tell us about inhibition in bilingual speech planning. Journal of Cognitive Psychology, 25, 568–585.
Broersma, M. (2009). Triggered codeswitching between cognate languages. Bilingualism: Language and Cognition,
12, 447–462.
Broersma, M. (2011). Triggered code-switching: Evidence from picture naming experiments. In M. S. Schmid
& W. Lowie (Eds.), Modeling bilingualism from structure to chaos: In honor of Kees de Bot (pp. 37–57). John
Benjamins.
Broersma, M., Carter, D., Donnelly, K., & Konopka, A. (2020). Triggered codeswitching: Lexical processing
and conversational dynamics. Bilingualism: Language and Cognition, 23, 295–308.
Broersma, M., & de Bot, K. (2006). Triggered codeswitching: A corpus-based evaluation of the original trig-
gering hypothesis and a new alternative. Bilingualism Language and Cognition, 9, 1–13.
Broersma, M., Isurin, L., Bultena, S., & De Bot, K. (2009). Triggered code switching: Evidence from Dutch-
English and Russian-English bilinguals. In L. Isurin, D. Winford, & K. de Bot (Eds.), Multidisciplinary
approaches to code switching (pp. 103–128). John Benjamins.
Bullock, B. E., & Toribio, A. J. (2009). The Cambridge handbook of linguistic code-switching. Cambridge University
Press.
Bultena, S., Dijkstra, T., & Van Hell, J. G. (2015a). Switch cost modulations in bilingual sentence processing:
Evidence from shadowing. Language, Cognition, and Neuroscience, 30, 586–605.
Bultena, S., Dijkstra, T., & Van Hell, J. G. (2015b). Language switch costs in comprehension depend on language
dominance: Evidence from self-paced reading. Bilingualism: Language and Cognition, 18, 453–469.
Clyne, M. (1967). Transference and triggering. Nijhof.
Clyne, M. (2003). Dynamics of language contact. Cambridge University Press.
Cohen, C., Nabi, S. W., Higham, C., Putnam, M., Kootstra, G. J., & Van Hell, J. G. (2021). Individual variation
in the structure of bilingual grammars. Language, 97(4), 752–792.
Deuchar, M., Davies, P., Herring, J., Parafta Couto, M. C., & Carter, D. (2014). Building bilingual corpora. In
E. M. Thomas & I. Mennen (Eds.), Advances in the study of bilingualism (pp. 93–110). Multilingual Matters.
Deuchar, M., Muysken, P., & Wang, S. (2007). Structured variation in codeswitching: Towards an empirically
based typology of bilingual speech patterns. The International Journal of Bilingual Education and Bilingualism,
10(3), 298–340.
Dussias, P. E. (2001). Psycholinguistic complexity in codeswitching. International Journal of Bilingualism, 5,
87–100.
Emmorey, K., Borinstein, H. B., Thompson, R., & Gollan, T. H. (2008). Bimodal bilingualism. Bilingualism:
Emmorey, K., Li, C., Petrich, J., & Gollan, T. H. (2020). Turning languages on and of: Switching into and out
of code-blends reveals the nature of bilingual language control. Journal of Experimental Psychology: Learning,
Memory, and Cognition, 46(3), 443–454.
Fairchild, S., & Van Hell, J. G. (2017). Determiner-noun code-switching in Spanish-English heritage speakers.
265
Janet G. van Hell
Fernandez, C. B., Litcofsky, K. A., & Van Hell, J. G. (2019). Neural correlates of intra-sentential code-switching
in the auditory modality. Journal of Neurolinguistics, 51, 17–41.
Fernandez, C., & Van Hell, J. G. (submitted). Accented speech modulates bilingual listeners’ comprehension of
code-switched sentences: Evidence from event-related potentials.
Fricke, M., & Kootstra, G. J. (2016). Primed codeswitching in spontaneous bilingual dialogue. Journal of Memory
Fricke, M., Kroll, J. F., & Dussias, P. E. (2016). Phonetic variation in bilingual speech: A lens for studying the
production-comprehension link. Journal of Memory and Language, 89, 110–137.
Gollan, T. H., & Ferreira, V. S. (2009). Should I stay or should I switch? A cost-beneft analysis of voluntary
language switching in young and aging bilinguals. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 35, 640–665.
Green, D. W. (2018). Language control and codeswitching. Languages, 3, 8.
Green, D. W., & Wei, L. (2014). A control process model of code-switching. Language, Cognition and Neurosci-
ence, 29, 499–511.
Grosjean, F. (1997). Processing mixed language: Issues, fndings, and models. In A. M. B. de Groot & J. F. Kroll
(Eds.), Tutorials in bilingualism: Psycholinguistic perspectives (pp. 225–253). Lawrence Erlbaum.
Grosjean, F. (2010). Bilingual: Life and reality. Cambridge University Press.
Gullberg, M., Indefrey, P., & Muysken, P. (2009). Research techniques for the study of code-switching. In B.
E. Bullock & A. J. Toribio (Eds.), The Cambridge handbook of linguistic code-switching (pp. 21–39). Cambridge
University Press.
Gullifer, J. W., Kroll, J. F., & Dussias, P. E. (2013). When language switching has no apparent cost: Lexical access
in sentence context. Frontiers in Psychology, 4, 1–13.
Gullifer, J. W., & Titone, D. (2019). The impact of a momentary language switch on bilingual reading: Intense
at the switch but merciful downstream for L2 but not L1 readers. Journal of Experimental Psychology: Learning,
Memory, and Cognition, 45, 236–250.
Guzzardo Tamargo, R. E., Mazak, C. M., & Parafta Couto, M. C. (2016). Spanish-English codeswitching in the
Caribbean and the US. John Benjamins.
Guzzardo Tamargo, R. E., Valdés Krof, J. R., & Dussias, P. E. (2016). Examining the relationship between
comprehension and production processes in codeswitched language. Journal of Memory and Language, 89,
138–161.
Hartanto, A., & Yang, H. (2016). Disparate bilingual experiences modulate task-switching advantages: A difu-
sion-model analysis of the efects of interactional context on switch costs. Cognition, 150, 10–19.
Heredia, R. R., & Altarriba, J. (2001). Bilingual language mixing: Why do bilinguals code-switch? Current
Directions in Psychological Science, 10, 164–168.
Hofweber, J., Marinis, T., & Trefers-Daller, J. (2020). How diferent code-switching types modulate bilinguals’
executive functions: A dual control mode perspective. Bilingualism, Lanuage, & Cognition, 23, 909–925.
Ibáñez, A. J., Macizo, P., & Bajo, M. T. (2010). Language access and language selection in professional translators.
Acta Psychologica, 135, 257–266.
Isurin, L., Winford, D., & De Bot, K. (2009). Multidisciplinary approaches to code switching. John Benjamins.
Johns, M. A., & Steuck, J. (2021). Is codeswitching easy or difcult? Testing processing cost through the prosodic
structure of bilingual speech. Cognition, 211, 104634.
Kaan, E., Kheder, S., Kreidler, A., Tomic, A., & Valdés Krof, J. R. (2020). Processing codeswitches in the pres-
ence of others: An ERP study. Frontiers in Psychology, 11, 1288.
Kapiley, K., & Mishra, R. K. (2018). What do I choose? Infuence of interlocutor awareness on bilingual lan-
guage choice during voluntary object naming. Bilingualism: Language & Cognition, 22, 1029–1051.
Kheder, S., & Kaan, E. (2016). Processing code-switching in Algerian bilinguals: Efects of language use and
semantic expectancy. Frontiers in Psychology, 7, 248.
Kiesel, A., Steinhauser, M., Wendt, M., Falkenstein, M., Jost, K., Phillip, A., & Koch, I. (2010). Control and
interference in task switching—A review. Psychological Bulletin, 136, 849–874.
Kootstra, G. J., Dijkstra, T., & Van Hell, J. G. (2020). Interactive alignment and lexical triggering of code-
switching in bilingual dialogue. Frontiers in Psychology, 11, 1747.
Kootstra, G. J., van Hell, J. G., & Dijkstra, T. (2010). Syntactic alignment and shared word order in code-
switched sentence production: Evidence from bilingual monologue and dialogue. Journal of Memory and
Language, 63, 210–231.
Kootstra, G. J., Van Hell, J. G., & Dijkstra, T. (2012). Priming of code-switches in sentences: The role of lexical
repetition, cognates, and language profciency. Bilingualism: Language and Cognition, 15, 797–819.
Kroll, J. F., Bobb, S., & Wodniekca, Z. (2006). Language selectivity is the exception, not the rule: Arguments
against a fxed locus of language selection in bilingual speech. Bilingualism Language and Cognition, 9, 119–135.
266
Code-switching
Lai, G., & O’Brien, B. A. (2020). Examining language switching and cognitive control through the adaptive
control hypothesis. Frontiers in Psychology, 11, 1171.
Li, C., & Gollan, T. (2018). Cognates facilitate switches and then confusion: Contrasting efects of cascade
versus feedback on language selection. Journal of Experimental Psychology: Learning, Memory, & Cognition, 44,
974–991.
Lipski, J. M. (2019). Field-testing code-switching constraints: A report on a strategic languages project. Lan-
guages, 4, 7.
Litcofsky, K. A., & Van Hell, J. G. (2017). Switching direction afects switching costs: Behavioral, ERP and
time-frequency analyses of intra-sentential codeswitching. Neuropsychologia, 97, 112–139.
MacSwan, J. (2000). The architecture of the bilingual language faculty: Evidence from intrasentential code
switching. Bilingualism: Language and Cognition, 3, 37–54.
Muysken, P. (2000). Bilingual speech: A typology of code mixing. Cambridge University Press.
Myers-Scotton, C. (1993). Duelling languages. Grammatical structure in code-switching. Oxford University Press.
Pérez, A., & Duñabeitia, J. A. (2019). Speech perception in bilingual contexts: Neuropsychological impact of
mixing languages at the inter-sentential level. Journal of Neurolinguistics, 51, 258–267.
Pfaf, C. (1979). Constraints on language-mixing: Intrasentential code-switching and borrowing in Spanish/
English. Language, 55, 291–318.
Poarch. G. J. (2023). Cognitive efects of bilingualism. In A. Godfroid & H. Hopp (Eds.). The Routledge handbook
Poplack, S. (1980). Sometimes I’ll start a sentence in Spanish Y TERMINO EN ESPAÑOL: Toward a typology
of code-switching. Linguistics, 18, 581–618.
Soto, V., Cestero, N., & Hirschberg, J. (2018). The role of cognate words, POS tags and entrainment in code-
switching. Proceedings Interspeech, 2018, 1938–1942.
Stell, G., & Yakpo, K. (2015). Code-switching between structural and sociolinguistic perspectives. De Gruyter.
Suurmeijer, L., Parafta Couto, M. C., & Gullberg, M. (2020). Structural and extralinguistic aspects of
codeswitching: Evidence from Papiamentu-Dutch auditory sentence matching. Frontiers in Psychology, 11,
592266.
Torres Cacoullos, R., & Travis, C. E. (2018). Bilingualism in the community: Code-switching and grammars in contact.
Trefers-Daller, J. (1994). Mixing two languages. French-Dutch contact in a comparative perspective. Mouton de Gruyter.
Tsoukala, C., Broersma, M., Van den Bosch, A., & Frank, S. L. (2021). Simulating code-switching using a neural
network model of bilingual sentence production. Computational Brain & Behavior, 4, 87–100.
Valdés Krof, J. R., Guzzardo Tamargo, R., & Dussias, P. E. (2018). Experimental contributions of eye-tracking
to the understanding of comprehension processes while hearing and reading code-switches. Linguistic
Valdés Krof, J. R., Román, P., & Dussias, P. E. (2020). Are all codeswitches processed alike? Examining semantic
versus language unexpectancy. Frontiers in Psychology, 11, 2136.
Van Hell, J. G., Fernandez, C. B., Kootstra, G. J., Litcofsky, K. A., and Ting, C. Y. (2018). Electrophysiological
and experimental-behavioral approaches to the study of intra-sentential code-switching. Linguistic Approaches
to Bilingualism, 8, 134–161.
Van Hell, J. G., Litcofsky, K. A., & Ting, C. Y. (2015). Sentential code-switching: Cognitive and neural
approaches. In J. W. Schwieter (Ed.), The Cambridge handbook of bilingual processing (pp 459–482). Cambridge
University Press.
Van Hell, J. G., & Tanner, D. (2012). Second language profciency and cross-language lexical activation. Language
Learning, 62, 148–171.
Vaughan-Evans, A., Parafto Couto, M. C., Boutonnet, B., Hoshino, N., Webb-Davies, P., Deuchar, M., &
Thierry, G. (2020). Switchmate! An electrophysiological attempt to adjudicate between competing accounts
of adjective-noun codeswitching. Frontiers in Psychology, 11, 549762.
Weinreich, U. (1953). Languages in contact: Findings and problems. Mouton.
267
22
SYNTHESIS
Transdisciplinary approaches to understanding
variability in bilingual processing and use
Paola E. Dussias1
Introduction
Historically, research on second language (L2) acquisition has asked whether adult L2 learners can
acquire and process diferent aspects of the L2 in the same way as native speakers do (Birdsong, 1999).
Many studies have suggested that L2 speakers are unable to acquire the subtle nuances of the L2 (e.g.,
Piske et al., 2001), and that L2 acquisition and L2 processing are fundamentally diferent from acquiring
and processing the frst language. Anecdotal observations are reported by many L2 learners who speak
of the notorious difculty with word fnding, assembling sentences “on the fy,” and understanding the
back-and-forth language interactions among native speakers. Results demonstrating failure to acquire
the L2 natively have been widespread (e.g., Clahsen & Muysken, 1986) and have fueled proposals on
the existence of hard constraints on late L2 learning (Bley-Vroman, 1990) and L2 processing (Clahsen
& Felser, 2006). The traditional account has been that individuals who learn a second language past
early childhood are unable to fully acquire the L2 (e.g., Clahsen & Muysken, 1986) or to process the L2
in an L1-like fashion (e.g., Clahsen & Felser, 2006). Although the ability to acquire and comprehend a
second language improves with profciency, discrepancies in L2 ability have been reported even at high
levels of profciency (White, 2003). Consequently, a question that has dominated the adult L2 psycho-
linguistic literature is: What are the apparent constraints on the level of profciency that adult learn-
ers typically reach in their ability to process words, sentences and discourse in their second language,
and which factors account for the mixed success? The answer depends on what aspect of language is
being studied, as becomes clear from the chapters in this section. The chapter by Kira Gor (2023 [this
volume]) surveys diferent approaches to word recognition and encoding, focusing on the difculties
that L2 learners experience with the encoding of words and collocations in a second language. Her
proposal, which is backed up by compelling argumentation and sound evidence, is that “fuzzy” lexi-
cal representations present at diferent aspects of lexical encoding—from phonology to morphological
structure—make lexical processing in the L2 less efcient than in the L1. The chapter by Kathy Conk-
lin and Rüdiger Thul (2023 [this volume]) discusses how comprehenders access the mental lexicon.
The evidence reviewed strongly suggests that reduced lexical entrenchment results in reduced speed
of lexical processing (whether it be for single-word or for multiword processing), and that the criti-
cal contributing factor is frequency: Frequently encountered single-word or multiword units that are
entrenched in memory will show the traditional processing advantages; those that are not will show
processing slow-down. Holger Hopp (2023 [this volume]) discusses key themes in linguistic research on
L2 sentence processing that have dominated the feld. His focus not on what information type is used
268 DOI: 10.4324/9781003018872-25

Synthesis
during L2 sentence processing but when it is used, supports the conclusion that diferences between
native and non-native sentence processing may not be as widespread as postulated in early proposal, and
that diferences in the level of performance, which are more apparent at the syntactic reanalysis, may
arise due to reliance on discourse information over syntactic information, as well as on increased cogni-
tive demands during L2 sentence processing. The chapter on cognitive approaches to sentence process-
ing by Nuria Sagarra (2023 [this volume]) illustrates how participant variables, such as working memory
capacity, and linguistic characteristics of the input (e.g., cue saliency, cue reliability, cue entrenchment)
account for diferences between L1 and L2 learners. Saveria Colonna (2023 [this volume]) examines
an important and understudied area in L2 psycholinguistics; namely, discourse processing. Successful
discourse processing requires the simultaneous access and control of multiple sources of information,
including of syntactic, semantic, and pragmatic information. In her chapter, Colonna presents evidence
that the orchestration of these types of information is slower and more difcult in L2 speakers compared
to L1 speakers, and presents diferent accounts that provide an explanation for the observed L2 process-
ing disadvantages: Reduced ability to generate expectations, difculty integrating syntactic and non-
syntactic information, and shallow processing conspire to make L2 discourse processing more efortful.
Finally, the chapter by Van Hell (2023 [this volume]) examines the cognitive and neural mechanisms
that underlie the comprehension and production of code-switched utterances. Costs that arise during
the processing of code-switched sentences are thought to derive from the direction of the code-switch
(e.g., from the dominant to the non-dominant language or vice versa), participants’ code-switching
practices, whether the code-switches are naturally produced, and the identity of the interlocutors.
Taken together, the chapters point to between-group variation in L1 and L2 processing in efciency,
cue weighting, and experience, which give rise to partially divergent processing.
The underlying assumption in most L2 psycholinguistics studies is that L2 performance refects
variability that is unique to the L2, and that such variability is uncharacteristic of native monolingual
speakers. Indeed, most research on L2 processing has compared the performance of frst and second
language speakers, with little consideration of the fact that variability also underlines much of the
L1 processing work. The assumption of a uniform and highly efcient L1 processor has long been
challenged in the L1 processing literature, with evidence for individual variation in L1 speakers (e.g.,
Caplan & Waters, 1999; Pakulak & Neville, 2010), and with strategies in L1 language processing that
are sometimes “good enough” (e.g., Ferreira et al., 2002). While speakers of the same variety tend
to converge on which cues and strategies they employ, there is great heterogeneity and variation in
how L1 comprehenders approach sentence processing in their native language. Indeed, the study of
individual diferences is a key component in psycholinguistic work on monolingual sentence process-
ing (King & Just, 1991). Variability in L1 language processing, then, is considered the norm, rather
than the exception. Nonetheless, despite the ubiquity of variability in monolingual sentence process-
ing studies, variability in individuals who speak two or more languages has only recently begun to
be explored. An approach that connects L2 language processing with language experience and basic
cognitive principles is more compatible with our current knowledge of the architectural underpin-
nings of the systems responsible for language acquisition and language processing, and a more fruitful
approach in future studies of L2 sentence processing.
The goal of this chapter is to discuss recent fndings that demonstrate how considering variability
in the linguistic experiences of L2 speakers brings about a nuanced understanding of our view on
L2 processing. The focus in this chapter will be on two variables that have received little attention
in the L2 processing literature but which have the potential to contribute signifcant, widespread
benefts in the refnement of L2 processing theories: (1) language regulation as a faculty that plays
a deciding role in mediating L2 speakers’ abilities to efciently process their second language; and
(2) the interactional contexts in which second language speakers fnd themselves, and which serves
an infuential role in L2 speaker language processing. This chapter will point out the signifcance of
these particular factors in L2 processing; on the other hand, it will also more generally illustrate how
269
Paola E. Dussias
considering individual diferences among L2 learners can contribute to an increased understanding

of between-group diferences between L1 and L2 users.
Variability due to the engagement of cognitive control

and language regulation
Perhaps the most obvious ability that profcient second language speakers display is the ability to
speak in either of their languages without making errors in language selection; what is less obvious,
yet striking, is that they can do this despite the constant presence of the language not in use (Kroll
et al., 2015; Kroll et al., 2021). L2 speakers activate information of the language not in use when
producing and comprehending their other language, (e.g., Hoshino & Kroll, 2008; Thierry & Wu,
2007), across various stages of L2 learning, and despite high profciency in the second language (e.g.,
Van Hell & Dijkstra, 2002). Most evidence for cross-language activation comes from studies exam-
ining lexical features (for reviews, see Dijkstra, 2005), but there is also evidence for cross-language
infuences when examining syntactic aspects and discourse aspects of the L2 (Contemori & Dussias,
2020; Dussias & Sagarra, 2007; Hopp, 2016, 2023 [this volume]; Colonna, 2023 [this volume]). In
some cases, particularly when features of the L1 and the L2 converge, cross-language activation can
result in facilitation, such as in the case of cross-language structural priming (for a review, see Ber-
nolet et al., 2016) or in the processing of cognates. In other cases, when the features enter in confict
with one another, cross-language activation can result in momentary interference. The availability
of both languages makes code-switching (Van Hell, 2023 [this volume]) between the two languages
a natural linguistic behavior in many bilingual communities (e.g., Blanco-Elorrieta & Pylkkänen,
2018; Torres-Cacoullos & Travis, 2018). Many past studies have shown that not only does the L1
infuence L2 processing, but knowledge of second language infuences L1 processing (Bice & Kroll,
2015). A consequence of the parallel activation of both languages is that L2 speakers are constantly
juggling the competition that results when one of the two languages alone must be selected, and this
skill has a number of consequences for what it can reveal about L2 language processing. The require-
ment to continually regulate competition between the two languages means that L2 speakers must
rely on a series of mechanisms to enable the efcient use of the intended language. On the one hand,
L2 speakers must learn to regulate cross-language activation, which allows adequate engagement and
suppression of information as needed, but they may additionally recruit domain-general cognitive
abilities when processing demands are increased (see also Poarch, 2023 [this volume]). The most typi-
cal manifestation of regulation is the inhibition of the more dominant language to enable the use of
the less dominant L2, which is most evident in L2 immersion contexts (Linck et al., 2009), but L2
speakers also need to efciently de-regulate their L1 when appropriate. Two recent studies illustrate
how the native language comes to be regulated and the repercussions that this has for L2 processing.
Zirnstein et al. (2018) examined how cognitive control and language regulation ability impact
L2 readers’ susceptibility to prediction error costs. EGG recordings were made while participants—a
group of native English speakers and L1 Mandarin speakers immersed in their L2 English—read sen-
tences in English containing a target word that was either highly expected by the semantic context of
the sentence (e.g., tip in “After their meal, they forgot to leave a tip for the waitress”) or plausible but
unexpected (e.g., ten in “After their meal, they forgot to leave a ten for the waitress”). Because the
bilingual participants were quite diverse in their linguistic experience with English (e.g., some were
fuent in both languages, some had switched dominance—from Chinese to English—by virtue of
being immersed in an English-speaking environment), the scores of a verbal fuency task were used
to quantify the participants’ profciency.
A comparison of the ERP responses of the native and non-native group showed what might be
regarded as an expected result given the widespread diferences reported in past studies between
the brainwave patterns of L1 and L2 speakers. The native English participants showed an N400 for
270
Synthesis
expected words—indexing lexical-semantic facilitation—and a late positivity for unexpected words,

refecting a recovery process from prediction error. Conversely, the L2 speakers only produced the
N400. Because the lack of a late positivity efect has been taken to indicate that prediction has not
occurred, the results of the study seemed to indicate that Mandarin-English speakers were not able
to generate predictions for unexpected words in their L2. This was until the authors conducted a
series of individual diferences analyses examining domain-general control—indexed by the AX
continuous performance task (AX-CPT), which is a measure of cognitive control engagement that
efectively pits efects of reactive response inhibition against those of proactive goal maintenance and
confict monitoring, and by language regulation ability, which was indexed by a semantic verbal
fuency task in the L1. These results showed that the native participants and the L2 speakers alike
recruited cognitive control to recover from prediction costs, with better control predicting a reduc-
tion in the late positivity. The exciting fnding was that for the L2 speakers, the positivity was modu-
lated by L1 semantic fuency, such that increased semantic fuency in Mandarin related to a larger
positivity (i.e., larger prediction costs in the L2). Although this fnding is counterintuitive, it suggests
that the Mandarin speakers had to overcome the challenge of regulating their L1 to engage prediction
mechanisms in ways similar to native English speakers. Two key observations about L2 processing can
be derived from these fndings. Firstly, during L1 regulation, the recruitment of language-specifc
and domain-general resources are implicated; these may be separate from one another but are also
dependent on one another. Secondly, and equally importantly, if the authors had only examined L2
profciency, or had only asked whether the L2 speakers were able to generate predictions in the L2,
they would have reached an entirely diferent conclusion (see Kroll et al., 2021).
The second study reviewed in this section to illustrate diferences in cognitive control engagement
and their impact on L2 processing comes from a recent study by Navarro-Torres et al. (2019). The
researchers tested a group of L2-immersed speakers and compared their eye movements to English
monolinguals. Both groups of participants were asked to act the instructions of auditorily presented,
syntactically ambiguous sentences such as, “Put the frog on the napkin onto the box.” Past research
has shown that adult listeners initially (and incorrectly) interpret “the napkin” as the destination
(the goal). This incorrect-goal interpretation is revised upon hearing “onto the box” to assign “the
napkin” a modifer role, and “the box” as the correct goal. Navarro-Torres et al. (2019) also used a
paradigm reported in Hsu and Novick (2016) that interleaved Stroop sequences with the sentence
comprehension task. This allowed for the examination of how Stroop-related confict afected recov-
ery from syntactic ambiguity.
They replicated the results reported in Hsu and Novick (2016) with monolingual speakers: In
the presence of a Stroop-related confict—which arose when participants encountered incongruent-
Stroop trials—monolingual and bilingual listeners alike were better able to engage correct-goal inter-
pretations during the revision of their initial misparse. At frst glance, the fndings seemed to make
only an incremental contribution to our understanding of L2 processing. However, the new fnding
was that decreased looks to the incorrect target emerged at diferent points for the monolingual and
bilingual speakers. For the monolingual group, the process of disengaging incorrect-goal interpreta-
tions following incongruent-Stroop trials occurred after the disambiguating region (e.g., onto the
box) was encountered. Because monolinguals seemed to have waited for an explicit linguistic cue
signaling that their initial interpretation was incorrect to begin the revision process, it tentatively sug-
gests that they employed cognitive control strategies reactively. Conversely, for the L2 group, the pro-
cess of disengaging incorrect-goal interpretations following incongruent-Stroop trials emerged before
the disambiguating region was heard; namely, immediately after hearing the syntactically ambiguous
prepositional phrase (e.g., on the napkin). This could mean that the L2 participants relied on a proac-
tive recruitment strategy, employing cognitive control resources to recover from the ambiguity but
also to identify early linguistic cues (i.e., the omission of the complementizer + complementizer that’s
during ambiguous trials) to aid them in the revision process (Navarro-Torres et al., 2019). What is
271
Paola E. Dussias
signifcant about this fnding is that it demonstrates how monolinguals and L2 speakers can display
similar endpoints in their processing routines but get there using diferent routes.
Variability due to interactional contexts

As mentioned earlier, the presence of two languages introduces processing costs when L2 speakers
are using one language alone. To provide an example, relative to monolinguals, profcient L2 speak-
ers are generally slower to produce words and sentences (for a review, see Kroll & Gollan, 2014) and
name fewer pictures correctly on picture naming tasks (e.g., Sandoval et al., 2010), even when speak-
ing in their L1. According to the frequency-lag account, the fndings are attributed to diferences
in the frequency with which L2 speakers use each of their two languages relative to monolingual
counterparts (e.g., Gollan & Silverberg, 2001). None of these studies, however, have examined how
the interactional context afects usage patterns, above and beyond the fact that L2 speakers may have
reduced experience with each language. But this question is important given that recent models of
language processing highlight the role of experience in guiding comprehension, which in turn can
afect production (Dell & Chang, 2014; MacDonald, 2013).
Critically, studies on L2 processing lack reference to the various contexts of language experience
that L2 speakers are exposed to and the recurrent forms of conversational exchanges that they engage
in when communicating in a natural setting. An emerging trend in psycholinguistic research seeks to
build a more nuanced view of variation in language processing. For L2 speakers, the shift in emphasis
towards the everyday conversational use of language is captured by the adaptive control hypothesis
(Green & Abutalebi, 2013), which postulates that diferent interactional contexts impose diferent
communicative demands on speakers’ language control processes (for discussion, Van Hell, 2023
[this volume]). Indeed, there is growing evidence that individuals learn and attend to distributional
variation in the input. However, while such adaptation is a fundamental process of individuals of
all language backgrounds, many questions remain concerning the role of experience in guiding L2
online processing. Moreover, understanding the interactional demands of diferent contexts calls for
a systematic assessment of the relationship between language processing and language use, and the
contexts in which these take place.
In this respect, the study of code-switching provides a unique lens to examine the interactions
underlying processing adaptation and variation. Because code-switching emerges in some bilingual
communities but not in others, it provides a venue to examine the consequences of exposure and
adaptation to variation in code-switched speech and text. Indeed, recent research has shown that dis-
tributional regularities involving attested code-switching patterns act as cues heightening the prob-
ability of upcoming switches (Guzzardo Tamargo et al., 2016; Fricke et al., 2016, Valdés Krof et al.,
2017). This observation predicts variability among speakers who engage and do not engage in code-
switching (e.g., Beatty-Martínez & Dussias, 2017). To illustrate, recent evidence shows that speakers
who code-switch are more sensitive to codeswitching structures that are consistent with attested
distributional patterns that arise in their interactions with other codeswitching speakers, and there-
fore demonstrate facilitated processing relative compared to unattested code-switches. Conversely,
speakers who do not frequently code-switch do not show diferential processing to attested vs. unat-
tested code-switches, since these code-switches are virtually all unattested to non-code-switchers. To
provide a specifc example, Adamou and Shen (2019) carried out a study examining the processing
of code-switched sentences as a function of the frequency of use of specifc code-switching patterns.
Using a bimodal picture-sentence matching task, they tested whether switching costs were modulated
by exposure to specifc code-switching patterns from a well-established code-switching community.
Stimuli included ecologically valid and ecologically non-valid code-switches that were created based
on an analysis of code-switching preferences in natural conversations from the community under
study. Although participants responded the most quickly to unilingual sentences overall, a more
272
Synthesis
fne-grained analysis, based on the statistical frequencies of the switches in natural speech, revealed
that switched trials that were based on naturally-occurring examples in the corpus were just as fast
as unilingual trials. This and other studies (Beatty-Martínez & Dussias, 2017; Guzzardo Tamargo
et al., 2016) illustrate the importance of taking the switching preferences of the speech community
into consideration. Recently, Johns and Steuck (2021) also used code-switching to demonstrate that
when researchers examine participants’ contexts of language use, “what may seem costly numerically
may ultimately be benefcial to the bilingual when their linguistic behavior is contextualized” (p. 5).
A second set of studies we will review here further demonstrates the need of a shift in the way L2
psycholinguistic research moves forward to consider variability in language experience not as a source
of noise but rather as a source of evidence. Firstly, Beatty-Martínez and colleagues (2020) examined
the consequence that contexts of language use holds for the way in which cognitive resources modu-
late language abilities. The authors conducted a series of individual diference analyses to examine
how diferent cognitive control strategies (i.e., proactive vs. reactive control), as measured by the AX-
CPT, related to picture naming performance in the L1 and the L2 in each group. Three groups of
highly profcient L1 Spanish-L2 English speakers who resided in diferent locations (Granada, Spain;
San Juan, Puerto Rico; Pennsylvania, US), each with unique language environments, were recruited.
In the Spain group, individuals live in a context wherein Spanish is the dominant language across
most social contexts and English is spoken as the L2 mostly in specifc environments. These indi-
viduals, then, use their languages in separated contexts, and it is the context that serves to cue which
language to speak. Individuals in the Puerto Rico group live in a context wherein there are greater
opportunities to use and hear both languages. The use of Spanish and English is highly integrated and
thus speakers use both languages opportunistically as an established form of communication. Lastly,
individuals in the Pennsylvania group live in a varied context. These were speakers who had previ-
ously lived in Puerto Rico and had used their two languages in an integrated manner, but by virtue
of being immersed in the L2 (itself a highly monolingual community), they have a restricted use of
their L1 and need to constantly monitor the environment to decide whether they can use their two
languages opportunistically, or whether they can only use their L1 or L2.
The results showed that although naming accuracy was high across the three groups (above 90%
accuracy)—demonstrating that these L2 speakers were profcient in both their languages—individual
diference analyses revealed diferent patterns of association between picture naming performance
and the AX-CPT task. For the L2 speakers living in Spain, participants with increased reactive
control strategies tended to perform better on the picture naming. Since conversational exchanges
in this context typically involve Spanish, its usage may become highly predictable. Thus, these L2
speakers may reactively adjust to the less predictable instances when English is encountered, allow-
ing them to detect and resolve cross-language interference as needed. In the case of the L2 speakers
in Puerto Rico, no reliable patterns of association emerged between cognitive control and naming
performance. The authors interpreted the null result be a consequence of these speakers living in an
environment that allows for the use of the two languages opportunistically, which in turn minimizes
the need to rely on cognitive control to engage language regulation. For the Pennsylvania group, L2
speakers who had greater proactive control tendencies showed better naming performance. Because
the diference between the Puerto Rico group and the Pennsylvania group was the L2 immersion
experience, the fndings suggest that, once a critical threshold of profciency has been achieved, the
engagement of cognitive control depends on the demands of the language environment. This fnding
is consistent with the adaptive control hypothesis (Green & Abutalebi, 2013) and with recent empiri-
cal evidence demonstrating that diferent contexts of language use afect cognitive control ability
diferentially (Gullifer et al., 2018). In most past research, these L2 speakers might well have been
aggregated into a single L2 group to be compared to monolingual speakers. But the result reported
in this study shows how aggregating the data that ignores the context of language use can mask the
relation between language and the cognitive control processes that support them.
273
Paola E. Dussias
Finally, a study by Pot et al. (2019) with L2 speakers in the Netherlands demonstrated that not
all L2 speakers’ experiences produce the same consequences. L2 speakers in the Netherlands inter-
act with other L2 speakers either in their native Dutch or in their L2, indicating that they maintain
their languages separate. However, most people are multilingual, which allowed the researchers to
examine the relationship between L2 knowledge and aging. The authors conducted a large-scale
study of older adults to examine which aspects of language use might be related to language func-
tion. The critical fnding was that performance on a fanker task was predicted by the diversity
of interactional contexts. In other words, older L2 speakers who switched between their diferent
languages frequently in diferent social and environmental contexts performed better on the fanker
task. Although the study may seem tangential to the ongoing arguments presented in this chapter, it
nevertheless suggests that the interactional contexts in which L2 speakers used their languages turns
out to be an important variable.
Concluding remarks
Research on second language acquisition and processing points to an ongoing interplay between the
frst and the second language that involves a dynamic set of adaptive changes across the lifespan (de
Bot et al., 2007). Despite this, the classic evidence on L2 ultimate attainment has been taken to show
that L2 acquisition and processing are largely constrained by pre-determined features of the linguistic
and cognitive system (Clahsen & Felser, 2006), by a lack of proceduralized grammatical knowledge
(Ullman, 2005), or by reduced availability of cognitive resources (McDonald, 2000). Although a
large number of L2 studies have led to diferent theoretical proposals and commitments in terms of
the relation between language and domain-general cognition, as reviewed in the chapters in this sec-
tion, most of the empirical studies have focused on simple comparisons between L1 and L2 speakers.
However, the fnding that L1 speakers and L2 learners sometimes diverge in terms of language
processing should not be at all surprising if we take into consideration the inherent variability of
language processing itself, and the infuential role of linguistic experience. Kaan (2014) has proposed
that the mechanisms that underlie L1 and L2 sentence processing are fundamentally the same. What
difers—or rather, what yields diferences—are “several interdependent sets of factors, all of which
are subject to individual diferences” (p. 261). This includes input-driven factors such as frequency
and context (i.e., immersion vs. classroom learning), language-specifc factors like competition, and
variation in cognitive abilities and resources. The linkage between language experience and process-
ing has several implications for the way psycholinguists design experiments and draw conclusions
with respect to issues of language and domain general cognition. The proposal here is not to dismiss
the signifcance of what we have discovered and know about second language processing, and of
the diferences that have been observed between L1 and L2 speakers. Rather, the suggestion is that
to open the feld to new discoveries, we need to ask what can be learned by adopting a focus on
variation and how the systematic variation in second language experience results in phenotypes with
diferent patterns of associations across language processing and cognitive outcomes (see Beatty-
Martínez & Titone, 2021; Grüter, 2023 [this volume]).
To do this, two things need to happen. Firstly, researchers need to add to their toolbox of methods
extensive language history questionnaires that include a language background dimension, a language
use dimension, and an afective dimension. The language background dimension could include
questions that gather information about age of acquisition, manner of acquisition, and dominance.
The language use dimension would get at aspects of the frequency and manner in which L2 speak-
ers use their languages. Questions can include probes to gauge participants’ habits of language use to
characterize diversity of language use across three social settings: at home, at university, and during
leisure time. Finally, the afective dimension targets motivation to keep and use the two languages,
and whether community environment values learning and maintenance of the L2 (for examples
274
Synthesis
see Gullifer et al., 2018; Ikizer & Ramírez-Esparza, 2018; Marian et al., 2007; López-Beltrán,
2021; Luk & Bialystok, 2013). Secondly, there needs to be shift to transdisciplinary studies (Beatty-
Martínez & Titone, 2021). One important aspect that is needed to better characterize the complexity
that comes with examining language variation in L2 experience is to cross traditional disciplinary
boundaries. In recent years, questions specifc to L2 processing research have been informed by
research in usage-based approaches (e.g., Beatty-Martínez et al., 2017; Valdés Krof et al., 2017; Guz-
zardo Tamargo et al., 2016), network sciences (e.g., Xu et al., 2021), and information theory (e.g.,
Gullifer & Titone, 2020). Conducting transdisciplinary research will add a layer of complication but
the gains promise to more efectively “extract signal from noise” (Beatty-Martínez & Titone, 2021)
in L2 processing research.
To conclude, many L2 processing studies that have observed mismatches between L1 and L2
processing routines have not considered the inherent variability that exists in language use both
across and within the groups. L2 speakers who use two languages actively but in diferent contexts
may difer in dramatic ways because they may not be necessarily alike in the way in which they
engage cognitive resources. An important implication is that much of the controversy that currently
surrounds the diferences between L1 and L2 speakers could be, in part, the result of a failure to
characterize the complexity associated with the context of language use. The systematic variation
in language use may help determine the pattern of consequences that are observed in the L2 pro-
cessing research.
Note
1 The writing of this chapter was supported in part by NSF OISE 1545900 to Paola E. Dussias. I would like to
thank Aline Godfroid, Holger Hopp, and Irina Elgort for helpful comments and suggestions. All errors remain
my own.
References
Adamou, E., & Shen, X. R. (2019). There are no language switching costs when codeswitching is frequent.
Beatty-Martínez, A. L., & Dussias, P. E. (2017). Bilingual experience shapes language processing: Evidence from
codeswitching. Journal of Memory and Language, 95, 173–189.
Beatty-Martínez, A. L., Navarro-Torres, C. A., Dussias, P. E., Bajo, M. T., Guzzardo Tamargo, R. E., & Kroll,
J. F. (2020). Interactional context mediates the consequences of bilingualism for language and cognition.
Journal of Experimental Psychology: Learning, Memory, and Cognition, 46(6), 1022–1047.
Beatty-Martínez, A. L., & Titone, D. A. (2021). The quest for signals in noise: Leveraging experiential variation
to identify bilingual phenotypes. Languages, 6(4), 168.
Bernolet, S., Collina, S., & Hartsuiker, R. J. (2016). The persistence of syntactic priming revisited. Journal of
Bice, K., & Kroll, J. F. (2015). Native language change during early stages of second language learning. Neuro-
Report: For Rapid Communication of Neuroscience Research, 26(16), 966–971.
Birdsong, D. (Ed.). (1999). Second language acquisition and the Critical Period Hypothesis. Lawrence Erlbaum As-
sociates Publishers.
Blanco-Elorrieta, E., & Pylkkänen, L. (2018). Ecological validity in bilingualism research and the bilingual
advantage. Trends in Cognitive Sciences, 22(12), 1117–1126.
Caplan, D., & Waters, G. S. (1999). Verbal working memory and sentence comprehension. Behavioral and Brain
Sciences, 22(1), 77–126.
3–42.
Clahsen, H., & Muysken, P. (1986). The accessibility of Universal Grammar to adult and child learners. A study
of the acquisition of German word order. Second Language Research, 2(2), 93–119.
275
Paola E. Dussias
Colonna, S. (2023). Discourse processing. In A. Godfroid & H. Hopp (Eds.). The Routledge handbook of second
language acquisition and psycholinguistics (pp. 242–254). Routledge.
Contemori, C., & Dussias, P. (2020). The processing of subject pronouns in highly profcient L2 speakers of
English. Glossa: A Journal of General Linguistics, 5(1), 38.
de Bot, K., Lowie, W., & Verspoor, M. (2007). A dynamic systems theory approach to second language acquisi-
tion. Bilingualism: Language and Cognition, 10(1), 7–21.
Dell, G. S., & Chang, F. (2014). The P-chain: Relating sentence production and its disorders to comprehension
and acquisition. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences, 369(1634),
20120394.
Dijkstra, T. (2005). Bilingual visual word recognition and lexical access. In J. F. Kroll & A. M. B. de Groot (Eds.),
Handbook of bilingualism: Psycholinguistic approaches (pp. 179–201). Oxford University Press.
Dussias, P. E., & Sagarra, N. (2007). The efect of exposure on syntactic parsing in Spanish- English bilinguals.
Ferreira, F., Bailey, K. G., & Ferraro, V. (2002). Good-enough representations in language comprehension. Cur-
rent Directions in Psychological Science, 11(1), 11–15.
Fricke, M., Kroll, J. F., & Dussias, P. E. (2016). Phonetic variation in bilingual speech: A lens for studying the
production—comprehension link. Journal of Memory and Language, 89, 110–137.
Gollan, T. H., & Silverberg, N. B. (2001). Tip-of-the-tongue states in Hebrew—English bilinguals. Bilingualism:
& H. Hopp (Eds.). The Routledge handbook of second language acquisition and psycholinguistics (pp. 191–202).
Routledge.
Green, D. W., & Abutalebi, J. (2013). Language control in bilinguals: The adaptive control hypothesis. Journal of
Cognitive Psychology, 25(5), 51–530.
Gullifer, J. W., Chai, X. J., Whitford, V., Pivneva, I., Baum, S., Klein, D., & Titone, D. (2018). Bilingual experi-
ence and resting-state brain connectivity: Impacts of L2 age of acquisition and social diversity of language use
on control networks. Neuropsychologia, 117, 123–134.
Guzzardo Tamargo, R. E., Valdés Krof, J. R., & Dussias, P. E. (2016). Examining the relationship between
comprehension and production processes in code-switched language. Journal of Memory and Language, 89,
138–161.
Hopp, H. (2016). The timing of lexical and syntactic processes in L2 sentence comprehension. Applied Psycho-
linguistics, 37(5), 1253–1280.
Hopp, H. (2023). Sentence processing in a second language: linguistic approaches. In A. Godfroid & H. Hopp
(Eds.). The Routledge handbook of second language acquisition and psycholinguistics (pp. 216–228). Routledge.
Hoshino, N., & Kroll, J. F. (2008). Cognate efects in picture naming: Does cross-language activation survive a
change of script? Cognition, 106(1), 501–511.
Hsu, N., & Novick, J. M. (2016). Dynamic engagement of cognitive control modulates recovery from misinter-
pretation during real time language processing. Psychological Science, 27(4), 572–582.
Ikizer, E., & Ramírez-Esparza, N. (2018). Bilinguals’ social fexibility. Bilingualism: Language and Cognition, 21(5),
957–969.
Johns, M. A., & Steuck, J. (2021). Is codeswitching easy or difcult? Testing processing cost through the prosodic
structure of bilingual speech. Cognition, 211.
Kaan, E. (2014). Predictive sentence processing in L2 and L1.What is diferent? Linguistic Approaches to Bilingual-
ism, 4(2), 257–282.
King, J., & Just, M. A. (1991). Individual diferences in syntactic processing: The role of working memory.
Journal of Memory and Language, 30(5), 580–602.
Kroll, J. F., Dussias, P. E., Bice, K., & Perrotti, L. (2015). Bilingualism, mind, and brain. Annual Review of Lin-
guistics, 1(1), 377–394.
Kroll, J. F., & Gollan, T. H. (2014). Speech planning in two languages: What bilinguals tell us about language
production. In M. Goldrick, V. Ferreira, & M. Miozzo (Eds.), The Oxford handbook of language production
(pp. 165–181). Oxford University Press.
Kroll, J. F., Takahesu-Tabori, A., & Navarro-Torres, C. (2021). Capturing the variation in language experience
to understand language processing and learning. Language, Interaction and Acquisition, 12(1), 84–111.
Linck, J. A., Kroll, J. F., & Sunderman, G. (2009). Losing access to the native language while immersed in a sec-
ond language: Evidence for the role of inhibition in second-language learning. Psychological Science, 20(12),
1507–1515.
276
Synthesis
López-Beltrán, P. (2021). Heritage speakers’ online processing of the Spanish subjunctive: A comprehensive usage-based
study [Unpublished Doctoral dissertation], Penn State University.
Luk, G., & Bialystok, E. (2013). Bilingualism is not a categorical variable: Interaction between the language
profciency and usage. Journal of Cognitive Psychology, 25(5), 605–621.
MacDonald, M. C. (2013). How language production shapes language form and comprehension. Frontiers in
Psychology, 4, 226.
Marian, V., Blumenfeld, H. K., & Kaushanskaya, M. (2007). The language experience and profciency question-
naire (LEAP-Q): Assessing language profles in bilinguals and multilinguals. Journal of Speech, Language, and
Hearing Research, 50, 940–967.
McDonald, J. L. (2000). Grammaticality judgments in a second language: Infuences of age of acquisition and
native language. Applied Psycholinguistics, 21(3), 395–423.
Navarro-Torres, C., Dalia, A., Garcia, L., Chidambaram, V., & Kroll, J. F. (2019). Cognitive control facilitates
attentional disengagement during second language comprehension. Brain Sciences, 9(5), 95.
Pakulak, E., & Neville, H. J. (2010). Profciency diferences in syntactic processing of native speakers indexed by
event-related potentials. Journal of Cognitive Neuroscience, 22(12), 2728–2744.
Piske, T., MacKay, I. R. A., & Flege, J. E. (2001). Factors afecting the degree of foreign accent in an L2: A
review. Journal of Phonetics, 29(2), 191–215.
Poarch. G. J. (2023). Cognitive efects of bilingualism. In A. Godfroid & H. Hopp (Eds.). The Routledge handbook
Pot, A., Porkert, J., & Keijzer, M. (2019). The bidirectional in bilingual: Cognitive, social and linguistic efects
of and on third-age language learning. Behavioral Sciences, 9(9), 98.
phological processing. In A. Godfroid & H. Hopp (Eds.). The Routledge handbook of second language acquisition
Sandoval, T. C., Gollan, T. H., Ferreira, V. S., & Salmon, D. P. (2010). What causes the bilingual disadvantage in
verbal fuency? The dual-task analogy. Bilingualism: Language and Cognition, 13(2), 231–252.
Thierry, G., & Wu, Y. J. (2007). Brain potentials reveal unconscious translation during foreign- language com-
prehension. Proceedings of the National Academy of Sciences of the United States of America, 104(30), 12530–12535.
Torres Cacoullos, R., & Travis, C. E. (2018). Bilingualism in the community: Code-switching and grammars in contact.
Ullman, M. T. (2005). A cognitive neuroscience perspective on second language acquisition: The declarative/
procedural model. In C. Sanz (Ed.), Mind and context in adult second language acquisition: Methods, theory and
practice (pp. 141–178). Georgetown University Press.
Valdés Krof, J. R., Dussias, P. E., Gerfen, C., Perrotti, L., & Bajo, M. T. (2017). Experience with code-switch-
ing modulates the use of grammatical gender during sentence processing. Linguistic Approaches Bilingualism,
7(2), 163–198.
van Hell, J. (2023). Code switching. In A. Godfroid & H. Hopp (Eds.). The Routledge handbook of second language
van Hell, J. G., & Dijkstra, T. (2002). Foreign language knowledge can infuence native language performance
in exclusively native contexts. Psychonomic Bulletin & Review, 9(4), 780–789.
White, L. (2003). Fossilization in steady state L2 grammars: Persistent problems with infectional morphology.
Xu, Q., Markowska, M., Chodorow, M., Xu, Q., & Chodorow, M. (2021). A network science approach to
bilingual code-switching. Proceedings of the Society for Computation in Linguistics, 4(1), 18–27.
Zirnstein, M., van Hell, J. G., & Kroll, J. F. (2018). Cognitive control ability mediates prediction costs in mono-
linguals and bilinguals. Cognition, 176, 87–106.
277
SECTION IV
The psycholinguistics of learning

23
IMPLICIT LEARNING AND
SECOND LANGUAGE
ACQUISITION
A cognitive psychology perspective
Introduction
There are two distinct yet related research strands within cognitive psychology that are directly relevant
to the study of how we learn a second language, namely implicit learning and statistical learning (see Text-
box 23.1). The two strands prioritize diferent research themes and questions, and they rely on diferent
experimental paradigms, as illustrated in the section “Implicit learning paradigms”, below. Despite these
diferences, implicit learning and statistical learning studies share many commonalities, and there have
been calls for closer alignment between the two lines of investigation (Rebuschat & Monaghan, 2019).
Second language research has much to gain from the theoretical insights and methodological advances
developed in implicit learning and statistical learning research. Implicit learning has placed the cognitive
unconscious back at the heart of research on language learning (Reber, 2022), and it has promoted experi-
mental tasks and paradigms that allow us to distinguish the contributions of implicit and explicit processes
in language learning. Statistical learning has demonstrated convincingly that infants and very young chil-
dren can acquire diferent aspects of language by tracking statistical information in the input, thus provid-
ing an essential proof of concept for the potential role of implicit learning in language acquisition (Frost
Implicit learning: Deriving information from the environment without awareness of what was learned.
Usually observed under incidental exposure conditions, i.e., when participants are not informed about
the learning target and the surprise test.
Explicit learning: Deriving information from the environment by means of conscious learning strategies
(e.g., hypothesis formation and testing). Usually associated with intentional exposure conditions, i.e.,
when participants are informed about the learning target and are actively trying to acquire it.
Statistical learning: Learning from the distributional properties of the environment. The process entails
computations based on units or patterns (sounds, syllables, syntactic categories, etc.).
DOI: 10.4324/9781003018872-27 281

et al., 2019). Implicit learning and statistical learning research contribute to our understanding of fundamen-
tal processes of (child and adult) language learning. In return, second language acquisition (SLA) research
contributes an applied perspective that is frequently missing in implicit and statistical learning research.
Experimental paradigms
In this section, we briefy review four paradigms that are widely used in implicit learning and sta-
tistical learning research. An introduction to these experimental tasks will provide us with a better
understanding of the evidence base discussed in the following sections.
Implicit learning paradigms

The experimental arrangements in implicit learning research generally have the following charac-
teristics. Participants (usually adults) are trained under incidental exposure conditions, i.e., they are
not informed of the learning target or of the subsequent test of learning. Experiments include a
measure of learning (to determine what was acquired following exposure) and a measure of awareness
(to determine if the acquired knowledge is conscious or not). The term implicit learning is usually
reserved for those instances in which the awareness measure suggests the presence of at least some
implicit (unconscious) knowledge. Otherwise, the term incidental learning is preferred (see Williams,
2009, for discussion). Finally, the learning target tends to be fairly complex. As Reber (1993) points
out, implicit learning is more likely to be observed when the learning target is too challenging to
acquire through explicit means (e.g., via hypothesis formation and testing). This type of arrangement
can be found in two classic paradigms that have been used extensively in implicit learning research,
namely artifcial grammar learning (AGL) and sequence learning.
In the classic AGL procedure (Reber, 1967, Expt. 2), participants are frst asked to memorize mean-
ingless letter strings (e.g., TPPTS, VXXVPXVS) that—unbeknownst to them—were generated by a
complex, artifcial grammar. They are then given a surprise test (usually a grammaticality judgment task,
GJT) to determine what aspects of the grammar they have learned. Participants also complete a measure
of awareness to determine if the acquired knowledge is conscious or not. Traditionally, this has been done
by prompting participants to report rules or patterns they might have noticed (retrospective verbal reports).
There are several types of sequence learning tasks. The best-known is the serial reaction time (SRT)
task, developed by Nissen and Bullemer (1987). Here, participants observe a visual cue (e.g., an asterisk)
that can appear in diferent locations (A-B-C-D) on the computer screen. Their task is to simply respond
to the cue by pressing on a button associated with the cue as quickly and accurately as possible. Participants
are not informed that, on the majority of trials, the location is determined by a repeating sequence (e.g.,
D-B-C-A-C-B-D-C-B-A). Learning is measured by comparing response times over runs of trials that do
and do not follow the fxed sequence. Awareness is typically measured by asking participants to generate
completions to sequences, or by recognition memory for fragments of the fxed versus random sequences.
Statistical learning paradigms

Experimental arrangements in statistical learning research are characterized by the careful manipula-
tion of input statistics, e.g., item frequencies, transitional probabilities, etc., to observe if, and how, this
manipulation afects learning. Participants (infants, children, adults) are exposed to artifcial languages
under incidental exposure conditions, though there are many statistical learning studies in which
exposure is intentional, i.e., participants are instructed to intentionally learn the meaning of novel
words (e.g., Yu & Smith, 2007) or to monitor a continuous stream of speech to fnd out where the
words begin and end (e.g., Newport & Aslin, 2004). The artifcial languages often resemble natural
languages more closely than the artifcial grammars used in implicit learning research, using a lexicon
282
Implicit learning and second language acquisition
consisting of pseudowords (dingep, tha, vinoy, noo) rather than letters, and phrase-structure grammars.
In contrast to implicit learning research, statistical learning studies often do not include a measure of
awareness. This is perhaps the most obvious methodological diference between implicit learning and
statistical learning research. The implicitness of the acquired knowledge is generally assumed but not
empirically verifed (for exceptions see “The implicitness of implicit statistical learning”, below).
The best-known paradigm in statistical learning research is the artifcial speech segmentation task
developed by Safran et al. (1996). In this task, participants are exposed to a continuous speech stream
(e.g., bupadapatubitutibudutabapidabu . . .) that consists of recurring pseudowords (here, bupada, patubi,
tutibu, dutaba, pidabu, etc.). Importantly, the stream does not contain acoustic or prosodic cues to demar-
cate word boundaries; the only cues to word boundaries are the transitional probabilities between sylla-
ble pairs, which are higher within words (bupa-, patu-, tuti-) than between words (-dapa-, -bitu-, -bapi-).
After exposure, participants are given a surprise test to determine if they are able to distinguish between
the target items and pseudowords that did not occur during training (e.g., batipa, dupitu, tipabu).
Finally, cross-situational learning is another important paradigm in statistical learning research. In the
classic version of this task (Yu & Smith, 2007), learners are presented with multiple objects on a computer
screen (two, three, or four objects) while listening to sequences of pseudowords (two, three, or four pseu-
dowords, depending on the number of objects), with no information as to which label goes with which
referent. Critically, the presentation sequence of the pseudowords is unrelated to the positioning of the
objects on the screen. To acquire the word-object mappings, learners thus have to keep track of potential
referents and labels over multiple trials. In the cross-situational learning task developed by Monaghan et
al. (2015), participants observe two dynamic scenes on the screen while listening to an artifcial language
sentence that describes only one of the scenes. No response was required during training. After training,
the participants’ task is to indicate, as quickly and accurately as possible, which scene the sentence refers
to. Successful completion requires keeping track of statistical information across multiple learning trials.
The implicitness of implicit statistical learning

The frst studies on implicit learning concluded that the acquired knowledge was completely unavail-
able to consciousness, given that participants failed to report any rules underpinning the artifcial
grammar (e.g., Reber, 1967). This early view has been challenged and revised over the past decades.
It soon became clear that participants in AGL or sequence learning experiments might not be
totally unaware of the underlying structure, with many participants frequently reporting partial
knowledge of the artifcial grammar or the repeating sequence in the debriefng session (e.g., Nissen
& Bullemer, 1987; Reber & Lewis, 1977). However, it also became clear that this conscious (explicit)
knowledge is generally insufcient to account for participants’ performance during the training and
test phases, i.e., the acquired knowledge tends to be richer than that which can be verbalized (e.g.,
Mathews et al., 1989; Reber & Lewis, 1977; Reber et al., 1980).
Early studies and interpretations have also been challenged on methodological grounds. There are
well-known limitations to using verbal reports as a direct measure of awareness, which led to a debate
concerning the implicitness of the knowledge acquired in AGL and sequence learning studies (for
discussion, see Newell & Shanks, 2014) and, as a result, other behavioral measures of awareness have
been proposed. For example, Dienes and Scott (2005) have advocated the use of subjective measures
of awareness (e.g., source attributions). For each grammaticality judgment, participants are prompted
to report whether the judgment was based on a guess, intuition, memory, or rule knowledge. If
participants perform signifcantly above chance on grammaticality judgments that were based on
guesses or intuition, this is taken as evidence for implicit knowledge. Above-chance performance
on grammaticality judgments based on memory or rule knowledge is taken as evidence for explicit
283
knowledge. A review of diferent awareness measures can be found in Rebuschat (2013) and Tim-
mermans and Cleeremans (2015).
Despite the limitations of diferent awareness measures (see Rebuschat, 2013), the past decades
of research have confrmed that participants in AGL and sequence learning experiments develop a
knowledge base that is tacit and very difcult to verbalize. It is also clear that adult participants often
develop both implicit (unconscious) and explicit (conscious) knowledge as a result of exposure (e.g.,
Guo et al., 2011). For example, when sequence learning data is reanalyzed based on participants’
ability to explicate the underlying sequence, a learning efect is often observed in both aware and
unaware participants (e.g., Curran & Keele, 1993; Monaghan et al., 2019). The fact that exposure to
a complex structure could lead to both types of knowledge (implicit and explicit) should not come
as a surprise; after all, we have at our disposal several ways of acquiring knowledge from the environ-
ment. The more interesting question, which has received too little attention until now, is how these
learning processes interact, and how the specifc task and exposure conditions afect the type(s) of
knowledge we develop (see also Godfroid, 2023 [this volume]; Morgan-Short & Ullman, 2023 [this
volume]; and the following).
Statistical learning experiments typically do not investigate whether exposure resulted in implicit
or explicit knowledge, but this gap has recently been addressed by studies that added awareness
measures. For example, Franco et al. (2011) found that participants acquired explicit knowledge
while completing a speech segmentation task similar to that in Safran et al. (1996). Batterink et
al. (2015) measured explicit knowledge by means of a direct test (forced-choice recognition of
syllable sequences, combined with subjective measures of awareness) and implicit knowledge by
means of an indirect test (speeded syllable detection task). Results indicated that participants readily
developed both types of knowledge, implicit and explicit, and that performance on the direct and
indirect tests did not correlate, suggesting that implicit and explicit knowledge accrue in parallel and
independently.
Hamrick and Rebuschat (2012) adopted the cross-situational statistical learning tasks devel-
oped by Yu and Smith (2007) and Monaghan et al. (2015) to investigate the incidental and
intentional learning of pseudowords via cross-situational statistics. They found that participants in
both groups developed both implicit and explicit knowledge. However, the exposure condition
played an important role in the type of knowledge that participants acquired primarily. Under
incidental exposure conditions, when participants were not informed about the learning target,
they developed primarily implicit knowledge. In contrast, participants instructed to learn the
pseudowords in the intentional exposure condition acquired primarily explicit knowledge. Inter-
estingly, these participants also acquired some unconscious knowledge, highlighting the complex
relationship between exposure condition (incidental vs. intentional) and acquired knowledge
(implicit vs. explicit) (see also Guo et al., 2011; Rebuschat & Williams, 2012). Franco et al.
(2016) also investigated the cross-situational learning of pseudowords but measured awareness
by means of the process dissociation procedure (Jacoby, 1991) combined with subjective mea-
sures (Dienes & Scott, 2005). They confrmed that statistical learning does not necessarily occur
unconsciously and that participants develop both implicit and explicit knowledge, even under
intentional learning conditions. Finally, Monaghan et al. (2019) investigated the role of aware-
ness in statistical learning. Participants learned novel words (nouns, verbs, function words) under
incidental or instructed exposure conditions. Awareness was measured by means of retrospective
verbal reports and subjective measures of awareness. They found signifcant learning efects in
both unaware participants (those that were unable to verbally account for their performance) and
in aware participants (those that reported partial knowledge of the learning target), with aware
participants outperforming unaware participants in terms of learning rate and learning outcomes.
It is clear that incidental and intentional learning tasks result in a complex mix of implicit and
explicit knowledge, at least as revealed by subjective measures.
284
The role of attention in implicit statistical learning

In an SLA context, implicit learning might be conceived as a process by which it is possible to auto-
matically pick up aspects of a language. If we think of automatic processes as requiring relatively little
attention (conceived as a limited processing resource), does this mean that implicit learning does not
require attention?
Consider the question of learning under distraction (e.g., being exposed to a language while
performing another attention-demanding task). In an AGL experiment, the participants’ primary
task is typically short-term recall of letter strings. If they are required to simultaneously hold random
digits in working memory (Hendricks et al., 2013) or continuously generate random digits (Dienes
& Scott, 2005) learning efects are not reduced. If the primary task is rule discovery then a second-
ary task can have a damaging efect, provided the system is simple enough to be learned explicitly
(Waldron & Ashby, 2001; see also the following), but implicit AGL appears to be robust under dual
task scenarios.
In contrast, auditory statistical learning of word boundaries appears to reduce to non-signifcant
levels by highly attention-demanding secondary tasks such as monitoring a concurrent rapid picture
sequence for repetitions (Toro et al., 2005). However, note that, unlike in the AGL task, in which
participants are required to make responses to the letter strings (i.e., recall them during the training
phase), no responses to the syllables are required in the statistical learning paradigm, and so they can
efectively be ignored. What the elimination of the learning efect might tell us then is that a suf-
fciently demanding secondary task can withdraw attention from stimulus encoding if no responses
are required, but not that the learning process itself requires attention.
The importance of attention in stimulus encoding is emphasized by Logan and Etherton’s (1994)
obligatory encoding assumption—“encoding into memory is an inevitable consequence of attending” (p.
1022). Conversely, there can be no encoding in memory without attention. They showed that novel
associations between words could be learned incidentally so long as both words were attended; the
learning efect disappeared if attention were oriented to just one of the words in advance. Hence,
although there is evidence that attention is not required for immediate processing of familiar stimuli,
there is good evidence that attention to stimuli is necessary for forming new associations (see Wil-
liams, 2013, for a review).
But it is not enough to attend to stimuli for learning relations between them; we have to attend
to the relevant dimensions of those stimuli. In an ingenious SRT experiment (Jiménez & Méndez,
1999) a stimulus could appear at one of four positions (A, B, C, D) as determined by an artifcial
grammar. In addition, the identity of the stimulus varied (e.g., the sequence A-C-D would appear
as ?-!-x). In fact, stimulus identity, as well as the AG, predicted the next stimulus position (in this
example, ? predicts position C, ! predicts position D). Both the underlying AG and the predictiveness
of stimulus identity could be simultaneously learned implicitly (without awareness of the regulari-
ties), but only when participants were required to keep a running count of character identities while
completing the standard SRT task. Without this additional requirement, requiring the stimulus iden-
tities to be noticed, participants only learned the AG structure. Hence it is not sufcient to attend
to stimuli for implicit learning to occur; we have to attend to them in the right way. Similarly, in
another condition of their auditory statistical learning experiment, described previously, Toro et al.
(2005) required participants to monitor the syllable stream for syllables of a relatively high pitch. The
learning efect was again reduced to a non-signifcant level, even though the stream itself was being
closely attended. This was presumably because participants were attending to the syllables in terms
of their pitch rather than phonemes, which were the elements between which associations occurred.
What is important, therefore, is that the relevant stimuli, or stimulus dimensions, are noticed,
and hence enter into awareness, but learners do not necessarily have to be aware of the role those
stimuli play in the underlying system. This relates to Schmidt’s (1994) distinction between awareness
285
at the level of noticing (e.g., forms) and awareness at the level of understanding (e.g., rules that explain the
distribution of forms). What the psychological literature suggests is that only the former is necessary
for implicit learning of underlying regularities. How attention is allocated depends critically upon
the task that the person is required to perform, and has a strong determining efect even on implicit
learning. At the same time, the dual-task AGL studies show that the actual process of deriving the
underlying regularities implicitly appears to operate with minimal demands on attention.
What is learned?
What kinds of regularities have been shown to be learnable implicitly? Are there linguistic constraints
on implicit learning, or can any regularity be acquired in this way?
There appears to be widespread agreement that some kind of implicit learning mechanism under-
lies the formation of chunks, e.g., multiword units in natural language, bigrams and trigrams in AGL,
triplets of syllables in statistical learning. AGL studies have shown that test strings are more likely
to be endorsed as grammatical if they share chunks with training items (Kinder & Assmann, 2000;
Knowlton & Squire, 1996), and, in statistical learning, recall of syllable sequences has been shown to
be chunk-based (Isbilen et al., 2020). The actual computations that produce these efects are clearly
unconscious; chunking, as a basic memory process, therefore qualifes as a mechanism of implicit
learning.
Non-adjacent dependencies (e.g., the association between “he” and “s” in “he walk/talk/ride-s”)
are one aspect of grammatical structure that goes beyond chunking. Studies have reported a failure
to learn analogous dependencies between syllables in continuous streams (Newport & Aslin, 2004;
Onnis et al., 2005). Other studies have reported more success but, in these cases, there was a phono-
logical cue that distinguished non-adjacent and adjacent elements, e.g., plosive versus continuant syl-
lables (Frost & Monaghan, 2016; Onnis et al., 2005) or mono- versus disyllabic nonwords (Gomez,
2002), or there was support from meaningful relations (Amato & MacDonald, 2010). This is an
example of how the power of implicit learning is enhanced by the convergence between multiple
cues, as is characteristic of natural languages.
Statistical learning studies have also revealed representational constraints on implicit learning. For
example, lexical segmentation and vocabulary learning seem to be driven more by the patterning of
consonants, such as the p_r_g pattern in pu-ra-gi and po-re-gi, than the patterning of vowels, such as
the u_e_a pattern in ku-me-pa and ru-me-ta (Bonatti et al., 2005; Nazzi & Cutler, 2019). In contrast,
abstract patterns are more learnable when instantiated over vowels than consonants (e.g., the ABA
structure over the vowels in ta-pe-na versus the consonants in ba-nu-be) possibly refecting the privi-
leged status of vowels in conveying grammatical information, at least in the frst language(L1)s of
the participants tested (Toro et al., 2008). Rats do not show diferential sensitivity to regularities over
vowels and consonants (de la Mora & Toro, 2013). Hence, it is not the case that we pick up any and
all regularities in the input, but our sensitivity to statistical patterns is conditioned to some extent by
our prior linguistic knowledge and expectations, which can actually make us “deaf ” to some patterns
(Endress & Hauser, 2009).
The foregoing studies represent rather synthetic abstractions from natural languages. Other studies
have explored implicit learning of more obviously linguistic, and rule-like, phenomena and, unlike
in statistical learning research, with more emphasis on whether linguistic regularities can be acquired
without awareness at the level of understanding. Such studies have shown implicit learning of pho-
notactic constraints (Dell et al., 2000) and stress patterns (Chan & Leung, 2014; Graham & Williams,
2018). In the domain of syntax, semi-artifcial language learning studies have shown implicit learn-
ing of phrasal patterns, such as the characteristic verb placement in German (Rebuschat & Williams,
2012). Other studies have shown implicit learning of semantically conditioned rules, such as whether
the novel determiners gi and ro are used before animate or inanimate nouns (Batterink et al., 2014;
286
Chen et al., 2011; Kerz et al., 2017; Leung & Williams, 2012; Williams, 2005), though the efect
using the Williams (2005) paradigm, or similar, has not always been obtained (Andringa, 2020;
Hama & Leow, 2010) for reasons that are as yet unclear. Other studies have shown implicit learning
of semantic preference rules—whether the novel verbs gouble and powter are followed by concrete or
abstract nouns (Paciorek & Williams, 2015a, 2015b). However, as in other areas of implicit learn-
ing, the acquisition of semantics-based generalizations does not appear to be unconstrained. Studies
have failed to fnd implicit learning efects for regularities based on relative size (Chen et al., 2011;
Leung & Williams, 2012) and Leung and Williams (2014) only found signifcant learning efects for
a long-fat distinction in Chinese-speaking participants (where this distinction is encoded by classi-
fers). Such studies demonstrate the diferential availability of conceptual distinctions for grammati-
cization, either due to the participants’ native language, language universals, or salience in the input,
as has been shown for the relative size distinction (Pham et al., 2020). Therefore, while studies have
demonstrated implicit learning of natural language phenomena relating to phonology, syntax, and
grammatical form-meaning connections, it has become increasingly evident that the learning process
is constrained.
Do implicit and explicit learning, knowledge, and instruction interact?

If implicit learning is an automatic process, the question arises as to how it might interact with, or be
afected by, any conscious learning strategies that a learner might employ, either spontaneously or as a
response to instruction. This issue has been discussed in the SLA literature in terms of either a strong
interaction between explicit and implicit learning (whereby they infuence each other directly), as
opposed to a weak interaction (mediated, for example by attentional processes), or complete inde-
pendence (see Godfroid, this volume, for a discussion of the interface issue in SLA). Here we focus on
studies using implicit learning paradigms which, although examining artifcial systems, we believe
ofer useful and generalizable insights.
Most research has focused on the efect of instructions to search for rules and how that might
impact underlying implicit learning. As one would expect, if a regularity is relatively simple, then
rule discovery will boost learning (Mathews et al., 1989). For more complex systems, such as those
used in AGL, early results suggested that participants given rule search instructions performed worse
than those given a memorization task (Reber, 1976). However, subsequent studies failed to fnd any
diference between intentional and incidental learners using artifcial grammars (Dienes et al., 1991;
Mathews et al., 1989). The weight of evidence, therefore, suggests that conscious attempts to work
out a complex system do not add anything over and above what is acquired by passive, implicit, learn-
ing. But at least trying is not detrimental, perhaps because participants simply can’t develop coherent
hypotheses.
But there may be other situations in which trying hurts. In the SRT task, the difculty of dis-
covering the sequence can be increased by interspersing the sequenced items with random events;
for example, a simple sequence such as 1,4,2,2 becomes 1,R,4,R,2,R,2,R (in which numbers rep-
resent possible screen positions and R represents randomly generated positions). Using this “alter-
nating” SRT, Fletcher et al. (2005) found that simply telling participants that there was a repeating
sequence eliminated the learning efect. They suggest that this was because the participants actually
looked for patterns involving adjacent elements, which is the most natural way of interpreting what
a “sequence” might be. In line with this, Howard and Howard (2001) found that telling participants
that there was an alternating pattern (but not what it was) did not impair learning, at least in young
adults, but it did not improve it either (see the following for why this might be). Eforts to learn
may be harmful because attention may be directed to the input in the wrong way, according to
naïve assumptions about the domain, warping the data that enters the implicit learning mechanism,
amounting to a negative efect of weak interaction.
287
Of course, even without any instructions to do so, some participants may spontaneously look
for structure. Could creating conditions that reduce the likelihood of this increase implicit learning
efects? Hypnosis disturbs executive function, and hence should suppress explicit learning activity,
and it has been found that using the alternating SRT task, learning efects were 2.5 times greater
under hypnosis than in the normal alert state (Nemeth et al., 2013). Interestingly, the improvement
was actually confned to participants with high executive function, suggesting that this is the group
that are most likely to try to spontaneously fgure out a system, and so they beneft the most from
interventions that suppress that tendency.
In cases in which conscious hypotheses are more likely to be helpful, could learning be boosted
by combining implicit and rule search tasks together? In Mathews et al. (1989) participants were
trained on a relatively simple artifcial grammar that some people are able to work out under rule-
search instructions. Diferent groups of participants performed diferent combinations of implicit
(memorization) and explicit (rule-discovery) tasks during training, followed by a GJT. The best (in
fact near-perfect) GJT performance was obtained when participants performed the implicit task in
the frst half of training, followed by the rule-discovery task in the second half. This combination
was far more efective than either explicit or implicit training alone, or explicit training in the frst
half followed by implicit training in the second. The authors argue for a synergy between implicit
and explicit learning modes, leading to over-additive learning efects, suggestive of a strong interac-
tion. However, implicit training on its own led to no signifcant learning, and so it is hard to see
what kind of knowledge the explicit process could have been interacting with. One possibility is
that the memorization task allowed the participants to build up a store of exemplars, or fragments,
that facilitated subsequent rule discovery through conscious recall. In this view, there was no actual
facilitation from implicit knowledge of the underlying grammar system, and no interaction between
implicit and explicit learning processes.
All of the foregoing studies examined explicit learning in the sense of rule discovery. But what about
explicit instruction? Can this have an impact on the implicit learning process? In Reber et al. (1980)
participants were shown a diagram of a complex AG, with an explanation of how it was used to gen-
erate fve example grammatical strings. This treatment on its own resulted in a relatively low level of
GJT performance (presumably because the instructional information was not fully internalized), which
was equivalent to that obtained after implicit training in the form of passive observation of 63 example
strings. However, combining instruction with implicit training signifcantly improved GJT performance.
The authors argue that instruction “served to establish cognitive ‘boundaries’ for the tacit induction
operations engaged during the observation [implicit training] period” (ibid., p. 500, our addition). This
implies at least a “weak” form of interaction between explicit knowledge and implicit learning whereby
conscious knowledge guides implicit learning through the appropriate allocation of attention to the
stimuli (see “The role of attention in implicit statistical learning”, above). However, given that the
implicit task was passive observation, we cannot rule out that participants were actually engaging in rule
discovery, especially since they were informed that they would subsequently be asked questions about
the strings. Without additional procedures that reduce contamination from explicit learning processes
(e.g., a demanding secondary task) it cannot be claimed that instruction afected implicit learning as such.
There are also general boundary conditions on the usefulness of explicit information. One is the
learners’ ability to remember it (as illustrated by Reber et al., 1980). The other is the time required
to mobilize it, since this depends on controlled processing. For example, in Sanchez and Reber’s
(2013) SRT study, participants who memorized the sequence through observation in a pretraining
phase showed the same learning efects, as measured by actual SRT performance, as a non-instructed
group, despite showing superior conscious knowledge of the sequence following training. Presum-
ably, conscious knowledge could not be mobilized quickly enough to afect this speeded task. But
what is interesting is that simply knowing that there was a repeating sequence did not facilitate sub-
sequent implicit learning, something that might have been expected on a weak interaction position.
288
These studies provide an indication of the difculties of addressing interactions between implicit
and explicit knowledge and learning processes even within the apparently tightly constrained envi-
ronment of implicit learning experiments. The challenge is to show that it is precisely implicit
knowledge that interacts with explicit learning (e.g., rule discovery), or that explicit knowledge (e.g.,
from instruction) actually infuences implicit learning, either directly or indirectly. However, what is
relatively clear is that consciously attempting to work out a simple system can be benefcial, and for
complex systems it will be either neutral or damaging, depending on whether it results in a coun-
terproductive allocation of attention. Providing explicit information may have no efect if the task
is speeded, and there is no convincing evidence from these studies that it has even an indirect efect
on implicit learning. Clearly, there is scope for exploring these issues further using implicit learning
paradigms (see Textbox 23.2).
Individual differences
Although Reber (1993) originally claimed that implicit learning should be relatively invariant across
individuals, participants in implicit and statistical learning studies clearly show diferential learning
efects. Recent research suggests that this variation is meaningful in that it can be related to individu-
als’ sensitivity to the probabilistic structure of natural language (Conway et al., 2010; Divjak & Milin,
2020), and to specifc sentence processing measures (Misyak & Christiansen, 2012). These studies
suggest that there is a common underlying implicit statistical learning ability engaged in laboratory
tasks and both adult and child (Kidd, 2012) language acquisition.
Consolidation
In recent years there has been an explosion of interest in the role of sleep in memory consolidation.
Sleep, even if only a brief nap, has been shown to enhance implicit learning of phonotactic constraints
(Gaskell et al., 2014), retention of statistical information (Durrant et al., 2011), and learning of abstract
structure in both statistical learning (Gomez et al., 2006) and AGL (Nieuwenhuis et al., 2013). These
studies are important in showing how single-session lab studies might actually underestimate the
power of implicit learning over an extended time (also see MacWhinney, 2023 [this volume]).
Do constraints on implicit learning derive from general cognition or domain-specifc knowledge? Does
the nature of the L1 infuence what can be learned implicitly? Are there universal constraints on
implicit learning?
How do exposure conditions (incidental, intentional, instructed) afect the development of implicit and
explicit knowledge?
What is the basis of individual diferences in implicit learning ability?
What is the role of sleep in implicit learning, especially in relation to the emergence of generalizations
(and, potentially, the emergence of conscious knowledge)?
Implicit learning experiments usually reveal efects in receptive tasks after brief exposure. How does this
relate to the kind of knowledge that underlies fuent language production in a second language?
289
Further reading
Christiansen, M. H. (2019). Implicit statistical learning: A tale of two literatures. Topics in Cognitive Science, 11,
468–481.
An article that charts the development of the two research strands, implicit learning and statistical learning,
from their inception to the present day.
Frost, R., Armstrong, B., & Christiansen, M. (2019). Statistical learning research: A critical review and possible
new directions. Psychological Bulletin, 145(12), 1128–1153.
A critical appraisal of the last two decades of statistical learning research, covering key theoretical and meth-
odological issues.
Monaghan, P., & Rebuschat, P. (Eds.). (2019). Aligning implicit learning and statistical learning: Two approaches,
one phenomenon. Special issue of Topics in Cognitive Science, 11(3).
A special issue that brings together leading researchers from implicit learning and statistical learning to
encourage the formulation of joint research agendas.
Rebuschat, P. (2013). Measuring implicit and explicit knowledge in second language research. Language Learn-
ing, 63(3), 595–626.
A review of several measures of implicit and explicit knowledge, their respective limitations, and basic guid-
ance on their application.
Williams, J. N. (2009). Implicit learning in second language acquisition. In W. C. Ritchie & T. K. Bhatia (Eds.),
The new handbook of second language acquisition (pp. 319–353). Emerald Press.
A thorough review article on implicit learning and its role in second language acquisition.
References
Amato, M. S., & MacDonald, M. C. (2010). Sentence processing in an artifcial language: Learning and using
combinatorial constraints. Cognition, 116, 143–148.
Andringa, S. (2020). The emergence of awareness in uninstructed L2 learning: A visual world eye tracking study.
Batterink, L. J., Oudiette, D., Reber, P. J., & Paller, K. A. (2014). Sleep facilitates learning a new linguistic rule.
Neuropsychologia, 65, 169–179.
Batterink, L. J., Reber, P. J., Neville, H. J., & Paller, K. A. (2015). Implicit and explicit contributions to statistical
learning. Journal of Memory and Language, 83, 62–78.
Bonatti, L. L., Peña, M., Nespor, M., & Mehler, J. (2005). Linguistic constraints on statistical computations: The
role of consonants and vowels in continuous speech processing. Psychological Science, 16, 451–459.
Chan, R. K., & Leung, J. H. (2014). Implicit learning of L2 word stress regularities. Second Language Research,
30, 463–484.
Chen, W. W., Guo, X. Y., Tang, J. H., Zhu, L., Yang, Z. L., & Dienes, Z. (2011). Unconscious structural
knowledge of form-meaning connections. Consciousness and Cognition, 20, 1751–1760.
Conway, C. M., Bauernschmidt, A., Huang, S. S., & Pisoni, D. B. (2010). Implicit statistical learning in language
processing: Word predictability is the key. Cognition, 114, 356–371.
Curran, T., & Keele, S. W. (1993). Attentional and nonattentional forms of sequence learning. Journal of Experi-
mental Psychology: Learning, Memory, and Cognition, 19, 189–202.
de la Mora, D. M., & Toro, J. M. (2013). Rule learning over consonants and vowels in a non-human animal.
Cognition, 126, 307–312.
Dell, G. S., Reed, K. D., Adams, D. R., & Meyer, A. S. (2000). Speech errors, phonotactic constraints, and
implicit learning: A study of the role of experience in language production. Journal of Experimental Psychology:
Learning Memory, and Cognition, 26, 1355–1367.
Dienes, Z., Broadbent, D., & Berry, D. C. (1991). Implicit and explicit knowledge bases in artifcial grammar
learning. Journal of Experimental Psychology: Learning, Memory, and Cognition, 17, 875–887.
Dienes, Z., & Scott, R. (2005). Measuring unconscious knowledge: Distinguishing structural knowledge and
judgment knowledge. Psychological Research, 69, 338–351.
290
Divjak, D., & Milin, P. (2020). Exploring and exploiting uncertainty: Statistical learning ability afects how we
learn to process language along multiple dimensions of experience. Cognitive Science, 44, e12835.
Durrant, S. J., Taylor, C., Cairney, S., & Lewis, P. A. (2011). Sleep-dependent consolidation of statistical learn-
ing. Neuropsychologia, 49, 1322–1331.
Endress, A. D., & Hauser, M. D. (2009). Syntax-induced pattern deafness. Proceedings of the National Academy of
Sciences, 106, 21001–21006.
Fletcher, P. C., Zafris, O., Frith, C. D., Honey, R. A. E., Corlett, P. R., Zilles, K., & Fink, G. R. (2005). On
the benefts of not trying: Brain activity and connectivity refecting the interactions of explicit and implicit
sequence learning. Cerebral Cortex, 15, 1002–1015.
Franco, A., Cleeremans, A., & Destrebecqz, A. (2011). Statistical learning of two artifcial languages presented
successively: How conscious? Frontiers in Psychology, 2.
Franco, A., Cleeremans, A., & Destrebecqz, A. (2016). Objective and subjective measures of cross-situational
learning. Acta Psychologica, 165, 16–23.
Frost, R. L. A., Armstrong, B., & Christiansen, M. (2019). Statistical learning research: A critical review and
possible new directions. Psychological Bulletin, 145(12), 1128–1153.
Frost, R. L. A., & Monaghan, P. (2016). Simultaneous segmentation and generalisation of non-adjacent depen-
dencies from continuous speech. Cognition, 147, 70–74.
Gaskell, M. G., Warker, J., Lindsay, S., Frost, R., Guest, J., Snowdon, R., & Stackhouse, A. (2014). Sleep under-
pins the plasticity of language production. Psychological Science, 25, 1457–1465.
guage acquisition. In A. Godfroid & H. Hopp (Eds.). The Routledge handbook of second language acquisition and
Gomez, R. L. (2002). Variability and detection of invariant structure. Psychological Science, 13, 431–436.
Gomez, R. L., Bootzin, R. R., & Nadel, L. (2006). Naps promote abstraction in language-learning infants.
Psychological Science, 17, 670–674.
Graham, C. R., & Williams, J. N. (2018). Implicit learning of Latin stress regularities. Studies in Second Language
Guo, X., Zheng, L., Zhu, L., Yang, Z., Chen, C., Zhang, L., Ma, W., & Dienes, Z. (2011). Acquisition of con-
scious and unconscious knowledge of semantic prosody. Consciousness and Cognition, 20, 417–425.
Hama, M., & Leow, R. P. (2010). Learning without awareness revisited. Studies in Second Language Acquisition,
32, 465–491.
Hamrick, P., & Rebuschat, P. (2012). How implicit is statistical learning? In P. Rebuschat & J. N. Williams
(Eds.), Statistical learning and language acquisition (pp. 365–382). de Gruyter.
Hendricks, M. A., Conway, C. M., & Kellogg, R. T. (2013). Using dual-task methodology to dissociate auto-
matic from nonautomatic processes involved in artifcial grammar learning. Journal of Experimental Psychology-
Learning Memory and Cognition, 39, 1491–1500.
Howard, D. V., & Howard, J. H. (2001). When it does hurt to try: Adult age diferences in the efects of instruc-
tions on implicit pattern learning. Psychonomic Bulletin & Review, 8, 798–805.
Isbilen, E. S., McCauley, S. M., Kidd, E., & Christiansen, M. H. (2020). Statistically induced chunking recall: A
memory-based approach to statistical learning. Cognitive Science, 44, e12848.
Jacoby, L. L. (1991). A process dissociation framework: separating automatic from intentional uses of memory.
Journal of Memory and Language, 30, 513–541.
Jiménez, L., & Méndez, C. (1999). Which attention is needed for implicit sequence learning? Journal of Experi-
mental Psychology: Learning, Memory, and Cognition, 25, 236–259.
Kerz, E., Wiechmann, D., & Riedel, F. B. (2017). Implicit learning in the crowd: Investigating the role of aware-
ness in the acquisition of L2 knowledge. Studies in Second Language Acquisition, 39, 711–734.
Kidd, E. (2012). Implicit statistical learning is directly associated with the acquisition of syntax. Developmental
Psychology, 48, 171–184.
Kinder, A., & Assmann, A. (2000). Learning artifcial grammars: No evidence for the acquisition of rules.
Memory and Cognition, 28, 1321–1332.
Knowlton, B. J., & Squire, L. R. (1996). Artifcial grammar learning depends on implicit acquisition of both
abstract and exemplar-specifc information. Journal of Experimental Psychology: Learning, Memory, and Cogni-
tion, 22, 169–181.
Leung, J. H. C., & Williams, J. N. (2012). Constraints on implicit learning of grammatical form-meaning con-
nections. Language Learning, 62, 634–662.
Leung, J. H. C., & Williams, J. N. (2014). Crosslinguistic diferences in implicit language learning. Studies in
Second Language Acquisition, 36, 733–755.
291
Logan, G. D., & Etherton, J. L. (1994). What is learned during automatization? The role of attention in con-
structing an instance. Journal of Experimental Psychology: Learning, Memory and Cognition, 20, 1022–1050.
MacWhinney, B. (2023). Synthesis: the psycholinguistics of learning. In A. Godfroid & H. Hopp (Eds.). The
Routledge handbook of second language acquisition and psycholinguistics (pp. 361–370). Routledge.
Mathews, R. C., Buss, R. R., Stanley, W. B., Blanchard-Fields, F., Cho, J.-R., & Druhan, B. (1989). The role of
implicit and explicit processes in learning from examples: A synergistic efect. Journal of Experimental Psychol-
ogy: Learning, Memory, and Cognition, 15, 1083–1100.
Misyak, J. B., & Christiansen, M. H. (2012). Statistical learning and language: An individual diferences study.
Monaghan, P., Mattock, K., Davies, R., & Smith, A. C. (2015). Gavagai is as gavagai does: Learning nouns and
verbs from cross-situational statistics. Cognitive Science, 39, 1099–1112.
Monaghan, P., Schoetensack, C., & Rebuschat, P. (2019). A single paradigm for implicit and statistical learning.
Topics in Cognitive Science, 11, 536–554.
psycholinguistic considerations. In A. Godfroid & H. Hopp (Eds.). The Routledge handbook of second language
Nazzi, T., & Cutler, A. (2019). How consonants and vowels shape spoken-language recognition. Annual Review
of Linguistics, 5, 25–47.
Nemeth, D., Janacsek, K., Polner, B., & Kovacs, Z. A. (2013). Boosting human learning by hypnosis. Cerebral
Cortex, 23, 801–805.
Newell, B. R., & Shanks, D. R. (2014). Unconscious infuences on decision making: A critical review. Behavioral
and Brain Sciences, 37, 1–18.
Newport, E. L., & Aslin, R. N. (2004). Learning at a distance I. Statistical learning of non-adjacent dependen-
cies. Cognitive Psychology, 48, 127–162.
Nieuwenhuis, I. L. C., Folia, V., Forkstam, C., Jensen, O., & Petersson, K. M. (2013). Sleep promotes the
extraction of grammatical rules. PLoS ONE, 8, e65046.
Nissen, M. J., & Bullemer, P. (1987). Attentional requirements of learning: Evidence from performance mea-
sures. Cognitive Psychology, 19, 1–32.
Onnis, L., Monaghan, P., Richmond, K., & Chater, N. (2005). Phonology impacts segmentation in online
speech processing. Journal of Memory and Language, 53, 225–237.
Paciorek, A., & Williams, J. N. (2015a). Implicit learning of semantic preferences of verbs. Studies in Second
Paciorek, A., & Williams, J. N. (2015b). Semantic generalization in implicit language learning. Journal of Experi-
mental Psychology: Learning Memory and Cognition, 41, 989–1002.
Pham, T., Kang, J. H., Johnson, A., & Archibald, L. M. D. (2020). Feature-focusing constraints on implicit
learning of function word and meaning associations. Applied Psycholinguistics, 41, 401–426.
Reber, A. S. (1967). Implicit learning of artifcial grammars. Journal of Verbal Learning and Verbal Behavior, 6,
855–863.
Reber, A. S. (1976). Implicit learning of synthetic languages: The role of instructional set. Journal of Experimental
Psychology: Human Learning and Memory, 2, 88–94.
Reber, A. S. (1993). Implicit learning and tacit knowledge. Oxford University Press.
Reber, A. S. (2022). Implicit learning: Background, history, theory. In A. S. Reber & R. Allen (Eds.), The cogni-
tive unconscious: The frst half-century (pp. 2–21). Oxford University Press.
Reber, A. S., Kassin, S. M., Lewis, S., & Cantor, G. (1980). On the relationship between implicit and explicit
modes in the learning of a complex rule structure. Journal of Experimental Psychology: Human Learning and
Memory, 6, 492–502.
Reber, A. S., & Lewis, S. (1977). Implicit learning: An analysis of the form and structure of a body of tacit
knowledge. Cognition, 5, 333–361.
Rebuschat, P. (2013). Measuring implicit and explicit knowledge in second language research. Language Learn-
ing, 63, 595–626.
Rebuschat, P., & Monaghan, P. (2019). Editors’ introduction: Aligning implicit learning and statistical learning:
Two approaches, one phenomenon. Topics in Cognitive Science, 11, 459–467.
Rebuschat, P., & Williams, J. N. (2012). Implicit and explicit knowledge in second language acquisition. Applied
Safran, J. R., Newport, E. L., & Aslin, R. N. (1996). Word segmentation: The role of distributional cues. Jour-
nal of Memory and Language, 35, 606–621.
Sanchez, D. J., & Reber, P. J. (2013). Explicit pre-training instruction does not improve implicit perceptual-
motor sequence learning. Cognition, 126, 341–351.
292
Schmidt, R. (1994). Implicit learning and the cognitive unconscious: Of artifcial grammars and SLA. In N. C.
Ellis (Ed.), Implicit and explicit learning of languages (pp. 165–209). Academic Press.
Timmermans, B., & Cleeremans, A. (2015). How can we measure awareness? An overview of current methods.
In M. Overgaard (Ed.), Behavioral methods in consciousness research (pp. 21–46). Oxford University Press.
Toro, J. M., Nespor, M., Mehler, J., & Bonatti, L. L. (2008). Finding words and rules in a speech stream: Func-
tional diferences between vowels and consonants. Psychological Science, 19, 137–144.
Toro, J. M., Sinnett, S., & Soto-Faraco, S. (2005). Speech segmentation by statistical learning depends on atten-
tion. Cognition, 97, B25–B34.
Waldron, E. M., & Ashby, F. G. (2001). The efects of concurrent task interference on category learning: Evi-
dence for multiple category learning systems. Psychonomic Bulletin & Review, 8, 168–176.
Williams, J. N. (2005). Learning without awareness. Studies in Second Language Acquisition, 27, 269–304.
Williams, J. N. (2009). Implicit learning and second language acquisition. In W. C. Ritchie & T. K. Bhatia
(Eds.), The new handbook of second language acquisition (pp. 319–353). Emerald.
Williams, J. N. (2013). Attention, awareness, and noticing in language processing and learning. In J. M. Berg-
sleithner, S. N. Frota, & J. K. Yoshioka (Eds.), Noticing and second language acquisition: Studies in Honor of Rich-
ard Schmidt (pp. 39–57). National Foreign Language Resource Centre.
Yu, C., & Smith, L. (2007). Rapid word learning under uncertainty via cross-situational statistics. Psychological
Science, 18, 414–420.
293
24
HYPOTHESES ABOUT THE
INTERFACE BETWEEN EXPLICIT
AND IMPLICIT KNOWLEDGE IN
SECOND LANGUAGE ACQUISITION
Aline Godfroid
Introduction
Researchers interested in the cognitive-psychological aspects of language learning often study the
contributions of attention and awareness to frst (L1) and second language (L2) learning. Attention
and especially awareness are fundamentally related to a basic distinction in knowledge, namely the
distinction between explicit knowledge and implicit knowledge. This distinction underlies many
domains of cognition, including social interaction, music perception, and language. Explicit knowl-
edge can be defned as conscious and verbalizable knowledge of regularities. Learners “know that
they know”; they are aware of the knowledge they have and this is why, in the case of explicit
linguistic knowledge, they are able to articulate it as a rule or an explanation of how the language
works. The explanation can be in layman’s terms; that is, explicit knowledge and metalanguage (the
use of technical terms) are not the same (R. Ellis, 2004). Implicit knowledge is tacit and unconscious
knowledge that is gained mainly through exposure to rich input, and therefore cannot be easily
taught. Implicit knowledge is difcult to verbalize and often manifests itself as a feeling or intuition
that something just sounds right or wrong when said in a certain way.
A central issue in second language acquisition (SLA), which is known as the interface hypothesis,
is how explicit and implicit knowledge relate.1 Does explicit knowledge of how a language works
help L2 learners acquire implicit knowledge? And assuming explicit knowledge helps, is that because
it transforms into implicit knowledge over time? Or are explicit knowledge and implicit knowledge
essentially unrelated? All of these questions are interface questions (see Textbox 24.1).
Consider the English article system as an example. For L2 speakers of English, it can be notoriously
difcult to learn how to use the, a(n), and ∅ correctly in speech and writing. Perhaps you were taught
a pedagogical rule such as to use the defnite article the before a noun when the referent is assumed to
be known knowledge to the hearer/reader. Learning such a rule would result in explicit knowledge
of the defnite article, but such knowledge is imperfect (there are many exceptions). Over time, how-
ever, speakers may acquire all the subtleties of article usage in English, which many researchers believe
refects their implicit knowledge. The interface hypothesis addresses the question of whether initial
rule knowledge ultimately results in learners’ ability to use articles correctly without giving it any
conscious thought and without thinking about the rule; that is, whether explicit knowledge directly
impacts or can be converted into implicit knowledge. Conversely, the reverse interface theorizes
294 DOI: 10.4324/9781003018872-28

The interface between explicit and implicit knowledge in second language acquisition
whether implicit knowledge can give rise to explicit knowledge, for instance whether L1 speakers will
be able to articulate the rules for when to use the, a(n), or ∅ even when they have not been trained in
linguistics or TESOL or have been taught the rules in some way.
Awareness: A conscious, subjective experience of a thought, feeling, or stimulus in the outside world that
the individual can report verbally. Awareness is a key criterion distinguishing explicit knowledge and
explicit learning from implicit knowledge and implicit learning.
Explicit/Implicit knowledge: Conscious and verbalizable knowledge vs. unconscious and tacit knowl-
edge of regularities in language. Whether knowledge is explicit or implicit can be determined through
participants’ self-reported awareness of their knowledge, subjective measures, or more indirectly by
manipulating the response time available on a language test.
Explicit/Implicit learning: Conscious-intentional vs. unconscious learning of regularities in language as
measured at the time at which the new information is encoded in memory. A third type of learning,
incidental learning, is unplanned and unintentional and can be either explicit or implicit.
Interface hypothesis: The question of whether explicit knowledge directly impacts or can be converted
into implicit knowledge.
Reverse interface hypothesis: The question of whether implicit knowledge can give rise to explicit
knowledge.
Theoretical perspectives
The interface hypothesis

Many questions in cognitive approaches to SLA fnd their origin in scholarly work from the 1980s
when Stephen Krashen put forward his monitor model of SLA (Krashen, 1981, 1982, 1985). Krashen
assumed that L2 knowledge develops through two independent routes, namely acquisition and learn-
ing. While language acquisition takes place subconsciously, in the sense that learners are mostly unaware
of their acquisition process, language learning is a conscious process in which learners become aware
of language rules. In contemporary terms, acquired knowledge refers to implicit knowledge, and
learned knowledge refers to explicit knowledge. Krashen maintained that L2 knowledge developed
via these independent routes serves distinct purposes and remains in two separate internalized sys-
tems. Therefore, one type of linguistic knowledge does not directly connect with, transform into, or
convert into the other (Hulstijn, 2002; Krashen, 1981; Paradis, 2009) as the aforementioned example
on English article use by English-as-a-second-language (ESL) speakers may have illustrated. Krashen
hypothesized that learned (explicit) knowledge, as promoted in many traditional second/foreign lan-
guage classrooms, may only act as a monitor or editor of L2 speakers’ spoken or written production
accuracy. To communicate fuently, L2 learners need to draw on their acquired (implicit) knowledge
instead. This viewpoint became known as the non-interface position.
Subsequent researchers responded to and took issue with Krashen’s non-interface position, giving
voice to the weak and strong interface positions, respectively. The strong interface position is associ-
ated with skill acquisition theory (DeKeyser, 1998, 2017, 2020; also see Suzuki, 2023 [this volume]).
From a skill acquisition perspective, explicit knowledge of how the language works gives learners a
base of declarative knowledge on which they may draw to develop procedural knowledge (a type of
implicit knowledge) through repeated practice. Proponents of the weak interface position, on the
295
Aline Godfroid
other hand, argue that explicit and implicit knowledge work cooperatively, with explicit knowledge
indirectly infuencing the development of implicit knowledge. Such infuence includes a facilitatory
role of (conscious) noticing ansd noticing the gap (consciously registering a mismatch between one’s
own output and target language input; e.g., R. Ellis, 1994) and explicit language processing (e.g., N.
Ellis, 2005, 2015) in the development of implicit or procedural knowledge. R. Ellis (1994) further
highlighted how output, produced with the help of L2 explicit knowledge, can serve as a source of
auto-input to the implicit acquisition system.
Although such infuences of explicit knowledge on implicit knowledge may indeed be pervasive
(e.g., DeKeyser, 2017; N. Ellis, 2015; R. Ellis, 2009; Hulstijn, 2015; Paradis, 2009), researchers agree
that all of these infuences are indirect, in that explicit knowledge does not change, make physi-
cal contact with, or transform into implicit knowledge. “An interface implies some form of direct
interaction, interconnection, intercommunication, data exchange between systems, sharing of infor-
mation, or signal conversion—none of which exists between explicit metalinguistic knowledge (or
learning thereof) and implicit linguistic competence (or acquisition thereof)” (Paradis, 2009, p. 63).
“An indirect infuence is not an interface” (Paradis, 2009, p. 76). Consequently, the commonalities
between the diferent theoretical interface positions may in fact be much larger than what distin-
guishes them and, I argue, the hard and fast distinction between the three interface positions may be
outdated. A more fruitful way forward, then, adopted in this chapter is to deconstruct the interface
hypothesis into its diferent components—instruction, processing, knowledge—and study the inter-
relationships between its components using the fne-grained empirical tools that have become avail-
able to psycholinguists. In the section “Critical issues and topics”, I take stock of the advances that
have been made in this area of L2 psycholinguistics.
The reverse interface hypothesis

The reverse interface hypothesis aims to account for how implicit knowledge may give rise to explicit
knowledge. It is rooted in developmental psychology, cognitive psychology, research on L1 acquisi-
tion, and bilingualism; and as such, it complements the interface hypothesis in SLA, discussed previ-
ously, which concerns the extent to which (adult) L2 learners may develop implicit knowledge from
explicit knowledge. One theoretical account of the reverse interface is the radical plasticity thesis
(Cleeremans, 2008, 2011). This thesis holds that knowledge progresses through three diferent stages
of frst increasing and then decreasing consciousness and control (see Figure 24.1). Initially, as the
mind is processing incoming information, the resulting knowledge representations are unconscious
or implicit (stage 1). At this stage, implicit knowledge is characterized by a low depth of processing
and weak, low-quality representations in memory (see Leow, 2015, for a similar viewpoint in SLA).
Because the implicit knowledge is so weak, it can infuence behavior only under very specifc task
conditions (e.g., when the knowledge primes associations in memory, via associative priming). With
increasing exposure, however, the strength and quality of the representations improves, and knowl-
edge gradually allows for conscious inspection, namely, when an explicit representation is formed
(see next paragraph for further details). This transition represents an implicit-explicit interface.
The process that enables the transition from implicit to explicit knowledge is called representational
redescription (Karmilof-Smith, 1992) or analysis (e.g., Bialystok, 2001). Representational redescrip-
tion is a general, cyclical process (not specifc to language) that denotes the “recoding [of] informa-
tion that is stored in one representational format into a diferent one” (Karmilof-Smith, 1992, p. 23)
that is “a more condensed or compressed version of the previous level” (ibid.). Analysis of knowl-
edge is a concrete example of representational redescription in the domain of language. It refers to
“the ability to represent increasingly explicit and abstract structures” (Bialystok, 2001, p. 131). Both
representational redescription and analysis result in conscious access to knowledge, enabling an indi-
vidual to have more fexible control over their knowledge so they may use the knowledge to pursue
296
Figure 24.1 A three-stage representation of implicit, explicit, and automatized explicit knowledge (Cleer-
emans, 2008) with representational redescription (Karmilof-Smith, 1992) and analysis (e.g., Bialystok, 2001)
marking the transition between diferent stages. Cleeremans (2008, Figure 1) reprinted and modifed with per-
mission from the author and from Elsevier.
diferent task goals. The important point here from an SLA perspective is that explicit knowledge is
seen as a more advanced stage of knowledge than implicit knowledge.
Yet, learning is not complete once knowledge has become conscious or explicit (Cleeremans,
2008, 2011). With continued exposure, explicit knowledge representations will eventually become
so strong that their infuence on behavior will no longer be subject to conscious control (i.e.,
the knowledge will infuence behavior automatically). This fnal stage of development represents
the automaticity of explicit knowledge (stage 3). In SLA approaches, automatized explicit knowledge
and implicit knowledge are often treated as “functionally equivalent,” or indistinguishable in language
use, in contrast with Figure 24.1, in which implicit and automatized explicit knowledge represent
the starting point and endpoint, respectively, of a single developmental continuum (Cleeremans,
2008, 2011). The radical plasticity thesis, then, encompasses all three aspects of implicit, explicit, and
automatized explicit knowledge and could thus ofer a unifying framework that captures all of these
aspects.

I next consider four bodies of research that all speak to diferent aspects of the interface and the
reverse interface hypothesis. Although no single line of work can provide any defnitive answers, the
picture to emerge from all four lines of work together is that infuences between explicit and implicit
knowledge are pervasive. Figure 24.2 is a diagram of the diferent components of the interface
297
Aline Godfroid
Figure 24.2 Main areas of research on the interface and the reverse interface hypothesis.
hypothesis and the reverse interface hypothesis addressed in this section. It is not meant to be a model
of the (reverse) interface, but rather can be used as a conceptual tool for understanding which inter-
face questions require further study. Readers may fnd it useful to refer back to this fgure as they
read the following sections.
Simultaneous acquisition of explicit and implicit knowledge from instruction

An interesting question from an interface perspective is whether one type of instruction (e.g., input
food, which is a type of implicit instruction) can promote implicit and explicit knowledge develop-
ment simultaneously. This line of research can illuminate diferent paths to knowledge development
even if the intervening learning processes (i.e., explicit or implicit learning) are assumed, rather than
observed directly (see Figure 24.2). As such, research on simultaneous acquisition can inform the
interface and the reverse interface hypothesis indirectly.
In natural language research, only a handful of researchers have tackled the question of simulta-
neous acquisition of implicit and explicit knowledge (e.g., Andringa et al., 2011; De Jong, 2005;
Godfroid, 2016), perhaps because it is difcult to develop valid and reliable measures of implicit and
explicit knowledge. De Jong (2005) examined whether the direction of training (comprehension-
only training or balanced, comprehension and production training) impacted learners’ comprehension
and production of Spanish noun-adjective gender agreement. Both types of training were considered
implicit. Nearly all participants detected the target structure and half of them induced at least par-
tial rule knowledge, suggesting they developed explicit knowledge from implicit instruction. Both
instructional groups also outperformed the rule-only, no-training control group in real-time com-
prehension posttests. De Jong concluded that “these groups either processed their explicit knowledge
quickly or used implicit knowledge” (p. 227), but she acknowledged that it was difcult to distinguish
between these two possibilities because of uncertainty about what the tests were measuring.
In another study on implicit instruction, Godfroid (2016) aimed to link instruction, processing,
and knowledge more directly. Godfroid administered measures of implicit and explicit knowledge at
pre-test and post-test and an online learning measure (reaction times on a sentence-picture matching
task) to track the learning during the sentence-picture matching task. Additionally, she used debrief-
ings to classify participants as aware or unaware, based on whether they reported noticing that the
experiment focused on verbs. The results showed that all learners became more target-like in their
298
real-time processing of German verbs and demonstrated implicit knowledge on the implicit knowl-
edge post-test as a result of training. “Unaware” participants showed evidence of implicit learning and
implicit knowledge only, but “aware” learners, who reported noticing the verbs, also demonstrated
explicit knowledge on the explicit knowledge post-test. Hence, simultaneous acquisition of explicit
and implicit knowledge (as measured independently from awareness, on the post-tests) coincided
with learners’ ability to detect and verbalize relevant linguistic information (i.e., the presence of
strong verbs in the input) during the debriefng interviews.
Debriefng interviews play an important role in classroom-based research to distinguish aware
and unaware learners. In lab-based research, and especially in artifcial and semi-artifcial language
studies, Dienes (2007) argued for the use of subjective measures, such as confdence ratings (asking
participants to judge how confdent they are in their decision) and source attributions (whether they
based their decision on a guess, feeling, memory, or rule). Explicit knowledge, it is argued, is knowl-
edge that people “know that they know” and hence, it manifests subjectively as a (conscious) rule or
memory, whereas the phenomenology of implicit knowledge is that of intuition or a guess (Dienes,
2007). Introducing this approach from psychology to SLA, Rebuschat and Williams (2012) exposed
participants to a semi-artifcial language that consisted of English sentences in which the verbs had
been rearranged according to German syntax, showcasing three verb placement rules. In their second
experiment, participants demonstrated statistically signifcant gains in knowledge of the syntactic
rules on an auditory grammaticality judgment test (GJT), for both items attributed to intuition
(a response type associated with implicit knowledge) and items attributed to rule (a response type
associated with explicit knowledge). These results of simultaneous implicit and explicit knowledge
were replicated by Kim and Fenn (2020), but not by Kim and Godfroid (2019), Miller and Godfroid
(2020), Bell (2017), or Tagarelli et al. (2016). Collectively, these studies demonstrate the potential
application of subjective measures in L2 research even if evidence from subjective measures for the
simultaneous acquisition of implicit and explicit knowledge in L2 research remains inconclusive.
In the area of morphosyntax, evidence for the simultaneous acquisition of implicit and explicit
knowledge has likewise been mixed. Studies typically show that responses on a GJT identifed by the
participants as guesses (i.e., a response type associated with implicit knowledge) do not reveal participants’
implicit knowledge because participants perform with chance-level accuracy on these items, indicating
no knowledge (e.g., Rebuschat et al., 2015; Rogers et al., 2016). Responses on a GJT attributed to intu-
ition do show implicit knowledge, at least in some studies (Rebuschat et al., 2015; Rogers et al., 2016).
In an attempt to further bring together research traditions in psychology and SLA, Maie and
DeKeyser (2020) triangulated subjective measures, namely confdence ratings and source attribu-
tions, with objective measures, i.e., an untimed auditory GJT for explicit knowledge and a word
monitoring task for implicit knowledge. Maie and DeKeyser found that the subjective and objective
measures yielded conficting evidence about what participants had learned as a result of instruction.
For instance, for the subjective measures, participants judged some structures accurately on the GJT
even when they attributed their judgments to intuition (a response category associated with implicit
knowledge), whereas the word monitoring task (the objective measure) did not reveal implicit knowl-
edge. Maie and DeKeyser’s study raised questions about the practical value of implicit knowledge as
refected in subjective measures. Specifcally, the authors showed that evidence for implicit knowledge
from subjective measures is not very strong because it does not have the functional and qualitative
characteristics commonly associated with implicit knowledge in SLA (i.e., knowledge available during
fuent and spontaneous language use). A possible interpretation in the context of Cleeremans’ (2008,
2011) model of cognition would be that subjective measures capture knowledge in the initial stages
of development, corresponding to stage 1 rather than stage 3 (see “The reverse interface hypothesis”).
In sum, natural language studies and (semi-)artifcial language studies provide some evidence for
the simultaneous acquisition of L2 implicit and explicit knowledge. Whether simultaneous acquisi-
tion actually occurs may depend on specifc factors such as the length and intensity of instruction,
299
Aline Godfroid
the participants’ awareness of the target structure in the input (as determined independently from the
outcome, in an interview or debriefng questionnaire), and how implicit and explicit knowledge are
measured (i.e., with subjective or objective measures). Notably, Maie and DeKeyser (2020) found
conficting evidence from subjective and objective measures, highlighting a need for more research
on the external validity of subjective measures of implicit knowledge in SLA. For parallel eforts
regarding internal validity, see Hamrick and Sachs (2018) and Sachs et al., (2020).
Effects of explicit instruction on online processing

In recent years, researchers have turned to processing measures, such as eye-tracking and self-paced
reading, as a means to gather fne-grained, real-time information about how L2 learners process
linguistic input. A question of interest is whether participants’ real-time processing changes under
the infuence of explicit instruction. This is an old interface question put in a new guise: Indeed, if
explicit instruction can be shown to infuence online processing, this may represent an example of an
explicit-implicit infuence, although the assumption that real-time processing is by its nature implicit
is itself a matter of debate (e.g., Godfroid, 2020a; Marsden et al., 2018).
In an eye-tracking study on verb temporality in Latin, Cintrón-Valentín and Ellis (2015) found
that explicit form-focused instruction on verb conjugation was an efective intervention for help-
ing L2 speakers overcome attentional biases from the L1 whereby they pay attention to temporal
adverbs at the expense of processing tense infections on verbs. After instruction, participants
used the adverbs and verb endings to comprehend and produce past, present, and future-tense
verbs. Explicit instruction was found to be more impactful in redirecting attention away from
the adverbs and towards the verbs in the visual than the auditory modality, possibly because verb
infections are more salient in writing than speech (Cintrón-Valentín & Ellis, 2016). Andringa and
Curcic (2015) also examined the efects of explicit instruction on real-time processing, but with
a diferent eye-tracking paradigm, namely visual world eye-tracking. They used the visual world
eye-tracking paradigm to study prediction in real-time language processing and interpreted pre-
diction (as revealed in the eye-movement record) as evidence of implicit processing. Participants
who were frst taught the target grammatical rule (i.e., diferential object marking) explicitly were
unable to use their explicit knowledge during real-time language processing to generate predic-
tions. The authors interpreted the lack of prediction in the eye-tracking data as evidence of a lack
of an explicit knowledge-implicit processing interface. Together, Cintrón-Valentín and Ellis (2015,
2016) and Andringa and Curcic (2015) provide interesting model studies of how eye-tracking
methodology can be used to study interface questions. More information on diferent eye-tracking
paradigms is available in Godfroid (2020a).
A second series of online processing studies was conducted within the theoretical framework
of VanPatten’s model of input processing (1996, 2004) and the intervention of processing instruc-
tion (PI) specifcally. The goal of PI is to “manipulate learner attention during input processing or
manipulate input data so that more and better form-meaning connections are made” (VanPatten,
2005, p. 272). Issa and Morgan-Short (2019) provided direct evidence from eye-tracking to show
how structured input (as one component of PI) together with corrective feedback might infuence
learners’ allocation of attention during processing. L2 learners interpreted Spanish pre-verbal clitics
more accurately after they engaged in structured input practice, which suggested they successfully
processed the clitics’ meanings. Similar results were obtained for the Spanish passive construction by
Lee and Doherty (2019) and the French causative construction by Wong and Ito (2018), both with
and without explicit instruction.
McManus and Marsden (2017, 2018) combined L1 explicit instruction and L1 practice in a
study designed to investigate the potential benefts of using the L1 in L2 instruction. The target
for acquisition was L2 verb aspectual distinctions for the French imparfait—a tense that can signal
300
either ongoingness or habituality. When L1 explicit instruction (i.e., instruction about the L1
features equivalent to the target L2 structure) was coupled with L1 PI consisting of form-mean-
ing mapping practice in interpreting L1 features, participants showed greater learning outcomes
than the learning from L2 explicit instruction with L2 practice alone. A group that received L1
practice alone did not show the same learning advantage (McManus & Marsden, 2018), high-
lighting the important role of L1 explicit instruction in promoting learning outcomes. As in
the previously reviewed studies, these efects held true for both ofine and online (self-paced
reading) measures.
Collectively, the fndings from these studies demonstrate that explicit instruction can be an
efective intervention for changing the ways in which L2 learners comprehend and process lan-
guage in real time.
Disentangling automatized explicit and implicit knowledge

The past decades have seen an increase in research designed to validate measures of explicit and,
especially, implicit knowledge. In a landmark study, Rod Ellis (2005) launched a productive line of
research into the construct validity of explicit and implicit L2 knowledge measures. He proposed a
list of task criteria that would make L2 learners more inclined to use explicit knowledge or implicit
knowledge on a given task. Using a principal component factor analysis, the researcher found there
were underlying commonalities between the elicited imitation test, oral production task, and a timed
written GJT, in line with the hypothesis that these measures tap into learners’ implicit knowledge.
The metalinguistic knowledge test and the ungrammatical items on an untimed GJT, on the other
hand, loaded onto a diferent component, consistent with the hypothesis that they measured learners’
explicit knowledge (see Ellis & Loewen, 2007, for a replication with confrmatory factor analysis).
In terms of task criteria, the implicit knowledge measures in the battery were timed and, with the
exception of the timed GJT, they focused participants’ attention on meaning. Hence, time pressure
and attentional focus (i.e., focus on form vs. meaning) emerged as the task criteria in the study that
future researchers would build on and/or call into question when designing and validating new mea-
sures of implicit knowledge.
Rod Ellis’s study inspired a series of follow-up studies that also examined the construct validity of
linguistic knowledge measures with diferent L2 and heritage learner groups (e.g., Bowles, 2011), in
diferent second- and foreign-language contexts (e.g., Zhang, 2015), and including diferent varia-
tions of the same tasks (e.g., Kim & Nam, 2017). By and large, these earlier studies confrmed
Ellis’s original results. The debate on how to measure implicit knowledge regained momentum
when online measures such as self-paced reading and visual world eye-tracking were introduced as a
novel alternative for measuring implicit knowledge (Suzuki & DeKeyser, 2015; Vafaee et al., 2017).
Online measures capture participants’ processing during real-time comprehension (also see “Efects
of explicit instruction on online processing”). They may be more appropriate to tap into learn-
ers’ implicit knowledge because they are processing measures, and the immediacy of measurement
leaves little room for conscious inspection (Suzuki & DeKeyser, 2015; Vafaee et al., 2017). Adding
time pressure to an ofine (product- or outcome-based) test like a GJT, on the other hand, makes
it more difcult for learners to access their explicit knowledge but it does not preclude them from
doing so (see DeKeyser, 2003). Based on these arguments, Suzuki and DeKeyser (2015) hypoth-
esized that timed tests (e.g., elicited imitation) may be more suited to tap into learners’ automatized
explicit knowledge, which they defned as “conscious [explicit] knowledge that is partially (not fully)
automatized [and] can theoretically be distinguished from implicit knowledge” (Suzuki & DeKeyser,
2017, p. 751; my addition; see also Suzuki, 2023 [this volume]). Hence, recent test validation research
has focused on the contrast between timed, ofine tests, such as the timed GJT and elicited imitation
test, and online reaction time measures, such as self-paced reading and the word monitoring task.
301
Aline Godfroid
In an efort to reconcile the diferent positions regarding the measurement of implicit knowledge,
Godfroid and Kim (2021) and Godfroid et al. (in prep.) recently brought together a battery of two
online reaction time measures, four time-pressured ofine tests, and three untimed ofine tests, along
with fve individual diferences measures. Using confrmatory factor analysis, the researchers found
that a two-factor model, termed implicit and explicit knowledge, accounted for the data best, as origi-
nally reported by R. Ellis (2005). Nonetheless, that model did not account for the reaction time mea-
sures well, suggesting the nature of these tasks is not yet well understood. Reaction-time-based tests
are promising measures of implicit knowledge, but they are currently less psychometrically robust
than timed, accuracy-based tasks (Draheim et al., 2019). New analytical approaches that rely on the
random slopes (i.e., statistical estimates capturing individual variation in how participants respond to
diferent conditions) from mixed-efects regression modeling (Rouder & Haaf, 2019) could improve
the reliability of reaction-time-based tests and, ultimately, help settle the question of how best to
measure implicit knowledge.
Reverse interface hypothesis

To my knowledge, Andringa (2020) was the frst lab-based study in SLA to test the reverse interface
hypothesis empirically by examining whether implicit learning, operationalized as learners’ ability to
make predictions from article-like words about which nouns they will hear next, gives rise to explicit
(verbalizable) knowledge in the form of a conscious insight or “aha moment” about the grammatical
properties of the article-like words. A conscious insight is a specifc moment in time when things just
click in the learner’s mind; it represents a form of self-generated awareness that is believed to occur as
a result of unconscious implicit learning (e.g., Wagner et al., 2004). An insight is thus a possible sign
of a reverse interface whereby unconscious implicit learning gives rise to conscious explicit knowl-
edge, which learners can then verbalize.
In the study by Andringa, Dutch native speakers participated in an untutored learning experi-
ment of an artifcial language, in which they saw simple visual scenes on the screen while listening to
sentences that described the visual scenes. Andringa (2020) found that participants showed evidence
of prediction, and made anticipatory eye movements to the correct objects in the scenes, only after
they became aware of the grammatical rules of the artifcial language. Thus, although there was evi-
dence of a spontaneous insight in the study, it crucially was not preceded by implicit learning, as the
reverse interface hypothesis would maintain. Instead, Andringa (2020) found that explicit knowledge
preceded linguistic prediction and may have been a precondition for learners to make predictions.
Hence, in learning experiments such as Andringa (2020), where participants are constructing new
knowledge, prediction may coincide with awareness associated with explicit knowledge (also see
Curcic et al., 2019, for similar results).
Kim (2020) built on the measurement research about implicit, automatized explicit, and
explicit L2 knowledge (see “Disentangling automatized explicit and implicit knowledge”) to
assess the interface hypothesis and the reverse interface hypothesis outside the lab (also see Suzuki
& DeKeyser, 2017). In a semester-long study, Kim tracked the development of 122 international
students’ implicit and explicit knowledge of L2 English. The students completed a battery of
tests of implicit knowledge (i.e., timed written GJT, oral production, and elicited imitation) and
explicit knowledge (i.e., untimed written GJT and metalinguistic knowledge) at the beginning
(Time 1) and the end (Time 2) of an academic semester. Test results were then submitted to a path
analysis that examined all possible relationships (i.e., between Explicit Time 1, Implicit Time 1,
Explicit Time 2, and Implicit Time 2) simultaneously. Accounting for all knowledge relationships
simultaneously, Kim found that students who had more explicit knowledge at Time 1 also had
more implicit knowledge at Time 2. It was also true that those who had more implicit knowl-
edge at Time 1 had more explicit knowledge at Time 2. These results map onto the interface and
302
the reverse interface, respectively, between explicit and implicit knowledge.2 Kim’s longitudinal
study therefore provided initial evidence to suggest that implicit and explicit knowledge may be
in a reciprocal relationship and infuence each other bi-directionally over time, although further
evidence will be needed to corroborate these claims.

As demonstrated in this chapter, research on the interface hypothesis has considerably broad-
ened over the past forty years. Traditionally framed in dichotomous terms regarding whether or
not there is an interface, contemporary approaches to the interface hypothesis address a range
of interconnected topics and questions regarding implicit/explicit instruction, online process-
ing, and implicit/explicit knowledge, and how these components may fit together in one big
picture (see Figure 24.2 and Textbox 24.2).
Work on implicit/explicit knowledge and implicit/explicit learning rests on a foundation of valid
and reliable measurement (see “Disentangling automatized explicit and implicit knowledge”). Eforts
at developing and validating new tests should extend beyond the current focus on grammar and
include implicit and explicit knowledge of vocabulary (Godfroid, 2020b), pronunciation (Saito &
Plonsky, 2019), and pragmatics as domains of inquiry. Across linguistic domains, there is a need for
more research contrasting ofine time-pressured tests (e.g., elicited imitation) and online reaction
time measures (e.g., self-paced reading) as hypothetical measures of implicit knowledge, using better
statistical techniques for analyzing reaction time data (Draheim et al., 2019; Rouder & Haaf, 2019).
Another area that needs more attention is linking research in the lab on semi-artifcial or natural lan-
guages with classroom-based studies (see Ettlinger et al., 2016, for an example). More researchers need to
take to the classroom in a deliberate attempt to bridge the divide between psychology and psycholinguis-
tics, on the one hand, and (instructed) SLA, on the other (see “Simultaneous acquisition of explicit and
implicit knowledge from instruction”). Research on the interface and the reverse interface hypothesis,
however, should not be limited to instructed (and often highly educated) L2 learners. Other L2 learners
and bilinguals, including non-college-educated adults, circumstantial bilinguals, and young learners, can
provide valuable perspectives on how language develops in informal contexts, with less explicit guidance,
with less metalinguistic awareness, and/or with more and richer input. Including these diferent popula-
tions will be especially important to study the potential progression from implicit to explicit knowledge,
as postulated by the reverse interface hypothesis (see “Reverse interface hypothesis”).
Lastly, the interface needs to be examined in its full complexity, preferably with longitudinal
research designs, so researchers can further explore how explicit knowledge relates to implicit knowl-
edge and vice versa over time, including real-time processing and instruction as key factors (see
Figure 24.2). To achieve this ambitious goal, more and better measures are needed, and a focus on
online processing measures such as eye-tracking and self-paced reading will remain key (see “Efects
of explicit instruction on online processing”). These measures also ofer the promise of tracking the
emergence of explicit and, especially, implicit knowledge in real time (see “Disentangling automa-
tized explicit and implicit knowledge” and “Reverse interface hypothesis”).
By studying explicit-implicit infuences in both directions, researchers will be able to shed more light
on the distinction between implicit and automatized explicit knowledge (see “Disentangling automa-
tized explicit and implicit knowledge”). Brain imaging studies (e.g., Suzuki et al., 2020) could provide
additional insight into this long-standing theoretical question and how it relates to diferent models of L2
acquisition such as Cleeremans’s radical plasticity thesis (Cleeremans, 2008, 2011) (see “Reverse interface
hypothesis”). Do implicit and automatized explicit knowledge represent the beginning and endpoint,
respectively, of how naturalistic language acquisition occurs, in line with the reverse interface hypothesis?
Or are the two types of knowledge behaviorally equivalent, as is the dominant view in instructed SLA?
Those are some of the larger questions that future interface researchers will need to address.
303
Aline Godfroid
What are valid and reliable measures of implicit, automatized explicit, and explicit L2 knowledge that
researchers can use in the classroom, the research lab, and online?
To what extent do the results of natural-language, classroom-based research converge with those of lab-
based language research?
Is it possible to track the emergence of implicit knowledge and explicit knowledge in real time, namely
during online processing?
Notes
1 Here, I refer to the interface hypothesis that developed out of Krashen’s writings about the distinction between
acquisition and learning (e.g., Krashen, 1981), as opposed to Sorace’s Interface Hypothesis, which addresses the
difculty of processing and acquiring linguistic structures in an L2 that are at the interface between syntax and
diferent cognitive domains (Sorace, 2011).
2 To my knowledge, Kim’s study was the frst longitudinal investigation of the interface hypothesis. The expe-
rience of repeated testing at the beginning and end of the semester may have increased the contribution of
explicit processes to the supposedly implicit knowledge tests at Time 2, rendering all the tests more similar.
This needs to be considered when interpreting the results, although statistically, the tests of explicit and
implicit knowledge remained distinct, also at Time 2.
Further reading
Andringa, S., & Rebuschat, P. (Eds.). (2015). New directions in the study of implicit and explicit learning [Spe-
cial issue]. Studies in Second Language Acquisition, 37(2).
This thematic issue marked the ten-year anniversary of Hulstijn and Ellis’s (2005) seminal publication on
implicit and explicit L2 learning. It features studies that showcase developments in research methodologies,
such as brain-based methods, online (real-time) methodologies, and the measurement of awareness.
Li, S., & DeKeyser, R. (Eds.). (2021). Implicit language aptitude: Conceptualizing the construct, validating the
measures, and examining the evidence [Special Issue]. Studies in Second Language Acquisition, 43(3).
This special issue brings together recent work on implicit language aptitude in SLA, including its role in
explicit and implicit instruction and the attainment of implicit knowledge.
Rebuschat, P. (Ed.). (2015). Implicit and explicit learning of languages. John Benjamins.
This volume represents an update to a 1994 edited volume by Nick Ellis carrying the same title. Both
volumes bring together psychologists and SLA researchers around the shared topic of implicit and explicit
learning.
References
Andringa, S. (2020). The emergence of awareness in uninstructed L2 learning: A visual world eye tracking study.
Andringa, S., & Curcic, M. (2015). How explicit knowledge afects online L2 processing: Evidence from dif-
ferential object marking acquisition. Studies in Second Language Acquisition, 37(2), 237–268.
Andringa, S., de Glopper, K., & Hacquebord, H. (2011). Efects of explicit and implicit instruction on free writ-
ten response task performance. Language Learning, 61(3), 868–903.
Bell, P. K. (2017). Explicit and implicit learning: Exploring their simultaneity and immediate efectiveness.
Applied Linguistics, 38(3), 297–317.
Bialystok, E. (2001). Bilingualism in development: Language, literacy, and cognition. Cambridge University Press.
Bowles, M. A. (2011). Measuring implicit and explicit linguistic knowledge: What can heritage language learn-
ers contribute? Studies in Second Language Acquisition, 33(2), 247–271.
304
Cintrón-Valentín, M. C., & Ellis, N. C. (2015). Exploring the interface: Explicit focus-on-form instruction and
learned attentional biases in L2 Latin. Studies in Second Language Acquisition, 37(2), 197–235.
Cintrón-Valentín, M. C., & Ellis, N. C. (2016). Salience in second language acquisition: Physical form, learner
attention, and instructional focus. Frontiers in Psychology, 7, 1284.
Cleeremans, A. (2008). Consciousness: The radical plasticity thesis. In R. Banerjee & B. K. Chakrabarti (Eds.),
Models of brain and mind: Physical, computational and psychological approaches. Progress in brain research (Vol. 168,
pp. 19–33). Elsevier.
Cleeremans, A. (2011). The radical plasticity thesis: How the brain learns to be conscious. Frontiers in Psychol-
ogy, 2, 86.
Curcic, M., Andringa, S., & Kuiken, F. (2019). The role of awareness and cognitive aptitudes in L2 predictive
language processing. Language Learning, 69, 42–71.
De Jong, N. (2005). Can second language grammar be learned through listening?: An experimental study. Studies
DeKeyser, R. (1998). Beyond focus on form. In C. Doughty & J. Williams (Eds.), Focus on form in classroom second
language acquisition (pp. 42–63). Cambridge University Press.
DeKeyser, R. (2003). Implicit and explicit learning. In C. Doughty & M. Long (Eds.), The handbook of second
language acquisition (pp. 313–348). Blackwell.
DeKeyser, R. (2017). Knowledge and skill in ISLA. In S. Loewen & M. Sato (Eds.), The Routledge handbook of
instructed second language acquisition (pp. 15–32). Routledge.
DeKeyser, R. (2020). Skill acquisition theory. In B. VanPatten, G. D. Keating, & S. Wulf. (Eds.), Theories in
second language acquisition: An introduction (3rd ed., pp. 83–104). Routledge.
Dienes, Z. (2007). Subjective measures of unconscious knowledge. Progress in Brain Research, 168, 49–269.
Draheim, C., Mashburn, C. A., Martin, J. D., & Engle, R. W. (2019). Reaction time in diferential and developmen-
tal research: A review and commentary on the problems and alternatives. Psychological Bulletin, 145(5), 508–535.
Ellis, N. C. (2005). At the interface: Dynamic interactions of explicit and implicit language knowledge. Studies
Ellis, N. C. (2015). Implicit and explicit language learning: Their dynamic interface and complexity. In P. Re-
buschat (Ed.), Implicit and explicit learning of languages (pp. 1–24). John Benjamins.
Ellis, R. (1994). The study of second language acquisition. Oxford University Press.
Ellis, R. (2004). The defnition and measurement of explicit knowledge. Language Learning, 54(2), 227–275.
Ellis, R. (2005). Measuring implicit and explicit knowledge of a second language: A psychometric study. Studies
Ellis, R. (2009). Implicit and explicit learning, knowledge and instruction. In R. Ellis, S. Loewen, C. Elder,
R. Erlam, J. Philp, & H. Reinders (Eds.), Implicit and explicit knowledge in second language learning, testing and
teaching (pp. 3–25). Multilingual Matters.
Ellis, R., & Loewen, S. (2007). Confrming the operational defnitions of explicit and implicit knowledge in
Ellis (2005): Responding to Isemonger. Studies in Second Language Acquisition, 29(1), 119–126.
Ettlinger, M., Morgan‐Short, K., Faretta‐Stutenberg, M., & Wong, P. C. (2016). The relationship between
artifcial and second language learning. Cognitive Science, 40(4), 822–847.
Godfroid, A. (2020a). Eye tracking in second language acquisition and bilingualism: A research synthesis and methodologi-
cal guide. Routledge.
Godfroid, A. (2020b). Sensitive measures of vocabulary knowledge and processing: Expanding Nation’s frame-
work. In S. Webb (Ed.), The Routledge handbook of vocabulary studies (pp. 433–453). Routledge.
Godfroid, A., & Kim, K. M. (2021). The contributions of implicit-statistical learning aptitude to implicit second
language knowledge. Studies in Second Language Acquisition, 43.
Godfroid, A., Kim, K. M., Hui, B., & Isbell, D. (in prep.). Synthesizing research on implicit, automatized
explicit, and explicit L2 knowledge: An empirical extension.
Hamrick, P., & Sachs, R. (2018). Establishing evidence of learning in experiments employing artifcial linguistic
systems. Studies in Second Language Acquisition, 40(1), 153–169.
Hulstijn, J. H. (2002). Towards a unifed account of the representation, processing and acquisition of second
language knowledge. Second Language Research, 18(3), 193–223.
Hulstijn, J. H. (2015). Explaining phenomena of frst and second language acquisition with the constructs of
implicit and explicit learning. The virtues and pitfalls of a two-system view. In P. Rebuschat (Ed.), Implicit
and explicit learning of languages (pp. 385–416). John Benjamins.
Hulstijn, J. H., & Ellis, R. (Eds.). (2005). Implicit and explicit second-language learning [Special issue]. Studies
in Second Language Acquisition, 27(2).
305
Aline Godfroid
Issa, B. I., & Morgan-Short, K. (2019). Efects of external and internal attentional manipulations on second lan-
guage grammar development: An eye-tracking study. Studies in Second Language Acquisition, 41(2), 389–417.
Karmilof-Smith, A. (1992). Beyond modularity: A developmental perspective on cognitive science. MIT Press: A Brad-
ford Book.
Kim, J., & Nam, H. (2017). Measures of implicit knowledge revisited: Processing modes, time pressure, and
modality. Studies in Second Language Acquisition, 39(3), 431–457.
Kim, K. M. (2020). Exploring the interface of explicit and implicit second-language knowledge: A longitudinal perspective
[Unpublished doctoral dissertation], Michigan State University.
Kim, K. M., & Fenn, K. M. (2020). Sleep-dependent consolidation of second language grammar knowledge:
The role of awareness. Studies in Second Language Acquisition, 42(5), 1107–1120.
Kim, K. M., & Godfroid, A. (2019). Should we listen or read? Modality efects in implicit and explicit knowl-
edge. The Modern Language Journal, 103(3), 648–664.
Krashen, S. D. (1981). Second language acquisition and second language learning. University of Southern California.
Krashen, S. D. (1982). Principles and practice in second language acquisition. Pergamon.
Krashen, S. D. (1985). The input hypothesis: Issues and implications. Longman.
Lee, J. F., & Doherty, S. (2019). Native and nonnative processing of active and passive sentences: The efects
of processing instruction on the allocation of visual attention. Studies in Second Language Acquisition, 41(4),
853–879.
Maie, R., & DeKeyser, R. M. (2020). Conficting evidence of explicit and implicit knowledge from objective
and subjective measures. Studies in Second Language Acquisition, 42(2), 359–382.
Marsden, E., Thompson, S., & Plonsky, L. (2018). A methodological synthesis of self-paced reading in second
language research. Applied Psycholinguistics, 39(5), 861–904.
McManus, K., & Marsden, E. (2017). L1 explicit instruction can improve L2 online and ofine performance.
McManus, K., & Marsden, E. (2018). Online and ofine efects of L1 practice in L2 grammar learning: A partial
replication. Studies in Second Language Acquisition, 40(2), 459–475.
Miller, Z. F., & Godfroid, A. (2020). Emotions in incidental language learning: An individual diferences
approach. Studies in Second Language Acquisition, 42(1), 115–141.
Paradis, M. (2009). Declarative and procedural determinants of second languages. John Benjamins.
Rebuschat, P., Hamrick, P., Sachs, R., Riestenberg, K., & Ziegler, N. (2015). Triangulating measures of aware-
ness: A contribution to the debate on learning without awareness. Studies in Second Language Acquisition, 37,
299–334.
Rebuschat, P., & Williams, J. N. (2012). Implicit and explicit knowledge in second language acquisition. Applied
Psycholinguistics, 33(4), 829–856.
Rogers, J., Révész, A., & Rebuschat, P. (2016). Implicit and explicit knowledge of infectional morphology.
Rouder, J. N., & Haaf, J. M. (2019). A psychometrics of individual diferences in experimental tasks. Psychonomic
Bulletin & Review, 26, 452–467.
Sachs, R., Hamrick, P., McCormick, T., & Leow, R. (2020). Exploring the veridicality and reactivity of subjec-
tive measures of awareness: Is a “guess” really a guess? Studies in Second Language Acquisition, 42(4), 919–932.
Saito, K., & Plonsky, L. (2019). Efects of second language pronunciation teaching revisited: A proposed mea-
surement framework and meta-analysis. Language Learning, 69(3), 652–708.
1(1), 1–33.
Suzuki, Y. (2023). Automatization and practice. In A. Godfroid & H. Hopp (Eds.), The Routledge handbook of
second language acquisition and psycholinguistics (pp. 308–321). Routledge.
Suzuki, Y., & DeKeyser, R. (2015). Comparing elicited imitation and word monitoring as measures of implicit
knowledge. Language Learning, 65(4), 860–895.
Suzuki, Y., & DeKeyser, R. (2017). The interface of explicit and implicit knowledge in a second language:
Insights from individual diferences in cognitive aptitudes. Language Learning, 67(4), 747–790.
Tagarelli, K. M., Ruiz, S., Vega, J. L. M., & Rebuschat, P. (2016). Variability in second language learning: The
roles of individual diferences, learning conditions, and linguistic complexity. Studies in Second Language
Acquisition, 38(2), 293–316.
Vafaee, P., Suzuki, Y., & Kachisnke, I. (2017). Validating grammaticality judgment tests: Evidence from two
psycholinguistic measures. Studies in Second Language Acquisition, 39(1), 59–95.
VanPatten, B. (1996). Input processing and grammar instruction in second language acquisition. Greenwood Publishing
Group.
306
VanPatten, B. (2004). Input processing in second language acquisition. In B. VanPatten (Ed.), Processing instruc-
tion: Theory, research, and commentary (pp. 5–32). Lawrence Erlbaum Associates.
VanPatten, B. (2005). Processing instruction. In C. Sanz (Ed.), Mind and context in adult second language acquisition
(pp. 267–281). Georgetown University Press.
Wagner, U., Gais, S., Haider, H., Verleger, R., & Born, J. (2004). Sleep inspires insight. Nature, 427(6972),
352–355.
Wong, W., & Ito, K. (2018). The efects of processing instruction and traditional instruction on L2 online pro-
cessing of the causative construction in French: An eye-tracking study. Studies in Second Language Acquisition,
40(2), 241–268.
Zhang, R. (2015). Measuring university-level L2 learners’ implicit and explicit linguistic knowledge. Studies in
Second Language Acquisition, 37(3), 457–486.
307
25
AUTOMATIZATION
AND PRACTICE
Yuichi Suzuki
Introduction
The concept of automaticity permeates everyday life. We can perform daily routines—such as using
a smartphone, typing, walking, and driving a car.—very smoothly and efciently without being
consciously aware of the mental processes these activities entail. Using our frst language is one of
the most automatized skills, since it has been practiced extensively since birth. Yet, automaticity
may seem like a far-reaching goal for many second language (L2) learners. In the beginning, L2
learners’ speaking performance is far from automatic, since it tends to be inaccurate, slow, halting,
and efortful. Because the ultimate goal of L2 learning is to attain automaticity—characterized as
fast, efortless, efcient, and stable language use—understanding automatization, i.e., the gradual
and extended learning process leading up to automaticity, is crucial (see Textbox 25.1). What is
agreed on among most L2 researchers, irrespective of their theoretical standpoints, is the essential
role of practice quantity for automatizing L2 knowledge and skills. What merits careful deliberation
and conceptualization is practice quality.
Traditionally, the term “practice” has been associated with mechanical drills and exercises, most
notably in audiolingualism in L2 teaching. Recently, the concept of practice was reconceptualized
by DeKeyser (2007) with the aim of integrating applied linguistics and cognitive/educational psy-
chology perspectives. Following this conceptualization, in this chapter, practice is defned as activi-
ties involving the use of an L2 repeatedly, systematically, and deliberately for developing knowledge
and skills. This broader conceptualization of practice can capture a wide range of L2 practice
activities on a continuum from form-focused drills and exercises to meaning-focused tasks, and
importantly, many activities that fall in between (e.g., oral reading, recitation, memorization, dialog
reading, skits, picture descriptions, reproduction/retelling). The importance of meaningful and
cognitively and emotionally engaging “repetition” is emphasized in the idea of practice aimed at
automatization, without unnecessarily reverting to the mindless, repetitive mechanical practice in
audiolingualism.
This new concept of practice has advanced L2 research pertaining to cognitive aspects of L2
learning, including input, interaction, and output practice, along with corrective feedback (DeKey-
ser, 2007). Suzuki et al. (2019b) have recently laid out a new research agenda of practice to fur-
ther stimulate the synergy with cognitive psychology. Just as cognitive psychologists and educational
researchers have provided a number of valuable empirical fndings over the decades that can serve
as useful guidelines for optimizing classroom instruction (e.g., Hattie & Yates, 2013), research on
308 DOI: 10.4324/9781003018872-29

Automatization and practice
practice from cognitive psychology perspectives can inform L2 classroom teaching and learning
(Lightbown, 2019). Practice lies at the intersection of many important L2 theoretical and practical
issues—automatization, automaticity, and declarative and procedural knowledge, as well as explicit
and implicit learning and knowledge.
Automatization: Development of knowledge and skills that aford superior performance both quantita-
tively (e.g., speed-up of underlying processes) and/or qualitatively (e.g., restructuring of processes).
Automaticity: An advanced stage in acquisition at which knowledge and skills can be deployed quickly,
efciently, efortlessly, stably, ballistically, and unconsciously.
Declarative knowledge: Knowledge of facts, events, lexical items, pedagogical grammatical rules, etc.
Procedural knowledge: Knowledge of how to (e.g., how to perform motor actions, calculation, compre-
hension, and production of language).
Practice: Activities that typically involve deliberate and/or systematic repetition aimed at developing L2
knowledge and skills.
Knowledge/skill transfer: Applying knowledge and skills that are acquired in one context to another.

Automaticity is a multi-faceted construct with several distinct features. It allows for elucidating the
nature of L2 learning outcomes on a continuum from controlled to automatic processing (Moors,
2016; Moors & De Houwer, 2006). The list is not exhaustive, but automaticity is characterized as
fast, efcient, efortless, stable, ballistic (unintentional), and even unconscious (implicit) processing
(DeKeyser, 2001; Segalowitz, 2003). These features can covary but should be considered separately,
because the presence of one feature (e.g., fast processing) does not necessarily guarantee the presence
of others (e.g., unconscious processing). The diferent roles of these features pertaining to automatic-
ity are rarely discussed in L2 research. Awareness (consciousness) has received focal attention in L2
research and has led to a debate regarding its role in the L2 process and the nature of explicit and
implicit L2 knowledge. For instance, researchers investigated the extent to which conscious (explicit)
knowledge can be used quickly or automatically (e.g., DeKeyser, 1997; see also Tzelgov, 1997). Per-
haps more importantly from theoretical standpoints, it remains to be established whether conscious
knowledge that is used quickly (automatized explicit knowledge) can be equated with the fast use of
knowledge without awareness (implicit knowledge) (e.g., Godfroid & Kim, 2021; Suzuki & DeKey-
ser, 2017b; see Isbell & Rogers, 2021 for a recent review).
The nature of L2 grammatical knowledge—explicit and implicit—has been extensively studied, par-
ticularly in relation to the central topic of instructed second language acquisition (SLA)—the role of
explicit instruction for the acquisition1 of grammatical knowledge that can be used for communication
(VanPatten & Williams, 2015). Yet, acquisition of implicit grammatical knowledge (without correlates
with consciousness) may be of much less concern for L2 teachers, because they are interested in how
knowledge that is used accurately and quickly in communication can be attained through L2 instruc-
tion and practice, irrespective of its consciousness (e.g., Spada, 2015; Suzuki & DeKeyser, 2017b). This
critical view is often overlooked but crucial. The role of explicit instruction is typically explored using
outcome measurements involving constrained-response ofine tasks that target non-automatized knowl-
edge (e.g., Goo et al., 2015; Norris & Ortega, 2000), which cannot uncover whether and how explicit
instruction infuences the development of automatized linguistic knowledge used for communication.
309
Yuichi Suzuki
The construct of automaticity provides a wider variety of criteria—speed, efciency, stability, and even
unconsciousness—for the assessment of L2 knowledge and skills. Automatization also ofers a useful lens
to study L2 learning processes that lead to automaticity, which can be explained in detail based on the
skill acquisition theory, originally derived from Anderson’s adaptive control of thought-rational (ACT-R)
theory in psychology (Anderson et al., 2004),
Skill acquisition theory (DeKeyser, 2015, 2017) stipulates three stages of development, namely
declarative, procedural, and automatization (also see Morgan-Short & Ullman, 2023 [this volume]).
Declarative knowledge about L2 (e.g., words, collocations, rules) is typically accessible to awareness
and is frst acquired by many L2 classroom learners through explicit instruction. Use of declarative
knowledge leads to proceduralization—generating target behaviors—resulting in procedural knowl-
edge that can be used for L2 comprehension and production. As procedural knowledge is further
fne-tuned for more efcient processing, this long and gradual process culminates in automatization.
This learning trajectory is often described as the power law of practice (see DeKeyser, 1997 for
empirical data on L2 learning). As illustrated in Figure 25.1, there is an initial rapid improvement
in the target skill through practice (proceduralization), and what follows is a very gradual but steady
improvement over a number of practice sessions (automatization) (DeKeyser, 2015). This power
law curve applies not only to simple tasks (e.g., rolling cigars, typing) but also complex skills like
L2 learning because each subcomponent required for mastering the complex skill obeys the power
law (Anderson, 2000).
While skill acquisition theory has been primarily associated with practice, the usage-based
approach (N. C. Ellis & Wulf, 2015; Tyler & Ortega, 2018) can also ofer complementary insights
into the nature of L2 practice. Akin to the skill acquisition perspective, the usage-based approach is
based on the premise that frequency is the main driver of L2 acquisition, which “takes tens of thou-
sands of hours of practice, practice that cannot be substituted for by provision of a few declarative
rules” (N. C. Ellis, 2002, p. 175). The usage-based approach emphasizes implicit-statistical learning
through L2 use for communication in social contexts. Furthermore, the unit of linguistic knowledge
is described as constructions with diferent levels of abstraction, consisting of single sounds, lexical
Figure 25.1 Power law curve: proceduralization and automatization.
310
items, collocations, and abstract rules. Accordingly, automatization occurs at the diferent levels of
construction, as evidenced by an experiment showing that abstract constructions (e.g., resultative
construction [Agent−Action−Theme−Result] denoting “X causes Y to Z”) can be processed more
automatically by more profcient L2 learners (H. Kim & Rah, 2019).
In sum, L2 practice contributes to L2 knowledge and skill development that goes through the
declarative−procedural−automatization stages. According to the skill acquisition theory, this process
relies primarily on explicit learning (DeKeyser, 2015). Usage-based approach complements this view
by emphasizing the implicit/statistical fne-tuning of L2 knowledge from concrete to more abstract
constructions (Tyler & Ortega, 2018). Critically, both theories acknowledge the complementary
mechanisms of explicit and implicit learning for achieving automaticity that is the eventual, target
outcome of L2 practice.

The chief objective of L2 practice research is to establish what kinds of practice are efective for
which learners, for what knowledge and skills, in what contexts, and when. Just like other profes-
sionals such as music teachers and sports coaches (Carlson, 2003), L2 researchers and instructors
have a vested interest in understanding how practice leads to automatization. An overview of key
topics and research questions pertaining to L2 practice and automatization is presented in Table 25.1.
Clearly, vocabulary or grammar acquisition has been extensively studied, while limited research has
been conducted on automatization of pronunciation (M. Li & DeKeyser, 2017, 2019) and pragmat-
ics (S. Li & Taguchi, 2014). Five of the key topics listed in Table 25.1 are discussed in detail in the
next section.
Table 25.1 Key topics and research questions in instructed L2 research on practice and automatization.
Topics Research questions
Incidental and Does incidental and intentional L2 learning lead to automatization? (e.g., Elgort,
intentional learning 2011; Elgort & Warren, 2014; Hui, 2020)
Role of declarative− Does declarative−explicit knowledge facilitate automatization in L2 practice? If
explicit knowledge so, what type of explicit metalinguistic information is benefcial? (e.g., McManus
& Marsden, 2019; Sato & McDonough, 2019)
Practice schedule What is the optimal practice interval for automatization? (e.g., M. Li &
DeKeyser, 2019; Nakata & Elgort, 2020; Suzuki, 2017; Suzuki & DeKeyser,
2017a)
Practice sequence What is the optimal practice sequence for automatization? (e.g., Suzuki &
Sunada, 2019)
Transfer of learning What kind of practice can maximize the transfer of a practiced skill to other skills
used in diferent contexts? (e.g., De Jong & Perfetti, 2011; M. Li & DeKeyser,
2017; Suzuki, 2021)
Types of corrective What type of oral corrective feedback is best for promoting automatization? (e.g.,
feedback Sato & Lyster, 2012)
Timing of corrective Is immediate corrective feedback more efective than delayed corrective feedback
feedback for the acquisition of automatized knowledge? (e.g., S. Li et al., 2016)
Individual diferences To what extent is the rate of automatization explained by individuals’ cognitive
in automatization abilities? (e.g., Pili-Moss et al., 2020; Suzuki, 2018)
311
Yuichi Suzuki
Vocabulary learning for automatization: speed and stability of L2 processing

Automaticity measurements are essential for a better understanding of the automatization process.
A variety of psycholinguistic research methods utilizing reaction time (RT) have already been
applied to elucidate L2 processing and automatization (see Hamrick, 2023 [this volume]; Jiang,
2011, for an overview). One of those is a representative vocabulary task—a lexical-decision task
in which participants judge, as quickly as possible, whether a single word belongs to the target L2
(see Elgort & Warren, 2023 [this volume]). Automatization cannot be simply equated to speeded-
up processing (indicated by shorter RT), because it also involves higher processing stability, which
is in this case assessed by coefcient of variance (CV), calculated as the ratio of an individual’s
standard deviation in RT and mean RT. Some researchers argue that CV can capture restruc-
turing or qualitative changes in underlying mental processes (e.g., elimination of an inefcient
sub-process) rather than quantitative changes, i.e., sheer process acceleration (e.g., Segalowitz &
Segalowitz, 1993; see also Saling & Phillips, 2007 for neurocognitive underpinnings). Nonethe-
less, both RT and CV are useful for examining automatization in multiple ways in L2 vocabulary
learning (Godfroid, 2019).
The central topic of L2 vocabulary research pertaining to automatization is intentional
and incidental learning. In one such investigation, Elgort (2011) asked participants to study
48 novel words intentionally (deliberately) using word cards for one week, after which they
were given a primed lexical decision task. The results showed that the studied words led not
only to faster access (shorter RT) but also to more stable lexical processing (smaller CV). The
lexical-developmental trajectory was further examined in deliberate paired-associate L2 word
learning in a laboratory setting by Solovyeva and DeKeyser (2018). Interestingly, their analy-
ses suggested that the CV values followed an inverted U-shaped curve (as opposed to a linear
decrease) over the course of training. In the initial learning stage, wherein new lexical items
are integrated into the lexicon, lexical processing becomes less stable (marked by an increase in
CV), whereas more extensive practice leads to the automatization of existing knowledge (and
thus CV decreases).
While intentional, deliberate vocabulary practice is consistently shown to facilitate automatiza-
tion, research fndings on incidental vocabulary learning remain inconclusive. For example, partici-
pants in Elgort and Warren’s (2014) study, who engaged in incidental vocabulary learning by reading
a non-fction book, did not show any priming efects on the lexical decision task (post-test) despite
encountering new words about 12 times on average (median) in the reading, suggesting an absence
of automatization. Similarly, the inverted U-shaped curve of CV values, originally suggested by
Solovyeva and DeKeyser (2018), was found only in the intentional condition and not in the inci-
dental learning condition in Hui (2020). The inconsistent results on incidental learning for automa-
tization have also been obtained in studies on L2 collocation (Sonbul & Schmitt, 2013; Toomer &
Elgort, 2019).
Perhaps, incidental learning may need to be supplemented by additional vocabulary instruction
to develop more robust lexical knowledge. Elgort et al. (2020) investigated the efects of teach-
ing vocabulary prior to contextual word learning through reading short passages. Their fndings
indicated that, although the vocabulary pre-teaching was efective in reducing meaning inference
errors during reading, post-teaching of word meaning after the reading resulted in better recall and
faster lexical processing than pre-teaching. This somewhat surprising fnding may suggest that pre-
teaching might lower the depth of learning by removing the need for contextual inferences for the
novel words while reading (cf., Pellicer-Sánchez, et al., 2021). Thus, the learning processes during
incidental learning should be explored further in relation to automatization.
312
The role of declarative knowledge in grammar practice for automatization

The key assumption underpinning the skill acquisition theory is that declarative knowledge is used
in proceduralization and further automatization. In laboratory intervention research, McManus and
Marsden (2019) trained French L2 learners on imparfait verbal morphology through four training
sessions delivered across a three-week period. The aim was to establish whether additional provision
of L1-explicit information about the target structure, in conjunction with L2-explicit information,
facilitated automatization in input-based grammar practice. Analyses of CV derived from RT during
input practice showed that L1 explicit information aided automatization in the later, third and fourth
training sessions.
On the other hand, Sato and McDonough (2019) conducted a classroom-based study to investi-
gate the role of declarative knowledge in L2 grammar practice for acquiring procedural knowledge
of wh- questions. Over fve weekly sessions, EFL learners engaged in a variety of communicative
grammar practice activities (e.g., spot-the-diference task) that obligatorily elicit the use of wh- ques-
tions. The analyses of speech during the grammar practice showed that prior declarative knowledge,
measured by an ofine paper-and-pencil test involving the target structure, allowed the learners to
use the target structure accurately in the initial stages of proceduralization. Furthermore, the use of
declarative knowledge was also implicated initially for speeding up the use of wh- questions in com-
municative speaking tasks (evidenced by higher speech rate and shorter pauses). The results yielded
by these two studies suggest that deliberate, repeated L2 practice—using declarative knowledge as a
crutch—leads to automatization, as evidenced by both global (speech data) and fne-grained (CV)
measures. Yet, the optimal integration of declarative knowledge in repeated grammar practice needs
to be further explored for a wider variety of linguistic features.
Distributed practice for automatization

A large body of cognitive psychology research has been conducted over the last century to identify an
optimal schedule for repeated practice (e.g., Cepeda et al., 2006). This age-old question in psychol-
ogy has recently motivated L2 researchers to investigate the extent to which L2 acquisition can be
enhanced by systematically manipulating practice schedules. In this context, massed, short-spaced,
and/or long-spaced practice are compared at diferent time scales (e.g., less than a day, over days,
weeks; see Serrano, 2011 for a comparison at the curriculum level):
• Massed practice: AAAA

• Short-spaced practice: A__A__A__A
• Long-spaced practice: A________A________A________A
One notable fnding from the skill acquisition perspective that has emerged from this research stream
is that longer spacing seems to facilitate retention of declarative knowledge (e.g., seven-day interval),
while shorter spacing (e.g., one-day interval) tends to be equally or more facilitative for the acqui-
sition of procedural/automatized knowledge (Kasprowicz et al., 2019; M. Li & DeKeyser, 2019;
Nakata & Elgort, 2020; Suzuki, 2017; Suzuki & DeKeyser, 2017a). Because proceduralization of L2
skills takes longer than the acquisition of declarative knowledge, repeated practice at short intervals
or even massed practice may assist in reaching a certain level of proceduralization/automatization
that is resistant to skill decay (J. W. Kim et al., 2013). For instance, Suzuki (2017) trained partici-
pants on morphological structures in a miniature language over four practice sessions under either
shorter-spaced (twice a week) or longer-spaced (once a week) conditions. The participants took an
oral, procedural grammar test on the target structures both before and after each practice session to
document the knowledge losses and gains, respectively. As shown in Figure 25.2, the shorter-spaced
313
Yuichi Suzuki
Figure 25.2 Performance changes (losses and gains) over four practice sessions under shorter-spaced and longer-
spaced conditions.
Note: The error bars indicate 95% confdence intervals.
practice led to higher retention of procedural knowledge gained in the previous session (i.e., before
sessions 3 and 4) than the longer-spaced practice. This outcome suggests that shorter-spaced practice
allowed learners to engage in repeated practice more efciently, before experiencing a more pro-
nounced deterioration of existing proceduralized knowledge, to further fne-tune their grammatical
knowledge. Given the incipient phase of this research domain, more research from multiple theoreti-
cal perspectives is clearly needed for gaining a more comprehensive understanding of this important
issue in repeated practice.
Blocked and interleaved practice for automatization

Another issue related to the practice schedule is whether to use blocked or interleaved schedules. In
blocked practice, the exemplars are blocked by category, whereas learners are exposed to multiple
exemplars from diferent categories at once in interleaved practice, as illustrated here:
• Blocked practice: AAAA−BBBB−CCCC−DDDD

• Interleaved practice: BACD−DABC−ABDC−CABD
In L2 grammar acquisition, when multiple structures are practiced at once, interleaving seems to
be superior to blocking for the acquisition of both declarative knowledge (Nakata & Suzuki, 2019;
Pan et al., 2019) and procedural knowledge (Suzuki & Sunada, 2019; Suzuki et al., 2020). Inter-
leaved practice can facilitate the discrimination of similar features that are present, for example, in
the English tense systems and relative clause constructions as well as Spanish morphological rules.
Furthermore, interleaved practice can potentially lessen the burden on working memory for the pro-
ceduralization of English relative clauses among EFL learners, because presenting exemplars of similar
314
grammatical structures immediately enhances the comparison of those recently presented exemplars
(Suzuki et al., 2020). Nonetheless, since the benefts of interleaved and blocked practice seem to vary
depending on the nature of target materials and skills (Brunmair & Richter, 2019), more research is
needed to investigate a variety of linguistic structures not limited to grammar, but also other linguistic
domains such as pronunciation and pragmatics (see Carpenter & Mueller, 2013, for the acquisition
of highly dissimilar lexico-phonological rules). Extending the bounds of these linguistic domains, it
would also be interesting to elucidate the efects of blocked and interleaved practice on the acquisi-
tion of constructions (e.g., “put A in,” “put B on,” “put D back”) from a usage-based perspective
(e.g., N. Ellis & Collins, 2009).
Transfer of learning
Transfer of learning, i.e., acquiring knowledge and skills in one context and applying them to a
new context, is the “holy grail” of education (e.g., Haskell, 2001; Singley & Anderson, 1989). An
important transfer question in L2 teaching is the extent to which grammar practice in a controlled
context (isolated from a meaningful context) can facilitate the acquisition of L2 knowledge that can
be used for communication (Tuz, 1993). To address this issue, Spada et al. (2014) compared two
versions of form-focused instruction (FFI) approaches in ESL classes: isolated FFI, whereby target
grammar was taught independently from communicative/content-based activities, versus integrated
FFI, whereby the instruction on target grammar was always embedded in communicative activities.
Their results showed that, while isolated FFI was more efective for the decontextualized written,
declarative grammar test, integrated FFI was more benefcial for automatized knowledge (measured by
an elicited imitation test) that is presumably more useful for communication. However, this fnding
was not statistically robust and it needs to be further corroborated given its high potential theoretical
and pedagogical relevance. Specifcally, this pattern fts the prediction of the transfer-appropriate pro-
cessing model, i.e., the transfer of learning increases when a practice condition has more similarities
to the outcome tasks (Lightbown, 2008, 2019). In line with the goal of developing communicative
skills, practice needs to involve processes that are similar to actual communication.
From the skill acquisition perspective, it is also important to explore the skill-specifc practice
efects that occur when a specifc skill (e.g., listening) is fne-tuned through practice, making it
harder to transfer procedural knowledge gained for that skill to other skills (e.g., speaking). This skill-
specifc practice efect is empirically supported by a number of L2 studies (e.g., DeKeyser, 1997; M.
Li & DeKeyser, 2017; S. Li & Taguchi, 2014; Suzuki & Sunada, 2019) in various practice modalities
(reading, writing, listening, speaking) for a variety of linguistic domains (e.g., vocabulary, grammar,
pronunciation, and pragmatics). While the aforementioned studies focused on relatively “near” skill
transfer, i.e., testing transfer within the same linguistic domain (e.g., comprehension and production
of one specifc linguistic target), Fukkink et al. (2005) posed an interesting cross-linguistic-domain
transfer question. In their EFL classroom intervention study, they examined the extent to which
vocabulary item training transfers to reading comprehension of passages containing those trained
vocabulary items. Their fndings indicated that faster lexical access did not contribute to better read-
ing comprehension. The limited evidence to date suggests that neither “near” nor “far” transfer is
robust for procedural knowledge.
Another interesting, and perhaps more pedagogically important, question pertains to how diferent
types of activities and the ways in which these activities are repeated may infuence the nature of L2
knowledge and skill development. An interesting perspective can be gleaned from motor and psychol-
ogy research on practice variability (Schmidt & Bjork, 1992). Kerr and Booth (1978), for instance,
demonstrated that variable throwing practice of bean bags (involving multiple targets) resulted in
superior performance on the transfer test relative to constant throwing practice (focusing on the same
target). In L2 research, the efects of practice variability were examined for the development of oral
315
Yuichi Suzuki
fuency, which is tied to underlying L2 proceduralization. In this context, De Jong and Perfetti (2011)
conducted an oral fuency training study in which English L2 learners in the United States engaged
in speech practice in a less variable condition (speech on the same three topics three times [AAA−
DDD−GGG]) or the more variable condition (speech on nine diferent topics [ABCDEFGHI]).
The transfer test using a diferent speech topic showed that only the less variable condition led to
meaningful fuency development or proceduralization (e.g., longer mean length of run, shorter pause
length, and higher phonation/time ratio).
In Suzuki’s (2021) study, EFL learners performed three oral narrative tasks for three consecutive
days outside the classroom. The practice variability was manipulated within the day, leading to either
a constant (Day 1: AAA, Day 2: BBB, Day 3: CCC) or a variable (Day 1: ABC, Day 2: ABC, Day
3: ABC) condition. These two practice conditions can also be considered as blocked and interleaved
practice, respectively (see earlier). On the transfer test using a diferent oral narrative task, learners
assigned to the constant (blocked) practice showed greater fuency gains (higher articulation rate)
than those in the variable (interleaved) practice condition. The evidence yielded by these two studies
suggests that constant practice leads to more transfer, which is countered by Kerr and Booth’s (1978)
fndings for motor practice (cf., Willey & Liu, 2018). The commonality in the practice conditions
for superior transfer appears to be the immediate identical repetition within one speaking training
session. Repeated practice en masse might have helped learners to use and proceduralize the same
linguistic constructions, resulting in greater fuency (Suzuki et al., 2022).
What is then the nature of linguistic knowledge that allowed more transfer from the narrowly
focused (constant and blocked) practice? From the skill acquisition perspective, declarative knowl-
edge is general and transferable across many domains/skills, but is not particularly useful due to its
inefciency, whereas procedural knowledge is efcient only for specifc processes and is less transfer-
able (Suzuki & Sunada, 2019). Perhaps procedural knowledge that is automatized at a “not too spe-
cifc” but “more abstract” level may be useful for promoting transfer (DeKeyser, 2018). For instance,
using trigrams with the same combinations of parts of speech (e.g., noun + preposition + noun)
seems to enhance L2 utterance fuency (De Jong & Tillman, 2018). These trigrams are considered
“constructions” in the usage-based sense, because they represent some abstractness of structures and
are neither lexically specifc nor too abstract (see Suzuki et al., 2022). The content of procedural
knowledge may need further scrutiny to better understand transfer and should be examined from
multiple, integrated theoretical perspectives.
The measurements and nature of automatization in L2 learning

Research on L2 automatization should go hand in hand with the development of automatization
measures. Recently, online grammar RT tasks such as sentence construction and maze tasks have
been proposed, yet there has been no consensus on the utility of CV in capturing the restructuring of
grammatical processing. Some researchers (Hulstijn et al., 2009) claim that CV may not be useful in
this context, because it is difcult to distinguish automatization (smaller RT and CV) from declarative
knowledge accumulation (accuracy improvement). The CV, however, may indicate some restructur-
ing of L2 grammatical processing among those with sufcient immersion experience in which L2
can be practiced extensively (Lim & Godfroid, 2015; Suzuki & Sunada, 2018). CV has also started
to be utilized in laboratory research on grammar learning (McManus & Marsden, 2019; Pili-Moss
et al., 2020; Suzuki, 2018) and its utility needs to be explored further.
Another set of research tools aimed at measuring automatization in grammar processing is strongly
linked to the validity issue of explicit and implicit knowledge tests (see Godfroid, 2023 [this volume]).
Real-time grammar comprehension tasks such as self-paced reading and word-monitoring tasks (see Jiang,
316
2011 for details) are arguably tapping into implicit knowledge, i.e., lack of awareness (e.g., Suzuki,
2017; Vafaee et al., 2017; cf., Godfroid & Kim, 2021). These most “implicit” tasks measure grammat-
ical sensitivity to errors of specifc grammatical structures during sentence processing rather than pro-
cessing speed (see Godfroid, 2016, for the application of implicit tests in L2 intervention research).
The validity of these measurements is presently being investigated to tackle some outstanding theo-
retically important questions (see Textbox 25.2 “Open questions and issues”). These fne-grained
measures can complement the data yielded by more global measurements, such as complexity−
accuracy−fuency in free production tasks, to better understand the nature of automatization.
The nature of L2 automatization can be further understood by investigating individual diference
factors. Laboratory intervention research using semi-artifcial languages has just started to uncover
the role of individual cognitive abilities in the rate of automatization (Pili-Moss et al., 2020; Suzuki,
2018); however, much less is known about the extent to which aptitudes can predict ultimate attain-
ment, i.e., the near-end state of automatization reached by highly advanced learners in naturalistic
settings. A thought-provoking hypothesis was proposed by Doughty (2019), who noted that “for
any given person, when motivation is high, personality facets are aligned, and the learning context
is excellent, diferences in aptitude determine ultimate attainment” (p. 101). Since automaticity is an
ultimate goal of L2 skill acquisition, this assertion should be investigated further by L2 researchers.
While it remains to be ascertained whether a relatively innate, stable cognitive ability is necessary for
expertise from a psychological perspective (see the special issue in Intelligence by Detterman, 2014),
further research is needed to establish to what extent cognitive abilities are necessary to achieve auto-
maticity in diferent aspects of L2 (see, e.g., Suzuki & DeKeyser, 2017b for the role of aptitudes in
automatized explicit and implicit grammatical knowledge among advanced naturalistic L2 learners).
Future of L2 practice in L2 education

When it comes to the big question of what practice is best suited for automatization, the answer can
never be simplistic. A plethora of factors need to be considered, and systematic empirical investigations
are essential to enable researchers and practitioners to tailor optimal practice for the learning of a par-
ticular linguistic feature for a group of learners with particular characteristics in a given setting. Suzuki
et al. (2019a) provided a theoretical framework for examining the interaction(s) of three key factors
(i.e., learner-related, linguistic difculty, and practice condition) and for better understanding the
optimal practice that poses desirable difculty for learners, which lays the ground for future research.
L2 research on practice is typically conducted in a laboratory, and/or practice activities are often
decontextualized for the sake of experimental control, despite the broad conceptualization of L2
practice as the ultimate goal of L2 learning (see Introduction of this chapter). This approach has its
advantages (e.g., high internal validity), but its shortcomings need to be acknowledged fully to make
L2 practice research more relevant for real classroom practice. More attention thus needs to be paid to
the quality of L2 practice, and more efort needs to be invested into increasing the ecological validity
of L2 practice research while maintaining high internal validity (see Sato & McDonough, 2019 for
an illustration).
The growing interest in task-based language teaching ofers a promising direction to make L2
practice more meaningful and relevant to real-life L2 use (R. Ellis et al., 2020). From the L2 practice
perspective, task repetition research is the most relevant research sub-domain (Bygate, 2018; DeKey-
ser, 2018). A recent attempt at manipulating the interval of repeated task practice demonstrates the
potential intersection across multiple research domains (Bui et al., 2019; Suzuki & Hanzawa, 2021),
as well as some of the fuency training studies reviewed above (De Jong & Perfetti, 2011; De Jong &
Tillman, 2018; Suzuki, 2021).
Automatization, which can be attained gradually through extended and repeated practice, should
be examined in longitudinal studies, ideally in classroom settings. Since automatized knowledge is
317
Yuichi Suzuki
retained better and longer than declarative knowledge, delayed posttests need to be administered to
ascertain that some automatized skills do not revert to an earlier stage. In addition to investigating
repeated practice using the same set of practice activities, an efective sequence and integration of
multiple instructions and practice activities should also be explored (e.g., Kachinske & DeKeyser,
2019). In order to answer the existing and emerging questions outlined in this chapter, automatiza-
tion must be studied further because it is an important theoretical construct for understanding the
nature of L2 knowledge development through practice.
How should a variety of automaticity features be assessed? How does one feature of automaticity (e.g.,
speed) relate to other criteria (e.g., stability, unawareness, ballistic)?
Are there any individual diference factors (e.g., age, cognitive aptitude, motivation) that determine the
rate and ultimate level of automatization? If so, which aspects of L2 acquisition are most vulnerable to
which individual diference factors?
When and how should practice activities be sequenced and repeated to facilitate automatization?
What are the practice conditions that promote transfer of learning that facilitates successful L2 use in
diferent situations?
How and when should explicit information be integrated with what kinds of activities?
Note
1 The terms “acquisition” and “learning” are used interchangeably in this chapter.
Further reading
This book presents the nuts and bolts of conducting reaction time research in L2. A variety of methodologi-
cal options are presented along with a tutorial of the free presentation software, DMDX.
Suzuki, Y., Nakata, T., & DeKeyser, R. M. (2019). Optimizing second language practice in the classroom: Per-
spectives from cognitive psychology. The Modern Language Journal, 103, 551−561.
This introductory paper to the special issue of The Modern Language Journal presents an overview of the state-
of-the-art empirical research on various practice-related topics.
References
Anderson, J. R. (2000). Learning and memory: An integrated approach. John Wiley & Sons Inc.
Anderson, J. R., Bothell, D., Byrne, M. D., Douglass, S., Lebiere, C., & Qin, Y. (2004). An integrated theory
of the mind. Psychological Review, 111(4), 1036–1060.
Brunmair, M., & Richter, T. (2019). Similarity matters: A meta-analysis of interleaved learning and its modera-
tors. Psychological Bulletin, 145(11), 1029–1052.
Bui, G., Ahmadian, M. J., & Hunter, A. M. (2019). Spacing efects on repeated L2 task performance. System,
81, 1–13.
Bygate, M. (2018). Learning language through task repetition. John Benjamins Publishing Company.
Carlson, R. A. (2003). Skill learning. In L. Nadel (Ed.), Encyclopedia of cognitive science (Vol. 4, pp. 36–42).
Macmillan.
Carpenter, S. K., & Mueller, F. E. (2013). The efects of interleaving versus blocking on foreign language pro-
nunciation learning. Memory & Cognition, 41(5), 671–682.
318
Cepeda, N. J., Pashler, H., Vul, E., Wixted, J. T., & Rohrer, D. (2006). Distributed practice in verbal recall tasks:
A review and quantitative synthesis. Psychological Bulletin, 132(3), 354–380.
De Jong, N., & Perfetti, C. A. (2011). Fluency training in the ESL classroom: An experimental study of fuency
development and proceduralization. Language Learning, 61(2), 533–568.
De Jong, N., & Tillman, P. C. (2018). Grammatical structures and oral fuency in immediate task repetition:
Trigrams across repeated performances. In M. Bygate (Ed.), Task-based language teaching (Vol. 11, pp. 43–73).
John Benjamins.
DeKeyser, R. M. (1997). Beyond explicit rule learning. Studies in Second Language Acquisition, 19(2), 195–221.
DeKeyser, R. M. (2001). Automaticity and automatization. In P. Robinson (Ed.), Cognition and second language
instruction (pp. 125–151). Cambridge University Press.
DeKeyser, R. M. (2007). Practice in a second language: Perspectives from applied linguistics and cognitive psychology.
DeKeyser, R. M. (2015). Skill acquisition theory. In B. VanPatten & J. Williams (Eds.), Theories in second language
acquisition: An introduction (2nd ed., pp. 94–112). Routledge.
DeKeyser, R. M. (2017). Knowledge and skill in ISLA. In S. Loewen & M. Sato (Eds.), The Routledge handbook
of instructed second language acquisition (pp. 15–32). Routledge.
DeKeyser, R. M. (2018). Task repetition for language learning: A perspective from skill acquisition theory. In
M. Bygate (Ed.), Learning language through task repetition (pp. 27–42). John Benjamins Publishing Company.
Detterman, D. K. (2014). Introduction to the intelligence special issue on the development of expertise: Is ability
necessary? Intelligence, 45, 1–5.
Doughty, C. J. (2019). Cognitive language aptitude. Language Learning, 69(S1), 101–126.
Elgort, I. (2011). Deliberate learning and vocabulary acquisition in a second language. Language Learning, 61(2),
367–413.
Elgort, I., Beliaeva, N., & Boers, F. (2020). Contextual word learning in the frst and second language: Defni-
tion placement and inference error efects on declarative and nondeclarative knowledge. Studies in Second
Elgort, I., & Warren, P. (2014). L2 vocabulary learning from reading: Explicit and tacit lexical knowledge and
the role of learner and item variables. Language Learning, 64(2), 365–414.
Ellis, N. C. (2002). Frequency efects in language processing: A review with implications for theories of implicit
and explicit language acquisition. Studies in Second Language Acquisition, 24(2), 143–188.
Ellis, N. C., & Collins, L. (2009). Input and second language acquisition: The roles of frequency, form, and
function. The Modern Language Journal, 93(3), 329–335.
Ellis, N. C., & Wulf, S. (2015). Usage-based approaches to SLA. In B. VanPatten & J. Williams (Eds.), Theories
in second language acquisition: An introduction (2nd ed., pp. 75–93). Routledge.
Ellis, R., Skehan, P., Li, S., Shintani, N., & Lambert, C. (2020). Task-based language teaching: Theory and practice.
Fukkink, R. G., Hulstijn, J. A. N., & Simis, A. (2005). Does training in second-language word recognition skills
afect reading comprehension? An experimental study. The Modern Language Journal, 89(1), 54–75.
Godfroid, A. (2019). Sensitive measures of vocabulary knowledge and processing: Expanding nation’s frame-
work. In S. Webb (Ed.), The Routledge handbook of vocabulary studies. Routledge.
Godfroid, A. (2023). Hypotheses about the interface between explicit and implicit knowledge in second lanuage
acquisition. In A. Godfroid & H. Hopp (Eds.), The Routledge handbook of second language acquisition and psycho-
linguistics (pp. 294–307). Routledge.
Godfroid, A., & Kim, K. M. (2021). The contributions of implicit-statistical learning aptitude to implicit sec-
ond-language knowledge. Studies in Second Language Acquisition, 43(3), 606–634.
Goo, J., Granena, G., Yilmaz, Y., & Novella, M. (2015). Implicit and explicit instruction in L2 learning: Norris
& Ortega (2000) revisited and updated. In P. Rebuschat (Ed.), Implicit and explicit learning of languages (pp.
443–482). John Benjamins Publishing Company.
Hamrick, P. (2023). Conducting reaction time research in second language psycholinguistics. In A. Godfroid & H.
Haskell, R. E. (2001). Transfer of learning: Cognition, instruction, and reasoning. Academic Press.
Hattie, J., & Yates, G. C. (2013). Visible learning and the science of how we learn. Routledge.
Hui, B. (2020). Processing variability in intentional and incidental word learning: An extension of Solovyeva and
DeKeyser (2018). Studies in Second Language Acquisition, 42(2), 237–357.
319
Yuichi Suzuki
Hulstijn, J. H., Van Gelderen, A., & Schoonen, R. (2009). Automatization in second language acquisition: What
does the coefcient of variation tell us? Applied Psycholinguistics, 30(4), 555–582.
Isbell, D. R., & Rogers, J. (2021). Measuring implicit and explicit learning and knowledge. In P. Winke &
T. Brunfaut (Eds.), The Routledge handbook of second language acquisition and language testing (pp. 304–313).
Routledge.
Kachinske, I., & DeKeyser, R. (2019). The interaction between timing of explicit grammar explanation
and individual differences in second language acquisition. Journal of Second Language Studies, 2(2),
197–232.
Kasprowicz, R., Marsden, E., & Sephton, N. (2019). Investigating distribution of practice efects for the learning
of foreign language verb morphology in the young learner classroom. The Modern Language Journal, 103(3),
580–606.
Kerr, R., & Booth, B. (1978). Specifc and varied practice of motor skill. Perceptual and Motor Skills, 46(2),
395–401.
Kim, H., & Rah, Y. (2019). Constructional processing in a second language: The role of constructional knowl-
edge in verb-construction integration. Language Learning, 69(4), 1022–1056.
Kim, J. W., Ritter, F. E., & Koubek, R. J. (2013). An integrated theory for improved skill acquisition and reten-
tion in the three stages of learning. Theoretical Issues in Ergonomics Science, 14(1), 22–37.
Li, M., & DeKeyser, R. M. (2017). Perception practice, production practice, and musical ability in L2 mandarin
tone-word learning. Studies in Second Language Acquisition, 39(4), 593–620.
Li, M., & DeKeyser, R. M. (2019). Distribution of practice efects in the acquisition and retention of L2 man-
darin tonal word production. The Modern Language Journal, 103(3), 607–628.
Li, S., & Taguchi, N. (2014). The efects of practice modality on pragmatic development in L2 Chinese. The
Modern Language Journal, 98(3), 794–812.
Li, S., Zhu, Y., & Ellis, R. (2016). The efects of the timing of corrective feedback on the acquisition of a new
linguistic structure. The Modern Language Journal, 100(1), 276–295.
Lightbown, P. (2008). Transfer appropriate processing as a model for classroom second language acquisition. In
Z. Han (Ed.), Understanding second language process (pp. 27–44). Multilingual Matters.
Lightbown, P. (2019). Perfecting practice. The Modern Language Journal, 103(3), 703–712.
Lim, H., & Godfroid, A. (2015). Automatization in second language sentence processing: A partial, conceptual
replication of Hulstijn, Van Gelderen, and Schoonen’s 2009 study. Applied Psycholinguistics, 36(5), 1247–1282.
McManus, K., & Marsden, E. (2019). Signatures of automaticity during practice: Explicit instruction about L1
processing routines can improve L2 grammatical processing. Applied Psycholinguistics, 40(1), 205–234.
Moors, A. (2016). Automaticity: Componential, causal, and mechanistic explanations. Annual Review of Psychol-
ogy, 67(1), 263–287.
Moors, A., & De Houwer, J. (2006). Automaticity: A theoretical and conceptual analysis. Psychological Bulletin,
132(2), 297–326.
psycholinguistic considerations In A. Godfroid & H. Hopp (Eds.), The Routledge handbook of second language
Nakata, T., & Elgort, I. (2020). Efects of spacing on contextual vocabulary learning: Spacing facilitates the
acquisition of explicit, but not tacit, vocabulary knowledge. Second Language Research, 37(2), 233–260.
Nakata, T., & Suzuki, Y. (2019). Mixing grammar exercises facilitates long-term retention: Efects of blocking,
interleaving, and increasing practice. The Modern Language Journal, 103(3), 629–647.
Norris, J. M., & Ortega, L. (2000). Efectiveness of L2 instruction: A research synthesis and quantitative meta-
analysis. Language Learning, 50(3), 417–528.
Pan, S. C., Tajran, J., Lovelett, J., Osuna, J., & Rickard, T. (2019). Does interleaved practice enhance foreign
language learning? The efects of training schedule on Spanish verb conjugation skills. Journal of Educational
Psychology, 111(7), 1172–1188.
Pellicer-Sánchez, A., Conklin, K., & Vilkaitė-Lozdienė, L. (2021). The efect of pre-reading instruction on
vocabulary learning: An investigation of L1 and L2 readers’ eye movements. Language Learning, 71(1), 162–203.
Pili-Moss, D., Brill-Schuetz, K. A., Faretta-Stutenberg, M., & Morgan-Short, K. (2020). Contributions of
declarative and procedural memory to accuracy and automatization during second language practice. Bilin-
Saling, L. L., & Phillips, J. G. (2007). Automatic behaviour: Efcient not mindless. Brain Research Bulletin, 73(1), 1–20.
Sato, M., & Lyster, R. (2012). Peer interaction and corrective feedback for accuracy and fuency development:
Monitoring, practice, and proceduralization. Studies in Second Language Acquisition, 34(4), 591–626.
Sato, M., & McDonough, K. (2019). Practice is important but how about its quality? Contextualized practice
in the classroom. Studies in Second Language Acquisition, 41(5), 999–1026.
320
Schmidt, R. A., & Bjork, R. A. (1992). New conceptualizations of practice: Common principles in three para-
digms suggest new concepts for training. Psychological Science, 3(4), 207–217.
Segalowitz, N. S. (2003). Automaticity and second languages. In C. J. Doughty & H. M. Long (Eds.), The hand-
book of second language acquisition (pp. 382–408). Blackwell Publishers.
Segalowitz, N. S., & Segalowitz, S. J. (1993). Skilled performance, practice, and the diferentiation of speed-
up from automatization efects: Evidence from second language word recognition. Applied Psycholinguistics,
14(3), 369–369.
Serrano, R. (2011). The time factor in EFL classroom practice. Language Learning, 61(1), 117–145.
Singley, M. K., & Anderson, J. R. (1989). The transfer of cognitive skill (Vol. 9). Harvard University Press.
Solovyeva, K., & DeKeyser, R. (2018). Response time variability signatures of novel word learning. Studies in
Sonbul, S., & Schmitt, N. (2013). Explicit and implicit lexical knowledge: Acquisition of collocations under
diferent input conditions. Language Learning, 63(1), 121–159.
Spada, N. (2015). SLA research and L2 pedagogy: Misapplications and questions of relevance. Language Teaching,
48(1), 69–81.
Spada, N., Jessop, L., Tomita, Y., Suzuki, W., & Valeo, A. (2014). Isolated and integrated form-focused instruc-
tion: Efects on diferent types of L2 knowledge. Language Teaching Research, 18(4), 453–473.
Suzuki, Y. (2017). The optimal distribution of practice for the acquisition of L2 morphology: A conceptual
replication and extension. Language Learning, 67(3), 512–545.
Suzuki, Y. (2018). The role of procedural learning ability in automatization of L2 morphology under diferent
learning schedules: An exploratory study. Studies in Second Language Acquisition, 40(4), 923–937.
Suzuki, Y. (2021). Optimizing fuency training for speaking skills transfer: Comparing the efects of blocked and
interleaved task repetition. Language Learning, 71, 285–325.
Suzuki, Y., Eguchi, M., & de Jong, N. (2022). Does the reuse of constructions promote fuency development in
task repetition? A usage-based perspective. TESOL Quarterly, Early View.
Suzuki, Y., & DeKeyser, R. M. (2017a). Efects of distributed practice on the proceduralization of morphology.
Language Teaching Research, 21(2), 166–188.
Suzuki, Y., & DeKeyser, R. M. (2017b). The interface of explicit and implicit knowledge in a second language:
Insights from individual diferences in cognitive aptitudes. Language Learning, 67(4), 747–790.
Suzuki, Y., & Hanzawa, K. (2021). Massed task repetition is a double-edged sword for fuency development: An
EFL classroom study. Studies in Second Language Acquisition, 1–26.
Suzuki, Y., Nakata, T., & DeKeyser, R. M. (2019a). The desirable difculty framework as a theoretical founda-
tion for optimizing and researching second language practice. The Modern Language Journal, 103(3), 713–720.
Suzuki, Y., Nakata, T., & DeKeyser, R. M. (2019b). Optimizing second language practice in the classroom:
Perspectives from cognitive psychology. The Modern Language Journal, 103(3), 551–561.
Suzuki, Y., & Sunada, M. (2018). Automatization in second language sentence processing: Relationship between
elicited imitation and maze tasks. Bilingualism: Language and Cognition, 21(1), 32–46.
Suzuki, Y., & Sunada, M. (2019). Dynamic interplay between practice type and practice schedule in a second
language: The potential and limits of skill transfer and practice schedule. Studies in Second Language Acquisi-
tion, 42(1), 169–197.
Suzuki, Y., Yokosawa, S., & Aline, D. (2020). The role of working memory in blocked and interleaved grammar
practice: Proceduralization of L2 syntax. Language Teaching Research, 26(4).
Toomer, M., & Elgort, I. (2019). The development of implicit and explicit knowledge of collocations: A con-
ceptual replication and extension of Sonbul and Schmitt (2013). Language Learning, 69(2), 405–439.
Tuz, E. (1993). From controlled practice to communicative ability: Does training transfer? Temple University Japan
Research Studies in TESOL, 1(1), 97–108.
Tyler, A. E., & Ortega, L. (2018). Usage-inspired L2 instruction: Some refections and a heuristic. In A. E.
Tyler, L. Ortega, M. Uno, & H. Park (Eds.), Usage-inspired L2 instruction: Researched pedagogy (Vol. 49). John
Benjamins Publishing Company.
Tzelgov, J. (1997). Automatic but conscious: That is how we act most of the time. In J. Wyer (Ed.), Advances in
social cognition (Vol. 10, pp. 217–230). Psychology Press.
Vafaee, P., Suzuki, Y., & Kachinske, I. (2017). Validating grammaticality judgment tests: Evidence from two new
psycholinguistic measures. Studies in Second Language Acquisition, 39(1), 59–95.
VanPatten, B., & Williams, J. (2015). Theories in second language acquisition: An introduction (2nd ed.). Routledge.
Willey, C. R., & Liu, Z. (2018). Long-term motor learning: Efects of varied and specifc practice. Vision
Research, 152, 10–16.
321
26
DECLARATIVE AND PROCEDURAL
MEMORY IN SECOND
LANGUAGE LEARNING
Psycholinguistic considerations
Introduction
The feld of second language (L2) acquisition aims to account for the learning, representation, and
processing of L2. Here we focus on L2 learning, while also touching on representation and pro-
cessing. Note that we defne learning broadly, as the cognitive process of gaining or strengthen-
ing knowledge. Much of the psycholinguistic research on L2 learning has examined the roles of
underlying cognitive factors such as attention and awareness (Godfroid, 2023 [this volume]) and
short-term/working memory (Li, 2023 [this volume]). These factors indeed afect L2 learning (e.g.,
Faretta-Stutenberg & Morgan-Short, 2018; Issa & Morgan-Short, 2019). However, they are not in
and of themselves learning mechanisms (Squire & Dede, 2015). Thus, researchers have increasingly
considered contributions of (neuro)cognitive processes that actually underlie learning and long-term
memory. Most of this work has examined declarative and procedural memory. In this chapter, we
focus on the declarative/procedural (DP) model of language, a neurocognitive account of L2 and L1
(frst language) (Ullman, 2001, 2020; see Textbox 26.1). In this chapter, we situate the DP model in a
historical research context; lay out the model, with a focus on learning; discuss what psycholinguistic
methodological approaches are appropriate (or not) for testing the role of declarative and procedural
memory in L2 (and L1) and summarize related evidence; and present gaps and future directions.
Declarative memory system: The DP model defnes declarative memory as the learning and memory that
rely on the medial temporal lobe and associated circuitry. Knowledge learned in this brain system can
be explicit or implicit.
Procedural memory system: The DP model defnes procedural memory as the learning and memory that
rely on the basal ganglia and associated circuitry. Knowledge learned in this brain system seems to be
entirely implicit.
322 DOI: 10.4324/9781003018872-30

Declarative and procedural memory in second language learning
Three major theoretical accounts have proposed roles for declarative and procedural memory or
knowledge in L2. All three accounts make the same basic assumption that domain-general learning
and memory mechanisms underlie the learning of L2. Nevertheless, they also difer in key respects,
largely because they are based on distinct conceptual foundations, with diferent historical roots.
DeKeyser’s skill acquisition theory (DeKeyser, 2020) emerged from cognitive architectures devel-
oped in the 1970s that aimed to computationally model skill learning. His theory relies particu-
larly on the infuential ACT-R architecture, an overarching model of cognition (Anderson et al.,
2019). According to ACT-R, skill learning begins with the acquisition and application of declarative
knowledge, which is referential information represented in a propositional network. Next, through
repeated application of declarative knowledge, a process called production compilation creates pro-
cedural knowledge, which consists of action rules that perform the task. Finally, the skill is tuned,
such that performance becomes automatic (highly accurate and rapidly deployed). Overall, it is sug-
gested that the three phases account for the power function observed in skill acquisition, with rapid
changes in performance at initial stages of learning followed by a shift to more gradual performance
improvements.
Although language was always central to ACT-R, the ideas were frst tested empirically in L2 by
DeKeyser (1997), whose study demonstrated that, with practice, improvement of L2 comprehension
and production refected the power function in skill acquisition. This soon led to the development of
DeKeyser’s Skill Acquisition Theory (e.g., DeKeyser, 2020), which suggests (a) a learner frst acquires
declarative knowledge, that is, explicit knowledge “about” a specifc L2 form, through instruction
or observation; (b) then, as declarative knowledge is applied through practice, the learner quickly
develops procedural knowledge about “how” to use the L2 form, e.g., in oral production; and (c)
fnally, with further practice, the learner develops automaticity for the specifc skill (also see Suzuki,
this volume).
Paradis’s neurolinguistic theory of bilingualism (Paradis, 2009) was inspired by a cognitive neu-
roscience perspective of learning and memory that emerged in the 1980s and 1990s. According to
this perspective (Eichenbaum et al., 1994), learning facts and events relies on declarative (explicit)
memory, which is hippocampus-based, subject to conscious recollection, and fexibly used, whereas
non-declarative learning (e.g., of perceptual, motor, or cognitive skills, as well as priming) relies on
procedural memory, which is independent of the hippocampus, not available to conscious recall,
“expressed implicitly” (p. 457) through performance, and used infexibly. Paradis (2009, p. 9) pins the
origins of his theory to an oral presentation of this view by Neal Cohen in 1991, which led Paradis to
note that the characteristics of procedural memory “were compatible with the processing of syntax”
and that declarative memory seemed to be associated with learning words.
Subsequently, Paradis developed the neurolinguistic theory of bilingualism (Paradis, 1994, 2009).
This theory posits that, in adult L1 speakers, communicative competence relies on several elements,
including (a) explicit metalinguistic knowledge, that is, conscious knowledge of language, including
knowing the form-meaning pairings of vocabulary; and (b) implicit linguistic competence, that is,
grammatical functions, which comprise phonology, morphology, and syntax as well as the implicit
grammatical properties and semantic constraints of lexical items. According to Paradis (2009, p. xi),
explicit metalinguistic knowledge relies on declarative memory, which is supported by the “hip-
pocampal system and extensive areas of tertiary associative cortex.” In contrast, implicit linguistic
competence relies on procedural memory, which is supported by the “cerebellum, striatum, and
focal cortical areas.” For L2, the theory posits that learning both vocabulary and grammatical func-
tions initially depends largely on metalinguistic knowledge. However, with increasing practice and
exposure, especially in communicative contexts, grammatical functions may gradually come to rely
on implicit linguistic competence.
323
Ullman’s declarative/procedural model (Ullman, 2004, 2016; Ullman et al., 1997; Ullman et al.,
2020) emerged epiphanously as he was walking across a bridge at the Audubon Ipswich River Wild-
life Sanctuary in 1993. This occurred while he was trying to make sense of data he had obtained from
patients with diferent types of aphasia or degenerative disease, while working with Steven Pinker
and Suzanne Corkin at MIT. This work was examining lexical and grammatical aspects of language
in these patients in order to investigate how Pinker’s words and rules theory (Pinker, 1991) might
map onto the brain. After collecting the data, Ullman had this “Aha!” realization that the pattern of
language, memory, and neuroanatomical anomalies in these patients could be explained by diferent
sorts of dysfunction of the declarative or procedural memory brain systems, whose conceptualizations
were based on contemporary cognitive neuroscience research (Ullman et al., 1993; Ullman et al.,
1997). These conceptions of declarative and procedural memory, detailed later, difer somewhat from
those espoused by DeKeyser and Paradis (as described earlier). For example, the declarative memory
brain system can underlie both explicit knowledge (which can be brought to conscious awareness)
and implicit knowledge (which cannot be brought to awareness). Initially, the model focused only on
L1, and made the simple assumption that the learning, storage, and processing of lexical knowledge
relies on declarative memory, whereas that of grammatical knowledge relies on procedural memory
(Ullman et al., 1997). Over the years, the model has been refned and extended. This includes the
proposal of more subtle relations between the memory systems and diferent aspects of language, and
the application of the model to various other populations, including L2 learners and a wide range
of neuroatypical populations (Ullman, 2001, 2004, 2016, 2020; Ullman et al., 2020). In the next
section we briefy summarize the current conception of the model, with a focus on L2 as well as the
model’s predictions that are most relevant to the methods and evidence discussed in the subsequent
section.
Theoretical approach: the Declarative/ Procedural model

The DP model is grounded in the notion of cooptation. According to this basic principle of evolu-
tion and biology, previously existing structures and mechanisms are constantly being reused—that
is, co-opted—for new purposes. Thus, it is likely that language depends importantly on previously
existing neurocognitive substrates that were coopted for this new function—whether or not those
substrates have become further specialized for language through evolution, development, or learning.
The DP model applies this principle of cooptation to language, with a focus on learning. Indeed,
the model is based on two premises related to learning. Firstly, most of language must be learned,
whether or not aspects are innately specifed. Secondly, declarative memory and procedural memory
are arguably the two most important learning and memory systems in the brain, in both animals
and humans (Squire & Dede, 2015; Ullman, 2004, 2016). Given these premises and the principle of
cooptation, the declarative and procedural memory systems are predicted to play critical roles in the
learning of language, across both L1 and L2. This is the essence of the DP model. Here we frst sum-
marize the nature of the two memory systems, and then briefy lay out key predictions for language
that follow from this independent understanding of the systems.
The memory systems

According to the neurocognitive perspective espoused by the DP model, the two memory systems
are neuroanatomically grounded in distinct brain circuits. The learning characteristics of each mem-
ory system quite simply emerge from the characteristics of its underlying circuit. For more detailed
information on the systems, see Ullman (2004, 2020), and Ullman et al. (2020).
Declarative memory is defned as the learning and memory that rely on the medial temporal lobe
(including the hippocampus) and its associated circuitry. This system seems to be critical for learning
324
arbitrary pieces of information and their associations and has traditionally been implicated in learning
information about events (episodic knowledge) and facts (semantic knowledge). However, declara-
tive memory is very fexible, and can learn just about anything. It can support both explicit infor-
mation and implicit information, though it appears to be the only system that can underlie explicit
knowledge (also see Godfroid, 2023 [this volume]; Williams & Rebuschat, 2023 [this volume]).
Learning in the system is very rapid and can occur after even a single presentation of information.
Also of interest here is that learning in declarative memory seems to improve during childhood and
then plateaus in adolescence and early adulthood, after which it declines.
Procedural memory is defned as the learning and memory that rely on the basal ganglia and its
associated circuitry (Ullman, 2020; Ullman et al., 2020). This system underlies the learning, storage,
and processing of information for a range of functions, including cognitive and (perceptuo-)motor
skills (such as motor sequences), perceptual sequences (often tested in statistical learning paradigms),
categories, habits, and routes (for navigation). Learning and knowledge in procedural memory
seem to be entirely implicit. Learning involves generating predictions and then evaluating these
predictions on the basis of the outcome (i.e., information about the correctness of the prediction);
rapid (dis)confrmation (feedback) of the prediction leads to learning (e.g., when predicting the next
element in a sequence, the subsequent presence or absence of that element can serve as feedback that
can lead to learning). Procedural learning occurs gradually across trials (though this is not always a
slow process), and may eventually result in more automatized (faster, reliable, infexible) processing
than does learning in declarative memory. Learning in procedural memory seems to already be robust
early in childhood, though it appears to decline around adolescence.
The declarative and procedural memory systems also interact (Ullman et al., 2020). Of primary
interest here is that they can complement each other by learning analogous functions, such as knowl-
edge of a given route, sequence, or category. That is, the two learning systems play at least partly
redundant roles, though they generally learn somewhat diferent types of knowledge. For example,
declarative memory may learn explicit knowledge of the order of elements of a sequence whereas
procedural memory may implicitly learn to predict subsequent elements in the sequence. Certain
types of knowledge, however, can be learned by only one or the other of the two systems; impor-
tantly, arbitrary pieces of information and their associations may only be learnable by declarative
memory.
A range of variables appear to modulate which of the two systems is relied on more, for those
tasks and functions that can be learned in either. Here we focus on just a few variables. Firstly, the
amount of input matters. Less input often leads to a greater reliance on declarative memory (since it
learns quickly), whereas more input can shift dependence to procedural memory (which can become
predominant due to automatization). Secondly, the type of input makes a diference. In particular,
explicit, instructed input should lead to learning in declarative memory, since that is the only mem-
ory system to support explicit knowledge. Thirdly, age of acquisition also matters. Children will tend
to rely more on procedural memory and younger adults more on declarative memory, simply because
procedural memory is stronger during childhood whereas declarative memory improves over the
course of childhood and adolescence. Additionally, and in the same vein, other subject-level factors
can also afect which system is relied on more, because those individuals or groups who are better
at learning in one versus the other system should depend more on the better system. For example,
learning in declarative memory (but not procedural memory) may generally be better in females than
males, and thus females will tend to rely more on declarative memory than males.
Predictions for language

As we have seen, the declarative and procedural memory systems have been studied extensively,
which has led to a substantial, independent understanding of the systems. Crucially, based on this
325
understanding, we can generate a wide range of specifc, testable, and often novel predictions for
language (see especially Ullman, 2016 and Ullman et al., 2020; for L2, see Ullman, 2020). The fol-
lowing predictions are especially relevant here.
Lexical memory is predicted to depend heavily on declarative memory, across populations,
including in L1 and L2. This follows from this memory system’s critical role in learning arbitrary
pieces of information and their associations, since such knowledge constitutes a substantial portion
of lexical information. Thus, knowledge about simple open-class (content) words, including their
phonological forms, meanings, subcategorization frames, and the mappings between them (e.g.,
sound-meaning mappings), should depend importantly on declarative memory. Similarly, some sort
of representation of irregular morphological forms, both infectional (e.g., dig–dug) and derivational
(e.g., solemn–solemnity), should be stored in declarative memory, as should knowledge about idio-
syncratic aspects of idioms, proverbs, collocations, and so on. In contrast, lexical knowledge should
depend much less on procedural memory, though it may do so in certain specifc circumstances, such
as during word segmentation, for closed-class words and morphemes, and for motor-related words
like hammer (Ullman et al., 2020).
Unlike lexical memory, grammar should be able to depend substantially on both memory systems
(Ullman et al., 2020). Firstly, grammar shares a number of characteristics with procedural memory.
Both grammar and procedural memory are involved in rules, sequences, categories, implicit knowl-
edge, and prediction; moreover, acquired knowledge can be automatized in both. Due to these
shared characteristics, procedural memory may be expected to play an important role in grammar,
across domains, including phonology, morphophonology, and (morpho)syntax. Secondly, because
declarative memory is highly fexible, it should also be able to support grammar. Yet the two learning
systems are predicted to support grammar in largely diferent ways. Procedural memory may underlie
the learning of rule-governed grammatical knowledge that underlies real-time combination, across
language domains (e.g., in the (de)composition of regular afxed forms such kick + -ed). In contrast,
thanks to its fexible nature, declarative memory may memorize individual chunks (e.g., the afxed
form kicked), generalize analogically across such forms to new forms (e.g., blicked), and/or learn
explicit rules.
Because the two memory systems have diferent characteristics, grammar (as well as other aspects
of language that can depend on both systems, such as speech-sound representations and perhaps
pragmatics) should depend more on one or the other system and in diferent ways as a function of
various factors (Ullman, 2020; Ullman et al., 2020). Firstly, because declarative memory learns faster
than procedural memory, grammar should depend more on declarative memory during early stages
of L2 (and L1) learning and more on procedural memory at later stages (after more exposure). Sec-
ondly, because explicit knowledge can only be supported by declarative memory, explicit instructed
grammatical information (e.g., in L2 classroom settings) should lead to a greater dependence of
grammar on declarative memory, whereas an absence of such input may shift reliance to procedural
memory—especially when instead the learner is exposed to a naturalistic/immersion environment
(in L1 or L2). This is because the rapid input of natural speech may be particularly well-suited for
procedural learning, given that this system learns by predicting subsequent elements and receiving
rapid feedback (e.g., subsequent linguistic elements can be predicted, and the rapid appearance of
such elements in the input in natural speech can constitute rapid feedback; Ullman, 2020). Thirdly,
because learning in procedural memory seems well-established early in life and then may decline,
whereas learning in declarative memory is poor early on, but then improves during childhood, gram-
mar should tend to rely more heavily on procedural memory for early-learned languages (L1 or L2),
and more on declarative memory for languages acquired comparatively later in life (at least to early
adulthood, after which declarative memory in particular declines, Reifegerste et al., 2021). Overall,
grammar in adult L2 should thus tend to depend more on declarative than procedural memory as
compared to adult L1 for the reasons laid out just previously: The L2 will be at an earlier stage of
326
learning (because it was learned later, and thus will have had less exposure); L2s are more likely than
L1s to be taught explicitly; and L2 learners have later ages of acquisition. Finally, any individual or
group with better declarative memory should rely more on this system (e.g., females as compared to
males), and conversely for those with better procedural memory.
Issues in testing the roles of declarative and procedural memory in L2 (and L1)
A key issue for any model is how its predictions can be tested. Here we lay out methodological
approaches that can be used to test the predictions of the DP model in L2 (and L1). We focus on
psycholinguistic approaches, with somewhat of an emphasis on those aimed at probing learning.
Before delving into specifc approaches, we emphasize that most psycholinguistic as well as neu-
rocognitive research of L2 and L1 has focused on processing (and to some extent representation)
rather than learning. This may be due to various reasons, including the empirical difculty of probing
learning. Importantly, a simple distinction between processing and learning is fawed, in that learning
likely often occurs during processing and vice versa (Hopp, 2023 [this volume]; Ullman, 2001). This
can complicate interpretation of evidence and is critical to keep in mind when evaluating research.
Not all methods are equally appropriate for testing the key predictions of the DP model.
Approaches that can reliably link language outcomes specifcally to one or the other learning and
memory system constitute particularly valid tests of the model. That is, such approaches have high
internal validity.
One especially valid approach is testing associations between learning abilities in declarative or
procedural memory and predicted learning, representation, or processing outcomes in L2 or L1 (for
more information on this approach, see Hamrick et al., 2018; Morgan-Short et al., 2022; Ullman,
2020). For example, one can test the predicted reliance of lexical abilities on declarative memory
by examining whether people who are better at non-verbal learning in declarative memory are also
better at word learning or retrieval or have larger vocabularies. This individual-diferences approach
is a particularly valid test of the predictions of the DP model to the extent that performance on the
declarative or procedural learning tasks has been independently linked to the brain circuits underly-
ing that specifc system. However, simple correlations do not imply causation. A correlation between
word learning and declarative memory could be explained not just by words being learned in declar-
ative memory but instead by factors (e.g., working memory) that may be common to both. This
problem can be minimized in various ways. For instance, one can hold such factors constant, for
example by controlling for them statistically. One can also employ learning tasks that are relatively
process-pure (minimizing the infuence of the other memory system or abilities such as working
memory) and non-verbal (minimizing the likelihood that correlations with language measures simply
refect verbal abilities). And one can demonstrate the specifcity of the correlation: Because the DP
model generates multiple specifc predictions regarding which language functions depend on which
memory systems in which circumstances (e.g., lexicon/grammar in L1/ L2 at low/high experience),
it is unlikely that the full pattern of predictions will be explained by other factors or by chance.
The extant evidence using the associational approach is largely consistent with the predictions of
the DP model. A meta-analysis of associational studies in L1 and L2 (Hamrick et al., 2018) tied lexi-
cal abilities in L1 (no relevant lexical L2 studies were found) to learning only in declarative memory.
In contrast, grammar was tied mainly to procedural memory in L1, but to both systems in L2. Spe-
cifcally, grammar was associated only with declarative memory at lower L2 experience but only with
procedural memory at higher experience. More recently, Morgan-Short et al. (2022) performed
a systematic review (without a meta-analysis) of such associational studies in L2. They examined
multiple language domains and the infuence of factors beyond the amount of L2 experience. The
review revealed the following. The L2 lexicon seems to be linked to learning in declarative but not
procedural memory. L2 phonology (which was not examined in Hamrick et al.) was linked only to
327
declarative memory, perhaps because studies thus far have tested phonology in classroom learners or
at early stages of learning. For L2 morphophonology, the evidence suggests that afxed forms are
associated with both memory systems (consistent with rule-governed combination in procedural
memory and chunking in declarative memory) whereas analogical forms appear to be supported
only by declarative memory. L2 morphosyntax is associated mainly with procedural memory, though
also to some extent with declarative memory, perhaps especially following classroom instruction or
when explicit knowledge is present. Finally, L2 syntax appears to be linked with declarative memory
mainly at earlier stages of learning but procedural memory at later stages, as well as with procedural
memory following uninstructed and immersion learning; surprisingly, syntax in instructed and class-
room contexts was linked to neither memory system. Thus overall, the evidence from correlational
studies supports the DP model, though some research gaps and inconsistencies suggest the need for
further research using this powerful approach.
Some psycholinguistic approaches are appropriate for testing links specifcally with declarative
memory. One such approach is the examination of frequency efects. If linguistic forms are stored as
wholes in declarative memory (e.g., lexical forms such as cat or dug, or chunks such as walked or
“the cat”), then those that are more frequent in the input should have stronger memory traces and
so should be retrieved more rapidly or accurately (e.g., in a production task) or yield higher accept-
ability judgments (Prado & Ullman, 2009). In contrast, if forms are composed in procedural memory
(e.g., walk + -ed), the input frequency of the whole complex form (walked) should not afect perfor-
mance, at least when holding constant the frequency of its parts. Thus, whole-form frequency efects
can be taken to refect storage (vs. composition), consistent with a reliance on declarative memory.
Additionally, independent evidence has linked frequency efects of both simple words and chunks to
declarative memory circuits (Fernandez et al., 2002; Lieberman et al., 2004). This strengthens the
interpretation that frequency efects refect access to representations stored in the declarative memory
system.
Evidence has revealed frequency efects largely where predicted by the DP model. For example,
whole-form frequency efects for morphologically infected forms are widely found for irregulars,
whereas for regulars, their presence varies, though in systematic ways (Babcock et al., 2012; Bowden
et al., 2010; Prado & Ullman, 2009). For example, in L1, frequency efects are found for regulars in
females (especially for higher-frequency regulars, which are more likely to be chunked) but not males
(Babcock et al., 2012; Prado & Ullman, 2009). In contrast, as expected with an increased reliance
of grammar on declarative memory in L2, frequency efects for regulars are observed consistently
in L2, and indeed are found for both sexes (Babcock et al., 2012; Bowden et al., 2010). Moreover,
within L2, frequency efects for regulars are stronger with less language exposure and at later ages
of acquisition (Babcock et al., 2012). As with the associational approach, support for the model is
further strengthened by the full pattern of fndings, which match the specifcity of the predictions of
the DP model.
Certain other psycholinguistic approaches may also test links with declarative memory in par-
ticular. More imageable forms are better remembered than less imageable forms (e.g., apple vs.
idea), an efect that seems to depend on declarative memory circuits (Klaver et al., 2005). Thus,
such imageability efects can be leveraged to test for the storage of linguistic forms in declarative
memory (Prado & Ullman, 2009). For example, if certain forms are stored in this system (e.g.,
irregular infected forms or chunked regular forms) then those that are more imageable may be
retrieved more rapidly or accurately in production tasks or judged to be more acceptable (Prado
& Ullman, 2009). Indeed, evidence suggests that, as expected by the DP model, imageability
efects in L1 are typically found for irregular infected forms, whereas for regulars such efects are
observed in females (especially for higher imageability forms) but not males (Prado & Ullman,
2009). To our knowledge, such imageability efects have not been investigated in L2, representing
a clear gap in the literature.
328
A similar logic regarding testing for the predicted dependence of infected forms on declarative
memory may hold for the examination of phonological neighborhood efects. That is, if the processing of
an irregularized form or a regularized infected form (e.g., spling–splang, or blow–blowed, respectively) is
afected by similar-sounding forms thought to be stored in declarative memory (e.g., irregulars such
as rang, sang; or chunked regulars such fowed, rowed), those irregularized/regularized forms may be
taken to also be processed in this system, through analogical generalization. (For an alternative inter-
pretation of neighborhood efects see Albright & Hayes, 2003.) Research using this approach has
largely supported the specifc pattern of predictions of the DP model, though again studies have thus
far focused on L1 (Hartshorne & Ullman, 2006; Johari et al., 2019; Prasada & Pinker, 1993). For,
example, in female but not male children and patients with Parkinson’s disease, over-regularizations
such as blowed that have many similar-sounding regular neighbors (e.g., fowed, rowed) are produced
more often than those with few neighbors, as expected if females in particular are storing regulars and
generalizing them analogically to new forms (Hartshorne & Ullman, 2006; Johari et al., 2019). The
apparent absence of studies using this technique in L2 suggests a clear motivation for future research.
Whereas frequency, imageability, and neighborhood efects examine links with storage in declar-
ative memory, other psycholinguistic approaches such as priming and working memory efects
can be used to probe composition, which should typically rely on procedural memory. Inferring
(de)composition from morphological priming efects relies on the notion that if morphologically
complex forms such as regular infected forms are decomposed, they should prime their stems more
than stored morphological forms do (see Gor, 2023 [this volume]). For example, walked should prime
walk more than dug primes dig if only walked is decomposed into stem and afx (walk and -ed), since
this would lead to direct access to its stem. Working memory efects are premised on the principle
that maintaining more items in working memory is harder, and hence forms undergoing decomposi-
tion into two or more units should be harder to maintain in working memory than forms that are not
decomposed or are decomposed into fewer units (Reifergerste et al., in prep.; Ullman et al., 2001).
For example, irregular (dug) or chunked regular infected forms (walked) should be easier to maintain
in working memory in a receptive language task (with each form maintained as a single element) than
regular forms that are decomposed (which would require maintenance of two elements, walk and -ed).
Neither efect has yet been directly linked to procedural memory, and indeed morphological
priming efects may be found for reasons other than decomposition (Marslen-Wilson, 2007). Nev-
ertheless, the pattern of results from these efects is largely consistent with the predictions of the DP
model, suggesting in turn that both morphological priming and working memory efects are in fact
linked to procedural memory. Morphological priming efects are found robustly for regulars in L1
but not in L2, whereas irregulars typically show only weak priming in both (Jacob et al., 2018; Jacob
et al., 2013; but see Feldman et al., 2010). Moreover, priming efects for regular infection are greater
when the age of L2 acquisition is earlier (Veríssimo et al., 2018). Working memory efects have been
observed in L1 for regular (vs. irregular) infected forms, especially in males and for lower frequency
infected forms, as expected by the DP model (Reifegerste et al., in prep.; Ullman et al., 2001). Such
working memory efects have not yet been examined in L2.
Importantly, some approaches are not appropriate for testing the predictions of the DP model,
primarily because they do not in and of themselves generate measures that link language specif-
cally to declarative or procedural memory. That is, they have low internal validity. For example,
fast response times, or responses in speeded tasks such as speeded grammaticality judgments, do not
necessarily implicate procedural (or even implicit) knowledge: Responses based on knowledge in
either system can be equally fast (Prado & Ullman, 2009) and speeded tasks may simply speed up
participants, but not afect which underlying mechanisms are relied on (Maie & Godfroid, 2022).
Likewise, leveraging the presence of explicit or implicit training conditions or knowledge to iden-
tify reliance on either memory system should be used with caution (Ullman, 2020). Although the
presence of explicit knowledge does indeed seem to indicate a reliance on declarative memory, we
329
emphasize that implicit knowledge does not specifcally refect a dependence on procedural memory,
since knowledge in declarative memory as well as various other memory systems can be implicit. One
should even be cautious regarding the interpretation of efects from within- or between-subject mod-
erating variables that are predicted to shift reliance between the two memory systems. For example,
though explicit instruction should lead to a greater dependence on declarative memory than unin-
structed input, both types of input can lead to learning in both systems (Morgan-Short et al., 2012;
Morgan-Short et al., 2015; Ullman et al., 2020). Similarly, though males may depend more than
females on procedural memory, both sexes appear to depend importantly on this memory system, at
least in L1 (Prado & Ullman, 2009). The methods laid out earlier (e.g., the associational approach in
individual-diferences research) more directly test dependence on one or the other system. Thus, studies
examining behavioral efects of instructed vs. uninstructed input, or the efect of sex diferences, should
ideally also use one of the methods laid out in this chapter, or an appropriate neurocognitive method
(Ullman, 2020), to provide more direct evidence linking language outcomes to the memory systems.
Finally, a critical point is that converging evidence is key, in several respects. Firstly, since every
methodological approach has weaknesses as well as strengths, converging evidence from diferent
approaches increases confdence in the pattern of fndings. Secondly, converging evidence across
researchers and research groups reduces the risk of bias, e.g., for or against a model or claim. Ideally,
multiple research groups would carry out informed studies testing the DP model. Thirdly, the mul-
titude of the DP model’s specifc yet interacting predictions provide an ideal opportunity to test for
converging evidence. In particular, as mentioned earlier, empirical support for the pattern of predic-
tions greatly strengthens confdence in the underlying theoretical framework.

We have seen that the DP model generates a wide range of predictions, with a number of the predictions
about L2 laid out here (also see Ullman, 2020; Ullman et al., 2020). The methods described just previ-
ously are appropriate for testing many of these predictions. Nevertheless, our summaries of the evidence
based on these methods revealed that many gaps remain. Here we highlight a number of the gaps and
associated future directions that we believe warrant further investigation (see also Textbox 26.2).
Numerous predictions of the DP model have not been examined in L2 (or for the most part in
L1), or they merit further investigation. These include predictions about input factors (e.g., fast/slow
feedback, type of exposure); individual/group diferences (e.g., age of acquisition, sex diferences);
aspects of language (e.g., beyond morphology and syntax); and, crucially, interactions of such factors
(e.g., fast/slow feedback efects interacting with age of acquisition on learning morphosyntax).
Methodological gaps or under-utilized methodological approaches in the investigation of declara-
tive/procedural memory in L2 (and L1) include examining language outcomes during learning
(e.g., Pili-Moss et al., 2020); employing artifcial linguistic paradigms (to probe learning mechanisms),
possibly in conjunction with natural language paradigms (to test for generalizability; Morgan-Short
et al., 2012; Morgan-Short, 2020); extending the examination of frequency, imageability, priming,
and working memory efects from morphology to syntax; using meta-analysis, which can yield more
reliable and generalizable results than single-study fndings; incorporating new methods well suited
to link language to either memory system (e.g., consolidation, in which sleep facilitates changes to
declarative but not procedural memory, Wilhelm et al., 2008); and, crucially, combining diferent
psycholinguistic as well as neurocognitive approaches, in particular in the examination of learning.
The relation between the DP model and issues of particular interest in the feld of second language
acquisition (SLA) should be further investigated. These include whether and how the memory systems
may contribute to L2 aptitude, L2 retention, and individual diferences in L2 abilities, and what roles they
may play with regard to the quality of L2 practice, the importance of attention and feedback in L2 learning,
and the nature of learning in statistical learning paradigms (Williams & Rebuschat, 2023 [this volume]).
330
The implications of the DP model for improving learning and pedagogy clearly warrant further
investigation (Karatas et al., 2021; Ullman & Lovelett, 2018). Both item-level manipulations (e.g.,
spaced repetition, retrieval practice, interleaving, deep encoding, fast/slow feedback) and learner-
level variables (e.g., sleep, exercise, diet, and mindfulness) should be probed.
A wide range of linguistic, behavioral, contextual, demographic, biological, and other factors are pre-
dicted to afect the learning, representation, or processing of L2 in each memory system. Just how do
these various factors and especially their interactions infuence L2?
How can we refne methods to tap into specifc memory systems? That is, how can we improve the
validity of such methods?
How can the various predictions of the DP model be further leveraged to provide insights into outstand-
ing questions in the feld of SLA?
How can pedagogical approaches, informed by the DP model, be employed to improve language learning?
Note
We thank Sayako Earle, Jana Reifegerste, and João Veríssimo for input on our discussion of methodological
approaches that can be used to test contributions of declarative and procedural memory to language. We also
thank Robert DeKeyser and Michel Paradis for confrming the accuracy of our descriptions of their models and
the historical contexts in which they arose, and Aline Godfroid and Holger Hopp for very useful comments that
improved this chapter.
Further reading
Bufngton, J., Demos, A., & Morgan-Short, K. (2021). The reliability and validity of procedural memory assess-
ments used in second language acquisition research. Studies in Second Language Acquisition, 43(3), 635–662.
This empirical study examines the validity and reliability of tasks that have been used to probe learning in
procedural memory in the examination of associations between L2 and this memory system.
Morgan-Short, K., Hamrick, P., & Ullman, M. T. (2022). Declarative and procedural memory as predictors of
second language development. In S. Li, P. Hiver, & M. Papi (Eds.), The Routledge handbook of second language
acquisition and individual diferences (pp. 67–81). Routledge.
This is a systematic review of studies using the associational approach to test links between the memory
systems and a wide range of L2 abilities.
Ullman, M. T. (2020). The declarative/procedural model: A neurobiologically-motivated theory of frst and
second language. In B. VanPatten, G. D. Keating, & S. Wulf (Eds.), Theories in second language acquisition
(3rd ed., pp. 128–161). Routledge.
This is the most recent full description of the DP model of SLA. It lays out the logic of the model; the nature
of the memory systems; the claims and predictions of the model for L2 (and L1); what counts as evidence
(with a focus on neurocognitive methods); common misunderstandings; and the explicit/implicit debate in
the context of the model.
Ullman, M. T., Earle, F. S., Walenski, M., & Janacsek, K. (2020). The neurocognition of developmental disorders
of language. Annual Review of Psychology, 71, 389–417.
This position and review paper on developmental language disorders also contains a recent in-depth exposi-
tion of the memory systems, focusing on procedural memory and how it interacts with declarative memory.
331
References
Albright, A., & Hayes, B. (2003). Rules vs. analogy in English past tenses: A computational/experimental study.
Cognition, 90(2), 119–161.
Anderson, J. R., Betts, S., Bothell, D., Hope, R., & Lebiere, C. (2019). Learning rapid and precise skills. Psy-
chological Review, 126(5), 727–760.
Babcock, L., Stowe, J. C., Maloof, C. J., Brovetto, C., & Ullman, M. T. (2012). The storage and composition
of infected forms in adult-learned second language: A study of the infuence of length of residence, age of
arrival, sex, and other factors. Bilingualism: Language and Cognition, 15(4), 820–840.
Bowden, H. W., Gelfand, M. P., Sanz, C., & Ullman, M. T. (2010). Verbal infectional morphology in L1
and L2 Spanish: A frequency efects study examining storage versus composition. Language Learning, 60(1),
44–87.
Bufngton, J., Demos, A., & Morgan-Short, K. (2021). The reliability and validity of procedural memory assess-
ments used in second language acquisition research. Studies in Second Language Acquisition, 43(3), 635–662.
DeKeyser, R. M. (1997). Beyond explicit rule learning: Automatizing second language morphosyntax. Studies
DeKeyser, R. M. (2020). Skill acquisition theory. In B. VanPatten, G. D. Keating, & S. Wulf (Eds.), Theories in
second language acquisition (3rd ed., pp. 83–104). Routledge.
Eichenbaum, H., Otto, T., & Cohen, N. J. (1994). Two functional components of the hippocampal memory
system. Behavioral and Brain Sciences, 17(3), 449–472.
Faretta-Stutenberg, M., & Morgan-Short, K. (2018). Contributions of initial profciency and language use to
second-language development during study abroad: Behavioral and event-related potential evidence. In C.
Sanz & A. Morales-Font (Eds.), Handbook of study abroad research (pp. 421–435). Routledge.
Feldman, L., Kostić, A., Basnight-Brown, D., Durdtević, D. F., & Pastizzo, M. J. (2010). Morphological facili-
tation for regular and irregular verb formations in native and non-native speakers: Little evidence for two
distinct mechanisms. Bilingualism: Language and Cognition, 13(2), 119–135.
Fernandez, G. S. E., Klaver, P., Fell, J., Grunwald, T., & Elger, C. E. (2002). Human declarative memory forma-
tion: Segregating rhinal and hippocampal contributions. Hippocampus, 12(4), 514–519.
Godfroid, A. (2023). Hypotheses about the interface between explicit and implicit knowledge in second language
acquisition. In A. Godfroid & H. Hopp (Eds.), The Routledge handbook of second language acquisition and psycho-
linguistics (pp. 294–307). Routledge.
Gor, K. (2023). Word and multi-word processing: Memory-based and linguistic approaches. In A. Godfroid & H.
Hamrick, P., Lum, J. A. G., & Ullman, M. T. (2018). Child frst language and adult second language are both
tied to general-purpose learning systems. Proceedings of the National Academy of Sciences of the United States of
America, 115(7), 1487–1492.
Hartshorne, J. K., & Ullman, M. T. (2006). Why girls say “holded” more than boys. Developmental Science, 9(1),
21–32.
Hopp, H. (2023). Sentence processing in a second language: Linguistic approaches. In A. Godfroid & H. Hopp
Issa, B. I., & Morgan-Short, K. (2019). Efects of external and internal attentional manipulations on second lan-
guage grammar development: An eye-tracking study. Studies in Second Language Acquisition, 41(2), 389–417.
Jacob, G., Flieschhauer, E., & Clashen, H. (2013). Allomorphy and afxation in morphological processing: A
cross-modal priming study with late bilinguals. Bilingualism: Language and Cognition, 16(4), 924–933.
Jacob, G., Heyer, V., & Veríssimo, J. (2018). Aiming at the same target: A masked priming study directly com-
paring derivation and infection in the second language. International Journal of Bilingualism, 22(6), 619–637.
Johari, K., Walenski, M., Reifegerste, J., Ashraf, F., & Ullman, M. T. (2019). Sex, dopamine, and hypokinesia:
A study of infectional morphology in parkinson’s disease. Neuropsychology, 33(4), 508–522.
Karatas, N. B., Özemir, O., Lovelett, J. T., Demir, B., Erkol, K., Veríssimo, J., Erçetin, G., & Ullman, M. T. (2021).
Improving second language vocabulary learning and retention by leveraging memory enhancement techniques: A
multidomain pedagogical approach. Language Teaching Research. Advance online publication.
Klaver, P., Fell, J., Dietl, T., Schür, S., Schaller, C., Elger, C. E., & Fernández, G. (2005). Word imageability
afects the hippocampus in recognition memory. Hippocampus, 15(6), 704–712.
Li, S. (2023). Working memory and second language learning: A critical and synthetic review. In A. Godfroid & H.
Lieberman, M. D., Chang, G. Y., Chiao, J., Bookheimer, S. Y., & Knowlton, B. J. (2004). An event-related
fMRI study of artifcial grammar learning in a balanced chunk strength design. Journal of Cognitive Neurosci-
ence, 16(3), 427–438.
332
Maie, R., & Godfroid, A. (2022). Controlled and automatic processing in the acceptability judgment task: An
eye-tracking study. Language Learning, 72(1) 158-197.
Marslen-Wilson, W. D. (2007). Morphological processes in language comprehension. In M. G. Gaskell (Ed.),
The Oxford handbook of psycholinguistics (1st ed., pp. 495–524). Oxford University Press.
Morgan-Short, K. (2020). Insights into the neural mechanisms of becoming bilingual: A brief synthesis of sec-
ond language research with artifcial linguistic systems. Bilingualism: Language and Cognition, 23(1), 87–91.
research to issues in explicit and implicit second language acquisition. In P. Rebuschat (Ed.), Explicit and
implicit learning of languages (1st ed., pp. 349–384). John Benjamins.
Morgan-Short, K., Hamrick, P., & Ullman, M. T. (2022). Declarative and procedural memory as predictors of
second language development. In S. Li, P. Hiver, & M. Papi (Eds.), The Routledge handbook of second language
acquisition and individual diferences (pp. 67–81). Routledge.
Morgan-Short, K., Steinhauer, K., Sanz, C., & Ullman, M. T. (2012). Explicit and implicit second language
training diferentially afect the achievement of native-like brain activation patterns. Journal of Cognitive Neu-
roscience, 24(4), 933–947.
Paradis, M. (1994). Neurolinguistic aspects of implicit and explicit memory: Implications for bilingualism and
SLA. In N. C. Ellis (Ed.), Implicit and explicit learning of languages (pp. 393–419). Academic Press.
Paradis, M. (2009). Declarative and procedural determinants of second languages. John Benjamins Publishing Company.
Pili-Moss, D., Brill-Schuetz, K. A., Faretta-Stutenberg, M., & Morgan-Short, K. (2020). Contributions of
declarative and procedural memory to accuracy and automatization during second language practice. Bilin-
Pinker, S. (1991). Rules of language. Science, 253(5019), 530–535.
Prado, E. L., & Ullman, M. T. (2009). Can imageability help us draw the line between storage and composition?
Journal of Experimental Psychology: Learning, Memory, and Cognition, 35(4), 849–866.
Prasada, S., & Pinker, S. (1993). Generalisation of regular and irregular morphological patterns. Language and
Cognitive Processes, 8(1), 1–56.
Reifegerste, J., Veríssimo, J., Rugg, M. D., Pullman, M. Y., Babcock, L., Glei, D. A., Weinstein, M., Goldman,
N., & Ullman, M. T. (2021). Early-life education may help bolster declarative memory in old age, especially
for women. Aging, Neuropsychology, and Cognition, 28(2), 218–252.
Reifergerste, J., Walenski, M., Prado, E. L., Steinhauer, K., & Ullman, M. T. (in prep.). Leveraging working
memory limitations to test for the composition of complex forms in language.
Squire, L. R., & Dede, A. J. O. (2015). Conscious and unconscious memory systems. Cold Spring Harbor Perspec-
tives in Biology, 7, a021667.
Suzuki, Y. (this volume). Automatization and practice. In A. Godfroid & H. Hopp (Eds.), The Routledge handbook
of second language acquisition and psycholinguistics. Routledge.
procedural model. Bilingualism: Language and Cognition, 4(2), 105–122.
Ullman, M. T. (2004). Contributions of memory circuits to language: The declarative/procedural model. Cogni-
tion, 92, 231–270.
Ullman, M. T. (2016). The declarative/procedural model: A neurobiological model of language learn-
ing, knowledge, and use. In G. Hickok & S. L. Small (Eds.), Neurobiology of language (pp. 953–968).
Elsevier Inc.
Ullman, M. T. (2020). The declarative/procedural model. In B. VanPatten, G. D. Keating, & S. Wulf (Eds.),
Theories in second language acquisition (3rd ed., pp. 128–161). Routledge.
Ullman, M. T., Corkin, S., Coppola, M., Hickok, G., Growdon, J. H., Koroshetz, W. J., & Pinker, S. (1997). A
neural dissociation within language: Evidence that the mental dictionary is part of declarative memory, and
that grammatical rules are processed by the procedural system. Journal of Cognitive Neuroscience, 9(2), 266–276.
Ullman, M. T., Corkin, S., Pinker, S., Coppola, M., Locascio, J., & Growdon, J. H. (1993). Neural modularity in
language: Evidence from Alzheimer’s and Parkinson’s diseases. 23rd Annual Meeting of the Society for Neurosci-
ence, Washington, DC.
Ullman, M. T., Earle, F. S., Walenski, M., & Janacsek, K. (2020). The neurocognition of developmental disorders
of language. Annual Review of Psychology, 71(1), 389–417.
Ullman, M. T., & Lovelett, J. T. (2018). Implications of the declarative/procedural model for improving second
language learning: The role of memory enhancement techniques. Second Language Research, 34(1), 39–65.
Ullman, M. T., Walenski, M., Prado, E. L., Ozawa, K., & Steinhauer, K. (2001). A new method for investigating the
real-time composition of complex linguistic forms. Proceedings of the 14th Annual CUNY Conference on Human
Sentence Processing.
333
Veríssimo, J., Heyer, V., Jacob, G., & Clahsen, H. (2018). Selective efects of age of acquisition on morphological
priming: Evidence for a sensitive period. Language Acquisition, 25(3), 315–326.
Wilhelm, I., Diekelmann, S., & Born, J. (2008). Sleep in children improves memory performance on declarative
but not procedural tasks. Learning & Memory (Cold Spring Harbor, N.Y.), 15(5), 373–377.
Williams, J. N., & Rebuschat, P. (2023). The psycholinguistics of L2 learning: A cognitive psychology perspec-
tive. In A. Godfroid & H. Hopp (Eds.), The Routledge handbook of second language acquisition and psycholinguistics
334
27
OF SECOND LANGUAGE
INTERACTION
Michael H. Long
Introduction
Passive exposure to speech intended for native speakers (NSs) produces little or no second language
(L2) development by elementary-level non-native speakers (NNSs) of a new language. Such input
is not comprehensible to them, so not processable, so not usable for learning. For input to become
intake, learners and their interlocutors must negotiate for meaning, i.e., do the interactional work
necessary to achieve mutual intelligibility (see Textbox 27.1). Work on NS-NNS conversation since
the mid-1970s has identifed numerous devices used to improve comprehensibility, as well as ways
that these devices can facilitate, or in some cases impede, acquisition. Current research issues include
relationships among (i) interactional modifcations to the structure of conversation, (ii) their efects
on linguistic salience of target structures and the kinds of input, output, and negative feedback learn-
ers experience, and (iii) cognitive processes that communicative L2 interaction induces in the learner:
incidental, explicit, and implicit learning, noticing, and detection. This chapter surveys these strands
of research in historical and contemporary perspective and sketches future directions.
Negotiation for meaning: Attempts to make meaning comprehensible following a breakdown in

communication.
Input elaboration: The modifcation of input by means of a wide variety of compensatory devices to
improve comprehensibility of what is said.
Noticing: Registration of new forms or form-meaning-function associations with conscious attention.
Detection: The registration of a stimulus without awareness.
Recast: A reformulation of all or part of a learner’s immediately preceding utterance in which one or
more non-target-like (phonological, lexical, grammatical, etc.) items are replaced by the correspond-
ing target language versions, and where, throughout, the interlocutors’ focus is on meaning, not
language as an object.
DOI: 10.4324/9781003018872-31 335

Michael H. Long
Input and interaction

Most second language acquisition (SLA) research in the 1960s and 1970s focused on the analysis
of elicited or spontaneous speech samples from child and adult learners, but it ignored the input
and thus glossed over potential explanations for acquisition, such as item frequency and perceptual
salience. Hatch (1978) called for SLA researchers to include “the other side of the page.”
[L]anguage learning evolves out of learning how to carry on conversations. . . . One learns
how to do conversation, one learns how to interact verbally, and out of this interaction
syntactic structures are developed.
(Hatch, 1978, p. 404)
Motivated by Hatch’s claim, a series of mostly laboratory and classroom studies through the 1980s
compared NS-NS, NS-NNS, and NNS-NNS conversation, documenting modifcations speakers
made when addressing (usually low-profciency) non-natives. An early laboratory study (Long, 1980)
examined transcripts and recordings of randomly formed dyads of 16 NS-NS and 16 NS-NNS who
performed six tasks. Aside from containing shorter utterances, linguistic input to NNSs did not dif-
fer much from input directed at NSs. Rather, through the task-induced need to negotiate for meaning,
which was often, but not only, triggered by a breakdown in communication, it was the interactional
structure of conversation that distinguished the conversations with NNSs. For example, NSs might talk
about the same things, but use a slower speech rate to do so with NNSs, and employ a variety of
devices to make topic shifts salient. They would often deliberately relinquish topic control, accept
unintentional topic-switches by NNSs, and allow them to say less about a topic than would be
expected of NSs before moving on to the next topic. The typical result was input that was not simpli-
fed, but elaborated by the more competent speaker, thereby remaining comprehensible for the learner
(Long, 1983a). Interactional modifcations were signifcantly more frequent in two-way than one-
way tasks, for instance more frequent in a two-way spot-the-diference task than a movie-description
task, and therefore highlighted some of the interactional benefts of a bidirectional exchange of
information between speakers.
There were potential implications for language teaching. Simplifed input, typical in commer-
cially published pedagogic materials, improves comprehensibility, but it is counter-productive for
acquisition because it entails removal of many unfamiliar items to which learners need exposure.
When talking to NNSs in real life, NSs tend to simplify the task, not the input. By adding redun-
dancy, they elaborate the input to compensate for linguistic complexity. In the best-case scenario,
learners are wrapped in a protective linguistic cocoon, rich in language-learning potential. By modi-
fying the interactional structure of conversation through negotiation for meaning, it was assumed that
interlocutors would achieve the same, or nearly the same, level of comprehensibility without sacrifc-
ing unknown items. All this prompted the frst (overly strong) version of the interaction hypothesis.
[P]articipation in conversation with NSs, made possible through the modifcation of inter-
action, is the necessary and sufcient condition for SLA.
(Long, 1981, p. 275)
Studies of L2 conversation gathered momentum in the 1980s (e.g., Chaudron, 1982; Gass & Varonis,
1985; Pica & Doughty, 1985; Pica et al., 1986). The research showed that NSs drew on a range of
devices that provided learners with extra opportunities to perceive and process problematic items.
They employed synonyms, lexical switches, appositional phrases, and diferent kinds of repetition.
336
The psycholinguistics of second language interaction
They preferred yes/no and or-choice questions over wh-questions, and through a process termed
“decomposition,” changed subject-predicate to topic-comment constructions (1).
(1) NS: Uh what does uh what does your father do in uh you’re from Kyoto, right?
NNS: Yeah.
NS: Yeah. What does your father do in Kyoto?
NSs monitored the learners’ and their own comprehension with liberal use of clarifcation requests,
comprehension checks, and confrmation checks. The devices were used retroactively, as tactics to
repair communicative trouble when things went wrong, or proactively, as strategies to prevent trouble
from occurring in the frst place, or both (Long, 1983a). For acquisition, interactional modifcations
provided learners with access to samples of the L2 modifed for them individually and with input at
least roughly tuned to their current stage of development that was comprehensible and processable
and thus usable for acquisition (Gass, 1997; Goo, 2019; Loewen & Sato, 2018; Long, 1983b; Long
& Porter, 1985; Mackey, 2012).
Output: scaffolding and pushed output

While most early research focused on the role of interactional modifcations in increasing input
comprehensibility and learning potential, another interest was the output opportunities interaction
provides. Firstly, learner output may require assistance in the form of conversational scafolding, i.e., the
joint construction of meaning across turns and speakers. In a longitudinal study of naturalistic L2
acquisition by two Vietnamese brothers, Tai and Thanh, aged 10 and 12, Sato (1988, 1990) showed
that scafolding can be especially valuable for facilitating initial forays into complex syntax (e.g., rela-
tive clauses), fostering development of what she termed “syntactic precursors” across turns. Such
scafolding was observed both in NS-NNS (2) and NNS-NNS (3) interactions.
(2) Thanh: In Vietnam they play cards

NS: They what?
Thanh: Play cards
NS: They play cards?
Thanh: Yeah, when Christmas
(Sato, 1990, p. 127)
(3) Tai: He looking um um

Thanh: at man
Tai: at man hi hi smokin
(Sato, 1990, p. 128)
Similarly, Bygate (1988) described how pairs of adult classroom learners working on problem-solving
tasks used intra-turn repairs and cooperative dialogue to build syntactically more complex freestand-
ing utterances out of shorter, syntactically dependent “satellite units.” Other work demonstrated how
an interlocutor’s use of a syntactic construction can prime learners to use the same construction in a
subsequent turn (e.g., McDonough, 2006).
Production also facilitates development by “pushing” output to meet communicative needs. The
output hypothesis (Swain, 1995) proposed that speech production obliged learners to shift from a
focus on decoding meaning, which they could often achieve with far less than native-like compe-
tence, to a more complex task: attending to grammatical structure to encode meanings of their own
(for some supportive fndings, see Loewen [2005] and Egi [2010]). However, Shehadeh (2002) and
337
Michael H. Long
Yilmaz (2016a) suggest there is little unambiguous empirical evidence for a relationship between
modifed output and acquisition. At a minimum, a defensible version of the output hypothesis seems
to be that spoken production is useful because it elicits negative feedback and encourages grammati-
cization; hence it is facilitative, but not necessary.
Negative feedback and recasts

Negotiated interaction typically includes instances of negative feedback (NF). Because of the primary
focus on communication, the corrective function of NF is likely to be incidental and implicit, usually
in the form of recasts (4; for a defnition, see Textbox 27.1).
(4) S: It means I am not familiar about the Jazz?

T: No, it doesn’t mean that you’re not familiar with Jazz. It means you’re familiar with Jazz, but
don’t like it. You tried that, but you don’t like it.
S: Oh, I see.
(Kim & Han, 2007, p. 281)
Recasts have many advantages, not least, as in the examples, that unlike forms of explicit NF such
as prompts and elicits, they include positive evidence concerning the correct version(s). Because
recasts are immediately contingent on what the learner has just said, they facilitate cognitive com-
parison (Saxton, 1997) of their output and the input (i.e., the recast) that their output triggered. This
potentially results in uptake of linguistic information contained in the NS’s recast. Being unobtru-
sive, moreover, recasts do not disrupt communicative L2 interaction the way explicit NF does. The
efectiveness of recasts in promoting L2 development is well established (for reviews, see Goo, 2019;
Loewen & Sato, 2018).
Two meta-analyses (Li, 2010; Lyster & Saito, 2010) found a trend for explicit feedback and
prompts to be more immediately benefcial than recasts in teacher-student interaction, and Lyster
and Ranta (1997) used immediate uptake to measure the efectiveness of diferent kinds of NF. It is a
mistake, however, to judge whether any kind of NF, implicit or explicit, succeeded or failed simply
by the presence or absence of immediate uptake. This is because most explicit NF (e.g., prompts,
elicits, and metalinguistic comments) leave classroom learners with no alternative but to attempt an
improved efort (i.e., uptake), even if it is merely echoic, whereas recasts do not require production.
Interlocutors may simply “move on,” so the next speaking turn may not fall to the learner (Oliver,
1995). Eye-tracking studies of computer chat show that learners often notice recasts and the informa-
tion they contain even when the behavioral record suggests otherwise (Smith, 2012).
The fndings of a statistical meta-analysis of 33 studies (Li, 2010, p. 343) suggested that, whereas
the immediate and short-term efect of explicit feedback was greater, the longer-term efect size for
recasts was slightly larger than the short-term efect, more efective than explicit feedback on long-
delayed post-tests, and more enduring, even increasing over time. Thus, the short-term efects of
explicit feedback tend to diminish over time, whereas the value of implicit NF tends to increase (Li,
2010; Mackey & Goo, 2007). Irrespective of whether or not the turn that follows recasts provides an
immediate opportunity for learners to produce the correct version, the tailored input recasts provide
results in learning and improved performance (Goo & Mackey, 2013; Mackey, 2012; Yilmaz, 2016a).
The suggested long-term efect of recasts is consistent with the fnding of another statistical meta-
analysis (Kang et al., 2019) which showed greater durability of learning from implicit instruction in
general. Such results are important, given the increasingly recognized value of implicit L2 knowledge
(see Section “Salience, noticing, and detection”).
Given this understanding, immediate uptake is not a valid indicator of greater efectiveness, and
the production opportunities typical of explicit feedback constitute a confound in several studies
338
purporting to compare learning from explicit NF and recasts (Goo & Mackey, 2013; Long, 2007;
Yilmaz, 2016a).
From a psycholinguistic perspective, recasts have seven potential advantages over models that pres-
ent correct exemplars in isolation (Long et al., 1998). Recasts convey needed information about the
target language (i) in context, (ii) when listeners and speakers share a joint attentional focus, (iii) when
the learner is vested in the exchange, (iv) and so is probably motivated, and (v) attending. In addition,
because the reformulation is of what learners just tried to say, they already have prior comprehension of
at least part of the message that the recast contains. This, in turn, (vi) frees up attentional resources and
facilitates form-function mapping. Indirect support for the role of prior comprehension may come
from the fndings of a study of the value of content familiarity. Revesz and Han (2006) showed that
learners who received recasts of errors they made with the past progressive while performing the
same narrative task (describing the same crime scene) three times improved more in oral and written
production than learners who received the recasts while describing three diferent crime scenes once
each. Finally, and crucially, (vii) the contingency of recasts on deviant output means that incorrect and
correct utterances are juxtaposed, allowing learners briefy to hold and compare the two versions in
working memory. That is an opportunity not presented by non-contingent utterances, like models or
instances of the same items or structures in ambient input, which are not triggered by learners’ own
production. The comparison can lead learners to focus their attention on language form, in context,
and consciously “notice the gap” (Schmidt, 2010) between their deviant utterance and the native-like
version in the input that follows.
Effects of input and interactional modifcations

on comprehension and acquisition
Experimental and quasi-experimental studies in the 1980s and 1990s set out to assess the efects of
various kinds of spoken and written input and interactional modifcations on L2 comprehension
and acquisition. Pica et al. (1987), for example, compared the comprehensibility of spoken direc-
tions for 16 NNSs of English under two conditions, pre-modifed and interactionally modifed. In
the pre-modifed condition, directions were less syntactically complex, but contained more lan-
guage and more redundancy. In the interactionally modifed condition, there were no changes to
the input, but instead, opportunities for interaction with the NS giving the directions. Their results
confrmed observations in earlier descriptive work. Comprehension and confrmation checks in
the interactional condition served to trigger repetition and rephrasing, increasing comprehensibility,
whereas lower syntactic complexity (i.e., input modifcation alone) had no efect. Similar fndings
were obtained in a classroom study by Ellis et al. (1994), in which Japanese high school students
learned more English vocabulary from interactionally modifed than pre-modifed input. A labora-
tory experiment by Yilmaz (2016b) found that beginning learners of Turkish who received explicit
NF on their errors while interacting with a NS during three communication games performed bet-
ter on the more salient of two target morphological structures than a group that only observed the
interaction.
The advantage of interactionally modifed input may be attributable to two reasons: (i) the oppor-
tunity for interaction means negotiation for meaning can take place, thereby rendering the input
appropriate for the individual learner, which is less likely with pre-modifed input, and (ii) oppor-
tunities for negotiation work provide the learner with more time to process the input, which can be
minimal if pre-modifed input is presented orally. An important laboratory study by Mackey (1999)
showed that some learners (termed “developmentally ready”) who interacted with an NS one on one
improved their command of English wh-questions from pre- to post-test, and improved more than
two other groups, one of which received pre-modifed input, while the other simply listened to the
frst group’s interactions. These fndings suggest that it is not so much the input alone that matters;
339
Michael H. Long
it is learners’ engagement with it through negotiation work, including benefts received from the
implicit NF the negotiation work elicits, mostly in the form of recasts.
Where written input is concerned, processing time is available if the reading task is unspeeded.
Under these conditions, even pre-modifed, elaborated versions of texts have usually been found as
comprehensible or almost as comprehensible for learners as linguistically simplifed ones, and both of
them are statistically signifcantly more comprehensible than the original versions produced by and
for NSs (e.g., Long, 2015, pp. 250–259, 2020; Oh, 2001; Yano et al., 1994).
The discussion so far has dealt with the relationship between L2 interaction and comprehension.
The relationship between L2 interaction and acquisition initially proved a harder nut to crack. Sato’s
study, reviewed in the section titled “Output: scafolding and push output,” provided the frst empiri-
cal test of Hatch’s position that interaction drives language acquisition. One focus was the brothers’
past time reference. Hatch (1983) had cautioned that some aspects of conversation might actually
inhibit learning, pointing out, for example, that “[m]istakes in the marking of verbs . . . would not
be caught by When questions, which would be more likely to elicit a time adverb rather than a verb
correction for morphology” (Hatch, 1983, p. 432). This proved to be the case. Using a function-to-
form analysis, Sato documented a series of four developmental stages. In stage 1, learners rely on their
interlocutors to establish past time reference, which allows them to say something about a past event
despite using only the base form of the verb (5).
(5) NS: Yeah, tell me about your goal. How did you make it? What was happening?
Than: [My] left foot (slaps left leg) I shot [with my] left foot
(Sato, 1990, p. 126)
That is, they use the scafolding that L2 interaction provides. In stage 2, they exploit their listener’s
background knowledge to interpret what they say as referring to past time (In Vietnam, he farmer).
In stage 3, they employ temporal adverbials for the purpose (Yesterday, I shopping). Finally, in stage 4,
morphological tense markers appear (I saw Mary yesterday. He played well). While conversational
scafolding opened the door for Thanh and Tai’s earliest attempts at complex syntax, the assistance
worked against the boys’ development of morphological marking, precisely because it obviated the
need for them to establish temporal reference. Neither boy ended the frst ten-month period of
observation using infectional morphology for the purpose. Sato (1986, 1990) concluded that con-
versation was selectively facilitative when it came to the emergence of L2 grammar.
Sato’s study revealed a potential limitation of learning through L2 interaction. Because speakers’ pri-
mary goal is to send and receive comprehensible messages, learners may ignore, or at least pay less atten-
tion to, less salient features in the input that do not play an essential role in getting meaning across. The
situation can be compounded by the L1, e.g., in cases like Vietnamese, whose open syllable structure,
Sato pointed out, hinders production of word-fnal consonants or consonant clusters for infectional
morphology, or Mandarin, whose NSs have become accustomed to attending to adverbs, due to their
importance as cues to temporal reference in their L1. The L1 can result in “blocking” their attention to
infectional morphology in morphologically rich languages as a part of “learned attention” (Cintrón-
Valentín & Ellis, 2015; see also Sagarra, 2023 [this volume]). It is not enough simply that such items
occur in the input; it is when learners have to negotiate for meaning that their attention is more likely to
be drawn to them, increasing the chances they will be perceived, whether consciously or unconsciously.
The updated version of the interaction hypothesis was an attempt to capture these relationships.
Negotiation for meaning, and especially negotiation that triggers interactional adjustments by
the native speaker or more competent interlocutor, facilitates acquisition because it connects
input, internal learner capacities, particularly selective attention, and output in productive ways.
(Long, 1996, p. 451–452)
340
While broader in scope, the updated version was weaker. Interaction was viewed as facilitative,
no longer as necessary and sufcient. The updated interaction hypothesis emphasized the poten-
tial for the NF that L2 interaction provides to draw learners’ attention to problematic forms,
especially those of low saliency, so that “negative feedback obtained during negotiation work
or elsewhere may be facilitative of L2 development, at least for vocabulary, morphology and
language-specifc syntax, and essential for learning certain specifable L1—L2 contrasts” (Long,
1996, p. 414).
Such benefts of interaction are not limited to NF. Negotiation for meaning involves higher-
than-usual frequencies of semantically contingent speech, including repetitions, reformulations, and
expansions, sometimes functioning simultaneously as recasts, but more generally as positive evidence.
Problematic forms are recycled in the process, increasing their salience and the likelihood that they
will be perceived by the learner. Mackey and Goo summarized their meta-analysis of 28 studies as
follows.
Interaction plays a strong facilitative role in the learning of lexical and grammatical target
items. The 28 interaction studies qualifed for the present meta-analysis showed large mean
efect sizes across immediate and delayed post-tests, providing evidence of short-term as well
as longer-term efects on language acquisition. . . . Another interesting fnding was that
the delayed efects of interaction may be even stronger than immediate efects, supporting
claims to this efect made by some researchers in the past.
(Mackey & Goo, 2007, pp. 438)
Eight years later, Gass and Mackey (2015, p. 181) concluded there existed “a robust connection
between interaction and learning.” Supportive studies and meta-analyses appear to this day (Goo,
2019; Loewen & Sato, 2018; Mackey, 2012; Yilmaz, 2016a; Ziegler, 2016).
A cognitive-interactionist theory of ISLA

The interaction hypothesis predicts that L2 interaction helps compensate for the adult’s language-
learning difculties. In my view, there are two main obstacles to native-like attainment in adult L2
acquisition. The frst, developmental sharpening (Cutler, 2001), stems, ironically, from the success of L1
acquisition, during which processing becomes optimized for the native language—a phenomenon
known as “learned attention” (Cintrón-Valentín & Ellis, 2015; N. Ellis, 2006). As a result, processing
is no longer tuned appropriately for a new language. An adult Spanish-speaking learner of English,
for example, “flters” two sounds, /ɪ/ and /i:/, in the L2 input through the L1-established phonetic
grid to the L1 attractor /i:/ (the single sound in Spanish, where English has two), blocking a focus on
the new /ɪ/–/i:/ distinction. The second factor, maturational constraints on language learning (Abra-
hamsson & Hyltenstam, 2009), includes a decline in the capacity for implicit learning, especially
instance learning of the thousands of arbitrary form—meaning associations a language contains, for
instance the implicit learning of vocabulary or collocations. The decline in implicit learning capacity
is compounded by the need for extensive time and L2 exposure for implicit learning—resources few
adult learners enjoy.1
As one component in a cognitive-interactionist theory of instructed SLA (Long, 2015, pp. 30–62;
Long, 2017), the interaction hypothesis predicts that L2 interaction helps compensate for these adult
language-learning difculties. In addition to scafolding, pushed output, and NF, negotiation for
meaning adds perceptual salience to problematic features and provides learning opportunities not
available from passive exposure to L2 input alone.
341
Michael H. Long
Critical issues and current work
Salience, noticing, and detection

L2 interaction provides opportunities for learners focused on meaning to pick up a new language
incidentally, as an unintended by-product of doing something else—communicating through the
L2—and often also implicitly, i.e., without awareness. Interacting in the L2 while focused on what is
said, learners sometimes perceive new forms or form-meaning-function associations consciously—a
process referred to as noticing (Schmidt, 1990, 2010). On other occasions, their focus on commu-
nication and attention to the task at hand is such that they will perceive new items in the input
unconsciously—a process known as detection (Tomlin & Villa, 1994).
While communicating through the L2, a learner’s attention is likely drawn to problematic items
by added salience resulting from typical characteristics of meaning-focused exchanges. Such salience
may facilitate noticing or detection. For instance, interlocutors will consciously or unconsciously
highlight important items, e.g., by pausing briefy before and/or after them, adding stress, repeating
them, providing synonyms and informal defnitions, moving them to more salient initial or fnal
positions in an utterance through left-dislocation or decomposition, and through comprehension
checks, confrmation checks, and clarifcation requests, all triggering a momentary switch of the
learner’s focal attention from meaning to linguistic form. In addition, NF—mostly implicit, mostly
by way of recasts—can have the same efect, while simultaneously providing whatever positive evi-
dence is needed, whether a missing item or a model of more target-like usage.
The same incidental learning process operates when learners read a book or a newspaper, listen to
a radio interview, or watch a movie. Whereas those activities involve attempting to understand and
learn from static spoken or written input intended for NSs over which learners have no control, face-
to-face L2 interaction is dynamic, ofering opportunities to negotiate for meaning. The negotiation
work increases the likelihood that salience will be added by the interlocutor, attention drawn to items
uniquely problematic for them, and communicative trouble repaired.
Conscious noticing of those items, which involves perception with awareness but without under-
standing (Schmidt, 2010), is explicit and may result in explicit knowledge. Unconscious detection,
registration without awareness, is not only incidental but also implicit, and absent some consciousness-
raising event, may end up as implicit knowledge. The two systems are stored in diferent parts of the
brain, and there is no evidence that explicit knowledge can “become” implicit through automatiza-
tion (see Godfroid, 2023 [this volume]), which would entail a loss of knowledge.
Since implicit knowledge is the ideal basis for fast, fuent, and efortless listening and speaking,
detection and implicit learning are highly desirable outcomes. As Whong et al. (2014, pp. 21–22)
point out, implicit learning is more basic and more important than explicit learning, and supe-
rior. Perhaps because of the greater depth of processing required, implicit learning has the further
advantage of being more durable. It obviates the need for explicit knowledge, freeing up attentional
resources for a speaker to focus on message content. Explicit treatments make form-meaning associa-
tions obvious, and thus learnable via shallower processing. While explicit treatments may facilitate
noticing, they potentially work against learners in the long run by providing items “on a plate,” ready
to be consumed with less efort on their part. Implicit learning, on the other hand, requires more
input and time, and may remain the default learning mechanism for adults, since learning a new lan-
guage is far too large and too complex a task to be handled explicitly (Ellis & Wulf, 2015).
There are problems, however. The scope of implicit learning may not extend to phenomena,
such as anaphora, that involve non-adjacent items (Hawkins was arrested, and so were several members of
his gang), which may no longer be learnable that way (Williams, 2009, p. 339). The decline in the
capacity of implicit learning is marked especially by an age-related deterioration in the efciency of
instance learning (Long, 2017), starting around age 12, the same period during which the capacity for
342
explicit, probabilistic sequence learning appears to increase. Instance learning is particularly impor-
tant for the detection of arbitrary, non-rule-governed form-meaning-function associations, such
as gender marking, lexical items, collocations, and other multiword units, and implicit learning is
more important generally for recognizing input strings as exemplars of common constructions. The
frequency with which such items are encountered will afect the rate at which they are perceived,
and then the mostly implicit processes of pattern recognition, abstraction, category formation, and
generalization follow. Acquisition will be more difcult and slower when L2 constructions are more
marked (i.e., occur in fewer of the world’s languages) than the L1 equivalent is (Eckman, 1977) and
when L2 input is “non-robust,” i.e., when relationships among L2 forms, meanings, and functions
are characterized by low frequency and high variability (Han & Lew, 2012). However, even when
exemplars are frequent and consistent, i.e., the input is “robust,” implicit learning requires time and
rich data of the sort L2 interaction can provide.
In sum, while L2 interaction can ofer plentiful opportunities for incidental and implicit learning
to operate, stimulation will be needed in some areas. For example, explicit interventions of various
kinds can kick-start implicit processing (Cintrón-Valentín & Ellis, 2015). According to the tallying
hypothesis (Ellis, 2002), ensuring that learners’ attention is drawn to learning targets in an explicit
way, especially to abstract, perceptually non-salient items, can modify entrenched, automatic L1
processing routines. This, in turn, may alter the way additional L2 input is processed implicitly.
Through explicit attention, an initial representation is established in long-term memory and this
functions as a selective cue priming the learner to attend to and perceive subsequent instances when
processing implicitly. Ellis identifes what he calls “the general principle of explicit learning in SLA:
changing the cues that learners focus on in their language processing changes what their implicit
learning processes tune” (Ellis, 2005, p. 327). Research is currently underway to determine whether
it is possible to achieve the same results by unobtrusive, less interventionist means, termed enhanced
incidental learning (Long, 2017). For instance, in a recent study of L2 formulaic sequences, Borro
(2021) found that enhanced incidental learning conditions, implemented through unobtrusive aural
enhancement, resulted in signifcant knowledge gains. Moreover, eye-tracking and retrospective
verbal reports showed that the learning process involved an increased attention allocation to the tar-
get items, which, crucially, took place without awareness. Such results are promising for enhanced
incidental learning.
Open issues and future directions

The efectiveness of interaction in general, and of NF and recasts in particular, appears to vary with
type of target feature (Li, 2014). For example, more perceptually salient items (e.g., adverb place-
ment) appear to respond better to recasts than less salient ones (e.g., clitic pronouns in Spanish;
Long et al., 1998). In English, progressive -ing and third-person possessive pronoun morphemes
are more conducive to recasts than plural -s and past -ed (Ono & Witzel, 2002), and a more salient
(plural) morpheme is more susceptible than a less salient (locative) morpheme to both implicit and
explicit feedback in Turkish (Yilmaz, 2012). A hypothesis yet to be confrmed in future research
is that the less perceptually salient the linguistic target is, the more explicit the NF needs to be,
and vice versa.
Comparisons of explicit and implicit NF have produced mixed results, possibly because studies
have difered methodologically in so many ways (Yilmaz, 2016a). Given the tendency of various
kinds of feedback to co-occur with other potentially confounding variables, such as the presence of
output opportunities and a general focus on form or meaning, Goo and Mackey (2013) recommend
researchers isolate a single feedback type and investigate its optimal characteristics and the conditions
under which it is more or less efective (see Textbox 27.2). This approach has already produced useful
fndings with recasts (Loewen & Philp, 2006; Sheen, 2006).
343
Michael H. Long
Most work examining efects of IDs on learning through L2 interaction has concerned learner
profciency, developmental stage, aptitudes, and working memory (WM). The main focus has
been how IDs predict uptake from NF rather than learning from participation in L2 interaction
in general (but see Li, 2017, for a review with broader scope). It would be logical to predict that
higher WM would be advantageous both for detection and noticing (see Li, 2023 [this volume]).
A review by Goo (2019), however, identifed fndings for and against almost every hypothesized
role for WM—indeed, for every ID variable investigated so far—and such was the case with early
studies of (usually explicit) language aptitude in learning from NF. However, Yilmaz and Granena
(2016) showed that explicit language aptitude (ELA)—a new combination of explicit inductive
learning ability, explicit associative learning, and rote memory ability—predicted learning from
explicit, but not implicit, feedback on an immediate post-test. A statistical meta-analysis and nar-
rative review of 24 studies involving WM and language aptitudes by Li (2017) concluded that
aptitude appears more important than working memory. However, the interpretation of fndings
of primary studies was made hard by variability in the salience of instructional treatments and in
the instruments used to measure noticing. Another crucial measurement issue that may complicate
the interpretation of fndings is whether some aptitude sub-tests really measure aptitude or implicit
memory and the ability to apply existing knowledge. Implicit memory ability, Granena suggests,
may be more important for accessing and retrieving previously learned or known information
efortlessly than for learning new material (Granena, 2019a, b). These issues merit attention in
future studies.
A valuable methodological improvement in interactionist research would be to incorporate learn-
ers’ developmental stage, either as a control or a moderator variable. Otherwise, results are hard to
interpret because the efects of instruction may be confounded with those of developmental stage.
The few studies (e.g., Mackey, 1999) where developmental readiness has been manipulated have
shown how decisive it can be. In addition, research would be more cost-efective if linguistic tar-
gets were selected to exemplify types of features, preferably defned in psycholinguistic, rather than
language-dependent, terms, e.g., as salient or non-salient infectional morphology, or marked or
unmarked features. Such changes could make fndings more generalizable not only to other features,
but across languages.
Which features of recasts and uptake improve their efectiveness?

How does type of linguistic target moderate the efects of L2 interaction and type of NF on learning?
Which individual diferences mediate the benefts of L2 interaction, and how?
Which interactional modifcations best promote detection, noticing, and implicit learning?
Note
1 Many other factors are involved, of course, including indirect-causal state variables, like attitude and motiva-
tion, and direct-causal trait variables, notably, language aptitudes (Doughty, 2019; Granena, 2019a; Li, 2017).
Diferential achievement among individuals and within diferent linguistic systems also requires explanation,
perhaps via a combination of input sensitivity (a key component of aptitude which varies across individuals) and
perceptual saliency (which varies across structures; Long, 2003).
344
Further reading
Ellis, N. C., & Wulf, S. (2019). Cognitive approaches to L2 acquisition. In J. W. Schwieter & A. Benati (Eds.),
The Cambridge handbook of language learning (pp. 41–61). Cambridge University Press.
This chapter outlines usage-based accounts that inform cognitive approaches to L2 acquisition of morphol-
ogy, particularly focusing on salience, contingency, and learned attention, and further describes educational
interventions taking into consideration the efects of L2 learners’ attentional biases.
Plonsky, L., & Gass, S. M. (2011). Quantitative research methods, study quality and outcomes: The case of
interaction research. Language Learning, 61(2), 325–366.
This article assesses the methodological quality of quantitative studies within the tradition of L2 interac-
tion research and provides important suggestions to improve the methods and reporting practices for future
interaction-based research.
References
Abrahamsson, N., & Hyltenstam, K. (2009). Age of onset and nativelikeness in a second language: Listener
perception versus linguistic scrutiny. Language Learning, 59(2), 249–306.
Borro, I. (2021). Enhanced incidental learning of formulaic sequences by Chinese learners of Italian L2 [Ph.D. disserta-
tion], University of Portsmouth (UK).
Bygate, M. (1988). Units of oral expression and language learning in small group interaction. Applied Linguistics,
9(1), 59–82.
Chaudron, C. (1982). Vocabulary elaboration in teachers’ speech to L2 learners. Studies in Second Language
Acquisition, 4(2), 170–180.
Cintrón-Valentín, M., & Ellis, N. (2015). Exploring the interface: Explicit focus-on-form instruction and
learned attentional biases in L2 Latin. Studies in Second Language Acquisition, 37(2), 197–235.
Cutler, A. (2001). Listening to a second language through the ears of a frst. Interpreting, 5(1), 1–18.
Doughty, C. J. (2019). Cognitive language aptitude. Language Learning, 69, 101–126.
Eckman, F. (1977). Markedness and the contrastive analysis hypothesis. Language Learning, 27(2), 315–330.
Egi, T. (2010). Uptake, modifed output, and learner perceptions of recasts: Learner responses as language aware-
ness. The Modern Language Journal, 94(1), 1–21.
Ellis, N. C. (2002). Frequency efects in language acquisition: A review with implications for theories of implicit
and explicit language acquisition. Studies in Second Language Acquisition, 24(2), 143–188.
Ellis, N. C. (2005). At the interface: Dynamic interactions of explicit and implicit language knowledge. Studies
Ellis, N. C. (2006). Selective attention and transfer phenomena in L2 acquisition: Contingency, cue competition,
salience, interference, overshadowing, blocking, and perceptual learning. Applied Linguistics, 27(2), 164–194.
Ellis, N. C., & Wulf, S. (2015). Usage-based approaches to SLA. In B. VanPatten & J. Williams (Eds.), Theories
in second language acquisition. An introduction (2nd ed., pp. 75–93). Erlbaum.
Ellis, R., Tanaka, Y., & Yamazaki, A. (1994). Classroom interaction, comprehension, and the acquisition of L2
word meanings. Language Learning, 44(3), 449–491.
Gass, S. M. (1997). Input, interaction, and the development of second languages. Erlbaum.
Gass, S. M., & Mackey, A. (2015). Input, interaction, and output in second language acquisition In B. Van Patten
& J. Williams (Eds.), Theories in second language acquisition (2nd ed., pp. 180–206). Routledge.
Gass, S. M., & Varonis, E. (1985). Variation in native speaker speech modifcation to non-native speakers. Studies
Godfroid, A. (2023). Hypotheses about the interface between explicit and implicit knowledge in second language
acquisition. In A. Godfroid & H. Hopp (Eds.), The Routledge handbook of second language acquisition and psycholin-
guistics (pp. 294–307). Routledge.
Goo, J. M. (2019). Interaction in L2 learning. In J. W. Schwieter & A. Benati (Eds.), The Cambridge handbook of
language learning (pp. 233–257). Cambridge University Press.
Goo, J. M., & Mackey, A. (2013). The case against the case against recasts. Studies in Second Language Acquisition,
35(1), 127–165.
Granena, G. (2019a). Language aptitudes in L2 acquisition. In J. W. Schwieter & A. Benati (Eds.), The Cambridge
handbook of language learning (pp. 390–408). Cambridge University Press.
345
Michael H. Long
Granena, G. (2019b). Cognitive aptitudes and L2 speaking profciency. Studies in Second Language Acquisition,
41(2), 313–336.
Han, Z. H., & Lew, W. M. (2012). Acquisitional complexity: What defes complete acquisition in SLA. In B.
Szmrecsanyi & B. Kortmann (Eds.), Linguistic complexity in interlanguage varieties, L2 varieties, and contact lan-
guages (pp. 192–217). Walter de Gruyter.
Hatch, E. M. (1978). Discourse analysis and second language acquisition. In E. M. Hatch (Ed.), Second language
acquisition: A book of readings (pp. 402–35). Newbury House.
Hatch, E. M. (1983). Psycholinguistics: A second language perspective. Newbury House.
Kang, E. Y., Sok, S., & Han, Z. (2019). Thirty-fve years of ISLA on form-focused instruction: A meta-analysis.
Language Teaching Research, 23(4), 428–453.
Kim, J. H., & Han, Z. H. (2007). Recasts in communicative EFL classes: Do teacher intent and learner inter-
pretation overlap? In A. Mackey (Ed.), Conversational interaction in second language acquisition (pp. 269–300).
Li, S. (2010). The efectiveness of corrective feedback in SLA: A meta-analysis. Language Learning, 60(2), 309–365.
Li, S. (2014). The interface between feedback type, L2 profciency, and the nature of the linguistic target. Lan-
guage Teaching Research, 18(3), 373–396.
Li, S. (2017). The efects of cognitive aptitudes on the process and product of L2 interaction: A synthetic review.
In L. Gurzynski-Weiss (Ed.), Expanding individual diference research in the interaction approach: Investigating learn-
ers, instructors, and other interlocutors (pp. 41–70). John Benjamins.
Li, S. (2023). Working memory and second language learning: A critical and synthetic review. In A. Godfroid
Routledge.
Loewen, S. (2005). Incidental focus on form and second language learning. Studies in Second Language Acquisition,
27(3), 361–386.
Loewen, S., & Philp, J. (2006). Recasts in the adult English L2 classroom: Characteristics, explicitness, and efec-
tiveness. The Modern Language Journal, 90(4), 536–556.
Loewen, S., & Sato, M. (2018). Interaction and instructed second language acquisition. Language Teaching, 51(3),
285–329.
Long, M. H. (1980). Input, interaction, and second language acquisition [Unpublished Ph.D. dissertation], University
of California, Los Angeles.
Long, M. H. (1981). Input, interaction and second language acquisition. In Winitz, H. (ed.), Native language and
foreign language acquisition. Annals of the New York academy of sciences, 379, 259–278.
Long, M. H. (1983a). Native speaker/non-native speaker conversation and the negotiation of comprehensible
input. Applied Linguistics, 4(2), 126–141.
Long, M. H. (1983b). Linguistic and conversational adjustments to non-native speakers. Studies in Second Lan-
guage Acquisition, 5(2), 177–193.
Long, M. H. (1996). The role of the linguistic environment in second language acquisition. In W. C. Ritchie &
T. K. Bhatia (Eds.), Handbook of second language acquisition (pp. 413–468). Academic Press.
Long, M. H. (2003). Stabilization and fossilization in interlanguage development. In C. J. Doughty & M. H.
Long (Eds.), Handbook of second language acquisition (pp. 487–535). Blackwell.
Long, M. H. (2007). Recasts in SLA: The story so far. In M. H. Long (Ed.), Problems in SLA (pp. 75–116).
Lawrence Erlbaum.
Long, M. H. (2015). Second language acquisition and task-based language teaching. Wiley-Blackwell.
Long, M. H. (2017). Instructed second language acquisition (ISLA): Geopolitics, methodological issues, and
some major research questions. Instructed Second Language Acquisition, 1(1), 7–44.
Long, M. H. (2020). Optimal input for language learning: Genuine, simplifed, elaborated, or modifed elabo-
rated? Language Teaching, 53(2), 169–182.
Long, M. H., Inagaki, S., & Ortega, L. (1998). The role of implicit negative feedback in SLA: Models and recasts
in Japanese and Spanish. The Modern Language Journal, 82(3), 357–371.
Long, M. H., & Porter, P. (1985). Group work, interlanguage talk, and second language acquisition. TESOL
Quarterly, 19(2), 207–227.
Lyster, R., & Ranta, L. (1997). Corrective feedback and learner uptake: Negotiation of form in communicative
classrooms. Studies in Second Language Acquisition, 19(1), 37–66.
Lyster, R., & Saito, K. (2010). Oral feedback in classroom SLA: A meta-analysis. Studies in Second Language
Acquisition, 32(2), 265–302.
Mackey, A. (1999). Input, interaction and second language development. Studies in Second Language Acquisition,
21(4), 557–587.
Mackey, A. (2012). Input, interaction, and corrective feedback in L2 learning. Oxford University Press.
346
Mackey, A., & Goo, J. (2007). Interaction research in SLA: A meta-analysis and research synthesis. In A. Mackey
(Ed.), Conversational interaction in second language acquisition (pp. 407–452). Oxford University Press.
McDonough, K. (2006). Interaction and syntactic priming: English L2 speakers’ production of dative construc-
tions. Studies in Second Language Acquisition, 28(2), 179–207.
Oh, S.-Y. (2001). Two types of input modifcation and EFL reading comprehension: Simplifcation versus elabo-
ration. TESOL Quarterly, 35(1), 69–96.
Oliver, R. (1995). Negative feedback in child NS-NNS conversation. Studies in Second Language Acquisition,
17(4), 459–481.
Ono, L., & Witzel, J. (2002). Recasts, salience, and morpheme acquisition. Scholarly Paper. University of Hawai’i,
Department of Second Language Studies.
Pica, T., & Doughty, C. J. (1985). Input and interaction in communicative language classrooms: A comparison
of teacher-fronted and group activities. In S. M. Gass & C. Madden (Eds.), Input and second language acquisition
(pp. 115–132). Newbury House.
Pica, T., Lincoln-Porter, F., Paninos, D., & Linnell, J. (1986). Language learners’ interaction: How does it address
the input, output, and feedback needs of language learners? TESOL Quarterly, 30(1), 59–84.
Pica, T., Young, R., & Doughty, C. (1987). The impact of interaction on comprehension. TESOL Quarterly,
21(4), 737–758.
Revesz, A., & Han, Z. (2006). Task content familiarity, task type and efcacy of recasts. Language Awareness,
15(3), 160–179.
phological processing. In A. Godfroid & H. Hopp (Eds.), The Routledge handbook of second language acquisition
Sato, C. J. (1986). Conversation and interlanguage development: Rethinking the connection. In R. R. Day
(Ed.), “Talking to learn”: Conversation in second language acquisition (pp. 23–45). Newbury House.
Sato, C. J. (1988). Origins of complex syntax in interlanguage development. Studies in Second Language Acquisi-
tion, 10(3), 371–395.
Sato, C. J. (1990). The syntax of conversation in interlanguage development. Gunter Narr.
Saxton, M. (1997). The contrast theory of negative input. Journal of Child Language, 24(1), 139–161.
Schmidt, R. W. (1990). The role of consciousness in second language learning. Applied Linguistics, 11(2), 129–158.
Schmidt, R. W. (2010). Attention, awareness, and individual diferences in language learning. In W. M. Chan,
et al. (Eds.), Proceedings of CLaSIC 2010, Singapore, December 2–4 (pp. 721–737). National University of Sin-
gapore, Centre for Language Studies.
Sheen, Y. (2006). Exploring the relationship between characteristics of recasts and learner uptake. Language
Teaching Research, 10(4), 361–392.
Shehadeh, A. (2002). Comprehensible output, from occurrence to acquisition: An agenda for acquisitional
research. Language Learning, 52(3), 597–647.
Smith, B. (2012). Eye-tracking as a measure of noticing: A study of explicit recasts in SCMC. Language Learning
and Technology, 16(3), 53–81.
Swain, M. (1995). Three functions of output in second language learning. In G. Cook & B. Seidlhofer (Eds.), Princi-
ple and practice in applied linguistics: Studies in honour of Henry G. Widdowson (pp. 125–144). Oxford University Press.
Tomlin, R., & Villa, V. (1994). Attention in cognitive science and SLA. Studies in Second Language Acquisition,
16(2), 183–204.
Whong, M., Gil, H.-G., & Marsden, E. (2014). Beyond paradigm: The “what” and the “how” of classroom
research. Second Language Research, 30(4), 551–568.
Williams, J. N. (2009). Implicit learning. In W. C. Ritchie & T. K. Bhatia (Eds.), The new handbook of second
language acquisition (pp. 319–353). Emerald Group Publishing.
Yano, Y., Long, M. H., & Ross, S. (1994). The efects of simplifed and elaborated texts on foreign language
reading comprehension. Language Learning, 44(2), 189–219.
Yilmaz, Y. (2012). The relative efects of explicit correction and recasts on two target structures via two com-
munication modes. Language Learning, 62(4), 1134–1169.
Yilmaz, Y. (2016a). The linguistic environment, interaction and negative feedback. Brill Research Perspectives in
Multilingualism and Second Language Acquisition, 1, 45–86.
Yilmaz, Y. (2016b). The role of exposure condition in the efectiveness of explicit correction. Studies in Second
Yilmaz, Y., & Granena, G. (2016). The role of cognitive aptitudes for explicit language learning in the relative
efects of explicit and implicit feedback. Bilingualism: Language and Cognition, 19(1), 147–161.
Ziegler, N. (2016). Synchronous computer-mediated communication and interaction: A meta-analysis. Studies
347
28
WORKING MEMORY AND
SECOND LANGUAGE LEARNING
A critical and synthetic review
Shaofeng Li
Introduction
Working memory refers to the ability to store, maintain, and process information in ongoing cog-
nitive tasks. Historically, the term “working memory” was coined to refect an updated view of
short-term memory as a cognitive resource for simultaneous information storage and manipulation
rather than merely a passive storage device. Major streams of research on working memory address
the relationships between the diferent components of working memory, the associations between
working memory and other cognitive (e.g., intelligence) and afective variables (e.g., anxiety), work-
ing memory’s associations with learning outcomes, and the trainability of working memory. Due to
its putative efects on the process and outcome of language learning, working memory is a widely
studied topic in second language acquisition (SLA). Pioneering studies on working memory in SLA
include Harrington and Sawyer (1992), which investigated the role of working memory in read-
ing comprehension, and Mackey et al. (2002), which examined the associations between working
memory and second language (L2) interaction. Since these seminal studies, there has been a steady
growth of interest in the mediating role of working memory in various aspects of L2 learning.
Despite the burgeoning interest in working memory in L2 research, there has been confusion over
its conceptualization, measurement, and mechanism, and the research has yielded disparate, and
sometimes contradictory, fndings. This chapter seeks to clarify the concept of working memory
and navigate through the empirical evidence that has accumulated over the past three decades with
a view to extracting meaningful patterns and trends from the research. The rest of the chapter starts
with a discussion of the architecture and measurement of working memory based on diferent theo-
retical models and a discussion of working memory in SLA theories, followed by a synthesis of the
research fndings in L2 learning. A conclusion is ofered at the end summarizing current knowledge
and pointing out future directions.
Working memory models

A number of theories have been proposed to account for the relationships between the diferent com-
ponents of working memory (Miyake & Shah, 1999). Fundamentally, these theories are represented
by two models: the multi-componential model and the unitary model. In the multi-componential
348 DOI: 10.4324/9781003018872-32

Working memory and second language learning
model, which is sometimes referred to as the European model (Wen, 2015), the components of
working memory are independent of each other and draw on diferent pools of resources. The
storage components in these models, which will be described in more detail below, are equated
with working memory. The multi-componential model was championed by Baddeley (2015, 2017),
according to whom working memory can be fractionated into four components: the central execu-
tive, the phonological loop, the visual-spatial sketchpad, and the episodic bufer (see Textbox 28.1).
The central executive is a domain-general attention control system with no storage capacity. Bad-
deley (2017) stated its main functions include coordinating between diferent components, focusing
and shifting attention, allocating resources, and liaising with long-term memory. The phonologi-
cal loop is a passive storage device for storing and rehearsing auditory information. It is a device
for vocabulary learning, and is important for learning novel vocabulary, not arbitrary associations
between known words. The visual-spatial sketchpad is responsible for storing and rehearsing infor-
mation in the form of images, shapes, colors, directions, locations, and their confgurations. The
episodic bufer is a transitional storage space for integrating discrete information bits into larger units,
linking short-term and long-term memory, and binding information from diferent sources and
information in diferent formats (e.g., auditory and visual; colors and shapes).
The unitary model originated in North America and was initiated by Daneman and Carpenter
(1980), who developed a reading span test that measures the storage and processing components
simultaneously. In this model, the storage and processing functions are inseparable, and they draw on
the same pool of resources. There is a trade-of between the storage and processing functions such
that allocating more resources to one component will lead to fewer resources for the other. Accord-
ing to the unitary model, the storage alone such as phonological loop and the visuo-spatial sketch
pad cannot represent working memory, nor can the executive control, although the executive control
plays a major role.
Textbox 28.1 Key terms and concepts in Baddeley’s working memory model
The central executive: The CEO of working memory; main functions include inhibition, shifting, and
updating.
The phonological loop: An inner ear; a storage device for storing and rehearsing auditory information.
The visual-spatial sketchpad: An inner eye; a storage device for information in the form of images, shapes,
locations, and directions.
The episodic bufer: A liaison department; a transitional storage device between the central executive and the
two domain-specifc storage systems for holding bound and integrated information from diferent sources.
Working memory measures

Measures of working memory fall into two broad categories, simple tasks and complex tasks, which
align with multi-componential and unitary models of working memory, respectively. Readers are
advised to consult the IRIS database (www.iris-database.org) for validated tests of working memory.
Simple tasks measure only the storage components of working memory, while complex tasks tap
both the storage and processing functions. For the purpose of clarity, in this chapter, simple tasks
measuring the two storage components are named “phonological short-term memory” tasks and
“visual-spatial short-term memory” tasks, respectively, and complex tasks tapping both storage and
processing are referred to as measures of “executive working memory” (Wen, 2015). In simple tasks,
349
Shaofeng Li
which measure only the storage function, participants are presented with lists of discrete, unrelated
items in the auditory or visual mode and are asked to recall the items in oral or written form at the
end of each list. The number of items in each list, which can be called span size, ranges from two
to nine, and each span size appears three or more times in the whole test. Simple tasks measuring
phonological short-term memory include nonword span, word span, digit span, and letter span, for
which the stimuli are artifcial words, real words, numbers, and letters, respectively. Among these
measures, nonword repetition is probably the most frequently used. Typically, in a nonword span test,
participants listen to word-like items coined based on the phonological structure of real words and
repeat them in the order in which they are presented, and scoring is based on participants’ accuracy
in repeating the nonwords. Nonwords can be created based on either the participants’ own/familiar
language or a foreign/unfamiliar language, and there has been evidence that nonword stimuli based
on unfamiliar languages have higher predictive validity for L2 learning than stimuli based on familiar
languages (Farnia & Geva, 2011).
Similar to tests of phonological short-term memory, simple tasks measuring visual-spatial short-
term memory require participants to memorize and recall isolated items such as shapes, locations, and
directions. Example tests include visual matrix, arrow span, and odd one out. In a visual matrix task,
participants are presented with a series of dots in a matrix and recall the locations of the dots after
the matrix is removed. In an arrow span task, participants see some short and long arrows in diferent
angles, draw the arrows, and then recall them at the end of the trial. In an odd-one-out task, partici-
pants see pictures with three shapes, one of which is diferent from the others, and they are required
to recall the diferent shapes in each picture in the order in which they appeared.
Tests of the episodic bufer typically require learners to integrate information. One example test
is sentence repetition in which the participant is simply asked to repeat an aurally presented sentence.
The rationale is that in the process of sentence repetition, the participant integrates received informa-
tion with long-term memory and integrates “the products of semantic and syntactic analysis” (Allo-
way et al., 2004, p. 89). In some studies, the episodic bufer is tested by asking participants to learn
the arbitrary associations between words and shapes, binding verbal and nonverbal information. For
example, in Wang et al. (2017), participants were presented with nonwords, followed by shapes that
were unusual or “non-nameable”; they were then presented with the pairings between nonwords
and shapes and asked to recall the pairings at the end of each trial. Similarly, in Gray et al. (2017),
participants listened to nonwords while seeing corresponding polygons and at the end of each trial
selected the right polygons upon hearing the nonwords. However, as Baddeley (2015) pointed out,
the episodic bufer has been measured in many diferent ways, and the extent to which these are valid
tests of the construct is unknown. The fundamental idea of the episodic bufer is binding or integrat-
ing diferent types of information, but binding syntactic and semantic information, as in a sentence
repetition task, is obviously diferent from binding auditory and visual information.
The central executive has been measured in two ways: integrated and independent. In integrated
measurement, which is based on the unitary model, the central executive is tested by using complex
working memory tasks which integrate both the processing and storage components. Typically, the
participant is required to perform two tasks: a (primary) memorization task and a (secondary) process-
ing task. For example, in a reading span test, the participant reads sentences divided into sets of two to
six sentences; half of the sentences are meaningful and half are not (Waters & Caplan, 1996). For each
sentence, the participant decides whether it is meaningful, memorizes the fnal word of the sentence,
and recalls the sentence-fnal words in the order they are displayed upon completion of the whole set.
Other variants of the complex task include listening span, in which stimuli are presented in the audi-
tory mode, and operation span, in which the participant performs simple mathematical calculations by
judging the correctness of the equations while trying to memorize an unrelated item such as a letter or
word. Tests of visual-spatial executive memory follow the same procedure except that the stimuli are
shapes and images. For example, in a rotation span test, the participant sees short and long arrows in
350
diferent angles (the same as the arrow task described earlier testing only the storage function), followed
by a processing task requiring the participant to judge whether a letter is normal or mirrored (fipped).
Independent measurement of the central executive is based on the multi-componential model,
in which the central executive is measured separately from the storage components. In this model,
the central executive is separated into three functions: inhibition, shifting, and updating (Miyake &
Friedman, 2012). Inhibition is tested by means of tasks in which participants suppress “dominant,
automatic, or prepotent responses” (Rezaei & Mousanezhad Jeddi, 2012, p. 32), such as not looking
to the left (but looking to the right instead) when one is told to look to the left. Example inhibition
tasks include antisaccade, the Simon task, the Stroop task, Go/No-Go, and the Flanker task. Shifting
refers to the ability to switch from one task to another. An example shifting task is the color-shape
task in which participants are asked to categorize objects by color or shape, and their performance is
represented by a “switching cost”—the diferences between reaction times for switched and repeated
trials. Updating refers to the ability to constantly monitor and update information in an ongoing
task, an example being the running memory task in which participants are presented with lists of an
unpredictable number of items and are asked to recall the last three items. Among the three executive
functions, inhibition has probably received the most attention in the research.
Working memory in SLA theories

Working memory is represented as a construct in several theoretical approaches to L2 learning.
The interaction hypothesis (Long, 2015) posits an important role for selective attention in notic-
ing and processing linguistic forms during negotiated interaction. The theory suggests that selec-
tive attention, which concerns the ability to select relevant input, draws heavily on the executive
functions of working memory—inhibition, shifting, and updating (see Textbox 28.1). Selection of
input involves focusing on relevant information and blocking irrelevant information (inhibition),
switching between form and meaning (shifting), and constantly monitoring and updating informa-
tion (updating). Research on the associations between working memory and interactional feedback
has been conducted in the framework of the interaction hypothesis (see Long, 2023 [this volume]).
Skill acquisition theory (DeKeyser, 2020) holds that language learning starts with declarative knowl-
edge—knowledge about rules and discrete items. Declarative knowledge is then integrated and
applied in speaking, writing, listening, and reading, and the integration and application of declara-
tive knowledge is called proceduralization. Proceduralized knowledge is automatized via repeated
practice to approximate the native-like state (also see Morgan-Short & Ullman, 2023 [this volume];
Suzuki, 2023 [this volume]). Based on the claims of skill acquisition theory, it can be inferred that as
a cognitive device for explicit/conscious learning, working memory is more likely to be implicated
in initial stages where learning is explicit and conscious; working memory is less likely to be drawn
on in more advanced stages where learning is more implicit and depends on cognitive abilities for
unconscious learning. The cognition hypothesis (Robinson, 2011) states that tasks that are more
cognitively demanding direct learners’ working memory resources to relevant linguistic resources
and lead to more complex and accurate linguistic production and more learning gains. Thus, learn-
ers are more likely to draw on their working memory when performing complex tasks compared
with simple tasks. Usage-based theories view L2 learning as associative learning on the grounds that
learning a language is essentially a process of learning the arbitrary connections between linguistic
units based on a large amount of input (see Sagarra, 2023 [this volume]). Therefore, phonologi-
cal short-term memory (phonological loop), an ability to memorize unrelated items, is crucial for
language learning. To conclude this section, it should be clarifed that most L2 theories, including
those discussed here, do not make overt, systematic hypotheses about the role of working memory.
Likewise, not all research on working memory in L2 learning has a solid theoretical basis. Therefore,
additional, contingent theoretical discussion will be provided in the research synthesis that follows.
351
Shaofeng Li

The research on the role of working memory in L2 learning can be categorized in two ways. One way
is to divide the research into correlational and experimental studies based on their research design. Cor-
relational research examines the associations between working memory and outcome variables, regard-
less of context and learning condition, using an individual diferences design. In a typical correlational
study, two sets of scores are obtained, one for working memory and one for an outcome variable such as
grammar learning (Serafni & Sanz, 2016). Correlational analyses or multiple regressions are performed
to ascertain whether working memory is a signifcant predictor of the outcome variable. Experimental
research investigates whether working memory plays diferent roles in diferent treatment or learning
conditions. In a typical experimental study examining the interaction between working memory and
the efectiveness of instructional treatments, learners are divided into two or more groups and receive
diferent types of instructional treatment (Li et al., 2019). In these studies, treatment type is consistently
and strictly manipulated. Treatment efects are tested using pre-tests and post-tests, and statistical analysis
is performed for each group, with working memory and pre-test scores serving as predictors and post-
test scores as the dependent variable. Another way is to classify the studies based on the domain of the
outcome variable, such as language skills, namely reading, speaking, listening, and writing, or based on
aspects of the linguistic system: grammar, vocabulary, and pronunciation.
The review of the research in the following sections is organized by language skills and aspects
of the linguistic system, and within each section, correlational and experimental studies are distin-
guished. Studies on pronunciation are discussed in the section on speaking given that pronunciation
has been examined as a dimension of speaking in the research. Due to a lack of research on writing
and listening, they are not included in the synthesis. Given that working memory has been examined
using complex and simple tasks, which measure executive working memory and phonological short-
term memory respectively, the two types of working memory are distinguished in the synthesis.
Before dividing the research into specifc streams, an overview of the research is provided based on
Linck et al.’s meta-analysis.
Overview
To provide an overall picture of the associations between working memory and L2 learning gains,
a brief summary of the fndings of Linck et al.’s (2014) meta-analysis is provided here. The meta-
analysis, which was based on 79 studies, showed an overall weak, positive correlation between all
working memory measures and outcome variables measuring L2 comprehension and production,
r = .25. The meta-analysis further showed that verbal working memory is more strongly correlated
with L2 comprehension and production than phonological short-term memory, r = .27 vs .17, and
that verbal measures that are based on linguistic stimuli showed stronger associations with learn-
ing outcomes than non-verbal measures such as digits and spatial information: r = .26 vs .18. The
meta-analysis also provided a methodological synthesis of the primary research, highlighting some
methodological issues. For instance, it showed that in 46 of the 79 included studies, working memory
was tested in learners’ L2 (rather than in their native languages), which causes a concern because of
the potential infuence of learners’ L2 profciency on their working memory performance. In all, the
meta-analysis provided a synthesis of the research on working memory in L2 learning, and thereby
provided valuable insights and inspiration.
Reading
In L2 research, the fndings on reading comprehension are mixed, and one can only extract some
trends based on the primary research. For verbal working memory, one appreciable pattern is that
352
almost all studies using L2 working memory tests showed signifcant, positive correlations between
working memory and reading comprehension (Alptekin & Ercetin, 2010; Harrington & Sawyer,
1992; Joh & Plakans, 2017). For studies using L1 working memory tests, however, the results are
disparate: While some studies found signifcant correlations between working memory and read-
ing comprehension (Engel de Abreu & Gathercole, 2012; Sagarra, 2017), other studies did not
(Alptekin & Ercetin, 2010; Jung, 2018). It is unclear what caused the diferent results between
studies using L2 and L1 measures. One theme, or rather concern, is that the consistent correlations
based on L2 working memory tests seem to be partly due to learners’ L2 profciency. Learners
who are more profcient in the L2 also score higher in the working memory tests created in the
L2, hence the signifcant correlations between working memory and L2 reading. Therefore, the
infuence of L2 profciency needs to be addressed in future research, such as by using an L1 test (see
Linck et al., 2014) or by including a profciency test and accounting for the variance attributable to
L2 profciency. It is also possible that L1 and L2 working memory have diferent mechanisms and
need to be tested diferently, but this speculation needs to be theoretically elaborated and empiri-
cally examined.
Several studies have examined the extent to which working memory interacts with background
knowledge in contributing to reading comprehension (Joh & Plakans, 2017; Shin et al., 2019).
Background knowledge has been operationalized as learners’ previous knowledge about the topics
of the reading texts or their existing lexical and grammatical knowledge about the L2. Overall, it
has been found that the role of working memory is signifcant only when readers have a certain
amount of previous knowledge. It would seem that learners with less previous knowledge likely
draw more on lower-order, bottom-up skills such as phonetic decoding and lexical and syntactic
processing rather than working memory during comprehension activities. The author would like
to propose a new hypothesis to account for the efects of background knowledge on the role of
working memory. This hypothesis can be called the threshold hypothesis; that is, in order for work-
ing memory to kick in, a certain threshold needs to be reached in terms of learners’ previous
topical and linguistic knowledge. There has been limited research on phonological short-term
memory (the phonological loop) and the fndings are equivocal. Some studies (Jung, 2018; Ser-
vice & Kohonen, 1995) found signifcant, positive correlations between phonological short-term
memory and L2 reading comprehension, while other studies (Geva & Ryan, 1993; Hummel,
2009) failed to fnd signifcant correlations. One possible explanation seems to be that the studies
that obtained signifcant fndings investigated either children or beginners and reading tasks that
only involved the understanding of short sentences and phrases. In these types of reading materials
or reading tasks, vocabulary knowledge is key, for which phonological short-term memory is cru-
cial. Thus, phonological short-term memory may contribute to reading via vocabulary learning.
Another way phonological short-term memory contributes to reading is by facilitating the learn-
ing of grammar knowledge, which is found to be highly correlated with L2 reading comprehen-
sion (Jeon & Yamashita, 2014).
Speaking
The mechanism through which working memory mediates speaking can be discussed in the frame-
work of Levelt’s (1989) model, whereby speech production is divided into three stages: conceptual-
ization, formulation, and articulation. The frst stage is conceptualization, in which the message of an
utterance is generated. The generated message is called the preverbal message, that is, a message before
it is encoded in language. The preverbal message is then fed into the second stage—formulation,
in which linguistic items are retrieved and organized into sentences based on syntactic procedures
and the relevant phonological features are activated. The formulated message is called a “phonetic
plan” or “internal message,” which is then delivered to the fnal stage—articulation. Articulation is
353
Shaofeng Li
the process of verbalizing the message using vocal organs. Where does working memory ft in this
system? According to Levelt, working memory is involved only in the conceptualization stage, when
the message of an utterance is generated. Formulation and articulation are “underground processes”
(p. 22) that are not subject to controlled processing implicating working memory. However, while
formulation and articulation are perhaps automatic in L1 speech production, they are not in L2
speech production, at least in initial stages of learning when speaking often involves efortful retrieval,
selection, and articulation of linguistic items. Thus, the importance of working memory in L2 speak-
ing is paramount (Li & Fu, 2018).
Executive working memory and phonological short-term memory (the phonological loop) may
play diferent roles at the diferent stages of L2 speech production, which in turn may afect difer-
ent dimensions of speech performance. It can be hypothesized that executive working memory is
primarily implicated in the conceptualization stage, which is associated with linguistic complexity
and fuency, and in the formulation stage where linguistic accuracy is monitored. Phonological
short-term memory is primarily involved in formulation and articulation, and its main function is to
facilitate speedy retrieval and activation of linguistic knowledge from long-term memory. Therefore,
phonological short-term memory is likely critical for the fuency aspects of speech production.
In L2 research, a number of studies have investigated the role of verbal working memory in medi-
ating task performance in terms of complexity, accuracy, and fuency. When learners simply per-
formed an oral task and the task was not manipulated, it was shown that executive working memory
contributed to fuency (Fehringer & Fry, 2007; Georgiadou & Roehr-Brackin, 2017). Studies that
examined task complexity showed varied fndings. These studies are based on Robinson’s (2011)
cognition hypothesis, which predicts that individual diference factors such as working memory are
more likely to be implicated in complex tasks than simple tasks (see “Working memory and SLA
theories”). Task complexity has been operationalized as more/fewer reasoning demands (Cho, 2018;
Zalbidea, 2017) or with/without task structure (Kormos & Trebits, 2011). In general, these studies
showed either no diference in learners’ performance between simple and complex tasks (Zalbidea,
2017) or a trade-of between linguistic accuracy and complexity across the two task conditions (Cho,
2018; Kormos & Trebits, 2011). Executive working memory was found to have a positive efect on
syntactic complexity in complex tasks rather than simple tasks (Kormos & Trebits, 2011; Zalbidea,
2017 yet see Cho, 2018). The third category of research concerns the relationship between execu-
tive working memory and task planning (Ahmadian, 2012; Li & Fu, 2018). The studies found that
working memory was associated with accuracy and fuency in within-task planning conditions but
not with task performance under strategic planning conditions.
With respect to phonological short-term memory (the phonological loop), the research has
focused on its role in (1) the development of fuency and accuracy and (2) rated or perceived oral
performance. O’Brien et al. (2007) reported that phonological short-term memory was predictive
of L2 fuency gains and grammatical accuracy measured at diferent time points. Payne and Whitney
(2002) compared the efects of online and face-to-face communication on learners’ rated oral per-
formance and its relationship with phonological short-term memory. Online communication was
more efective than face-to-face communication in facilitating oral gains. Phonological short-term
memory was signifcantly and positively correlated with the efects of both types of instruction, but
the positive correlation was stronger for face-to-face instruction than online instruction, suggesting
that learners with weaker phonological short-term memory were disadvantaged to a greater extent in
face-to-face instruction. While phonological short-term memory was found to be facilitative of the
development of fuency and global oral gains, it was not predictive of rated accentedness (Hu et al.,
2013), pronunciation (Slevc & Miyake, 2006), or comprehensibility (Venkatagiri & Levis, 2007),
nor was it important for learners’ ability to recognize sounds (Slevc & Miyake, 2006). These fndings
suggest that phonological short-term memory is important for the development of oral ability but
it is less or not important for sound production or the phonological aspects of speech production.
354
In other words, it may constitute a storage device for auditory information, but it is not responsible
for acquiring the phonetic and phonological aspects of linguistic materials.
Grammar
Correlational research shows that phonological short-term memory (the phonological loop) is pre-
dictive of L2 grammar knowledge (French & O’Brien, 2008; Hummel, 2009; Slevc & Miyake,
2006). Phonological short-term memory may contribute to grammar learning either directly, by
extracting regularities from memorized linguistic stimuli, or indirectly, via vocabulary learning that
provides the fodder for rule learning (Martin & Ellis, 2012). One important fnding in this line
of research is that the role of phonological working memory seems to become progressively less
important as learners move from beginning to more advanced levels (Serafni & Sanz, 2016). The
research also shows that verbal working memory is a signifcant predictor of L2 grammar (Engel
de Abreu & Gathercole, 2012; Sagarra, 2017). In contrast, neither seems to be important for L2
metalinguistic knowledge, i.e., to acquire knowledge about grammar terms or knowledge used to
describe grammar rules or understand rule descriptions (Martin & Ellis, 2012; Roehr & Gánem-
Gutiérrez, 2009).
Experimental research examines the interaction between working memory and instructional
treatments in grammar learning (also see Sachs et al., 2023 [this volume]). Therefore, the results
inform us of not only the link between working memory and learning outcomes but also the process
that may underlie the link. Experimental research has primarily examined verbal working memory,
probably premised on the assumption that the ability to simultaneously store and process informa-
tion is crucial for learners to beneft from instructional tasks involving online information process-
ing. One instructional device that may have links with verbal working memory is oral corrective
feedback, which refers to responses to learners’ errors in their L2 utterances. Over the past two
decades, there has been a large body of research on various aspects of corrective feedback, one per-
spective being whether the efectiveness of feedback is mediated by individual diference variables
such as working memory. Using both meta-analysis and narrative synthesis, Li (2017) synthesized the
research examining the role of working memory and language aptitude in mediating the efectiveness
and noticing of feedback and learners’ responses after feedback. Li’s synthesis showed the following
fndings. Firstly, working memory showed an overall weak correlation with the efects of feedback,
r = .23, confrming the weak correlation with L2 learning in general found by Linck et al.’s (2014)
meta-analysis. Secondly, when feedback was divided into explicit and implicit feedback according
to whether learners’ attention is drawn to the presence of errors, explicit feedback showed stronger
correlations with executive working memory than implicit feedback, r = .34 versus .23, respectively.
This fnding suggests that working memory is a cognitive ability for conscious learning. Thirdly,
studies on noticing—whether the corrective intention of feedback is recognized—did not provide
conclusive evidence for the importance of working memory in the noticing of feedback. This was
attributed to the heterogeneity in the measures of noticing and to the salience of the feedback exam-
ined in the studies. Fourthly, with regard to modifed output, which refers to whether feedback is
followed by learner responses, studies showed that working memory was predictive of the modifed
output after prompts but not recasts. The responses after prompts may represent deeper cognitive
processing than the responses after recasts, given that some responses to recasts may be mere repeti-
tions of the received feedback, which may not tax working memory.
Other experimental studies examined the role of executive working memory in various
instructional types. These studies converge on one point; that is, working memory is implicated
in treatments that pose a heavy processing load on learners’ cognitive resources. For example, Li
et al. (2019) found that among fve instructional conditions manipulated based on whether and
when learners received grammar instruction, only the conditions where feedback was provided
355
Shaofeng Li
during communicative tasks showed signifcant associations with working memory; other condi-
tions in which learners received only pre-task or post-task grammar instruction, or no grammar
instruction, did not draw on learners’ working memory. Suzuki and DeKeyser (2017) found that
working memory was positively correlated with the efects of massed instruction, in which two
practice sessions were separated by one day, but not with the efects of spaced instruction, in which
the interval between the two sessions was seven days. Sanz et al. (2016) showed that the efect of
working memory was evident in practice activities performed in the absence of prior grammar
instruction; however, when practice was preceded by grammar instruction, the infuence of work-
ing memory disappeared. The researchers interpreted the fnding as suggesting that pre-practice
grammar instruction alleviated learners’ processing burden, thus neutralizing the efects of working
memory.
Vocabulary
Because of the theoretical importance of phonological short-term memory (the phonological loop)
as a device for vocabulary learning, particularly for the learning of novel vocabulary (Baddeley,
2003), phonological short-term memory and L2 vocabulary learning are likely to be related. Thus
far, the evidence corroborates this claim in adults (Hummel, 2009) as well as in children (Farnia &
Geva, 2011). One important theme that has appeared in the research but needs further empirical
investigation is that phonological short-term memory is crucial for vocabulary learning by beginners,
but as learners proceed to more advanced stages (higher profciency), its predictive power weakens.
For example, in L1 research, Gathercole et al. (1992) found that phonological short-term memory
was a strong predictor of the vocabulary development of children aged four to fve, but when they
reached the age of eight, it was no longer a signifcant predictor. Similar fndings have been observed
in L2 research. Some treatment studies examining vocabulary learning in novel languages provided
evidence for the consistent efect of phonological short-term memory at initial stages of L2 learning
(Speciale et al., 2004). However, studies involving advanced learners failed to fnd a signifcant efect
for phonological short-term memory. For example, Slevc and Miyake (2006) found that phonologi-
cal short-term memory was not a signifcant predictor of the vocabulary size of advanced L2 English
learners who had lived in the U.S. for more than twenty years. Other studies reported that phono-
logical short-term memory is predictive of vocabulary size but not learning rate (the development
of vocabulary knowledge across two time points); for learning rate, previous vocabulary knowledge
may be a better predictor (Masoura & Gathercole, 2005; O’Brien et al., 2006). This points to a need
to include long-term memory or previous vocabulary knowledge when investigating whether pho-
nological short-term memory predicts learning rate, operationalized as gains over a period between
two time points.
There has been limited research on the role of executive working memory in vocabulary learning.
Several correlational studies reported inconsistent or no signifcant correlations between executive
working memory and child L2 learners’ vocabulary knowledge (Engel de Abreu & Gathercole, 2012;
Jean & Geva, 2009). However, executive working memory has proven to be a consistent predictor in
experimental studies with adults, including both studies investigating vocabulary learning in learning
novel languages (Kempe et al., 2010; Martin & Ellis, 2012) and studies involving familiar languages
(Lee et al., 2020; Malone, 2018). Therefore, it would seem that the efects of working memory can
be more easily observed in studies in which variables are carefully controlled and manipulated than
in correlational studies in which variables are not manipulated. It is also possible that executive work-
ing memory is not important for vocabulary size, but it is important for learning new vocabulary—a
pattern opposite to phonological short-term memory, which is more likely to predict vocabulary size
but not learning rate.
356
Summary
Taken together, research on working memory in L2 learning shows the following patterns. For
reading comprehension, executive working memory is a consistent predictor when it is measured in
learners’ L2; the fndings for L1 working memory tests are mixed. The infuence of executive work-
ing is more evident when learners have some threshold knowledge in the L2 and about the topic of
the reading text. Phonological short-term memory seems to predict low-level reading by means of
vocabulary, and it may also be linked with reading through knowledge of grammar—a hypothesis
to be tested empirically. For speaking, executive working memory is correlated with fuency, and it
is more likely to be implicated in complex than simple speaking tasks and in within-task planning
than pre-task planning. Phonological short-term memory facilitates fuency and the development
of oral competence, but it is not important for rated accent and perceived pronunciation ability. For
grammar learning, phonological short-term memory is more likely to be predictive of initial than
advanced learning, and it is not important for metalinguistic knowledge. Executive working memory
is more predictive of the outcomes of explicit instruction than implicit instruction and more predic-
tive of instruction types that are cognitively more demanding. For vocabulary learning, phonological
short-term memory is more likely predictive of initial than advanced vocabulary learning, which
may draw more on long-term memory (prior vocabulary knowledge). It is more likely predictive
of vocabulary size than learning new vocabulary knowledge. Executive working memory, however,
shows the opposite pattern.

Working memory has been found to be a signifcant predictor for various aspects of L2 learning.
Research on working memory contributes to a better understanding of the mechanism of language
learning and provides valuable pedagogical implications for how to adapt instruction to ft learners’
working memory profles or how to support learners with weaker working memory abilities. Over-
all, the research on the role of working memory in L2 learning is limited, especially regarding its role
for specifc aspects of learning (see Textbox 28.2). While there has been a certain amount of research
on reading, speaking, grammar, and vocabulary, there is very little research on listening and writing.
Yet, even in areas that have been studied, many open issues also remain. For reading, future research
needs to address the infuence of learners’ L2 profciency on their L2 working memory scores (or
simply avoid using L2 working memory tests), and to examine whether and how the role of work-
ing memory is constrained by text type (genre) and task type (type of comprehension questions).
For speaking, more research is needed on the relationship between working memory and diferent
task features such as task complexity and task planning. For grammar, further research may examine
factors constraining the role of the associations between working memory and L2 grammar learning,
such as L2 profciency, age, and instruction type. For vocabulary, a promising topic is the unique
and joint contributions of phonological short-term memory and long-term memory to vocabulary
learning at diferent learning stages. It is also necessary to examine whether working memory plays
diferent roles in predicting vocabulary size and learning rate and in mediating the efects of diferent
treatment tasks.
To better understand the nature of working memory, more research is warranted on the dif-
ferential efects of diferent components of working memory, particularly executive functions and
visual-spatial working memory. Although executive working memory has been found to have greater
predictive validity, the importance of phonological short-term memory in L2 learning may have
been underestimated. More research is in order on the importance of the two types of working
memory on diferent aspects of L2 learning. There is also a need to explore the joint and unique
357
Shaofeng Li
contributions of working memory and other individual diference variables such as motivation and
anxiety to the process and outcome of L2 learning.
Textbox 28.2 Open questions and future directions
How does working memory afect L2 listening and writing?

Is visual-spatial working memory relevant to L2 learning?
In what ways are executive functions (updating, shifting, and inhibition) important for language learning?
What are the comparative efects of working memory and other individual diference variables such as
anxiety, motivation, etc.?
To what extent is the role of working memory mediated by learners’ profciency level, instruction type, and age?
Further reading
Baddeley, A. (2017). Modularity, working memory and language acquisition. Second Language Research, 33,
299–311.
In this article, Baddeley synthesizes his theory on working memory and explains how it relates to second
language learning.
Jufs, A., & Harrington, M. (2011). Aspects of working memory in L2 learning. Language Teaching, 44, 137–166.
This article provides an overview of the research on working memory in second language learning.
In L. Gurzynski (Ed.), Expanding individual diference research in the interaction approach: Investigating learners,
instructors and researchers (pp. 41–70). Benjamins.
This study is a research synthesis integrating meta-analysis and narrative review examining the role of lan-
guage aptitude and working memory in second language interaction.
References
Ahmadian, M. (2012). The relationship between working memory capacity and L2 oral performance under
task-based careful online planning condition. TESOL Quarterly, 46, 165–175.
Alloway, T., Gathercole, S., Willis, C., & Adams, A. (2004). A structural analysis of working memory and related
cognitive skills in young children. Journal of Experimental Child Psychology, 87, 85–106.
Alptekin, C., & Ercetin, G. (2010). The role of L1 and L2 working memory in literal and inferential comprehen-
sion in L2 reading. Journal of Research in Reading, 22, 206–219.
Baddeley, A. (2003). Working memory and language: An overview. Journal of Communication Disorders, 36,
189–208.
Baddeley, A. (2015). Working memory in second language learning. In Z. Wen, M. Mota, & A. McNeil (Eds.),
Working memory in second language acquisition and processing (pp. 17–28). Multilingual Matters.
Baddeley, A. (2017). Modularity, working memory and language acquisition. Second Language Research, 33,
299–311.
Cho, M. (2018). Task complexity, modality, and working memory in L2 task performance. System, 72, 85–98.
Daneman, M., & Carpenter, P. (1980). Individual diferences in working memory and reading. Journal of Verbal
Learning and Verbal Behavior, 19, 450–466.
DeKeyser, R. (2020). Skill acquisition theory. In B. VanPatten, G. Keating, & S. Wulf (Eds.), Theories in second
Engel de Abreu, P., & Gathercole, S. (2012). Executive and phonological processes in second-language acquisi-
tion. Journal of Educational Psychology, 104, 974–986.
Farnia, F., & Geva, E. (2011). Cognitive correlates of vocabulary growth in English language learners. Applied
358
Fehringer, C., & Fry, C. (2007). Hesitation phenomena in the language production of bilingual speakers: The
role of working memory. Folia Linguistica, 41(1–2), 37–72.
French, L. M., & O’Brien, I. (2008). Phonological memory and children’s second language grammar learning.
Gathercole, S. E., Willis, C. S., Emslie, H., & Baddeley, A. D. (1992). Phonological memory and vocabulary
development during the early school years: A longitudinal study. Developmental Psychology, 28, 887–898.
Georgiadou, E., & Roehr-Brackin, K. (2017). Investigating executive working memory and phonological
short-term memory in relation to fuency and self-repair behavior in L2 speech. Journal of Psycholinguistic
Research, 46, 877–895.
Geva, E., & Ryan, E. (1993). Linguistic and cognitive correlates of academic skills in frst and second languages.
Gray, S., Green, S., Alt, M., Hogan, T., Kuo, T., Brinkley, S., & Cowan, N. (2017). The structure of working
memory in young children and its relation to intelligence. Journal of Memory and Language, 92, 183–201.
Harrington, M., & Sawyer, M. (1992). L2 working memory capacity and L2 reading skill. Studies in Second
Hu, X., Ackermann, H., Martin, J., Erb, M., Winklere, S., & Reiterer, S. (2013). Language aptitude for pro-
nunciation in advanced second language (L2) learners: Behavioural predictors and neural substrates. Brain
Hummel, K. (2009). Aptitude, phonological memory, and second language profciency in nonnovice adult
learners. Applied Psycholinguistics, 30, 225–249.
Jean, M., & Geva, E. (2009). The development of vocabulary in English as a second language and its role in
predicting word recognition ability. Applied Psycholinguistics, 30, 153–185.
Jeon, E. H., & Yamashita, J. (2014). L2 Reading comprehension and its correlates: A meta-analysis. Language
Learning, 64(1), 160–212.
Joh, J., & Plakans, L. (2017). Working memory in L2 reading comprehension: The infuence of prior knowl-
edge. System, 70, 107–120.
Jung, J. (2018). Efects of task complexity and working memory capacity on L2 reading comprehension. System,
74, 21–37.
Kempe, V., Brooks, P., & Kharkhurin, A. (2010). Cognitive predictors of generalization of Russian grammatical
gender categories. Language Learning, 60, 127–153.
Kormos, J., & Trebits, A. (2011). Working memory capacity and narrative task performance. In P. Robinson
(Ed.), Second language task complexity: Researching the cognition hypothesis of language learning and performance
Lee, H., Warschauer, M., & Lee, J. (2020). Toward the establishment of a data-driven learning model: Role of
learner factors in corpus-based second language vocabulary learning. The Modern Language Journal, 104, 345–362.
Levelt, W. (1989). Speaking: From intention to articulation. The MIT Press.
Li, S. (2017). The Efects of cognitive aptitudes on the process and product of L2 interaction: A synthetic review.
In L. Gurzynski (Ed.), Expanding individual diference research in the interaction approach: Investigating learners,
instructors and researchers (pp. 41–70). John Benjamins.
Li, S., Ellis, R., & Zhu, Y. (2019). The associations between cognitive ability and L2 development under fve
diferent instructional conditions. Applied Psycholinguistics, 40, 693–722.
Li, S., & Fu, M. (2018). Strategic and unpressured within-task planning and their associations with working
memory. Language Teaching Research, 22, 230–253.
Linck, J., Osthus, P., Koeth, J., & Bunting, M. (2014). Working memory and second language comprehension
and production: A meta-analysis. Psychonomic Bulletin & Review, 21, 861–883.
Long, M. (2015). Second language acquisition and task-based language teaching. Blackwell.
Long, M. (2023). The psycholinguistics of second language interaction. In A. Godfroid & H. Hopp (Eds.), The
Routledge handbook of second language acquisition and psycholinguistics (pp. 335–347). Routledge.
Mackey, A., Philp, J., Fujii, A., Egi, T., & Tatsumi, T. (2002). Individual diferences in working memory, notic-
ing of interactional feedback and L2 development. In P. Robinson (Ed.), Individual diferences and instructed
language learning (pp. 181–208). Benjamins.
Malone, J. (2018). Incidental vocabulary learning in SLA: Efects of frequency, aural enhancement, and working
memory. Studies in Second Language Acquisition, 40(3), 651–675.
Martin, K. I., & Ellis, N. C. (2012). The roles of phonological short-term memory and working memory in L2
grammar and vocabulary learning. Studies in Second Language Acquisition, 34, 179–413.
Masoura, E. V., & Gathercole, S. (2005). Contrasting contributions of phonological short-term memory and
long-term knowledge to vocabulary learning in a foreign language. Memory, 13, 422–429.
Miyake, A., & Friedman, N. (2012). The nature and organization of individual diferences in executive func-
tions: Four general conclusions. Current Directions in Psychological Science, 21, 8–14.
359
Shaofeng Li
Miyake, A., & Shah, P. (Eds.). (1999). Models of working memory. Cambridge University Press.
Morgan-Short. K. & Ullman, M.T. (2023). Declarative and procedural memory in second language learning:
O’Brien, I., Segalowitz, N., Collentine, J., & Freed, B. (2006). Phonological memory and lexical, narrative, and
grammatical skills in second language oral production by adult learners. Applied Psycholinguistics, 27, 377–402.
O’Brien, I., Segalowitz, N., Freed, B., & Collentine, J. (2007). Phonological memory predicts second language
oral fuency gains in adults. Studies in Second Language Acquisition, 29, 557–582.
Payne, J., & Whitney, P. (2002). Developing L2 oral profciency through synchronous CMC: Output, working
memory and interlanguage development. CALICO Journal, 20, 7–32.
Rezaei, A., & Mousanezhad Jeddi, E. (2012). The contributions of attentional control components, phonologi-
cal awareness, and working memory to reading ability. Journal of Psycholinguistic Research, 49, 31–40.
Robinson, P. (2011). Second language task complexity, the cognition hypothesis, language learning, and per-
formance. In P. Robinson (Ed.), Second language task complexity: Researching the cognition hypothesis of language
learning and performance (pp. 3–38). John Benjamins.
Roehr, K., & Gánem-Gutiérrez, G. (2009). The status of metalinguistic knowledge in instructed adult L2 learn-
ing. Language Awareness, 18, 165–181.
Sachs, R., Baralt, M., & Gurzyniski-Weiss, L. (2023). Psycholinguistics for the language classroom. In A.
Godfroid & H. Hopp (Eds.), The Routledge handbook of second language acquisition and psycholinguistics
Sagarra, N. (2017). Longitudinal efects of working memory on L2 grammar and reading abilities. Second Lan-
guage Research, 33(3), 341–363.
Sagarra, N. (2023). Sentence processing: cognitive approaches to second language morphosyntactic and morpho-
logical processing. In A. Godfroid & H. Hopp (Eds.), The Routledge handbook of second language acquisition and
Sanz, C., Lin, H., Lado, B., Staford, C., & Bowden, H. (2016). One size fts all? Learning conditions and work-
ing memory capacity in ab initio language development. Applied Linguistics, 37, 669–692.
Serafni, E., & Sanz, C. (2016). Evidence for the decreasing impact of cognitive ability on second language
development as profciency increases. Studies in Second Language Acquisition, 38, 607–646.
Service, E., & Kohonen, V. (1995). Is the relation between phonological memory and foreign language learning
accounted for by vocabulary acquisition? Applied Psycholinguistics, 16, 155–172.
Shin, J., Dronjic, V., & Park, B. (2019). The interplay between working memory and background knowledge.
TESOL Quarterly, 53, 320–347.
Slevc, L. R., & Miyake, A. (2006). Individual diferences in second-language profciency: Does musical ability
matter? Psychological Science, 17(8), 675–681.
Speciale, G., Ellis, N. C., & Bywater, T. (2004). Phonological sequence learning and short-term store capacity
determine second language vocabulary acquisition. Applied Psycholinguistics, 25, 293–232.
Suzuki, Y., & DeKeyser, R. (2017). Exploratory research on second language practice distribution: An Aptitude
× Treatment interaction. Applied Psycholinguistics, 38, 27–56.
Venkatagiri, H., & Levis, J. M. (2007). Phonological awareness and speech comprehensibility: An exploratory
study. Language Awareness, 16, 263–277.
Wang, S., Allen, R., Fang, S., & Li, P. (2017). Cross-modal working memory binding and L1-L2 word learning.
Memory and Cognition, 45, 1371–1383.
Waters, G. S., & Caplan, D. (1996). The measurement of verbal working memory capacity and its relation to
reading comprehension. Quarterly Journal of Experimental Psychology, 49A, 51–74.
Wen, Z. (2015). Working memory in second language acquisition and processing: The phonological/executive
model. In Z. Wen, B. Mailce, & A. McNeill (Eds.), Working memory in second language acquisition and processing
(pp. 41–62). Multilingual Matters.
Zalbidea, J. (2017). One task fts all? The role of task complexity, modality, and working memory in L2 perfor-
mance. The Modern Language Journal, 101(2), 335–352.
360
29
SYNTHESIS
The psycholinguistics of learning
Brian MacWhinney
Introduction
The six chapters in this section contribute state-of-the-art examinations of some of the core psycho-
linguistic processes underpinning second language (L2) learning. These include working memory,
proceduralization, and forms of implicit learning. The focus on proceduralization is motivated in
large part by the observation that adult learners seldom achieve complete L2 fuency. The emphasis
on implicit processes is in tune with a major shift in the feld away from explicit instruction. This
movement was triggered decades ago by Krashen’s emphasis on a distinction between language learn-
ing and language acquisition (Krashen, 1994). According to this view, if adults could learn language
implicitly, much as children do, then their fnal attainment might be more native-like.
Although this analysis from Krashen seems plausible, it fails to take into consideration the other
factors that stand in the way of smooth second language learning by adults. These include reduced
exposure, lessened social support, competition of L2 with the frst language (L1), lower motivation,
and competing time demands (MacWhinney, 2017b). Moreover, as each of these chapters recog-
nizes, there are important relations between explicit and implicit processes in L2 learning. There is
good evidence for the short-term efects of implicit processes in various perceptual domains, and it
is reasonable to imagine that some of these efects can also lead to long-term learning. However, to
understand the efect of these processes on L2 learning, we need to measure long-term consolidation.
One possibility is that implicit processes lead directly to long-term learning. Another possibility is
that implicit processes run in parallel with more explicit processes and that work together to achieve
consolidation. Yet another possibility is that language learning, particularly by adults, relies less on
implicit learning than the experimental work might suggest.
Current approaches to understanding implicit L2 learning

Williams and Rebuschat (2023 [this volume]) set the stage for this examination of this issue by focusing
on two infuential current approaches to understanding implicit L2 learning. The frst tradition uses
artifcial grammar learning (AGL) to track learning without explicit information (Reber, 1967, 1976;
Reber & Lewis, 1977). The stimuli in these tasks can often be generated through a fnite-state grammar
concatenating meaningless letter strings or nonsense words with no clear relation to natural language.
More recently, researchers have also employed AGL grammars with meaningful strings (Friederici et
al., 2002). In such cases, the focus of the research is not on implicit learning throughout the course of
DOI: 10.4324/9781003018872-33 361

Brian MacWhinney
L2 learning, but on implicit processes during initial learning of L2 (Morgan-Short, 2020). In the
original tradition, the claim was that learners could pick up grammatical patterns without awareness.
This fnding was used to support the idea that L2 learning could also proceed without awareness.
However, it eventually became clear that participants became aware of parts of repeating patterns. It
was also not clear that there was any long-term learning of the patterns in AGL experiments with
meaningless stimuli. Furthermore, given the perceptual and conceptual distance between meaning-
less AGL stimuli and meaningful L2 forms, it was not clear how well one can connect the results of
this work with SLA theory.
As with AGL experiments, studies in the statistical learning (SL) paradigm expose participants to
a repeated set of stimuli to see if they can pick up implicit patterns without explicit instruction. The
work most clearly related to language learning (Safran et al., 1996) focuses on pulling out poten-
tial word forms from a string of syllables, although other experimental combinations and stimulus
types have also been used. As with research in the AGL tradition, there has been little attention to
the long-term efects of exposure to the meaningless syllable strings. For example, one often-cited
demonstration of long-term efects (Kim et al., 2009) uses visual stimuli and only checks retention
after 24 hours.
Although the evidence is still incomplete, it is possible that the long-term efects of SL may be
concentrated on the acquisition of the phonotactics of a new language (Safran & Thiessen, 2003).
This perceptual efect could operate in both children and adults (Smalle et al., 2018) as they become
accustomed to the sounds of a new language. To a large degree, this learning could be passive and
implicit. However, it could also involve explicit processing and noticing. For example, successful
learning of tones in Cantonese requires attention to both tone height (high, mid, low) and tone
contour (rising, falling, level). Much of this information is encoded on individual lexical items, but
general control of the system requires attention to pairwise combinations of tones independent of
specifc lexical items. Noticing these pairwise patterns can produce explicit encoding of target forms
in auditory working memory for subsequent repetition and imitation (Guenther & Perkell, 2003).
This shows that learning of sound patterns depends not just on SL, but also on the extraction of
words as chunks (Perruchet & Pacton, 2006), as well as explicit attention to sound patterns.
It is also not clear that SL is fundamental to the learning of new vocabulary. Some models of word
learning (Ambridge, 2020) postulate complete storage of all language input. Under that account, SL
would provide the beginning learner with thousands of candidate words in their phonological form,
but without any associated meaning. However, we have no evidence that input storage of this type
happens in either L1 or L2 learning. On the contrary, the usual fnding is that auditory short-term
memory fades quickly (Cowan, 1992; Goldinger, 1998). A more likely alternative is that word learn-
ing depends on the association of a sound with a meaning. In that case, a few new word forms might
be available through implicit SL processes for a short time, but these will decay unless they are soon
linked to a semantic representation (Goldinger, 2007; Schlichting & Preston, 2015). Here, again, we
see that the implicit working of SL is eventually coupled with subsequent more explicit processing.
Both Williams and Rebuschat (2023 [this volume]) and Godfroid (2023 [this volume]) note that
exposure to AGL and SL stimuli can produce both implicit and explicit knowledge. One way in which
learners can pick up explicit knowledge in these experiments is through the encoding of chunks of mate-
rial. In AGL studies, participants may pick up fragments of the larger fnite-state grammar as units. For
example, from strings like TPPTS, VXXVPXVS, and VXVPXTS, the learner may pick up the VPX
string. This is not enough to acquire the full grammar, but it is enough to demonstrate some learning.
Similarly, in SL, when exposed to bupadapatubitutibudutabapidabubupadapatubi the learner may pull out
bupada or patubi as the recurrent chunk. As Perruchet and Pacton (2006) note, the learning of these
chunks may be the backbone of what is going on in both AGL and SL learning. During the processing of
serial input in either AGL or SL tasks, the attentional mechanism may briefy encode one of these strings.
If that string recurs a bit later, the second occurrence can strengthen its status as a new chunk. To show
362
Synthesis
above-chance learning in these experiments, it is not necessary that the chunk or pattern be encoded as
a part of a complete fnite-state grammar or set of transitional probabilities.
Interactions between implicit and explicit processes

Godfroid (2023 [this volume]) approaches this issue of learning from a rather diferent angle. Instead
of just asking whether implicit processes underlie L2 learning, she asks whether there are demon-
strable interactions between implicit and explicit processes during both exposure and subsequent
consolidation. She conceptualizes these possible interactions as arising from an interface between a
module for implicit learning and a separate module for explicit learning. It is true that, if one fol-
lows the characterization of modules introduced by Fodor (1983) and invoked by Paradis (2009),
and if one confnes the theory to the initial moments of online processing, then these two forms of
knowledge do seem to act as modules and there would be no interface between implicit and explicit
processes given these parameters. However, learning involves more than immediate processing. It
also involves consolidation of patterns and linkages between comprehension and production. If one
looks at information processing across longer timeframes (MacWhinney, 2005), then there is far
more interaction and non-modularity and more evidence for the interaction of implicit and explicit
processes. Given these interactions at longer timeframes, we cannot regard learning as involving
either purely implicit or purely explicit processes. Moreover, as Conway (2020) has shown, SL is not
computed by a single module. Rather, it involves the interaction of several neural areas or modules.
Viewing learning as involving a single implicit module and a single explicit module fails to recognize
the complexity of these interactions.
Godfroid explores the interactions of explicit and implicit processes from two viewpoints. Firstly,
she examines Cleeremans’ radical plasticity lifespan model as one way of understanding these rela-
tions. However, that model passes over the fact that proceduralization, representational redescrip-
tion, and associative consolidation (MacWhinney, 2017b) must be initiated immediately after an
attended form is being received, although these processes could then continue through mechanisms
that refresh the original perception (Edelman & Gally, 2013; Wittenberg et al., 2002). These initial
moments can arise through practice, refection, error detection, or even covert processes during sleep
(James et al., 2017). In all these cases, diverse cortical areas and consolidation processes operate in
tandem to move representations closer to the target language. Godfroid cites studies suggesting that
both implicit and explicit knowledge could be acquired simultaneously. However, she also notes that
the evidence for such simultaneity is difcult to replicate. This difculty may refect problems with
devoting attention to two streams of processing simultaneously.
Second, Godfroid examines the interface between explicit and implicit processes in terms of the
ways in which explicit instruction and knowledge can guide attention and learning. The fndings
here seem quite consistent. Explicit methods such as processing instruction (VanPatten, 2004) or
eCALL (MacWhinney, 2017a) can lead to rapid and robust learning of new language patterns. As
Morgan-Short and Ullman note in their DP model, it could also be the case that learning which is
initially explicit and declarative becomes implicit and proceduralized over time. Here, again, under-
standing the relation between implicit and explicit processes involves focusing on efects across vary-
ing timescales.
To better understand the online interplay between implicit and explicit representations, it could be
helpful to study specifc processing paths longitudinally in individual learners. This would allow us to
understand not only whether implicit and explicit knowledge interact, but also the processes through
which they interact. For example, one could track the L2 learning of the system of German gender-case-
number (GCN) marking by looking at developments in article marking over time for specifc words in
specifc GCN confgurations. For some German nouns, such as Mutter “mother” with inherent feminine
gender or Ente “duck” with a fnal -e cue, the cues to use of die as the nominative singular feminine
363
Brian MacWhinney
defnite article are so clear that their gender status can be proceduralized very early on. However,
these well-learned nouns may also appear in a less common context, such as in the dative after the
preposition mit “with” when they take the form mit der Ente. In that case, the learner can notice
the use of der and begin to sense the ways in which feminine nouns take this alternative form of the
article which also marks a singular masculine noun in the nominative. This is the type of noticing
emphasized by Long in his chapter. To the degree that L1 child learners acquire phrases such as mit
der Ente as units (MacWhinney, 1978; Peters, 1983), they can compare successive versions of feminine
nouns with this and other prepositions to solidify the pattern. Accurate processing of this pattern can
support both comprehension and production of dative articles with new nouns and new prepositions.
The fact that young German children make errors when extending patterns like this to new nouns
and prepositions indicates that they are acquiring such patterns productively. Given their weaker
encoding of prepositional phrases as wholes, adult L2 German learners may require more explicit
attention to these relations than do German children. However, adult L2 learners can rely in part on
explicit statements about these patterns to scafold their solidifcation, whereas children must rely on
basic memory comparison processes for detection of similarities (McClelland, 2015). To evaluate this
process, experiments could display lexical items such as mit, der, die, Ente, or Mutter and their use in
various patterns. In the implicit learning condition, noun phrases would be given without cueing or
attentional focus. In the explicit learning condition, the relations between noun phrases with similar
genders would be explicitly noted. It would then be possible to evaluate the degree to which learning
is facilitated by explicit noticing and methods for overtly juxtaposing forms.
The role of working memory

Godfroid’s emphasis on the interaction between implicit and explicit processes leads us into the fur-
ther consideration of the role of working memory as a likely locus of the interaction of these pro-
cesses. In his chapter, Li (2023 [this volume]) explores in some detail the Baddeley model of working
memory with its four components: the phonological loop, the visual-spatial sketchpad, the episodic
bufer, and the central executive. A variety of experimental tasks have been used to measure individual
diferences in the functioning of these four components, as well as the basic operation of each. More-
over, each of these four components can be further dissected into smaller components. For example,
executive processes can be dissected into inhibition, shifting, and updating (Miyake et al., 2000) and
the phonological loop involves rehearsal, storage, and consolidation (Gupta & MacWhinney, 1997).
Within this four-component framework, the phonological loop may play an important role for
L2 learning. Gupta and MacWhinney (1997) presented a neurocomputational model of process-
ing in the phonological loop that accounted for data on learning of new vocabulary, as well as
individual diferences in learning. They argued that the operation of this loop could be funda-
mental to the human ability to acquire new vocabulary. Baddeley (2015) seconds this view. In a
sense, working memory can be viewed as occupying a central role that coordinates the efects of
noticing, attention, explicit cues, interactive activation (Schlichting & Preston, 2015), and gradual
proceduralization.
Given its potential centrality, it is remarkable that it has been difcult to demonstrate a strong and
across-the-board impact of working memory on L2 learning (Linck et al., 2014). One possible reason
for this could be that nearly all learners have sufcient access to working memory resources to allow
for basic L2 lexical learning. The exception to this arises in people with neural damage impacting
the operation of the phonological loop component of working memory. For example, Gupta and
MacWhinney (1997) found that children with early focal lesions were less successful at acquiring
new lexical forms than children with normal brain development. However, within the normal popu-
lation, both phonological short-term memory and executive working memory provide adequate
support for L2 learning. Executive memory processes also support reading in L2 and provide fuent
364
Synthesis
control of complex L2 speaking tasks. But these efects are only indirectly related to the acquisition
of new knowledge.
Understanding proceduralization and automaticity in L2 acquisition

The chapters by Suzuki (2023 [this volume]) and Morgan-Short and Ullman (2023 [this volume])
examine the ways in which L2 knowledge becomes automatized or proceduralized as L2 skill. As
noted earlier, the emphasis on this issue is stimulated by the fact that adult learners typically fail to
achieve full L2 fuency. One possibility is that this is due to a reduction in procedural learning ability
in adulthood. However, there are several other factors that may lead to this same age-related result,
including progressive entrenchment of L1, lessened time for L2 learning, exclusion from L2 social
groups, lower levels of motivation, conficting social responsibilities and roles, and inappropriate
learning methods (MacWhinney, 2017b). However, there are also good reasons to explore declines
in procedural memory as one source of age-related problems in acquiring fuent control of L2.
Following DeKeyser (2007), Suzuki examines which aspects of practice contribute most efec-
tively to L2 fuency. His approach shares much in common with the emphasis on deliberative practice
for the acquisition of expertise in domains such as science, medicine, art, and sport (Ericsson, 2006).
This theory holds that it is not enough to merely practice through repetition. Instead, practice must
include specifc attentional focusing on detailed aspects of the skill to be learned. As Long notes in his
chapter, the same must hold true for L2 learning. Research on automatization often considers ways
in which automatization can best be promoted. This type of research has clear implications for SLA
theory and pedagogy. One issue is whether vocabulary acquisition can arise from incidental exposure
to new words during reading. As Suzuki notes, the fnding is that incidental exposure is not efective
in promoting automaticity, whereas deliberative vocabulary practice leads consistently to automatiza-
tion. Although deliberative vocabulary practice may work well for acquiring an initial mapping of a
new sound to a meaning, learners will need further encounters with the word to appreciate its use
across contexts.
Theories of automatization or proceduralization all view automatic processes as fast, efcient,
efortless, stable, and relatively independent of conscious control (Anderson et al., 2019). These
theories identify the basal ganglia as the controllers of automatic skill (Stocco et al., 2010). This
linkage of skills to the basal ganglia works well when we think about those aspects of language that
involve serial positioning, such as grammatical relations, syntactic constructions, and fuent sequence
production. It could also apply to the sequencing of articulatory gestures (Browman & Goldstein,
1992). However, it is less clear that the basal ganglia would be involved in the acquisition of lexical
skills which are processed within the temporal cortex (Kemmerer, 2015; Paradis, 2009). For such
skills, models that deal with consolidation through the hippocampal system (Kumaran et al., 2016)
seem more appropriate.
Another issue regards the optimal schedule for repeated practice of new skills. Research in this
area has consistently demonstrated the value of practice that is spaced increasingly further apart over
time (Pavlik et al., 2008). However, Suzuki notes that shorter-spaced practice with forms in a min-
iature linguistic system (MLS) is particularly favorable to higher retention of procedural knowledge.
For declarative knowledge a spacing of seven days is optimal, whereas for procedural knowledge in
the MLS a delay of one day is best (Suzuki, 2017) Suzuki suggests that proceduralization of MLS
structures depends on repetition before existing knowledge deteriorates. It would be interesting
to investigate the neurological basis of this efect as a way of comparing consolidation in the basal
ganglia procedural system with consolidation in the hippocampal system. In another series of stud-
ies, Suzuki examined the efects of spacing on the development of fuency in a narrative task. In
this case, constant practice showed greater fuency gains than variable practice. These studies of
spacing efects for promoting proceduralization are interesting. However, the materials involved are
365
Brian MacWhinney
extremely variable, ranging from vocabulary and miniature linguistic systems to spoken narratives. It
is unlikely that a single theory of proceduralization could cover such varied structures consistently.
Moreover, it is not clear that all these activities result in proceduralization or how these results would
apply to natural L2 learning.
The chapter from Morgan-Short and Ullman (2023 [this volume]) examines the implications of
the declarative/procedural (DP) model for L2 learning. The model’s formulation of the operations
of the hippocampal-cortical declarative system and the striatal-cerebellar-thalamic procedural system
is well supported by both behavioral and brain studies. However, some aspects of the model could
receive further clarifcation. For example, the model claims that the two systems operate in com-
petition. This claim is in line with the dual-route model of lexical formation (MacWhinney, 1978;
Pinker, 1998; Stemberger & MacWhinney, 1986) which recognizes a distinction in and competition
between rote retrieval and combinatorial formation. However, the competition between rote or
whole form memory and combinatorial morphological formation involves, not so much blocking
of one route by the other, as much as a competition to reach threshold frst, as specifed in the drift-
difusion model (Ratclif et al., 2016). The DP model also treats hippocampus-based learning as
uniformly rapid. However, recent accounts of the complementary systems approach to hippocampal
functioning (Kumaran et al., 2016) emphasize hippocampal control of both fast and slow memory
consolidation, often arising after periods of sleep. In addition, there are mechanisms of perceptual
learning that operate through direct cortical encoding (Hebscher et al., 2019). Because direct cortical
learning depends on repetition of known stimuli, it could be involved in processes such as statistical
learning of phonotactics (Safran & Thiessen, 2003) for both L1 and L2.
The DP model emphasizes the arbitrary nature of declarative learning. It is true that the relation
between the sound and meaning of a monomorphemic lexical item is largely arbitrary. However, the
fact that sound is being hooked to meaning makes lexical learning very diferent from the learning
of meaningless strings, such as those involved in AGL experiments. The DP model also provides
roles for both pattern formation and imageability in declarative learning. Given this, it seems that the
distinguishing feature of declarative learning is not so much its arbitrariness, but rather its associative,
as opposed to sequential, quality. Neuroimaging studies (Schlichting & Preston, 2015) and related
computational models (Wittenberg et al., 2002) have shown that it is the ability of hippocampal pro-
cessing to form associations that leads most clearly to cortical consolidation of declarative memories
(MacWhinney, 2017b).
Although the DP model holds that older learners rely more on declarative learning than proce-
dural learning, it allows for some level of procedural learning in adulthood. In that regard, the model
makes less of a commitment to critical period limitations for L2 learning than do similar accounts
from DeKeyser (2007) and Paradis (2009). Given increasing evidence from neuroscience regarding
mechanisms of cortical plasticity (Gervain et al., 2013; Werker & Hensch, 2014), this refocusing
seems appropriate.
Interaction and the Competition Model

Long’s chapter (2023 [this volume]) builds a bridge between current psycholinguistic approaches to
L2 learning and earlier developments in the feld. He reviews evidence that emphasizes the impor-
tance of face-to-face interaction as the most powerful driver of L2 learning. This emphasis then leads
directly to a consideration of the specifc features of conversation that drive learning. Here, Long
points to central roles for (1) negotiation for meaning, (2) input elaboration, (3) noticing of new
forms and knowledge gaps, and (4) recasting. Studies have shown that utilization of each of these
dimensions can facilitate L2 learning.
Somewhat remarkably, these fndings from the SLA literature align exactly with identical fndings
for L1 child language learning. Firstly, Tomasello (2003) and Bloom (2002) cite dozens of studies
366
Synthesis
showing how children’s word learning depends on shared focus of attention and negotiation for mean-
ing, much as Long describes it. For the early acquisition of phonological contrasts, Kuhl (2007) shows
that learning only occurs when infants interact directly and meaningfully with other humans. Sec-
ondly, as Sokolov (1993) showed using data from the CHILDES database (MacWhinney, 2000), par-
ents also engage in input elaboration of the type described by Long. Thirdly, child language researchers
have argued that children acquire new grammatical patterns by noticing failures to parse (Berwick,
1987), divergences between input and their internal grammar (MacWhinney, 2004; Saxton, 1997;
Yang, 2016), and links between new forms and functions (Golinkof et al., 1999). Fourthly, child
language researchers have identifed recasts and scafolding as powerful methods for extending children’s
grammars (Merriman, 1999; Waterfall et al., 2010).
These parallels between L1 and L2 learning provide strong support for the claim of the Unifed
Competition Model (UCM) that L1 and L2 learning rely on fundamentally similar processes. In line
with the emphasis on noticing from Suzuki, Long, and many others, the UCM views attention as
crucial for new learning. The gateway function of noticing for learning extends to both comprehen-
sion and production, as well as writing (Hanaoka & Izumi, 2012; Uggen, 2012). The UCM holds
that, if the cues predicting patterns are clear and simple, explicit rule formulation can facilitate learn-
ing (MacWhinney, 1997; Presson et al., 2014).
The most recent emergentist formulation of the model aligns well with Ullman’s DP model in
terms of separating out various forms of neural support for diferent levels of linguistic structure.
In accordance with the overall framework of emergentism (MacWhinney, 2015), the UCM holds
that the emergence of structures on diferent linguistic levels during L2 learning is constrained by
processes operative on those levels. This means, for example, that the learning of lexical structures is
constrained by the slow and fast operations of the hippocampal system, that learning of phonotactics
is constrained by SL in the auditory system, that learning of syntactic constructions is constrained
by sequential processing in the striatal-cerebellar-thalamic procedural system, and that learning of
conversational and narrative patterns is constrained by processes in the frontal system (Koechlin &
Summerfeld, 2007). The UCM views implicit learning as primarily arising during consolidation
and pattern linking over time in these various neuronal systems. This consolidation can arise either
through memory pruning during sleep (James et al., 2017) or repeated use of patterns in support of
comprehension and production.
Methodological considerations
In their survey of the evidence in support of the DP model, Morgan-Short and Ullman note that
psycholinguistic studies of L2 learning have focused far more on processing than learning. Processing
involves decisions made online as sentences are comprehended or produced. On the other hand,
learning of even a single form can extend over days, weeks, or months. In the laboratory, we have
a much greater ability to control stimuli and monitor results in processing studies. However, if our
goal is to understand L2 learning, we need to focus more directly on learning. Recent advances in
computer technology now make it possible to do just this in ways not possible before. Both learners
and instructors are now making increasing use of online language learning systems (Li & Lan, 2022;
MacWhinney, 2022). When these systems are created and deployed by researchers (MacWhinney,
2017a), learners’ progress can be tracked over time. Using counterbalanced Latin square designs, users
can be assigned to conditions that allow for within-participant statistical evaluation and hence better
control of individual diferences. We can also use ABBA designs to see whether exposure to some
pattern in block A results in improvements in performance in the following block B and then check
for order efects by looking at the BA order. Designs of this type can track learning within individual
learners. Improved methods for transmitting audio over the web allow for interactive methods and
for studies of changes in fuency. Increases in bandwidth allow for delivery of video methods, as well
367
Brian MacWhinney
as increasingly accurate measurement of stimulus presentation times as well as reaction times. These
methods can be applied to learning of individual target language structures in auditory perception,
articulation, lexicon, morphology, syntax, narrative, and pragmatics. Or they can be used to provide
instruction and input integrated across all domains. Eventually, these methods can be linked to virtual
reality delivery instruments (Li & Lan, 2022) and they can track the use of L2 in real life interactions,
as well as processing of information through QR codes. Data collected through these methods can be
stored in databases in ways that protect privacy, while still allowing for personalization of instruction
and the construction of individual user models (Kowalski et al., 2014).
We can adapt these methods to examine the core issues considered in these six chapters. The
materials can vary implicit and explicit presentation with and without corrective feedback or further
explanations. They can measure short-term memory to study the impact of individual variations in
memory on learning of alternative structures. They can track the course of declarative and procedural
learning in learners of varying genders, ages, and L1 backgrounds. These new methods will not fully
replace the laboratory methods currently in use. We will still need to rely on laboratory methods for
assessing neural functioning. However, these online methods can provide us with a fuller understand-
ing of the longer-term aspects of L2 learning. They will also allow us to reach a more diverse group
of learners with a wider set of instructional materials to understand the many paths of L2 learning.
References
Ambridge, B. (2020). Against stored abstractions: A radical exemplar model of language acquisition. First Lan-
guage, 40(5–6), 509–559.
Anderson, J. R., Betts, S., Bothell, D., Hope, R., & Lebiere, C. (2019). Learning rapid and precise skills. Psy-
chological Review, 126(5), 727.
Baddeley, A. (2015). Working memory in second language learning. In Z. Wen, M. Mota, & A. McNeil (Eds.),
Working memory in second language acquisition and processing (pp. 17–28). Multilingual Matters.
Berwick, R. (1987). Parsability and learnability. In B. MacWhinney (Ed.), Mechanisms of language acquisition.
Lawrence Erlbaum Associates.
Bloom, P. (2002). How children learn the meanings of words (Learning, Development, and Conceptual Change). MIT
Press.
Browman, C. P., & Goldstein, L. (1992). Articulatory phonology: An overview. Phonetica, 49, 155–180.
Conway, C. M. (2020). How does the brain learn environmental structure? Ten core principles for understand-
ing the neurocognitive mechanisms of statistical learning. Neuroscience & Biobehavioral Reviews, 112, 279–299.
Cowan, N. (1992). Verbal memory span and the timing of spoken recall. Journal of Memory and Language, 31,
668–684.
DeKeyser, R. (Ed.). (2007). Practice in a second language: Perspectives from applied linguistics and cognitive psychology.
Edelman, G. M., & Gally, J. A. (2013). Reentry: A key mechanism for integration of brain function. Frontiers in
Integrative Neuroscience, 7, 63.
Ericsson, K. A. (2006). The infuence of experience and deliberate practice on the development of superior
expert performance. The Cambridge Handbook of Expertise and Expert Performance, 38(685–705), 2–2.3.
Fodor, J. (1983). The modularity of mind: An essay on faculty psychology. MIT Press.
Friederici, A., Steinhauer, K., & Pfeifer, E. (2002). Brain signatures of artifcial language processing: Evidence
challenging the critical period hypothesis. Proceedings of the National Academy of Sciences, 99, 529–534.
Gervain, J., Vines, B. W., Chen, L. M., Seo, R. J., Hensch, T. K., Werker, J. F., & Young, A. H. (2013). Valproate
reopens critical-period learning of absolute pitch. Frontiers in Systems Neuroscience, 7, 102.
Goldinger, S. D. (1998). Echoes of echoes? An episodic theory of lexical access. Psychological Review, 105(2), 251.
Goldinger, S. D. (2007). A complementary-systems approach to abstract and episodic speech perception. Proceedings of
the 16th International Congress of Phonetic Sciences (pp. 49–54). Saarbrücken University Press.
Golinkof, R., Hirsh-Pasek, K., & Hollich, G. (1999). Emergent cues for early word learning. In B. MacWhin-
ney (Ed.), The emergence of language (pp. 305–330). Lawrence Erlbaum Associates.
368
Synthesis
Guenther, F., & Perkell, J. (2003). A neural model of speech production and its application to studies of the role
of auditory feedback in speech. In B. Maasen, R. D. Kent, H. Peters, P. van Lieshout, & W. Hulstijn (Eds.),
Speech motor control in normal and disordered speech (pp. 29–50). Oxford University Press.
Gupta, P., & MacWhinney, B. (1997). Vocabulary acquisition and verbal short-term memory: Computational
and neural bases. Brain and Language, 59(2), 267–333.
Hanaoka, O., & Izumi, S. (2012). Noticing and uptake: Addressing pre-articulated covert problems in L2 writ-
ing. Journal of Second Language Writing, 21(4), 332–347.
Hebscher, M., Wing, E., Ryan, J., & Gilboa, A. (2019). Rapid cortical plasticity supports long-term memory
formation. Trends in Cognitive Sciences, 23(12), 989–1002.
James, E., Gaskell, M. G., Weighall, A., & Henderson, L. (2017). Consolidation of vocabulary during sleep: The
rich get richer? Neuroscience & Biobehavioral Reviews, 77, 1–13.
Kemmerer, D. (2015). The cognitive neuroscience of language. Psychology Press.
Kim, R., Seitz, A., Feenstra, H., & Shams, L. (2009). Testing assumptions of statistical learning: Is it long-term
and implicit? Neuroscience Letters, 461(2), 145–149.
Koechlin, E., & Summerfeld, C. (2007). An information theoretical approach to prefrontal executive function.
Trends in Cognitive Sciences, 11, 229–235.
Kowalski, J., Gordon, G., & MacWhinney, B. (2014). Statistical modeling of student performance to improve
Chinese dictation skills with an intelligent tutor. Journal of Educational Data Mining, 6, 3–27.
Krashen, S. (1994). The input hypothesis and its rivals. In N. C. Ellis (Ed.), Implicit and explicit learning of languages
(pp. 45–78). Academic.
Kuhl, P. (2007). Is speech learning “gated” by the social brain? Developmental Science, 10(1), 110–120.
Kumaran, D., Hassabis, D., & McClelland, J. L. (2016). What learning systems do intelligent agents need?
Complementary learning systems theory updated. Trends in Cognitive Sciences, 20(7), 512–534.
Li, S. (2023). Working memory and second language learning: a critical and synthetic review. In A. Godfroid & H.
Li, P., & Lan, Y.-J. (2022). Digital language learning (DLL): Insights from behavior, cognition, and the brain.
Linck, J. A., Osthus, P., Koeth, J. T., & Bunting, M. F. (2014). Working memory and second language compre-
hension and production: A meta-analysis. Psychonomic Bulletin & Review, 21(4), 861–883.
Long, M.H. (2023). The psycholinguistics of L2 interaction. In A. Godfroid & H. Hopp (Eds.), The Routledge
MacWhinney, B. (1978). The acquisition of morphophonology. Monographs of the Society for Research in Child
Development, 43(Whole no. 1), 1–123.
MacWhinney, B. (1997). Implicit and explicit processes. Studies in Second Language Acquisition, 19, 277–281.
MacWhinney, B. (2000). The CHILDES project: Tools for analyzing talk (3rd ed.). Lawrence Erlbaum Associates.
MacWhinney, B. (2004). A multiple process solution to the logical problem of language acquisition. Journal of
Child Language, 31(4), 883–914.
MacWhinney, B. (2005). The emergence of linguistic form in time. Connection Science, 17, 191–211.
MacWhinney, B. (2015). Emergentism. In E. Dabrowsksa & D. Divjak (Eds.), Handbook of cognitive linguistics (pp.
689–706). Mouton-DeGruyter.
MacWhinney, B. (2017a). A shared platform for studying second language acquisition. Language Learning, 67,
254–275.
MacWhinney, B. (2017b). A unifed model of frst and second language learning. In M. Hickmann, E. Vene-
ziano, & H. Jisa (Eds.), Sources of variation in frst language acquisition: Languages, contexts, and learners (pp. 287–310).
John Benjamins.
MacWhinney, B. (2022). The future of DLL. Bilingualism: Language and Cognition, 25(3), 394–395.
McClelland, J. (2015). Capturing gradience, continuous change, and quasi-regularity in sound, word, phrase, and
meaning. In B. MacWhinney & W. O’Grady (Eds.), The handbook of language emergence (pp. 53–80). Wiley.
Merriman, W. (1999). Competition, attention, and young children’s lexical processing. In B. MacWhinney
(Ed.), The emergence of language (pp. 331–358). Lawrence Erlbaum.
Miyake, A., Friedman, N., Emerson, M., Witzki, A., Howerter, A., & Wager, T. (2000). The unity and diver-
sity of executive functions and their contributions to complex “frontal lobe” tasks: A latent variable analysis.
Cognitive Psychology, 41, 49–100.
Morgan-Short, K. (2020). Insights into the neural mechanisms of becoming bilingual: A brief synthesis of sec-
ond language research with artifcial linguistic systems. Bilingualism: Language and Cognition, 23(1), 87–91.
369
Brian MacWhinney
Paradis, M. (2009). Declarative and procedural determinants of second languages (Vol. 40). John Benjamins Publishing.
Pavlik, P., Bolster, T., Wu, S.-M., Koedinger, K., & MacWhinney, B. (2008). Using optimally selected drill prac-
tice to train basic facts. In B. Woolf, E. Aimeur, R. Nkambou, & S. Lajoie (Eds.), Intelligent tutoring systems
Perruchet, P., & Pacton, S. (2006). Implicit learning and statistical learning: One phenomenon, two approaches.
Trends in Cognitive Sciences, 10, 233–238.
Peters, A. (1983). The units of language acquisition. Cambridge University Press.
Pinker, S. (1998). Words and rules. Lingua, 106(1–4), 219–242.
Presson, N., MacWhinney, B., & Tokowicz, N. (2014). Learning grammatical gender: The use of rules by nov-
ice learners. Applied Psycholinguistics, 35(4), 709–737.
Ratclif, R., Smith, P. L., Brown, S. D., & McKoon, G. (2016). Difusion decision model: Current issues and
history. Trends in Cognitive Sciences, 20(4), 260–281.
Reber, A. (1967). Implicit learning of artifcial grammars. Journal of Verbal Learning and Verbal Behavior, 6,
855–863.
Reber, A. (1976). Implicit learning of synthetic languages: The role of instructional set. Journal of Experimental
Psychology: Human Learning and Memory, 2, 88–93.
Reber, A., & Lewis, S. (1977). Implicit learning: An analysis of the form and structure of a body of tacit knowl-
edge. Cognition, 5, 333–362.
Safran, J. R., Newport, E. L., & Aslin, R. N. (1996). Word segmentation: The role of distributional cues. Jour-
nal of Memory and Language, 35, 606–621.
Safran, J. R., & Thiessen, E. (2003). Pattern induction by infant language learners. Developmental Psychology,
39(3), 484.
Saxton, M. (1997). The contrast theory of negative input. Journal of Child Language, 24, 139–161.
Schlichting, M. L., & Preston, A. R. (2015). Memory integration: Neural mechanisms and implications for
behavior. Current Opinion in Behavioral Sciences, 1, 1–8.
Smalle, E. H., Page, M. P., Duyck, W., Edwards, M., & Szmalec, A. (2018). Children retain implicitly learned
phonological sequences better than adults: A longitudinal study. Developmental Science, 21(5), e12634.
Sokolov, J. L. (1993). A local contingency analysis of the fne-tuning hypothesis. Developmental Psychology, 29,
1008–1023.
Stemberger, J., & MacWhinney, B. (1986). Frequency and the lexical storage of regularly infected forms.
Memory and Cognition, 14(1), 17–26.
Stocco, A., Lebiere, C., & Anderson, J. R. (2010). Conditional routing of information to the cortex: A model
of the basal ganglia’s role in cognitive coordination. Psychological Review, 117(2), 541.
Suzuki, Y. (2017). The optimal distribution of practice for the acquisition of L2 morphology: A conceptual
replication and extension. Language Learning, 67(3), 512–545.
Tomasello, M. (2003). Constructing a frst language: A usage-based theory of language acquisition. Harvard University
Press.
Uggen, M. S. (2012). Reinvestigating the noticing function of output. Language Learning, 62(2), 506–540.
VanPatten, B. (2004). Processing instruction: Theory, research, and commentary. Routledge.
Waterfall, H., Sandbank, B., Onnis, L., & Edelman, S. (2010). An empirical generative framework for compu-
tational modeling of language acquisition. Journal of Child Language, 37, 671–703.
Werker, J., & Hensch, T. (2014). Critical periods in speech perception: New directions. Annual Review of Psy-
chology, 66, 173–196.
Williams, J. & Rebuschat, P. (2023). Implicit learning and second language acquisition: A cognitive psychol-
ogy perspective. In A. Godfroid & H. Hopp (Eds.), The Routledge handbook of second language acquisition and
Wittenberg, G., Sullivan, M., & Tsien, J. (2002). Synaptic reentry reinforcement based network model for long-
term memory consolidation. Hippocampus, 12, 637–647.
Yang, C. (2016). The price of linguistic productivity: How children learn to break the rules of language. MIT Press.
370
SECTION V
Transdisciplinary perspectives
Contexts and future directions
30
THE INFLUENCE OF
LEARNING CONTEXTS ON THE
PSYCHOLINGUISTICS OF SECOND
LANGUAGE LEARNING
Ronald P. Leow
Introduction
The role of context in learning has been a topic of much discussion over many decades in the educa-
tional literature (e.g., Dewey, 1985; Piaget, 1995; Vygotsky, 1978). For Dewey, context is the taken-
for-granted spatial and temporal background and selective interest of the agent, upon which the
content of thought stands out (Dohn et al., 2018). On the other hand, Piaget (1995) and Vygotsky
(1978) both espoused, albeit in diferent ways, the significance of social context for children’s cog-
nitive development. Others (e.g., Bateson, 1972) view context from a communicative perspective,
that is, “Without context, there is no communication” (p. 408). Given that learning inherently is
a communicative phenomenon, the role context plays in shaping cognition is major and remains a
recurrent theme within contemporary educational research, both theoretical (e.g., Guile, 2006) and
empirical (e.g., Greeno, 2011). At the same time, Dohn et al. (2018) point out that there exists a lack
of consensus on what context is, despite common acceptance of the claim that learning is context-
dependent to some degree as exemplified in research on learning in several diferent contexts such as
museums, workplaces, and the classroom.
The role of context in (instructed) second language acquisition (ISLA), interestingly, provides
a comparative issue that is clearly premised on the primary context of learning, that is, whether
learning is shaped by outside factors (e.g., viewed from a sociocultural or transdisciplinary, multi-
perspectival perspective) or whether learning occurs within the learner (e.g., viewed from a psycho-
linguistic perspective). For example, transdisciplinary, multi-perspectival accounts of SLA “attend to
multilingual development and use across multiple scales: from macro-societal and ideological ones to
micro-interactional ones, and across diferent temporal and spatial scales” (Duf & Byrnes, 2019, p. 3).
Prominent among these is the Douglas Fir Group (DFG, 2016) article that “assumes the embedding,
at all levels, of social, sociocultural, sociocognitive, sociomaterial, ecosocial, ideological, and emo-
tional dimensions” (p. 24) to address SLA (see Hulstijn, 2023 [this volume], for more elaboration).
DFG’s multiple foci of inquiry appear to be both ecologically valid and laudable in their trans-
disciplinary scope to bring several diferent fields under the same theoretical umbrella. However,
the framework seemingly grounds itself less in the active psycholinguistic role of the second/foreign
language (L2) learner (Han, 2016) and more on multilingualism (that may not apply easily to all
populations of L2 learners) and societal dimensions and their contributions to language learning. It
may also be methodologically challenging to address the infuence of this transdisciplinary context
DOI: 10.4324/9781003018872-35 373

Ronald P. Leow
on the psycholinguistics of L2 learning, when learning is viewed from a process-oriented perspec-

tive (see Textbox 30.1), especially given the many variables postulated to play a role in L2 learning.
On the other hand, a psycholinguistic perspective of L2 learning focuses on the internal cogni-
tive processes assumed to play an important role in the learning process. Aside from the recent role
of the aforementioned transdisciplinary context in SLA, the role of context, and more specifically,
a comparison of learning contexts, in relation to how L2 learners process L2 data, has not enjoyed
much discussion despite the fact that the concept of “learning context” in itself lies subsumed in
many aspects of the ISLA literature and may play an important role in the L2 learning process. To
this end, this chapter aims to identify, in the ISLA literature, several diferent perspectives of what
comprises learning contexts and their comparative potential infuence on the psycholinguistics of
L2 learning viewed from a process-oriented perspective while at the same time addressing theoreti-
cal, empirical, methodological, and pedagogical/curricular issues that arise out of this identifica-
tion. Each section on type of learning context will be followed by pertinent empirical research
(whenever possible) and a brief summary on where the contextual research lies with respect to its
role or infuence in the psycholinguistics of L2 learning. The chapter concludes with suggestions
for future research to address the infuence of the varied types of learning contexts in relation to
a better understanding of how L2 learners process L2 information. Textbox 30.1 provides the key
terms and concepts of this chapter.
Learning contexts: These contexts may be viewed from both a local (comparisons within relatively similar
learning environments) and global (comparisons between diferent learning environments) perspective
in relation to the afordances (e.g., amount of exposure to and interaction with the L2, resources or
support) that learning environments provide to the L2 learner.1
The psycholinguistics of L2 learning: A process-oriented perspective of the use of mental or cognitive
processes (e.g., attention, noticing, awareness, activation of prior knowledge, hypothesis and rule
formulations, metacognition, etc.) during the act or process of language learning (encoding or decod-
ing incoming L2 data) as opposed to a product-oriented perspective (knowledge) of learning. This
process-oriented perspective focuses on how L2 learners process L2 data.
Theoretical psycholinguistic perspectives in ISLA

ISLA theorists have postulated several cognitive processes composing the psycholinguistics of L2
learning (see Textbox 30.1) to account for the L2 learning process (e.g., N. Ellis, 2007; Gass, 1997;
Leow, 2015; Robinson, 1995; Schmidt, 1990; Tomlin & Villa, 1994). For example, Leow’s (2015)
recent fine-grained model of the L2 learning process in ISLA, incorporating several tenets of previ-
ous underpinnings, postulates a crucial role for how L2 learners cognitively process L2 data across
several stages from input processing to intake processing to knowledge processing (see Godfroid,
2019, for the contributions of eye-tracking methodology to these stages).
The importance of understanding the psycholinguistics of L2 learning as defined in this chapter is
undebatable from several perspectives, namely, theoretical, methodological, and pedagogical. Theo-
retically, it is noteworthy in the ISLA literature that there exist two definitions of what composes
the psycholinguistics of L2 learning, namely, a process-oriented vs. a product-oriented perspective,
with the latter perspective dominant in the ISLA literature. Methodologically, there exists a large
374
The infuence of learning contexts on psycholinguistics
imbalance between psycholinguistics studies that address the product of learning and those that
investigate the process of learning. Pedagogically, there is a paucity of psycholinguistics studies, from
both product- and process-oriented perspectives, that aim to provide robust classroom ramifications.
In addition, what infuence the learning context has on the psycholinguistics of L2 learning, from
a process-oriented perspective, remains largely unknown due to both the inadequacy of research
methodology (focusing on the product to make assumptions about the process) and the definition
of such a context. For example, a learning context may be simplistically viewed as a place where
learning takes place yet, similar to the discussion in the educational literature, what composes such a
context in ISLA is open for speculation. In an efort to elucidate this issue, a learning context in this
chapter is viewed and discussed from both a local and global perspective in relation to the psycholin-
guistics of L2 learning viewed from a process-oriented perspective.
Learning context from a local perspective

A local perspective of the learning context lies in a comparison of settings within a learning environ-
ment, be it within SLA or ISLA, that are viewed as sharing overall similar afordances but diferenti-
ated by contexts that may be, for example, curricular (e.g., type of program, programmatic levels),
pedagogical (e.g., type of delivery mode of instruction and type of feedback), and methodological
(e.g., the research setting as in laboratory- vs. classroom-based).
Curricular context
The type of language curriculum (e.g., language-based vs. content-based) within both study-abroad/
immersion (SLA) and the instructed (ISLA) settings may play a role in the psycholinguistics of L2
learning. The language curriculum may impact the type of syllabus and the subsequent instructional
procedures provided in the formal setting, which may implicitly underscore the theoretical under-
pinning of L2 learning. For example, a language-based curriculum may opt for a synthetic syllabus
(analysis of language and language use) that promotes explicit learning while a content-based cur-
riculum may prefer an analytic syllabus (derived from human learning/L2 learning and task-based,
e.g., Long and Crookes’s (1993) task-based language teaching) that may refect learning of a more
implicit nature (see Tomlinson & Masahara, 2021).
The impact of either type of language curriculum or syllabus on L2 learning has not been robustly
investigated although one recent meta-analysis (Bryfonski & McKay, 2019) has reported quantitative
efectiveness of authentic task-based language teaching (TBLT) programs. However, a re-analysis of
this study (Boers et al., 2021) that adopted a narrower interpretation of what composes “task” (as the
basic unit in TBLT) recommended much caution on the original finding. Consequently, whether the
type of curriculum impacts the psycholinguistics of L2 learning remains to be empirically pursued,
that is, whether such processing will be diferential given the instructed setting.
Programmatic levels
A curricular context may also occur within a departmental language program. For example, many
language curricula in the United States have both a writing component in a language curriculum
(writing-to-learn) and an upper-level writing course (learning-to-write). In the former context, the
objective of writing in many language classes is primarily to ofer students the opportunity to practice
what has been covered in a chapter, lesson, or unit specifically in relation to grammatical point(s),
vocabulary, and content (usually the topic of the chapter just covered). Writing is also integrated
in many language curricula with the other three skills (listening, reading, speaking), once again in
375
Ronald P. Leow
relation to grammar, vocabulary, and content. In the upper-level context, the main focus of instruc-
tion is on writing in relation to diferent genres (e.g., argumentative, narrative, expository, descrip-
tive, etc.). Written corrective feedback provided may be diferential at the two levels.
To my knowledge, there is no empirical study that has addressed the infuence of this type of pro-
grammatic level from a process-oriented perspective, perhaps given the huge diference in language
proficiency of learners enrolled at these two levels. At the same time, it is of both empirical and
pedagogical interest to address how, for example, L2 writers at these two programmatic levels process
diferent types of feedback, linguistic items, and any relationship between these two variables with
respect to L2 learning (see Manchón, 2023 [this volume]).
Pedagogical context
A local learning context may also be viewed as some pedagogical efort to manipulate how the L2
input is provided to L2 learners in ISLA to promote robust learning. Following this is a report on two
of the pedagogical contexts that have attracted a relatively large amount of interest and investigation
in ISLA, namely the type of delivery mode of instruction and type of feedback.
Delivery mode of instruction

The typical delivery mode of instruction found in most ISLA environments is one taught by an
instructor, usually face-to-face (FTF) although, interestingly, this FTF format has recently changed
to mainly an online format in response to the COVID-19 pandemic in the early 2020s. Since its
origins in the 1960s, computer-assisted language learning (CALL) has led to an explosion of the
use of manipulated CALL tasks or activities (including e-tutors and videogames) or synchronous
computer-mediated communication (SCMC) in which some target form or structure underlies the
practice or communication being promoted. These strands of technology-based contexts are pre-
mised on the construct of minimal attention paid to specific linguistic items or structures in the L2
that are designed to promote L2 development (see for example, Baralt, 2013 for SCMC and Leow et
al., 2019a for CALL) and, especially in the case of several CALL tasks, to also elicit information on
the role of learner cognitive processes employed during instruction or exposure (Hsieh et al., 2016).
Recent meta-analytic research (e.g., Grgurović et al., 2013; Ziegler, 2016) tackling this technology-based
context in contrast to the FTF context has concluded that learners in CALL/SCMC conditions can par-
allel, and in some cases supersede, the learning outcomes of students in “traditional” FTF conditions in
language learning. To glean more pertinent information on how L2 learners processed feedback during
the SCMC mode, Baralt’s (2013, p. 717) analysis of learner behavior (via screen recording videos) revealed
that some participants were processing the recasts deeper as demonstrated here, in their scrolling behavior:
(1) tuvo tuvieron tuviera [scrolled up to confirm form] tuvieran [sent message]
Guided induction studies within CALL attempt to guide the learning process by manipulating how
learners explicitly process the L2 data (e.g., Cerezo et al., 2016) as they perform psycholinguistically
motivated e-tutors or video games. These studies report superior performances of the guided induc-
tion group over traditional classroom instruction. The researchers attributed these superior perfor-
mances to the high depth of processing revealed in the think-aloud protocols, namely, high levels of
awareness that include hypothesis testing and rule formulation, activation of both old and new prior
knowledge, and metacognition, which all appear to play important roles in learning targeted dif-
cult L2 structures (e.g., Spanish gustar with its five substructures; see similar findings for the Spanish
376
prepositions para and por in Leow et al., 2019 and the Chinese ba structure with its four types in
Zhuang, 2019). In addition, these psycholinguistically-motivated instructional tasks led to robust L2
retention (over 80%) beyond immediate efects (between one and two weeks after treatment). These
processes align with those involved in explicit learning of new information postulated in the intake
processing stage of Leow’s (2015) model.
These process-oriented studies, which employed concurrent verbal reports to establish learners’
cognitive processes before addressing their subsequent performances, have directly addressed the
psycholinguistics of L2 learning, as defined in this chapter. The strength of these studies lies in their
reporting of both process and product; that is, these studies reported the infuence a given instruc-
tional context had on the cognitive processes employed by L2 learners as they interacted with the L2
data during the exposure (process) and its potential impact on subsequent L2 learning or knowledge
(product). In sum, the revelations of the think-aloud protocols reported in the guided induction
(CALL) and screen recording videos in SCMC studies appear to indicate that deeper processing led
to superior performances (see Leow, 2019b for further empirical support).
Type of feedback
The role of feedback, and more specifically, corrective feedback (CF), in the L2 learning process has
permeated the ISLA literature for decades. Corrective feedback is a response (oral, written, comput-
erized, or digital, in the L1 or L2) that is provided by a teacher, a researcher, or a peer in reaction
to an error committed by the L2 learner (Leow, 2020). Many studies have relied on L2 learners’
performances on post-exposure assessment tasks to infer how some type of CF was processed while
others, especially in the written or computerized mode, have also employed concurrent data elicita-
tion procedures (e.g., think-aloud protocols, eye-tracking, and screenshots) to gather some data on
the actual feedback processing.
Meta-analyses conducted to investigate the pedagogical efectiveness of oral feedback (e.g., Mackey
& Goo, 2007; Li, 2010; Lyster & Saito, 2010; Russell & Spada, 2006) and written feedback (Kang
& Han, 2015; Russell & Spada, 2006) have generally reported an overall medium efect size for its
beneficial role in L2 learning. In addition, while written corrective feedback (WCF) appears to be
more beneficial than oral CF (Russell & Spada, 2006), there are inconsistent results for type of oral
CF, whether implicit versus explicit (Li, 2010) or prompts versus recasts (Li, 2010; Lyster & Saito,
2010; Mackey & Goo, 2007). However, to address how this type of pedagogical context impacts how
the feedback was indeed processed, CALL (e.g., Hsieh, 2019; Moreno, 2019) and WCF (e.g., Qi &
Lapkin, 2001; Caras, 2019) studies employing think aloud protocols have revealed more pertinent
information on such processes, for example, by showing that higher depth of processing, even in
assumed implicit learning conditions, leads to superior performances.
The overarching finding reported by these feedback studies that methodologically addressed the
psycholinguistics of L2 learning from a process-oriented perspective appears to indicate that L2 learn-
ers process diferentially while exposed to diferent types of feedback. Regardless of type of feedback,
however, a higher depth of feedback processing appears to contribute to superior retention in L2
learning, a finding that aligns with those of the delivery mode of instruction studies discussed already.
Methodological context
The location of where ISLA research is conducted provides yet another type of learning context that
may have an impact on the psycholinguistics of L2 learning. This can be viewed from the research
setting, design of the study, and a curricular perspective that addresses the direction of the research
findings.
377
Ronald P. Leow
Research setting and design

The classic distinction lies between studies conducted in the classroom setting and those in a lab-
oratory-like setting. A classroom-based context can also be divided into one in which there is an
intervention related or unrelated to the syllabus and curricular goals. The advantages of conduct-
ing research in a laboratory-like setting, unlike the classroom setting, lie in creating a more robust
research design that controls many variables while allowing the collection of concurrent data that
may not be possible in the natural classroom setting. The major disadvantage lies in whether any
potential pedagogical implication can be derived from the findings in relation to the very classroom-
based context it aspires to replicate (Rogers & Cheung, 2020; Rogers & Révész, 2020).
Similarly, type of research design (one-shot versus longitudinal) may also constitute another type
of learning context. A one-shot design, as so aptly named, is one that attempts to gather data in a
learning context that does not have any connection to the syllabus or curricular goals of the class-
room setting. One-shot studies may be conducted within a laboratory setting or in the actual class-
room setting but the classroom is used solely for data collection. A longitudinal design is one that
takes place over a long and extended period of time and is usually viewed as quasi-experimental. A
curricular longitudinal design, on the other hand, is a part of the syllabus and is built to address more
faithfully what transpires in an authentic learning context. It is situated within the language curricu-
lum with pedagogical ramifications sought from the findings.
Applied ISLA vs. ISLA applied

A curricular perspective of the methodological context has, according to Leow (2019a), created two
empirical sub-contexts, which he calls applied ISLA and ISLA applied, with regard to the research
settings and designs of the studies. According to Leow, applied ISLA comprises studies that inves-
tigate the potential efects of many variables in the instructed setting (e.g., the role of WCF in L2
development in a laboratory setting) but are usually not situated within the language curriculum.
ISLA applied (which Leow prefers to call instructed language learning or ILL) seeks to inform
practical pedagogical practice by situating this same provision of WCF within the syllabus or lan-
guage curriculum. Both sub-contexts of ISLA research are important in their respective way and
can also potentially feed each other regarding L2 learning and the potential success of instructional
interventions.
Empirical research
Meta-analyses (Mackey & Goo, 2007; Li, 2010) have reported larger efect sizes for the laboratory
setting when compared to the classroom setting. However, it is not feasible to empirically address the
efects of type of research setting and type of design on the psycholinguistics of L2 learning viewed
from a process-oriented perspective due to methodological issues (e.g., control of time, exposure,
concurrent data gathering, etc.). At the same time, it is quite important to support a move away from
the laboratory-like setting and one-shot design in some strands of research (e.g., WCF) to studies
situated within the language curriculum that can potentially provide direct pedagogical ramifica-
tions (Leow, 2020; Manchón & Leow, 2020; Rogers & Cheung, 2020; Rogers & Révész, 2020).
Similarly, based on the robust learning outcomes evidenced in the guided induction studies (Cerezo
et al., 2016; Leow et al., 2019; Zhuang, 2019), well-designed video game-based instruction could
replace FTF instruction, which would allow teachers to migrate complex L2 material online to free
up classroom time for communicative practice (Leow, 2018).
378
The learning context from a global perspective

A global perspective of the learning context lies in a comparison of learning environments that are
viewed as providing afordances that clearly diferentiate the potential roles of learning contexts in the
psycholinguistics of L2 learning, viewed from a process-oriented perspective. These learning con-
texts include comparisons between naturalistic/informal vs. instructed/formal, study abroad (SA) or
immersion vs. at home (AH), and the educational environment that comprises elementary vs. high
school vs. tertiary levels, and foreign vs. second language learning.
Naturalistic vs. instructed environment

The naturalistic or informal environment represents the place in which first, second, third, etc. lan-
guage acquisition is assumed to take place, an environment studied under the SLA field. This environ-
ment is “natural in the sense that the language (or languages, as in the case of bilingualism, etc.) to
be acquired is used for communication between the speakers of that language. It possesses all of the
cultural, pragmatic, and social ingredients that assist in developing one’s membership in this com-
munity, and the amount of exposure to and interaction with the language(s) are arguably ideal and
unparalleled for natural acquisition.” (Leow, 2019a, p. 485).
The instructed or formal environment is often but not always subsumed under ISLA. It is well
acknowledged that afordances provided in this formal environment, for example, the amount of
and exposure to the L2, are clearly incomparable to those found in the naturalistic environment. At
higher educational levels, the instructed environment is also constrained by being situated within a
language curriculum that has learning outcomes to be achieved by students (Leow, 2019a). If viewed
from an acquisition perspective, as discussed later, the quantity of input and the opportunity to use the
language freely are typically described as being inadequate or extremely limited (e.g., Leow, 2019a;
Lightbown & Spada, 2019).
With respect to the psycholinguistics of L2 learning, the crucial distinctions between the natu-
ralistic and instructed environments lie in how the L2 is processed and where the environment
is situated, given the huge contextual disparity regarding the two contexts’ afordances (Leow &
Cerezo, 2016; Leow, 2019a). Leow and Cerezo call for a distinction to be clearly made between
acquisition (SLA) and learning (ISLA) (see Krashen, 1982, for this distinction related to implicit vs.
explicit knowledge), constructs typically confated in the literature. If seriously viewed from a
processing perspective, it is not debatable that the early stages of the acquisitional process are puta-
tively efortless, with a minimal amount of cognitive or mental efort, and occurs in a naturalistic
L1 environment in which exposure to and interaction with the L1 is, as mentioned earlier, ideal
and unparalleled.
From a contextual perspective, ISLA is typically situated within a language curriculum with its
learning outcomes, which logically warrants serious consideration of the type of processing that usu-
ally takes place in this environment. More specifically, the typical formal environment situated within
a language curriculum is designed to promote more explicit and intentional learning than implicit
and incidental learning and acquisition (Leow & Zamora, 2017).
Study abroad/immersion vs. at home

Studying abroad (SA) is generally defined as the ability to partially complete a degree program
through educational courses outside the home country. While studying abroad subsumes immer-
sion environments, the latter can take place in one’s country in language programs that simulate the
379
Ronald P. Leow
naturalistic environment. For example, in the United States, Middlebury College holds a highly
respected immersion language program annually during the summer. At the same time, there are
other variables that diferentiate this SA environment, for example, length, place, or type of resi-
dence, levels of language proficiency, type of curriculum (e.g., content-based vs. language-based),
type of student population (e.g., L2 learners vs. heritage learners) and so on. The SA context also
provides important afordances such as multiple types of exposure to the L2 and opportunities to
engage in natural interactions with native speakers.
Empirical research
To my knowledge and not surprisingly for obvious methodological issues, there is no published
empirical study that has compared the infuence of the naturalistic environment with that of the
instructed environment from a process-oriented perspective of L2 learning. Early comparative stud-
ies between these two learning environments addressed the product (naturalistic and classroom-based
data) of L2 learning (e.g., German word order rules) in relation to the route of acquisition (e.g.,
Lightbown, 1983) and the impact of instruction on the sequence of acquisition (e.g., R. Ellis, 1989).
Studies have also addressed mostly the potential impact of studying abroad (SA) versus at home
(AH) on adult L2 learners’ ofine performances (e.g., Laford & Uscinski, 2014; Shively, 2014). While
overall results appear to indicate some impact of the SA environment on learners’ performances when
compared to the AH context, how such L2 learning took place is typically not addressed. Indeed, it
is also noted that in many instances the adult L2 learners who participate in these SA or immersion
environments already possess some advanced knowledge of the L2, which in turn may impact how
they process the L2 in that environment when compared to the process of L1 acquisition in the same
natural learning environment.
Educational environment
The types of educational environments (e.g., elementary vs. high school vs. tertiary levels and foreign
vs. second language learning) may also provide another perspective of a global learning context. At
the same time, these two environments may arguably overlap somewhat with the perspective of a
local learning context.
Elementary vs. high school vs. tertiary levels

The first educational environment is observed in the comparison between elementary versus high
school versus tertiary educational levels that are likely to infuence how the L2 is processed and
learned. For example, at the elementary level of foreign language instruction in the United States,
exposure to the L2 may be less constrained by curricular demands such as obtaining a passing grade
in the L2 when compared to high school or tertiary levels. This, in turn, may allow the L2 to be
processed at a lower depth of processing, or with less cognitive efort employed during exposure to
the L2, especially if learners are less motivated. On the other hand, the majority of L2 learners at
the tertiary level typically take a language requirement as part of their academic program and will
be more cognitively engaged in learning that L2 to obtain minimally a passing grade. Potentially
also playing an important role in these three educational environments may be that of age that, for
example, may play a maturational or attentional role in L2 learning.
Only one meta-analysis on corrective feedback (Lyster & Saito, 2010) addressed the potential impact
of this type of educational level on L2 learning but did not compare these three levels as three discrete
380
groups. They instead considered age as a continuous variable in a simple regression analysis designed
to examine negative linear relationships between the efects of CF and learners’ age. They reported
that younger learners benefited more from CF than older learners. However, whether this type of
educational learning context infuenced how learners processed the L2 data remains unexplored.
Foreign language vs. second language environments

The educational context of learning a foreign language (FL) versus learning a second language (SL)
reveals a huge diference in the amount of exposure to and interaction with the L2. The first envi-
ronment is the typical academic setting that usually falls under the auspices of FL requirements and
is populated by a group of students from which many ISLA studies draw for their study populations.
Students attend FL classes several times a week for approximately one hour each session in their own
country. The major portion of their FL exposure is usually in the classroom setting, outside which
they are once more surrounded by their L1, and completing homework assignments based on their
textbooks and workbooks. Students in SL classes are non-native L2 speakers attending classes in
the L2 country and arguably more motivated when compared to FL learners. While they are also
exposed to the L2 in the formal setting, this exposure continues outside the classroom setting, which
can be viewed as an immersion or a naturalistic learning environment, albeit still situated within a
language curriculum and guided by formal instruction.
Meta-analyses on corrective feedback (Mackey & Goo, 2007; Li, 2010; Lyster & Saito, 2010) that
have addressed the potential impact of the FL versus SL educational contexts on L2 learning report
inconsistent findings in the oral mode with either larger efect sizes for the FL context (Li, 2010;
Mackey & Goo, 2007) or similar efect sizes for the FL and SL contexts (Lyster & Saito, 2010). In
the written mode (Kang & Han, 2015), a larger efect for SL was reported. Once again, whether
this type of educational learning context infuenced how learners processed the L2 data remains
unexplored.

In the ISLA field, the majority of studies continue to focus primarily on the product of assumed
cognitive processes infuenced by learning context and only a relatively small number has sought to
elicit concurrent data to explicate the psycholinguistics of L2 learning, namely, how L2 data were
processed. At the same time, there has been a noticeable uptick in ISLA studies that probe deeper
into the psycholinguistics of L2 learning, as defined in this chapter (see Sachs, Baralt, & Gurzynski-
Weiss, 2023 [this volume]). However, as observed earlier, there is a tremendous paucity of data on the
infuence of learning contexts on the psycholinguistics of L2 learning in both SLA and ISLA. While
this may be more evident within a global perspective of a naturalistic learning environment, there
are some global contexts that do provide avenues for future research to address in relation to their
impact on the psycholinguistics of L2 learning. For example, previous studies that have employed
a product-oriented approach to educational contexts (e.g., elementary vs. high school vs. tertiary
levels, foreign vs. second language learning) on corrective feedback may be replicated via an online
platform while employing concurrent data elicitation procedures such as think-aloud protocols and
eye-tracking. There is also a plethora of independent variables (e.g., linguistic item, language profi-
ciency, individual diferences, age, etc.) that need to be addressed urgently in these two educational
environments, including the addition of heritage speakers and how they process their L1 in a formal
setting (see DeRobles, 2019 for a recent study on heritage speakers and the psycholinguistics of
learning).
381
Ronald P. Leow
While it is indeed methodologically challenging to investigate the infuence of naturalistic ver-

sus instructed environments on L2 learning from a process-oriented perspective, investigating these
two environments within the same research design is important given that they do address and
can improve our understanding of how learners process the L2, for example, what role awareness
plays in L2 learning. Furthermore, such studies could also shed light on macro- (or meso-) and
micro-level interactions and enable researchers to address the relationship between social (macro
or meso) and cognitive (micro) factors (Ellis, 2019). Mixed-methods procedures may be one way
of attempting to address the diferential efects of these two environments on how L2 learners
process the L2.
The local learning contexts, especially the pedagogical context, also provide several avenues for
future research given the paucity of process-oriented studies in this area. Like previous studies in the
global learning context, future studies may replicate the product-oriented studies in the local learning
contexts to address the cognitive processes employed during exposure to or interaction with the L2
data while including additional learner populations (outside the popular college-level) and expand-
ing the range of independent variables. In addition, approaching local learning contexts from an
ISLA applied lens would entail using previous findings from laboratory settings (applied ISLA studies)
as a baseline to move and situate the research design within the real language classroom and its syl-
labus while adopting a longitudinal perspective in recognition of the temporal nature of a language
curriculum.
In sum, the relative paucity of studies that have probed deeply into the psycholinguistics of L2
learning from a process-oriented perspective infuenced by both local and global contexts provides a
worthy direction for future studies.
Conclusion
The infuence that the learning context plays in the psycholinguistics of L2 learning viewed from a
process-oriented perspective has not been fully explored or explicated in the ISLA literature, perhaps
due to three major reasons: (1) an unclear definition of what composes a learning context, (2) the
methodological infeasibility of doing so, and (3) a relative paucity of studies that have empirically and
methodologically addressed learner cognitive processes or the process of learning as compared to the
product of learning. At the same time, it is clear that shifting from a product-oriented and an applied
ISLA perspective to a process-oriented and ISLA applied one opens many new avenues for future
research and will likely include diferent populations.
This chapter has approached the learning context from both a local and global perspective
mainly in relation to the afordances provided by the learning environment. While it is challeng-
ing methodologically to address the infuence of the naturalistic versus instructed learning envi-
ronment on the psycholinguistics of L2 learning from a process-oriented perspective, it is noted
that there are several strands within both local (e.g., pedagogical and methodological) and global
learning contexts (e.g., study abroad and the educational and foreign language environments) that
do ofer the potential to do comparative studies empirically and methodologically. These strands
should continue to be fully explored with appropriate concurrent data elicitation procedures
given that a better understanding of the cognitive processes employed by L2 learners opens the
door to address the potential infuences of more (multi)layered contexts postulated to play a role
in L2 learning. This, in turn, may promote a more informed manipulation of L2 data designed
to promote more robust learning in these instructed learning contexts. Textbox 30.2 ofers some
directions for future research.
382
Do diferences in cultural background between learning environments play a role in the psycholinguistics
of L2 learning?
Given the diferent afordances between the naturalistic and instructed learning environments, do study
abroad students process the L2 diferently when compared to stay-at-home students?
Given the programmatic context of writing components (writing-to-learn vs. learning-to-write) in a language
program, does level of proficiency play a role in the processing of type of corrective feedback (written
or computerized)? In addition, does type of linguistic items (lexical, morphological, syntactic) play a
role at these two levels?
Does educational level (elementary vs. high school vs. tertiary) play a role in the psycholinguistics of L2
learning, as measured by a concurrent data elicitation method?
Does the educational environment (foreign vs. second language) play a role in the psycholinguistics of L2
learning, as measured by a concurrent data elicitation method?
Does learner profile (FL vs. SL vs. heritage) play a role in the psycholinguistics of L2 learning in relation
to (a) type of feedback, (b) type of instruction, (c) type of interaction, etc.?
Note
1 The term “environment” will be used to indicate broad diferences in afordances between learning contexts,
while “context” and “setting” will be used interchangeably.
Further reading
Bowles, M. A. (2010.) The think-aloud controversy in second language research. Taylor and Francis.
This book answers key questions about the validity and uses of think-alouds or concurrent verbal reports
completed by research participants while they perform a task. It serves as a valuable guide for any language
researcher who is considering using think-alouds.
guide. Routledge.
This book provides foundational knowledge and hands-on advice for designing, conducting, and analyzing
eye-tracking research in applied linguistics. It is an indispensable guide for scholars interested in employing
this methodology.
Jiang, N. (2011). Conducting reaction time research in second language acquisition studies. Routledge.
This book comprehensively addresses the use of reaction time in experimental research and ofers several meth-
odological options for the use of this measure. It is a valuable guide for anyone interested in using this measure.
Leow, R. P. (Ed.). (2019). The Routledge handbook of second language research in classroom learning. Routledge.
This recent handbook is an important source of over a dozen and a half empirical studies methodologically
addressing the psycholinguistics of L2 learning from diferent local contexts.
Leow, R. P., Grey, S., Marijuan, S., & Moorman, C. (2014). Concurrent data elicitation procedures, processes,
and the early stages of L2 learning: A critical overview. Second Language Research, 30(2), 111–127.
This is a critical review of the three popular concurrent data elicitation procedures (think alouds, eye-tracking,
and reaction times) currently employed in the ISLA field. Their strengths and limitations are fully discussed.
383
Ronald P. Leow
References
Baralt, M. (2013). The impact of cognitive complexity on feedback efcacy during online versus face-to-face
interactive tasks. Studies in Second Language Acquisition, 35(4), 689–725.
Bateson, G. (1972). Steps to an ecology of mind: Collected essays in anthropology, psychiatry, evolution, and epistemology.
University of Chicago Press.
Boers, F., Bryfonski, L., Faez, F., & McKay, T. (2021). A call for cautious interpretation of meta-analytic reviews.
Bryfonski, L., & McKay, T. H. (2019). TBLT implementation and evaluation: A meta-analysis. Language Teaching
Research, 23(5), 603–632.
Caras, A. (2019). Written corrective feedback in compositions and the role of depth of processing. In R. P.
Leow (Ed.), The Routledge handbook of second language research in classroom learning (pp. 188–200). Routledge.
Cerezo, L., Caras, A., & Leow, R. P. (2016). Efectiveness of guided induction versus deductive instruction on
the development of complex Spanish “gustar” structures: An analysis of learning outcomes and processes.
DeRobles, G. (2019). The efects of type of written corrective feedback and level of profciency on processing and accuracy in
heritage language learners of Spanish [Unpublished doctoral dissertation], Georgetown University.
Dewey, J. (1985). Context and thought. In J. A. Boydston (Ed.), The collected works of John Dewey, 1882–1953.
The later works of John Dewey, 1925–1953: Essays: First published in 1931 (pp. 3–22). Southern Illinois Uni-
versity Press.
Dohn, N. B., Hansen, S. B., & Klausen, S. H. (2018). On the concept of context. Education Sciences, 8(3), 1–17.
Journal, 100(Supplement), 19–47.
Duf, P. A., & Byrnes, H. (2019). SLA across diferent borders: Introduction to the special issue. The Modern
Language Journal, 103(Supplement), 3–5.
Ellis, N. C. (2007). The associative-cognitive CREED. In B. VanPatten & J. Williams (Eds.), Theories in second
language acquisition (pp. 77–95). Lawrence Erlbaum.
Ellis, N. C. (2019). Essentials of a theory of language cognition. The Modern Language Journal, 103, 39–60.
Ellis, R. (1989). Are classroom and naturalistic acquisition the same? A study of the classroom acquisition of
German word order rules. Studies in Second Language Acquisition, 11(3), 305–328.
Gass, S. M. (1997). Input, interaction, and the second language learner. Lawrence Erlbaum.
Godfroid, A. (2019). Investigating instructed second language acquisition using L2 learners’ eye-tracking data. In
R. P. Leow (Ed.), The Routledge handbook of second language research in classroom learning (pp. 44–57). Routledge.
Greeno, J. G. (2011). A situative perspective on cognition and learning in interaction. In T. Koschmann (Ed.),
Theories of learning and studies of instructional practice (pp. 41–71). Springer.
Grgurović, M., Chapelle, C. A., & Shelley, M. (2013). A meta-analysis of efectiveness studies on computer
technology-supported language learning. ReCALL: The Journal of Eurocall, 25(2), 165–198.
Guile, D. (2006). Learning across contexts. Educational Philosophy Theory, 38(3), 251–268.
Han, Z. (2016). A “reimagined SLA” or an expanded SLA? A rejoinder to The Douglas Fir Group (2016). The
Modern Language Journal, 100(4), 736-740.
Hsieh, H.-C. (2019). Computerized type of feedback and depth of processing during a computerized problem-
solving task. In R. P. Leow (Ed.), The Routledge handbook of second language research in classroom learning (pp.
Hsieh, H.-C., Moreno, N., & Leow, R. P. (2016). Awareness, type of medium, and L2 development: Revisit-
ing Hsieh (2008). In R. P. Leow, L. Cerezo, & M. Baralt (Eds.), A psycholinguistic approach to technology and
language learning (pp. 131–150). De Gruyter Mouton.
Hulstijn, J. (2023). Synthesis: transdisciplinary perspectives on second language acquisition: exploration versus
explanation. In A. Godfroid & H. Hopp (Eds.), The Routledge handbook of second language acquisition and
Kang, E. Y., & Han, Z. (2015). The efcacy of written corrective feedback in improving L2 written accuracy:
A meta-analysis. The Modern Language Journal, 99(1), 1–18.
Krashen, S. D. (1982). Principles and practice in second language acquisition. Pergamon Press.
Laford, B., & Uscinski, I. (2014). Study abroad and second language Spanish. In K. L. Geeslin (Ed.), The hand-
book of Spanish second language acquisition (pp. 386–403). Wiley-Blackwell.
Leow, R. P. (2018). Explicit learning and depth of processing in the instructed setting: Theory, research, and
practice. Studies in English Education, 23, 769–801.
384
Leow, R. P., Cerezo, L., Caras, A., & Cruz, G. (2019). CALL in ISLA: Promoting depth of processing of
complex L2 Spanish “Para/Por” prepositions. In R. DeKeyser & G. Prieto Botana (Eds.), SLA research with
implications for the classroom: Reconciling methodological demands and pedagogical applicability (pp. 155–78). John
Benjamins.
Leow, R. P. (2019a). From SLA > ISLA > ILL: A curricular perspective. In R. P. Leow (Ed.), The Routledge
handbook of second language research in classroom learning (pp. 483–491). Routledge.
Leow, R. P. (Ed.). (2019b). The Routledge handbook of second language research in classroom learning. Routledge.
Leow, R. P. (2020). L2 writing-to-learn: Theory, research, and a curricular approach. In Rosa M. Manchón
(Ed.), Writing and language learning. Advancing research agendas (pp. 95–117). John Benjamins.
Leow, R. P., & Cerezo, L. (2016). Deconstructing the “I” and “SLA” in ISLA: One curricular approach. Studies
in Second Language Learning and Teaching, 6(1), 43–63.
Leow, R. P., Cerezo, L., Caras, A., & Cruz, G. (2019). CALL in ISLA: Promoting depth of processing of
complex L2 Spanish “Para/Por” prepositions. In R. DeKeyser & G. Prieto Botana (Eds.), SLA research with
implications for the classroom: Reconciling methodological demands and pedagogical applicability (pp. 155–178). John
Benjamins.
Leow, R. P., & Zamora, C. (2017). Intentional and incidental learning. In S. Loewen & M. Sato (Eds.), Routledge
handbook of instructed second language acquisition (pp. 33–49). Routledge.
Li, S. (2010). The efectiveness of corrective feedback in SLA: A meta-analysis. Language Learning, 60(2),
309–365.
Lightbown, P. M. (1983). Exploring relationships between developmental and instructional sequences in L2
acquisition. In H. Seliger & M. Long (Eds.), Classroom oriented research in second language acquisition (pp.
217–245). Newbury House.
Lightbown, P. M., & Spada, N. (2019). Teaching and learning L2 in the classroom: It’s about time. Language
Teaching, 53(4), 422–432.
Long, M., & Crookes, G. (1993). Units of analysis in syllabus design: The case for task. In G. Crookes & S. Gass
(Eds.), Tasks in a pedagogical context: Integrating theory & practice (pp. 9–54). Multilingual Matters.
Lyster, R., & Saito, K. (2010). Oral feedback in classroom SLA: A meta-analysis. Studies in Second Language
Acquisition, 32(2), 265–302.
Mackey, A., & Goo, J. (2007). Interaction research in SLA: A meta-analysis and research synthesis. In A. Mackey
(Ed.), Conversational interaction in second language acquisition: A collection of empirical studies (pp. 407–452). Ox-
ford University Press.
Manchón, R. M. (2023). The psycholinguistics of second language writing. In A. Godfroid & H. Hopp (Eds.),
Manchón, R. M., & Leow, R. P. (2020). Investigating the language learning potential of L2 writing: Method-
ological considerations for future research agendas. In R, M. Manchón (Ed.), Writing and language learning.
Advancing research agendas (pp. 335–355). John Benjamins.
Moreno, N. (2019). Type of feedback and assessment task modality: The role of depth of processing. In R. P.
Leow (Ed.), The Routledge handbook of second language research in classroom learning (pp. 138–155). Routledge.
Piaget, J. (1995). Sociological studies. Routledge.
Qi, D. S., & Lapkin, S. (2001). Exploring the role of noticing in a three-stage second language writing task.
Journal of Second Language Writing, 10(4), 277–303.
Robinson, P. (1995). Attention, memory and the “noticing” hypothesis. Language Learning, 45(2), 283–331.
Rogers, J., & Cheung, A. (2020). Input spacing and the learning of L2 vocabulary in an L2 context. Language
Teaching Research, 24, 616–641.
Rogers, J., & Révész, A. (2020). Experimental and quasi-experimental designs. In J. McKinley & H. Rose
(Eds.), The Routledge handbook of research methods in applied linguistics (pp. 133–143). Routledge.
Russell, J., & Spada, N. (2006). The efectiveness of corrective feedback for the acquisition of L2 grammar. A
meta-analysis of the research. In J. M. Norris & L. Ortega (Eds.), Synthesizing research on language learning and
teaching (pp. 133–164). John Benjamins.
Sachs, R., Baralt, M., & Gurzynski-Weiss, L. (2023). Psycholinguistics for the language classroom. In A. God-
froid & H. Hopp (Eds.), The Routledge handbook of second language acquisition and psycholinguistics (pp. 387–399).
Routledge.
Schmidt, R. (1990). The role of consciousness in second language learning. Applied Linguistics, 11(2), 129–158.
Shively, R. (2014). Language in context: Pragmatics in SL Spanish. In K. L. Geeslin (Ed.), The handbook of Span-
ish second language acquisition (pp. 331–350). Wiley-Blackwell.
Tomlin, R. S., & Villa, V. (1994). Attention in cognitive science and second language acquisition. Studies in
385
Ronald P. Leow
Tomlinson, B., & Masahara, H. (2021). SLA applied: Connecting theory and practice. Cambridge University Press.
Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes. Harvard University Press.
Zhuang, J. (2019). Computer-assisted guided induction and deductive instruction on the development of com-
plex Chinese ba structures: Extending Cerezo et al. (2016). In R. P. Leow (Ed.), The Routledge handbook of
second language research in classroom learning (pp. 391–406). Routledge.
Ziegler, N. (2016). Synchronous computer-mediated communication and interaction: A meta-analysis. Studies
386
31
PSYCHOLINGUISTICS FOR THE
LANGUAGE CLASSROOM
Introduction
Pedagogically oriented psycholinguistics research has made substantial contributions to the field’s mul-
tifaceted understanding of how language learners and teachers can maximize the benefits of instruc-
tion in a second/additional language (L2). With in-depth knowledge of the psycholinguistic rationales
behind a variety of pedagogical techniques, combined with awareness of the roles that individual
diferences can play in the processes of teaching and learning, educators can be even better equipped
to help L2 learners reach their goals. In this chapter, given the wide range of ways in which psycholin-
guistics is relevant to the language classroom, we highlight a selection of areas (introduced in Textbox
31.1) where research has been especially fruitful and preview exciting directions for future work.
Aptitude-treatment interaction: A finding that the efectiveness of particular instructional techniques dif-
fers according to learner diferences in cognitive abilities that are more or less suited to learning under
certain conditions.
Cognitive task complexity: The mental (e.g., attentional, conceptual, reasoning) demands involved in
engaging in a communicative pedagogical task.
Focal attention, awareness, and depth of processing: in the context of language learning, respectively, the
direction of mental resources toward language items, the subjective experience of noticing and (pos-
sibly) understanding them, and the level of cognitive efort, elaboration, analysis, hypothesis testing,
and/or attempts at rule formation involved in processing the items.
Interactions between anxiety and cognition in the language classroom: Relationships between learners’
experiences of worry, unease, or nervousness—whether due to stable personality traits or specific situ-
ations—and their ability to pay attention, process information, engage in reasoning, and so on, in the
context of language learning and use.
Teacher and interlocutor individual diferences: Linguistic (e.g., proficiency, dialect), cognitive (e.g.,
working memory, beliefs), afective (e.g., anxiety), background (e.g., training, experience), and other
factors that vary among learners’ instructors and other communicative partners.
DOI: 10.4324/9781003018872-36 387


As educators, we know that learners with diferent psychological, cognitive, social, and afective pro-
files may achieve diferent levels of success in language learning in part because they engage in difer-
ent learning processes and take advantage of diferent instructional conditions in diferent ways—that
is, because of how their aptitudes, motivations, and emotions interact, not only with each other and
with features of instruction such as the cognitive complexity of tasks, but also with aspects of the
social context, including their teachers and other interlocutors within and beyond the classroom. As
such, the deepest and most meaningful psycholinguistic insights with relevance to language pedagogy
are likely to come, not from macro-oriented comparisons of treatment conditions or approaches that
examine general correlations between individual diferences (IDs) and learning outcomes, but rather
from micro-oriented approaches that examine the complex interplay between learner-external and
-internal variables in ecologically valid contexts (Skehan, 2015). It is crucial to explore the roles of
IDs and cognitive processes in learning under diferent instructional conditions at diferent stages of
L2 development, focusing not just on tasks and pedagogical techniques, and not just on the char-
acteristics and experiences of learners or teachers themselves, but on the combined infuence of a
myriad of interacting task, learner, and teacher variables. Methodologically, this will require purpose-
ful, complementary mixing of quantitative and qualitative approaches and will benefit from the more
in-depth, comprehensive methodological triangulation enabled by advances in a variety of research
techniques discussed in this chapter.
The following sections ofer a necessarily brief treatment of a few related areas of pedagogically
oriented psycholinguistics research, beginning with a foundation in aptitude-treatment interactions
(ATI) since an ATI mindset can provide a productive framework for investigating interrelationships.
Many L2 practitioners interested in creating optimal conditions for language development focus on
how they can manipulate task features (e.g., complexity, explicitness) and their associated cognitive
demands in order to guide learners’ allocation of attentional resources, awareness, depth of processing
(DoP), and learning. Broadly, research suggests that increases in task complexity can lead to more
interaction, more noticing, and more accurate and complex L2 production (see Baralt et al., 2014,
for review), and that more explicit instructional conditions tend to produce higher levels of aware-
ness, which is associated with greater initial learning gains (see Leow, 2015, for review). However,
learner IDs can moderate these relationships. They may help to explain why awareness and DoP
levels do not always align, why DoP seems to be more closely related to retention of learning than
awareness is, and why the patterns of task complexity results noted earlier may not hold true in all
contexts (e.g., synchronous computer-mediated communication). Growing strands of research are
devoted to exploring how cognitive, afective, and experiential variables can interact dynamically
with features of instruction and exposure to afect learning processes and outcomes (Dörnyei &
Ryan, 2015; Gurzynski-Weiss, 2020), and this is contributing to a growing potential for psycholin-
guistic research to inform language pedagogy.
Aptitude-treatment interactions
Decades of research into the efectiveness of diferent instructional methods and the roles of cognitive
IDs in language learning have produced a wealth of insights with relevance for pedagogy (Dörnyei
& Ryan, 2015). However, group diferences and group correlations do not tell the whole story of
which treatments work for whom and why. The efectiveness of diferent instructional approaches
can depend on learners’ aptitudes—that is, “stable cognitive abilities that difer among people and
may allow a person to excel in a specific learning area under particular learning conditions” (Vatz et
388
Psycholinguistics for the language classroom
al., 2013, pp. 274–5). In other words, because of aptitude-treatment interactions (ATI), learners with
diferent cognitive profiles may achieve diferent learning outcomes due to diferences in how they
experience and take advantage of diferent instructional conditions (DeKeyser, 2012).
A variety of approaches exist for investigating the roles of cognitive IDs in contextualized ways.
A framework proposed by Skehan (2016), for example, takes stages of L2 processing that are well
known to classroom practitioners (e.g., input, output) as a starting point and hypothesizes aptitude
components that seem useful for specific subprocesses during these stages. For instance, Skehan
(2016) argues that phonemic coding ability and attention control are relevant for segmenting audi-
tory input and noticing particular forms, whereas grammatical sensitivity and inductive learning
ability may support identifying patterns and complexifying the developing L2 system.1 A proposal
by Robinson (2012), on the other hand, theorizes aptitudes that seem likely to be involved in taking
advantage of specific pedagogical techniques. Reasoning that “instructional, interventionist tech-
niques must be matched to the cognitive resources and abilities learners bring to the classroom to be
optimally efective” (p. 60), he outlines how, for instance, oral recasts (i.e., corrective reformulations
of speech) might be most efective for learners who have strengths in cognitive resources such as
perceptual speed, pattern recognition, and phonological short-term memory (PSTM), which in turn
equip them to remember contingent utterances and notice diferences between what they have said
and what their interlocutors have said.
Taking inspiration from these models, ATI researchers have explored relationships with learning
for a range of IDs in a range of instructional conditions (for reviews, see, e.g., Skehan, 2015; Vatz
et al., 2013). For instance, in an early study investigating the roles of grammatical sensitivity (GS) and
rote memory (RM) in adult L2 English learners’ awareness and learning of easy and hard rules in four
instructional conditions, Robinson (1997) found that none of the aptitude or awareness measures
were related to learning in the incidental condition. However, GS and RM predicted learning in the
(explicit) instructed and rule-search conditions, and GS also predicted learning in the implicit condi-
tion. In fact, correlations between GS and learning were strongest in the implicit condition, where
GS was associated with looking for rules and being able to verbalize them even though this was not
a part of the instructions. Beyond contributing several empirically based insights, this seminal study
drew attention to the fact that conditions intended to be implicit may not involve implicit learning.
In a classroom-based ATI study investigating the roles of GS, phonemic coding ability, and PSTM
in high-school L2 French students’ learning of direct object pronouns, Erlam (2005) found that GS
predicted learning in inductive and structured-input conditions, and PSTM additionally predicted
learning from structured input. However, in a deductive condition, in which learners showed better
performance overall, the only ID that predicted learning was phonemic coding ability. This led Erlam
to conclude that explicit instruction with production practice (as found in the deductive condition)
may serve as an equalizer that helps minimize the efects of IDs in grammatical sensitivity.
In the years since these early studies, ATI research has branched out to explore the roles of IDs
in learners’ ability to benefit from diferent types of oral (e.g., Li, 2013), written (e.g., Shintani &
Ellis, 2015), and computer-mediated (e.g., Sachs et al., 2019) feedback; as well as from input that
has been manipulated in various ways (e.g., low vs. high variability, as in Brooks et al., 2006; dis-
tributed vs. massed practice, as in Suzuki & DeKeyser, 2017). The rich body of work on feedback,
in particular, has produced mixed results that point toward the need for systematic mixed-methods
conceptual replications to shed clearer light on how language analytic ability (LAA) and/or work-
ing memory (WM), for example, are likely to be related to learners’ ability to benefit from explicit
and implicit corrections (see Li, 2017, for review). Moving forward, qualitative measures of learners’
thought processes will be crucial in advancing understanding of the complex interplay among IDs
that can lead learners with particular profiles of abilities to excel or experience difculty with difer-
ent instructional techniques (Sachs et al., 2019). The observation of multiple IDs simultaneously in
action reveals how learners with similar strengths in one ID of interest (e.g., LAA) may nonetheless
389
achieve diferent learning outcomes due to other IDs (e.g., memory). Such findings can stimulate
ideas about how to adapt instruction to individuals’ unique combinations of strengths and needs.
Cognitive task complexity

Within task-based language teaching (TBLT), another area of rigorous research examines how teach-
ers can engineer the cognitive demands of task features to guide learners’ attentional resources and
promote DoP and learning. The idea is that learners should first perform a cognitively simple ver-
sion of a task and then gradually progress through increasingly complex tasks to prepare them for the
complexity level of the real-world target task. While task has been espoused for decades as a produc-
tive unit for teaching, learning, and assessing, and while numerous studies have explored and con-
firmed the design features that render tasks cognitively complex, to date, there are no agreed-upon
criteria with which to sequence tasks based on their complexity (Baralt et al., 2014).
One theory that has informed much task complexity research is Robinson’s triadic componential
framework (TCF) and SSARC (Simplify-Stabilize, Automatize, Restructure-Complexify) model
(e.g., Baralt et al., 2014). The TCF proposes two dimensions of cognitive task complexity: resource-
dispersing variables, such as planning time; and resource-directing variables, such as having to com-
municate in the here-and-now (versus there-and-then). Resource-dispersing variables promote the
gradual automatization of a learner’s linguistic resources, while resource-directing variables promote
interlanguage development. The SSARC model encourages teachers to use the task-design features
in the TCF to guide task sequencing in a lesson or syllabus, increasing cognitive complexity along
resource-dispersing variables first, and then along resource-directing variables. The TCF also outlines
task difculty factors such as ability variables (e.g., WM) and afective variables (e.g., motivation).
Thus, much as in ATI research, particular combinations of design features can be especially well
suited for individual learners (e.g., those with high WM and low anxiety might be able and inclined
to push themselves to produce more complex language when faced with high reasoning demands,
whereas those with low WM and high anxiety might not). These considerations can guide teach-
ers in determining the types of tasks that students may need. Skehan’s (2009) trade-of hypothesis
discusses many of the same variables, but makes diferent predictions about how task complexity will
afect learners’ noticing, interaction, and learning. Ellis (2003) has also proposed a model for real-
izing and sequencing task complexity in the L2 classroom, considering the psycholinguistic processes
involved in task performance and providing illustrative examples related to task input, code complex-
ity, processes, and outcomes.
A significant amount of research has examined the efects of increased task complexity on learn-
ers’ noticing, interactions, oral and written production, processing and incorporation of feedback,
and learning, all with concrete implications for the L2 classroom setting (e.g., Frear & Bitch-
ener, 2015; Kim, 2012; Révész, 2009; Zalbidea, 2017). Overall—albeit with diferent theoretical
interpretations—this body of literature indicates that more cognitively complex tasks lead to more
interaction, more accurate and complex L2 production, and greater learning gains. That said, incipi-
ent research indicates that this may only apply to face-to-face classroom settings, and not to online
contexts (e.g., Baralt, 2013; Torres, 2018).
Research on task sequencing in the classroom setting is also starting to grow. One example is
Baralt (2014), who examined the efects of having L2 Spanish learners perform sequences of tasks
that ranged from cognitively simple to complex or vice versa with regard to their intentional reason-
ing demands. She found that the sequences with more complex tasks resulted in more interaction
and learning. In a similar design with L2 English learners, Malicka (2020) explored increases in com-
plexity along both resource-dispersing (number of elements) and resource-directing (+/- reasoning
demands) dimensions and found that, when compared to individual task performance, the simple-
to-complex sequencing led to a higher speech rate, greater structural complexity, enhanced accuracy,
390
and greater disfuency. Similarly, Allaw and McDonough (2019) operationalized task complexity via
both resource-dispersing (+/- task structure) and resource-directing (+/- spatial reasoning) dimen-
sions and examined French learners’ written lexical complexity, grammatical accuracy, and fuency.
While all learners improved, those who performed tasks sequenced from simple to complex main-
tained their gains best over time.
Taking both task design and sequencing into account, a burgeoning research agenda is aiming to
improve techniques for measuring cognitive complexity (e.g., Révész et al., 2016; Sasayama, 2016).
So far, the most common method—and one which can easily be used to inform instruction—is to
have learners fill out a questionnaire asking about their perceptions of task difculty (e.g., Gilabert
& Barón, 2018). Several other psycholinguistic approaches to validate cognitive complexity are gain-
ing traction. These have included interviews (e.g., Pang & Skehan, 2014), expert judgments (e.g.,
Révész et al., 2016), think-alouds (e.g., Sangarun, 2005), time-on-task (Lee, 2019), retrospective
estimations of time-on-task (Baralt, 2013), and stimulated recall (e.g., Torres, 2018). This research
base will continue to inform how to diagnose sources of task difculty so that instruction can be
adjusted to meet learners’ needs better, including how best to sequence tasks in the classroom so that
well-timed increases in complexity can help to promote skill development.
Attention, awareness, and depth of processing

Resonating strongly with both ATI and task complexity research, a rich body of work on learners’
attention, awareness, and DoP has sought to understand what cognitive processes are promoted by dif-
ferent pedagogical techniques, how they are related to L2 performance and learning, and how and why
learners might benefit diferently from the same instructional methods. There is now broad recognition
of the difculties involved in defining, operationalizing, and measuring attention and awareness (Leow
et al., 2011), as well as depth of processing, which Leow (2015) defines as involving cognitive efort,
elaboration of input, analysis, hypothesis testing, and rule formation. However, over the past decades,
the field has refined the use of a wide array of instruments, from concurrent think-alouds (e.g., Leow,
2000), immediate oral reports (e.g., Philp, 2003), and eye-tracking (e.g., Godfroid, 2020); to text mark-
ing (e.g., Izumi & Bigelow, 2000), uptake charts (e.g., Mackey, 2006), and lesson refection sheets (e.g.,
Kartchava & Ammar, 2014); to retrospective questionnaires (e.g., Robinson, 1997), interviews (e.g.,
Rebuschat et al., 2015), and stimulated recall (SR) protocols (e.g., Mackey et al., 2000).
Drawing on explicit psycholinguistic rationales for the efcacy of recasts (e.g., Long, 2007) and
the value of output (e.g., Izumi, 2003), many researchers have explored learners’ cognitive pro-
cesses and perceptions of feedback during communicative interactions using methods such as stimu-
lated recall (SR), in which learners are shown videoclips of their conversations and asked to report
what they were thinking during the interactions. In an infuential early study, Mackey et al. (2000)
employed SR to examine learners’ recognition of interactional feedback. The researchers found that
both L2 English and L2 Italian learners tended to perceive lexical feedback accurately (at rates of
83% and 66%, respectively), but the English learners were much better at recognizing phonological
(60%) than morphosyntactic (13%) feedback, whereas the Italian learners showed relatively equal but
low accuracy in those areas (18% and 24%, respectively). Since then, SR has been used alongside
other methods to explore the degree of overlap between teachers’ intentions and learners’ percep-
tions depending on characteristics of the feedback (e.g., Egi, 2007), learners’ awareness of feedback in
synchronous computer-mediated communication (Smith, 2012), and relationships between learners’
awareness and responses to recasts (e.g., Egi, 2010), among other topics. Together, such studies have
suggested, for example, that features of recasts, such as their length, focus, prosodic emphasis, and
degree of change to learners’ original utterances, can afect learners’ perceptions.
Studies exploring DoP and learning across diferent pedagogical conditions are also helping to
shed light on unexpected findings, such as a lack of maintenance of learning gains despite high levels
391
of awareness of linguistic targets in explicit conditions. For instance, Hsieh et al. (2015) discovered
that, although participants in a more explicit condition demonstrated higher levels of awareness,
those in a more implicit condition tended to show greater DoP and maintenance of gains on a
delayed post-test. Incorporating a focus on IDs, Sachs and Nakatsukasa (2019) found complex rela-
tionships between DoP and a range of IDs (e.g., GS, WM, metalinguistic knowledge) that difered
depending on the type of computer-mediated feedback learners received. Zamora (2017) found that,
compared to heritage language learners, L2 learners relied more on metalinguistic prior knowledge
when analyzing input, and that relationships between DoP and learning tended to be stronger for
the L2 learners. Such work suggests that a broader integration of awareness/DoP research with ID-
treatment interaction research (and vice versa) will be fruitful.
In L2 writing research as well, introspective measures have traditionally been used to explore
learners’ cognitive processes (see Manchón, 2023 [this volume]). In the context of multistage writ-
ing and revision tasks, there has been interest in discovering what learners focus on while working
collaboratively with models, reformulations, and/or direct error corrections (e.g., Santos et al., 2010;
Yang & Zhang, 2010), and in tracking connections between learners’ initial compositions, awareness
of feedback, and changes made to revisions (e.g., Sachs & Polio, 2007). Hanaoka and Izumi (2012)
have taken the intriguing approach of providing learners with both reformulations and models, then
cross-examining the compositions, models, reformulations, revisions, and learners’ notes on prob-
lematic features at each stage. Finding that learners frequently used model texts to solve “covert prob-
lems” that were not addressed in feedback, the authors highlighted the value of self-initiated focus on
form and suggested that models and reformulations can play complementary roles in meeting learner
needs that might otherwise go unrecognized.
Researchers are increasingly triangulating verbal reports with less obtrusive methods such as key-
stroke logging, screen capture, and eye-tracking to gain insights about the dynamic, recursive nature
of writing, including its temporal dimensions, and to obtain a more complete and veridical picture
of the behavioral and cognitive processes involved (Révész & Michel, 2019). This is enabling them
to link directly observable writing behaviors with complementary information on learners’ moment-
by-moment allocations of attention, perceptions, and strategy use, thereby disambiguating the mean-
ings of pauses and regressions. In addition to providing theoretical insights relevant to both “learning
to write” and “writing to learn” (Manchón, 2011), such studies can help to diagnose and address
sources of difculty and generate ideas regarding pedagogical applications, such as the teaching of
beneficial metacognitive strategies.
Anxiety and cognitive processing

Anxiety has been one of the most investigated IDs in L2 classroom research, and understandably
so. It has been described as both “an emotional reaction and disruptive to ongoing cognition and
behavior—thus making it part of a continuous cycle of sometimes infuencing other variables and
sometimes being infuenced” (Gregersen, 2020, p. 79). L2 research increasingly diferentiates whether
the anxiety at play is a personality trait (e.g., I am an overall anxious person, see Spielberger, 1989), a
state or situational anxiety (e.g., I am anxious in this moment/when I complete a specifc task, see Baralt
& Gurzynski-Weiss, 2011), or even whether the anxiety is specific to using the L2 in the classroom
(see Horwitz et al., 1986). Most recently, with increased recognition of the dynamic nature of IDs
(Dörnyei & Ryan, 2015), language anxiety has also been found to change in moment-to-moment
interaction (Gregersen, 2020), rendering awareness of its role particularly important, given its poten-
tial impact within the L2 classroom. Regardless of the type of anxiety, it may have diferent efects
on L2 learning processes: (1) beneficial, given that a certain amount of anxiety may promote or be
indicative of cognitive engagement in a task, or (2) detrimental, because too much anxiety can nega-
tively impact learners’ exposure to L2 input, processing of information, and language production.
392
Early research found that low or moderate levels of anxiety may have a facilitating role in SLA. For
example, Neville et al. (1967) found a moderate level of anxiety was associated with silent reading
comprehension gains, a finding researchers have since explained with reference to the importance of
motivation and focused attention for L2 learning. Alternatively—and much more commonly cited
in the literature—too much anxiety can inhibit learners’ ability to process L2 input or their inclina-
tion to produce output. For example, in a quasi-experimental study involving recasts on English
articles, Sheen (2008) found that more anxious learners produced less modified output and showed
smaller improvements in accuracy compared to their less anxious peers. In a repeated-measures study
examining the role of anxiety in the quantity of beginning French learners’ output, Côté and Gaf-
ney (2018) found that learners showed less anxiety during synchronous computer-mediated com-
munication than face to face and that learners with higher anxiety in the face-to-face environment
produced significantly fewer turns and words than their lower-anxiety counterparts. Teachers’ actions
and task characteristics can also shape the efects of anxiety in the classroom. MacIntyre (2017), for
instance, found that learners stopped producing language after very direct and/or repetitive feedback
from teachers, while Gregersen et al. (2014) found that learners froze when a video camera was
introduced. This line of research suggests that anxiety may not only cause learners to miss valuable
opportunities to convert input to intake, but also reduce their production of output, thereby limiting
the amount of feedback they receive and ultimately resulting in less L2 development.
Importantly, there is no known threshold for what level of anxiety is “optimally low” or con-
versely “too high” for L2 development. This makes sense, since one would imagine anxiety to be
highly idiosyncratic and specific to individual learners and their constellations of additional IDs
in particular learning situations. For example, anxiety has been found to correlate directly with
perfectionist tendencies (Pishghadam & Akhondpoor, 2011) and indirectly with motivation and
personality traits, such as extraversion; however, the patterns can difer according to social context
(Dewaele, 2002). In Dewaele’s (2002) study of Flemish learners of L2 French and L3 English, more
extraverted and less neurotic learners reported significantly lower levels of communicative anxiety
than their less extraverted and more neurotic counterparts in English, but sociodemographic factors
such as language perceptions and social class seemed to play a stronger role than personality in the
case of communicative anxiety in French, perhaps due to its prestige status in Belgium. Moreover, it
is important to keep in mind that what might be a prohibitive level of anxiety for one learner may be
just the amount another needs to be cognitively and/or behaviorally engaged in a given learning task.
For some learners, for example, exceptionally high motivation might counteract anxiety and allow
them to continue engaging with L2 input and producing output. What is critical for our purposes is
the fact that L2 anxiety is subject to complex interactions of factors within the learner, within the L2
classroom, and beyond, and it can infuence learners in all moments of L2 development.
Teacher/interlocutor individual differences

The IDs of teachers and other interlocutors significantly infuence their own cognitive processes and
behaviors in the classroom, and these ultimately infuence the L2 opportunities available to learners,
including how learners are encouraged to engage with them and how they actually engage with and
learn from them. By interlocutors, we mean the learners’ communicative partners, including not
only their teachers and peers, but also individuals who are not physically present, including past,
anticipated, or hypothetical interlocutors (Back, 2020; Lantolf, 2020).
IDs in prior experience have been found to play a multifaceted role in teachers’ prioritization
and allocation of attention. Mackey et al. (2004), for instance, found English language educators
with more teaching experience and pedagogical coursework to provide more feedback than their
less experienced colleagues, while Polio et al. (2006) found teacher experience to infuence the
direction of their attention; language instructors with more experience focused on student abilities,
393
the elicitation of L2 production, and learning, whereas inexperienced teachers focused on them-
selves, student feelings, and procedural demands such as time management. This resonates with
Gurzynski-Weiss’s (2016) finding that novice teachers may be more cognitively preoccupied with
classroom management, whereas more experienced teachers may be able to simultaneously direct
their attentional resources to learner errors and feedback as well. Prior training has also been found
to direct teachers’ attention to specific factors in learner production, most notably in teachers’ pro-
cessing of learners’ errors and their subsequent decisions as to whether, when, and how to correct
them. For instance, Gurzynski-Weiss (2016) found teachers to have a dominant ID that directed
their sensitivity to learner errors (e.g., a teacher specializing in L2 phonology was more sensitive to
pronunciation errors).
Teacher IDs have also been found to determine the type of L2 exposure learners receive. For
example, both Gurzynski-Weiss et al. (2018) and Long and Geeslin (2020) found teacher dialect to
infuence their expression of subject pronouns in the L2 Spanish classroom. As L1 speakers of mor-
phologically poor languages such as English tend to be delayed in shifting their cue reliance to verbal
morphology due to an over-focus on Spanish pronouns, such dialect diferences on the part of the
teacher might either inhibit or promote attention to the critical cues for Spanish, thereby intensifying
or mitigating learners’ processing biases.
The subdomain of cognitive IDs is ripe for additional research, particularly with respect to their
role in teachers’ processing during class time. The only study that has examined teacher WM, for
example, occurred outside the classroom context. Ziegler and Smith (2017) examined whether
teachers’ WM predicted the amount of feedback they provided to learners during text-based chat,
the hypothesis being that teachers with greater WM would be able to provide more feedback. While
this hypothesis was not borne out in their modest dataset, the study pointed to directions for future
study, including into the mediating roles of beliefs, metacognition, and contextual factors. Much
more research is needed to examine whether and how teachers’ cognitive IDs might infuence their
processing during face-to-face class time, when there are many more moving parts, and how these
IDs shape students’ L2 learning opportunities.
One particularly useful methodology to capture teacher and learner insights on the roles of IDs
is the idiodynamic method developed by MacIntyre (2012). In this technique, which is similar to
stimulated recall, an interaction (dyadic or even full-class) is video recorded. Following the interac-
tion, all interested parties review the recording, pause it when they wish to comment, and share what
they were thinking or feeling at the original moment of interaction. What is particularly powerful
about the idiodynamic method is the ability to compare researcher notions of what is occurring in
an interaction with teacher and learner insights. This technique can be used by teachers and students
for valuable pedagogical purposes as well, helping learners become more aware of how their IDs,
including anxiety, impact moment-to-moment opportunities for learning, and providing opportuni-
ties for teachers to see if they are correctly interpreting the relationship between students’ IDs and
observable behaviors.
All in all, this body of work continues to inform what we know about the interplay of learner and
teacher IDs, and crucially, how teacher IDs modulate the ways in which teachers shape opportunities
for psycholinguistic processes that promote L2 development.
Future directions
Pedagogically oriented psycholinguistics research has provided numerous insights into the com-
plex, multifaceted, nonlinear, context-dependent interrelationships among learner and teacher
IDs, instructional conditions, cognitive processes, afective variables, linguistic targets, and learn-
ing outcomes (DeKeyser, 2012; Robinson, 2012). Building on the substantial theoretical, meth-
odological, and empirical advances achieved to date, research in these areas is generating new
394
directions on multiple fronts. Textbox 31.2 ofers a small taste of some open questions. As we
broaden our conceptualizations of aptitude and other IDs, researchers are also beginning to
investigate abilities involved in implicit learning (e.g., Li & DeKeyser, 2021) and psychological
constructs such as beliefs, growth mindsets, and self-efcacy (Kartchava & Ammar, 2014; Papi
et al., 2019); as well as characteristics potentially relevant to learning pragmatic aspects of lan-
guage, such as emotional intelligence, nonverbal sensitivity, social insight, and various aspects of
motivation (Robinson, 2005; Takahashi, 2005). In parallel with these developments, researchers
are broadening their focus to investigate learners’ perceptions of a wider range of instructional
techniques that are commonly used in language classrooms, such as non-verbal forms of feedback
(e.g., Nakatsukasa, 2016).
Methodologically, the growing practices of triangulating multiple measures of learners’ cognitive
processes (e.g., Egi, 2007; Rebuschat et al., 2015; Révész & Michel, 2019; Smith, 2012) and taking
additional steps to validate experimental manipulations (e.g., Révész et al., 2016) are very positive
developments which should be adopted in other areas of research as well. A common refrain in the
limitations sections of many studies is the caveat that, in the absence of process measures or other
forms of validation, multiple speculative interpretations are possible. With the inclusion of new tools
and the combination of qualitative and quantitative analyses, this can change in ways that will lead to
more credible conclusions, richer insights, and deeper understanding.
Pedagogically, in addition to developments in task sequencing, there have been renewed forays into
instruction matching (e.g., Granena & Yilmaz, 2019), as well as initiatives building on styles- and strat-
egy-based instruction (Cohen & Weaver, 2006), as interdisciplinary research teams have begun exploring
the benefits of providing learners with guidance on how to take best advantage of their ability profiles
(e.g., Doughty, 2013). A particularly exciting area of work involves teaching learners to value, provide,
and attend to feedback from both teachers and peers through metacognitive instruction (e.g., Sato &
Loewen, 2018). Emerging technologies also hold great promise for enabling learners to take proactive
advantage of individualized opportunities for language exposure, instruction, and meaningful practice
(Petersen & Sachs, 2015). Teachers familiar with the results of pedagogically oriented psycholinguistics
research will be well positioned to provide learners guidance and assist them in those endeavors.
How can teachers provide metacognitive training that will empower learners with particular ID profiles
to optimize the benefits they derive from feedback?
How can teachers determine what constitutes cognitive task complexity for specific learners?
How can this inform the way that tasks are sequenced in the classroom setting?
How does learner anxiety interact with teacher and peer IDs in the language classroom?
How do teacher IDs relate to the ways in which they encourage learners to take advantage of L2 learning
opportunities? How can this impact learners’ L2 development?
Note
1 Grammatical sensitivity can be defined as the ability to recognize the functions of linguistic entities in senten-
tial contexts, while inductive learning ability refers to the capacity to make generalizations (or induce rules)
based on a set of examples. Phonemic coding ability is the capacity to mentally represent (or encode) sounds
so that they can be retained in memory. For definitions of working memory and phonological short-term
memory, see Li (2023 [this volume]).
395
Further reading
Baralt, M., Gilabert, R., & Robinson, P. (Eds.). (2014). Task sequencing and instructed second language learning.
Bloomsbury.
This volume presents a detailed synthesis of theoretical approaches to the construct of cognitive task com-
plexity and presents several empirical studies examining the role of various features of task complexity in
language learning.
Dörnyei, Z., & Ryan, S. (2015). The psychology of the language learner revisited. Routledge.
This text provides an overview of theory and research on several individual diferences, including personal-
ity, aptitude, motivation, anxiety, beliefs, and learning strategies, and emphasizes the dynamic interactions
between internal and external variables.
Gurzynski-Weiss, L. (Ed.). (2020). Cross-theoretical explorations of interlocutors and their individual diferences.
Benjamins.
This volume ofers a comprehensive discussion of the role of interlocutor IDs in learners’ opportunities for
L2 development as seen through four theoretical lenses: complex dynamic systems, cognitive-interactionist,
sociocultural, and variationist.
Leow, R. P. (2015). Explicit learning in the L2 classroom: a student-centered approach. Routledge.
With a special focus on the role of cognitive processes such as attention, awareness, and depth of processing,
this text provides not only implications for L2 curriculum development and pedagogy, but also an emphasis
on research methods.
Leow, R. P., Cerezo, L., & Baralt, M. (Eds.). (2015). A psycholinguistic approach to technology and language learning.
Mouton De Gruyter.
This edited collection takes a four-pronged approach to the use of technology in language learning and
teaching, contextualizing empirical studies with chapters on theoretical, methodological, and pedagogical
issues.
References
Allaw, E., & McDonough, K. (2019). The efect of task sequencing on second language written lexical com-
plexity, accuracy and fuency. System, 85, 102104.
Back, M. (2020). Interlocutor diferences and the role of social others in a Spanish peer tutoring context. In L.
Gurzynski-Weiss (Ed.), Cross-theoretical explorations of interlocutors and their individual diferences (pp. 99–123).
Benjamins.
Baralt, M. (2013). The impact of cognitive complexity on feedback efcacy during online versus face-to-face
interactive tasks. Studies in Second Language Acquisition, 35, 689–725.
Baralt, M. (2014). Task sequencing and task complexity in traditional versus online classes. In M. Baralt, R. Gi-
labert, & P. Robinson (Eds.), Task sequencing and instructed second language learning (pp. 95–122). Bloomsbury.
Baralt, M., Gilabert, R., & Robinson, P. (Eds.). (2014). Task sequencing and instructed second language learning.
Bloomsbury.
Baralt, M., & Gurzynski-Weiss, L. (2011). Comparing learners’ state anxiety during task-based interaction in
computer-mediated and face-to-face communication. Language Teaching Research, 15, 201–229.
Brooks, P. J., Kempe, V., & Sionov, A. (2006). The role of learner and input variables in learning infectional
morphology. Applied Psycholinguistics, 27, 185–209.
Cohen, A. D., & Weaver, S. J. (2006). Styles-and strategies-based instruction: A teachers’ guide. CARLA Working Paper
Series. Center for Advanced Research on Language Acquisition, University of Minnesota.
Côté, S., & Gafney, C. (2018). The efect of synchronous computer-mediated communication on beginner L2
learners’ foreign language anxiety and participation. The Language Learning Journal, 49(1), 105–116.
DeKeyser, R. (2012). Interactions between individual diferences, treatments, and structures in SLA. Language
Learning, 62, 189–210.
Dewaele, J.-M. (2002). Psychological and sociodemographic correlates of communicative anxiety in L2 and L3
production. International Journal of Bilingualism, 6, 23–38.
Dörnyei, Z., & Ryan, S. (2015). The psychology of the language learner revisited. Routledge.
396
Doughty, C. J. (2013). Optimizing post-critical-period language learning. In G. Granena & M. H. Long (Eds.),
Sensitive periods, language aptitude, and ultimate L2 attainment (pp. 153–175). Benjamins.
Egi, T. (2007). Interpreting recasts as linguistic evidence: The roles of linguistic target, length, and degree of
change. Studies in Second Language Acquisition, 29, 511–537.
Egi, T. (2010). Uptake, modified output, and learner perceptions of recasts: Learner responses as language aware-
ness. The Modern Language Journal, 94, 1–21.
Ellis, R. (2003). Task-based language learning and teaching. Oxford University Press.
Erlam, R. (2005). Language aptitude and its relationship to instructional efectiveness in second language acqui-
sition. Language Teaching Research, 9, 147–171.
Frear, M. W., & Bitchener, J. (2015). The efects of cognitive task complexity on writing complexity. Journal of
Second Language Writing, 30, 45–57.
Gilabert, R., & Barón, J. (2018). Independently measuring cognitive complexity in task design for interlanguage
pragmatics development. In N. Taguchi & Y. Kim (Eds.), Task-based approaches to teaching and assessing pragmat-
ics (pp. 159–186). Benjamins.
guide. Routledge.
Granena, G., & Yilmaz, Y. (2019). Language aptitude profiles and the efectiveness of implicit and explicit cor-
rective feedback. In R. P. Leow (Ed.), Routledge handbook of second language research in classroom learning (pp.
Gregersen, T. (2020). Dynamic properties of language anxiety. In L. Gurzynski-Weiss (Ed.), In pursuit of the
dynamic nature of learner individual diferences [Special issue]. Studies in Second Language Learning and Lan-
guage Teaching, 10, 67–87.
Gregersen, T., MacIntyre, P. D., & Meza, M. D. (2014). The motion of emotion: Idiodynamic case studies of
learners’ foreign language anxiety. The Modern Language Journal, 98, 574–588.
Gurzynski-Weiss, L. (2016). Factors infuencing Spanish instructors’ in-class feedback decisions. The Modern
Language Journal, 100, 255–275.
Gurzynski-Weiss, L. (Ed.). (2020). Cross-theoretical explorations of interlocutors and their individual diferences.
Benjamins.
Gurzynski-Weiss, L., Geeslin, K. L., Daidone, D., Linford, B., Long, A. Y., Michalski, I., & Solon, M. (2018).
L2 classrooms as multifaceted sources of input: The synergy of variationist and usage-based approaches. In A.
Tyler, L. Ortega, M. Uno, & H. I. Park (Eds.), Usage-inspired L2 instruction: Researched pedagogy (pp. 293–313).
Benjamins.
Hanaoka, O., & Izumi, S. (2012). Noticing and uptake: Addressing pre-articulated covert problems in L2 writ-
ing. Journal of Second Language Writing, 21, 332–347.
Horwitz, E. K., Horwitz, M. B., & Cope, J. (1986). Foreign language classroom anxiety. The Modern Language
Journal, 70, 125–132.
Hsieh, H.-C., Moreno, N., & Leow, R. P. (2015). Awareness, type of medium, and L2 development: Revisit-
ing Hsieh (2008). In R. P. Leow, L. Cerezo, & M. Baralt (Eds.), A psycholinguistic approach to technology and
language learning (pp. 131–150). de Gruyter.
Izumi, S. (2003). Comprehension and production processes in second language learning: In search of the psy-
cholinguistic rationale of the output hypothesis. Applied Linguistics, 24, 168–196.
Izumi, S., & Bigelow, M. (2000). Does output promote noticing and second language acquisition? TESOL
Quarterly, 34, 239–278.
Kartchava, E., & Ammar, A. (2014). Learners’ beliefs as mediators of what is noticed and learned in the language
classroom. TESOL Quarterly, 48, 86–109.
Kim, Y. (2012). Task complexity, learning opportunities, and Korean EFL learners’ question development. Stud-
ies in Second Language Acquisition, 34, 627–658.
Lantolf, J. P. (2020). I~ You> I~ Me. In L. Gurzynski-Weiss (Ed.), Cross-theoretical explorations of interlocutors and
their individual diferences (pp. 79–97). Benjamins.
Lee, J. (2019). Time-on-task as a measure of cognitive load in TBLT. Journal of Asia TEFL, 16, 958–969.
Leow, R. P. (2000). A study of the role of awareness in foreign language behavior. Studies in Second Language
Leow, R. P., Johnson, E., & Zárate-Sández, G. (2011). Getting a grip on the slippery construct of awareness:
Toward a finer-grained methodological perspective. In C. Sanz & R. P. Leow (Eds.), Implicit and explicit condi-
tions, processes and knowledge in SLA and bilingualism (pp. 61–72). Georgetown University Press.
Li, S. (2013). The interactions between the efects of implicit and explicit feedback and individual diferences in
language analytic ability and working memory. The Modern Language Journal, 97, 634–654.
397
In L. Gurzynski-Weiss (Ed.), Expanding individual diference research in the interaction approach: Investigating learn-
ers, instructors, and other interlocutors (pp. 41–70). Benjamins.
Li, S. (2023). Working memory and second language learning: a critical and synthetic review. In A. Godfroid
Routledge.
Li, S., & DeKeyser, R. (2021). Implicit language aptitude: Conceptualizing the construct, validating the mea-
sures, and examining the evidence. Studies in Second Language Acquisition, 43(3), 473–497.
Long, A. Y., & Geeslin, K. L. (2020). Examining the role of instructor first language in classroom-based oral
input. In L. Gurzynski-Weiss (Ed.), Cross-theoretical explorations of interlocutors and their individual diferences (pp.
159–185). Benjamins.
Long, M. H. (2007). Problems in SLA. Erlbaum.
MacIntyre, P. D. (2012). The idiodynamic method: A closer look at the dynamics of communication traits. Com-
munication Research Reports, 29, 361–367.
MacIntyre, P. D. (2017). An overview of language anxiety research and trends in its development. In C. Gkonou,
M. Daubney, & J. M. Dewaele (Eds.), New insights into language anxiety: Theory, research and educational implica-
tions (pp. 11–30). Multilingual Matters.
Mackey, A. (2006). Feedback, noticing and instructed second language learning. Applied Linguistics, 27, 405–430.
Mackey, A., Gass, S. M., & McDonough, K. (2000). How do learners perceive interactional feedback? Studies in
Mackey, A., Polio, C., & McDonough, K. (2004). The relationship between experience, education and teachers’
use of incidental focus-on-form techniques. Language Teaching Research, 8, 301–327.
Malicka, A. (2020). The role of task sequencing in fuency, accuracy, and complexity: Investigating the SSARC
model of pedagogic task sequencing. Language Teaching Research, 24(5), 642–665.
Manchón, R. M. (Ed.). (2011). Learning-to-write and writing-to-learn in an additional language. Benjamins.
Nakatsukasa, K. (2016). Efcacy of recasts and gestures on the acquisition of locative prepositions. Studies in
Neville, D., Pfost, P., & Dobbs, V. (1967). The relationship between test anxiety and silent reading gain. American
Educational Research Journal, 4, 45–50.
Pang, F., & Skehan, P. (2014). Self-reported planning behavior and second language performance in narrative
retelling. In P. Skehan (Ed.), Processing perspectives on task performance (pp. 95–128). Benjamins.
Papi, M., Rios, A., Pelt, H., & Ozdemir, E. (2019). Feedback-seeking behavior in language learning: Basic
components and motivational antecedents. The Modern Language Journal, 103, 205–226.
Petersen, K., & Sachs, R. (2015). The language classroom in the age of networked learning. In R. P. Leow, L.
Cerezo, & M. Baralt (Eds.), A psycholinguistic approach to technology and language learning (pp. 3–22). Mouton
De Gruyter.
Philp, J. (2003). Constraints on “noticing the gap”: Nonnative speakers’ noticing of recasts in NS-NNS interac-
tion. Studies in Second Language Acquisition, 25, 99–126.
Pishghadam, R., & Akhondpoor, F. (2011). Learner perfectionism and its role in foreign language learning suc-
cess, academic achievement, and learner anxiety. Journal of Language Teaching & Research, 2, 432–440.
Polio, C., Gass, S. M., & Chapin, L. (2006). Using stimulated recall to investigate native speaker perceptions in
native-nonnative speaker interaction. Studies in Second Language Acquisition, 28, 237–267.
Rebuschat, P., Hamrick, P., Riestenberg, K., Sachs, R., & Ziegler, N. (2015). Triangulating measures of awareness:
A contribution to the debate on learning without awareness. Studies in Second Language Acquisition, 37, 299–334.
Révész, A. (2009). Task complexity, focus on form, and second language development. Studies in Second Lan-
guage Acquisition, 31, 437–470.
Révész, A., & Michel, M. (2019). Methodological advances in investigating L2 writing processes. Studies in
Révész, A., Michel, M., & Gilabert, R. (2016). Measuring cognitive task demands using dual-task methodol-
ogy, subjective self-ratings, and expert judgments: A validation study. Studies in Second Language Acquisition,
38, 703–737.
Robinson, P. (1997). Individual diferences and the fundamental similarity of implicit and explicit adult second
language learning. Language Learning, 47, 45–99.
Robinson, P. (2005). Aptitude and second language acquisition. Annual Review of Applied Linguistics, 25, 46–73.
Robinson, P. (2012). Individual diferences, aptitude complexes, SLA processes, and aptitude test develop-
ment. In M. Pawlak (Ed.), New perspectives on individual diferences in language learning and teaching (pp. 57–75).
Springer.
398
Sachs, R., Akiyama, Y., & Nakatsukasa, K. (2019). The value of introspective measures in aptitude-treatment
interaction research: A window on individual diferences in action. Journal of Second Language Studies, 2,
336–364.
Sachs, R., & Nakatsukasa, K. (2019). Aptitude-treatment interactions in depth of processing: Individual difer-
ences and prior linguistics coursework predict learners’ approaches to computer-mediated language learning
activities. In R. P. Leow (Ed.), Routledge handbook of second language research in classroom learning (pp. 422–437).
Routledge.
Sachs, R., & Polio, C. (2007). Learners’ uses of two types of written feedback on a L2 writing revision task.
Studies in Second Language Acquisition, 29, 67–100.
Sangarun, J. (2005). The efects of focusing on meaning and form in strategic planning. In R. Ellis (Ed.), Plan-
ning and task performance in a second language (pp. 111–142). John Benjamins.
Santos, M., López-Serrano, S., & Manchón, R. M. (2010). The diferential efect of two types of direct writ-
ten corrective feedback on noticing and uptake: Reformulation vs. error correction. International Journal of
English Studies, 10, 131–154.
Sasayama, S. (2016). Is a ‘complex’ task really complex? Validating the assumption of cognitive task complexity.
The Modern Language Journal, 100, 231–254.
Sato, M., & Loewen, S. (2018). Metacognitive instruction enhances the efectiveness of corrective feedback:
Variable efects of feedback types and linguistic targets. Language Learning, 68(2), 507–545.
Sheen, Y. (2008). Recasts, language anxiety, modified output, and L2 learning. Language Learning, 58, 835–874.
Shintani, N., & Ellis, R. (2015). Does language analytical ability mediate the efect of written feedback on gram-
matical accuracy in second language writing? System, 49, 110–119.
Skehan, P. (2009). Modelling second language performance: Integrating complexity, accuracy, fuency, and lexis.
Skehan, P. (2015). Foreign language aptitude and its relationship with grammar: A critical overview. Applied
Linguistics, 36, 367–384.
Skehan, P. (2016). Foreign language aptitude, acquisitional sequences, and psycholinguistic processes. In G.
Granena, D. O. Jackson, & Y. Yilmaz (Eds.), Cognitive individual diferences in second language processing and
acquisition (pp. 17–40). Benjamins.
Smith, B. (2012). Eye tracking as a measure of noticing: A study of explicit recasts in SCMC. Language Learning
& Technology, 16, 53–81.
Spielberger, C. D. (1989). State-trait anxiety inventory (2nd ed.). Consulting Psychologists Press.
Suzuki, Y., & DeKeyser, R. (2017). Exploratory research on second language practice distribution: An aptitude
x treatment interaction. Applied Psycholinguistics, 38, 27–56.
Takahashi, S. (2005). Pragmalinguistic awareness: Is it related to motivation and proficiency? Applied Linguistics,
26, 90–120.
Torres, J. (2018). The efects of task complexity on heritage and L2 Spanish. Canadian Modern Language Review,
74, 128–152.
Vatz, K., Tare, M., Jackson, S. R., & Doughty, C. J. (2013). Aptitude-treatment interaction studies in second
language acquisition: Findings and methodology. In G. Granena & M. H. Long (Eds.), Sensitive periods, lan-
guage aptitude, and ultimate L2 attainment (pp. 271–290). Benjamins.
Yang, L., & Zhang, L. (2010). Exploring the role of reformulations and a model text in EFL students’ writing
performance. Language Teaching Research, 14, 464–484.
Zalbidea, J. (2017). One task fits all? The roles of task complexity, modality, and working memory capacity in
L2 performance. The Modern Language Journal, 101, 335–352.
Zamora, C. C. (2017). The secret is in the processing: A study of levels of explicit computerized feedback in heritage and
L2 learners of Spanish [Doctoral dissertation], Georgetown University. https://repository.library.georgetown.
edu/handle/10822/1044648
Ziegler, N., & Smith, G. (2017). Teachers’ provision of feedback in L2 text-chat. In L. Gurzynski-Weiss (Ed.),
Expanding individual diference research in the interaction approach: Investigating learners, instructors, and other interlocu-
tors (pp. 255–279). Benjamins.
399
32
SECOND LANGUAGE WRITING
Rosa M. Manchón
Introduction
The inquiry into the psycholinguistics of writing, i.e., the processing dimension of the act of
composing, has been at the heart of research interests in the fields of both first (L1) and additional
language (L2) writing. A body of L2 writing studies attempts to illuminate the language learning
potential that may derive from this processing dimension, i.e., from the implementation of the
cognitive operations (planning, formulation, revision, monitoring; see Textbox 32.1) that charac-
terize the act of composing itself, as well as from the cognitive processing of the feedback received
on one’s own writing. This work is a central focus in current discussions of the psycholinguistics
of L2 writing. This chapter traces developments in this incipient but fast-growing research domain
at the intersection between second language acquisition (SLA) and L2 writing studies and does so
by synthesizing key theoretical and empirical investigations of the relation between writing and
language learning.
Writing processes: The set of underlying cognitive operations—planning, formulation, revision, moni-
toring—that characterize composing in paper-based and screen-based environments in individual and
collaborative writing conditions.
Planning: The phase of the writing process when writers make decisions about their task approach, pro-
cedures for task completion, and the content and organization of their writing.
Formulation: The process of linguistic encoding, or the conversion of ideas into language.
Revision: The process whereby writers assess the match/mismatch between their intentions and their
linguistic expression.
Feedback processing: The part of L2 writers’ engagement with the feedback received on their writing in
which they invest cognitive efort into the analysis and understanding of the feedback.
Depth of processing (DoP): The amount and dimensions of cognitive efort expended by the writer when
processing the feedback received on their writing.
400 DOI: 10.4324/9781003018872-37

The psycholinguistics of second language writing
The study of the psycholinguistics of writing as a site for language learning can be traced back to
several empirically-based proposals informed by L1 writing and SLA theorizing (see Manchón, 2020,
for a more detailed discussion). Chronologically, the first of these initiatives was Alister Cumming’s
(1990) reanalysis and reinterpretation of the think-aloud data collected for his PhD thesis involving
23 adult Francophone learners of English who composed L2 texts of diferent complexity (Cum-
ming, 1989). This reanalysis led Cumming to articulate the language learning potential of writing
as follows:
Composing might function broadly as a psycholinguistic output condition wherein learners

analyze and consolidate second language knowledge that they have previously (but not fully)
acquired. . . . Composition writing elicits an attention to form-meaning relations that may
prompt learners to refine their linguistic expression—and hence their control over their
linguistic knowledge—so that it is more accurately representative of their thoughts and of
standard usage.
(Cumming, 1990, p. 483)
While acknowledging that “writing alone” is not “a sufcient or necessary condition for second lan-
guage learning” (Cumming, 1990, p. 503), and referring to Swain’s output hypothesis (Swain, 1985),
Cumming made a strong case for the language learning gains that could result from those forms of writ-
ing that require an intense linguistic meaning-making activity. Accordingly, intentional linguistic process-
ing while composing was suggested as the necessary condition for writing to result in language learning
gains, which underscores the relevance of explicit, intentional processes in learning through writing. This
position has remained central in current accounts of the psycholinguistics of L2 writing (see Bitchener,
2019; Leow, 2020; Manchón & Williams, 2016; Roca de Larios, 2013; Williams, 2012). In a more
recent elaboration of his original position, Cumming (2020) argued that writing leads to language learn-
ing “through processes of composing” (p. 32), adding that “conspicuous opportunities for learning the
L2 appear when writers evaluate forms of the L2 in relation to their intended meanings, search earnestly
to find the best words to express ideas, and switch purposefully between languages to make principled
decisions” (p. 32). As discussed later in the chapter, new lines of research aim to elucidate the nature of
this linguistic processing while composing and the corresponding language learning afordances.
The second empirically based initiative that was instrumental in advancing disciplinary debates
was Harklau’s (2002) classroom-based research in American schools, which revealed the relevance
and ever-present role of the printed word in instructed L2 learning. Accordingly, Harklau made a
call to make literacy more central in instructed SLA (ISLA) theory and research with her invita-
tion to the field to “interrogate research and theories of second language acquisition that do not
adequately account for the role of literacy in classroom learning” (p. 345), and she advocated more
modality-sensitive SLA research agendas. As we will also see in subsequent sections, a classroom/
curriculum-based orientation is central in recent proposals for moving forward research on the con-
nection between writing and language learning (e.g., Leow & Manchón, 2021), and some advances
have been made in Harklau’s proposed modality-sensitive research agenda.
A few years later, Manchón and Roca de Larios (2008) provided an SLA-oriented, problem-solving
articulation of the language learning afordances of L2 writing, which was based on the empirical insights
obtained in a long-term research program (synthesized in Manchón et al., 2009). In that research pro-
gram, they investigated cognitive activity while writing as well as the manner in which writing processes
and strategies varied as a function of L2 proficiency and language of writing (L1 or L2). Three groups
of participants were invited to write narrative and argumentative texts in their L1 and L2 while thinking
aloud in timed writing conditions. A key concern in the global program of research was to inspect the
401
Rosa M. Manchón
allocation of attentional resources, which was operationalized as the percentage of total time-on-task
spent on diferent composing activities, i.e., planning, formulation, revision, monitoring. In addition,
the program investigated whether, and, if so, how, writing in an L2 imposes competing demands for
attention, which, as a result, infuence the kind of problems posed and the nature of the problem-solving
strategies implemented. The researchers were especially interested in testing the theoretically grounded
assumption that the process of formulation (i.e., the process of transforming ideas into language) would
have particular characteristics in L2 (compared to L1) writing and infuence the writer’s problem-solving
behavior as a result of diferences in the status and accessibility of L1 and L2 knowledge.
The various studies conducted within this global research program provided a wealth of empiri-
cal data that showed the rich linguistic processing and, correspondingly, the rich and most intrigu-
ing problem-solving behavior that Cumming (1990) had anticipated as an integral component of
composing. Especially relevant was the finding that, regardless of language of composition and level
of proficiency, writers spent most of their time on the process of formulation, i.e., their attentional
resources were devoted primarily to linguistic encoding. Yet, the ratio of fuent formulation versus
problem-solving formulation varied across languages: Transforming ideas into language was five
times more fuent in L1, as opposed to a 2:1 fuency by problem-solving ratio in L2 writing, which
showed that writing in an L2 imposes a heavier burden on writers. Furthermore, these studies
showed that the amount and nature of the linguistic problems tackled during problem-solving for-
mulation episodes varied as a function of proficiency: As proficiency increased, writers spent more
time on refining their linguistic choices and less time on solving problems that resulted from lack of
access to the linguistic resources they needed. Proficiency also afected the allocation of attentional
resources to macro writing processes, both globally and at diferent times in the composing process
(see, for instance, Manchón & Roca de Larios, 2007, for the process of planning), and higher profi-
ciency also allowed writers to be able to strategically decide which attentional resources to devote to
which composing activity at any given point in the composing process.
Based on this empirical evidence, problem-solving activity while writing was central in Manchón
and Roca de Larios’s (2008) psycholinguistic rationale for the language learning potential of L2 writ-
ing. Grounded in cognitive approaches to SLA, their argument was that the psycholinguistic mecha-
nisms L2 writers implement when transforming ideas into language (lexical searches, reformulations,
etc.) could lead to changes in the their underlying linguistic system. More recently, López-Serrano
et al. (2019, 2020) provided a fuller analysis of problem-solving activity while writing, as discussed
in a later section. These three empirically-based initiatives constituted the foundation for subsequent
disciplinary developments. In her epilogue to the special issue Exploring L2-writing-SLA interfaces in
the Journal of Second Language Writing in 2012, Ortega (2012) anticipated “a new dialogue between
the L2 writing and the SLA research communities” (p. 413), one that would “eventually contribute
to change in the landscape of both fields, by leaving a trail of valuable intellectual bridges among
the relevant L2 writing and SLA research communities” (p. 413). Her predictions proved to be well
grounded because worthy theoretical and empirical eforts have been put into building these “intel-
lectual bridges,” as synthesized in the sections that follow.

Theoretical accounts of L2 writing as a site for language learning have been elaborated and expanded on
the basis of both models of writing and SLA theorizing (see Leow & Su, 2022; Storch, 2022, for compre-
hensive reviews). Additionally, empirical studies have tested the theoretical tenets and predictions on how
writing and feedback processing can foster language learning. To date, given the recency of this research,
these studies have been mainly descriptive and exploratory in nature rather than hypothesis testing.
Importantly, theoretical and empirical research agendas include both writing and feedback, even though
the latter was not part of the initial formulations of writing as a potential site for language learning.
402
As regards writing, important advances have taken place in what Cumming (2020) considers the
“narrow” perspective, i.e., understanding learning through writing in terms of cognitive processing.
Two key questions have guided theoretical initiatives: (a) what is unique about and/or characteristic
of writing and of writers’ engagement with feedback that can lead to advancing language compe-
tences; and (b) what kind of learning can be expected to result. The global answer to these questions
was summarized as follows:
L2 writing . . . can be a privileged site of language learning as a result of the deeper lin-
guistic processing that may be brought about by three conditions: first, the time pace that
characterizes writing and feedback processing, which, with minor exceptions, always takes
place ofine, in contrast to the on-line nature of oral communication; second, the visibility
and permanence of both the written text and the feedback on it, which contrasts with
the ephemeral nature of spoken language and oral interactional exchanges; and third, the
challenging, problem-solving nature of many forms of writing, together with the problem-
solving and reprocessing processes that may result from the active engagement with feed-
back that is possible in the environment of writing.
(Manchón & Cerezo, 2018, p. 2)
The pace and permanence of writing and feedback processing allow for L2 writers to be more in
control of their attentional resources, more prone to prioritizing linguistic concerns and, accordingly,
more likely to attend to incoming input and engage in noticing processes, to focus on language-
related concerns during both their composing activity and their engagement with feedback and, as
a result, to engage in metalinguistic refection and analysis. In essence, writing is predicted to ofer
favorable conditions for learning from output, as predicted in the original formulation of Swain’s
Output Hypothesis (Swain, 1985). Empirical evidence for these potential learning afordances of
L2 writing has been provided in task-related studies and studies of writing processes, which will be
reviewed in the next section.
Manchón and Williams (2016) also attempted to theorize likely learning outcomes through
writing. They distinguished between “indirect” and “direct” language learning efects. The former
are associated to learning processes such as cognitive comparison, noticing, and focus on form/
metalinguistic refection that may be facilitated by both the greater need for precision in most
forms of writing (as opposed to speaking) and the greater availability of time in the writing condi-
tion. Direct efects refer to changes in the underlying linguistic system and could come about in
two ways:
Seen through an SLA lens, within the broad and quite plausible claim that writing can
promote language development, lie two potentially conficting possibilities. Thus, writing
can actually lead to a change in linguistic knowledge by encouraging learners to analyze
their (possibly inaccurate) implicit L2 knowledge (Bialystok, 1994) or, more significantly,
by encouraging them to draw repeatedly on explicit knowledge, thereby promoting the
restructuring of their L2 knowledge in the opposite direction—from explicit to implicit
(N. Ellis, 2011). The second potential claim is that writing might, as Cumming (1990)
originally suggested, simply encourage retrieval of explicit, analyzed L2 knowledge, perhaps
improving the accuracy of output, or causing L2 writers to automatize their explicit knowl-
edge (e.g. DeKeyser, 2007), but leaving the developing L2 system qualitatively unchanged.
(Manchón & Williams, 2016, pp. 569–570)
As suggested by Godfroid (personal communication), this opens interesting possibilities to explore

the interaction of explicit and implicit processes during composition as well as the relevance of
403
Rosa M. Manchón
explicit knowledge in writing. Empirical findings (e.g., in Manchón and Roca de Larios’s program of
research reported earlier) suggest that greater L2 competence frees up conscious attentional resources
for engaging with the conceptual aspects of writing and problem-solving behavior. For instance,
Manchón et al. (2009) observed that problem solving among more proficient writers improved lin-
guistic choices as opposed to merely compensating for linguistic deficits, which was the kind of prob-
lem-solving activity they found to be prevalent in the composing process of less proficient writers.
Yet, it is important to note that L2 writing studies have pointed to an intriguing interaction
of L2 proficiency and general writing competence. L2 proficiency and general writing compe-
tence may develop simultaneously in L1 and L2 writing for some L2 users, whereas others may
develop writing competence and abilities in one of their languages (either the L1 or sometimes,
in an academic context, the L2), which can then be transferred to their writing in any of the
other languages in their linguistic repertoire. This means that problem-solving behavior while
writing should not be taken to be simply a question of the status of the writer’s L2 linguistic
knowledge. In short, as noted by Godfroid (personal communication), writing likely does not
result immediately in the creation of new implicit knowledge, but the prediction is that it may
prompt learners to try to call on the implicit knowledge that they already possess, as well as draw
on and consolidate their explicit knowledge, as originally postulated by Manchón and Williams
(2016):
It has been claimed that the act of writing prompts learners to consult their explicit knowl-
edge. . . . In other words, the interaction of the two knowledge stores is bidirectional;
indeed, any direct role for writing in knowledge creation depends on this claim. The first
direction—the claim that implicit knowledge can be analyzed and made explicit—is prob-
ably not terribly controversial. More controversial is the other direction: Can the creation,
retrieval, or use of explicit knowledge result in a change to the developing L2 system, as
claimed by proponents of the strong and weak interface positions?
(Manchón & Williams, 2016, p. 578)
Another line of research concerns the language learning potential of feedback processing (as reviewed
in Leow & Su, 2022, and Storch, 2022). Of special note are two recent model building attempts,
namely, Leow’s (2020) framework based on his 2015 model of SLA processes, and Bitchener’s (2019)
model of written corrective feedback (WCF) processing based on Gass’s (1997) cognitive model of
input processing.
Leow’s (2020) fine-grained WCF processing framework provides a cognitive account of the role
of WCF in subsequent L2 development with a strong focus on the L2 writer’s processing of the
feedback received. The framework postulates stages of processes and makes predictions about condi-
tions for successful WCF processing, including depth of processing, level of awareness, activation of
prior knowledge, hypothesis testing, rule formation, and metacognition. The basic premise is that the
degree of cognitive engagement with the feedback would enable the restructuring of L2 knowledge,
allowing for the possibility of only partial restructuring, i.e., immediate uptake (better text revisions),
or complete restructuring (longer-term learning, as manifested in accurate delayed performance).
The model further stipulates that:
whether the feedback has been attended to, detected, or noticed once again depends upon
the attentional resources allocated to the feedback by the learner in addition to the depth
of processing and level of awareness involved to make the connection between the learner’s
prior inaccurate knowledge or output and the information in the feedback received. In
other words, whether the feedback processing allows for potential restructuring of the
inaccurate knowledge may depend upon how deeply the feedback is processed or the level
404
of awareness in relation to the mismatch between the learner’s prior knowledge and the
feedback.
(Leow, 2020, p. 104)
In line with the essence of Leow’s model, Bitchener’s (2019) model also postulates stages of input
processing as well as factors infuencing them. The stages include paying attention to WCF input,
noticing the gap between one’s own output and the WCF received, understanding WCF input,
analyzing/comparing WCF input and knowledge in long-term memory, and hypothesis formation
and testing. The model also makes predictions about the range of feedback-related variables that
may infuence some of these stages, especially those of noticing and understanding WCF, and the
moderating efects of a range of cognitive and afective individual diferences, especially in con-
nection with understanding and analyzing WCF and when formulating and testing hypotheses.
Similar to Leow’s suggestions, Bitchener includes the temporal dimension of cognitive processing
of the feedback by virtue of a distinction between processing within and beyond the initial feedback
episode.
Given that the body of empirical data on feedback processing is scarce (see review in Roca de Lar-
ios & Coyle, 2022), additional descriptive analyses of feedback processing would be relevant to test
the theoretical predictions of Leow’s and Bitchener’s models. Several dimensions could be explored:
First, both models emphasize the relevance of the use of prior knowledge during various
stages of feedback processing. Accordingly, it would be worth comparing the use of prior
knowledge in L2 writers with diferent educational backgrounds and the resulting varying pro-
files, especially in terms of degree of previous background (or lack of background) in language
and linguistics.
Second, both models also predict an infuence of additional individual diferences and highlight
the crucial role of motivation and the resulting attention to focus on form in infuencing the degree
of cognitive engagement with the feedback provided. If, as Leow’s model predicts, L2 writers need to
go beyond simple noticing (i.e., detection) of items in the feedback for L2 knowledge restructuring
to take place, future correlational studies are needed that look into the connection between motiva-
tion and depth of processing or the degree of cognitive involvement.

Two relevant lines of research on the language learning afordances of writing itself are task-related
studies, on the one hand, and studies of macro- and micro-writing processes, on the other. Regard-
ing WCF, psycholinguistic interests revolve primarily around (i) L2 writers’ depth of processing
(DoP) of the feedback received and moderating factors, especially the role of individual diferences
(see Ahmadian & Vasylets, 2021, and Papi, 2021 for comprehensive reviews); and (ii) the relationship
between DoP and subsequent revisions.
Task-related studies
A growing line of SLA-oriented research is concerned with task modality, task complexity, and task
repetition efects in terms of three main outcome variables: (i) how language learning is fostered via
processes such as noticing, focus on form, hypothesis testing, monitoring (the “indirect efects” of
writing on learning), and their efects on written output; (ii) what the characteristics of the texts are;
(iii) what afordances writing provides in these diverse task conditions for learning gains in terms of
either the expansion or the consolidation of L2 knowledge (“direct efects” of writing on L2 learning;
see Manchón & Williams, 2016). This rapidly developing body of work (see Johnson, 2022; Manchón
& Vasylets, 2019; Vasylets & Gilabert, 2022, for reviews) has targeted writing in diferent time-on-task
405
Rosa M. Manchón
conditions (i.e., in time-compressed and in more time-extended conditions), writing with and with-
out access to external sources, writing individually and collaboratively, and writing in pen and paper
and in digital environments.
Task-modality studies and task complexity studies attempt to shed light on the way in which
modality (oral vs. written) or the cognitive complexity of writing tasks can impact learning oppor-
tunities. This research is particularly relevant for testing whether the language learning potential of
writing is higher compared to speaking, on account of, firstly, the greater availability of time in the
writing condition, and, secondly, the problem-solving nature of composing. In order to test these
predictions, some researchers have collected data on the performance of the same task across modali-
ties (e.g., Tavakoli, 2014; Zalbidea, 2021), others have experimentally manipulated the complexity of
the task to be performed (e.g., Révész et al., 2017), whereas others have investigated the interaction
of task modality and task complexity (e.g., Vasylets et al., 2017, 2020; Zalbidea, 2017, 2021).
Regarding task modality, Cumming (1990) anticipated that “[w]hile composing in a second lan-
guage, learners may be obliged to monitor their language production in a way that is not necessary or
feasible under the time constraints of comprehending or conversing in the second language” (p. 483).
Along the same lines, Manchón and Vasylets (2019) observed that “because of the pressures during
on-line communication, there might not be enough time for restructuring to occur as processing
resources may be devoted in full to achieving communicative intentions. Writing, in contrast, allows
for a more relaxed and strategic production whose outcome may be higher complexity of written
discourse” (p. 346). Following the same line of thinking, Manchón and Williams (2016) suggested
that the act of producing written output entails more opportunity (and probably greater need) for
focus on form than does speaking. This opportunity, they argued, comes from the greater availability
of time in the written mode compared to the oral mode, which allows for more deliberate retrieval
of explicit knowledge, as well as pre-task and during-task opportunities for focus on form.
Research shows that the language being used in writing is more complex. For instance, Vasyl-
ets et al. (2017) looked at the interaction of task complexity and task modality and found that the
written mode led to the production of more complex, informationally dense ideas, as well as the
production of more linguistically complex discourse. In fact, studies looking into these interaction
efects have also systematically reported that task modality appears to play “a more robust role than
task complexity in promoting improved linguistic performance” (Zalbidea, 2017, p. 348). In this
respect, Vasylets et al. (2017) found that mode exerted greater efects on performance than task
complexity, and that task complexity afected oral and written production in distinct ways: While
the linguistic complexity of production was similar across two modes when writers performed more
and less complex tasks, modality-related efects emerged in the areas of propositional complexity,
accuracy, and time on task. Another relevant finding was that more variation between the complex
and simple versions of the task was found in the written mode. These insights can be linked to
Cumming’s (1990) original postulation of the task-dependency of composing processes. Yet, task
modality and task complexity studies to date have primarily focused on efects on performance,
understood in terms of complexity, accuracy, and fuency measures, and hence open questions
exist about how task modalities and the degree of complexity of the task diferentially infuence L2
composing processes and learning.
Studies of writing processes

Recent studies of L2 writing processes have provided abundant empirical evidence on online writing
behaviors, especially with respect to fuency and pausing (e.g., Chukharev-Hudilainen et al., 2019), whereas
other studies are more directly concerned with higher-order writing processes and the purported problem-
solving nature of composing (e.g., López-Serrano et al., 2019). Some of the studies looking into online
writing behaviors have been guided by test validation concerns (e.g., Michel et al., 2020), whereas others
406
have investigated the relationship between cognitive processes and online behaviors, and/or text
quality (e.g., Barkaoui, 2019; Révész et al., 2017, 2019). Collectively, they have used a range of
data collection procedures intended to provide clues about internal processes by capturing external,
observable behavior, mainly keystroke logging, eye-tracking, and screen capture technologies, at
times supplemented with introspection data, collected often through post-task stimulated recalls.
For example, in their study of cognitive processes underlying pauses at diferent textual locations
(e.g., within/between words) and various levels of revision (e.g., below word/clause), Révész et al.
(2019) triangulated keystroke logging and eye-tracking data (obtained during task performance),
together with stimulated recall (collected after task performance). Their results pointed to a con-
nection between engagement in what the authors call “higher-order” or “lower-order” cognitive
processes and the textual locations of pauses: Pausing at larger textual units went hand in hand with
the implementation of higher-order writing processes such as planning content, whereas pauses at
lower textual units were related to writers’ engagement in lower-order writing processes such as lexi-
cal retrieval or syntactic encoding.
Other studies have inspected the nature and temporal distribution of macro-writing processes
(planning, formulation, revision, and monitoring; see Michel et al., 2022, for review). For instance,
Gánem-Gutiérrez and Gilmore (2018) triangulated digital screen capture and eye-tracking technolo-
gies, together with retrospective recalls, to look into the role of proficiency in L2 writing processes
during real-time writing. They found clear diferences in L2 writing activity at diferent points in the
total composition time as well as proficiency-related diferences in cognitive activity in terms of both
frequency and duration of diferent writing processes, especially regarding revision, which became
more frequent with increasing proficiency.
López-Serrano et al. (2019) developed a psycholinguistically informed coding system intended
“to capture in a comprehensive way the language processing activity learners engage in when solving
the linguistic problems they face during individual writing tasks” (p. 504). Building on previous work
on the processing dimension of writing, the basic premise of their program of research was:
L2 writers activate search processes (accessing and retrieving lexical and morphosyntactic
items from memory and constructing connections between them) that enable them to
generate alternative linguistic options and to assess the forms generated. These search and
assessment cycles may also result in the modification and consolidation of L2 knowledge or
even in the internalization of new knowledge.
(López-Serrano et al., 2019, p. 505)
To shed light on these issues, they analyzed the think-aloud data provided by L2 writers at diferent
proficiency levels while writing in their L2 (English) in time-constrained conditions. They pursued
two aims. First, to provide a detailed description of the participants’ linguistic processing via an analy-
sis “situated in the inner workings of the text generation processes responsible for translating ideas
into linear strings of language” (p. 506) and the production of the coding scheme mentioned earlier.
Second, to theorize the potential connection between writing and language learning on the basis of
the analysis of their empirical data and the resulting coding scheme. To achieve their first aim, their
analytical procedure entailed a number of steps. First, the identification of problem-solving units,
operationalized as “language-related episodes” (LRE, verbalizations in the think-aloud data that indi-
cated problem-solving behavior related to linguistic encoding), and the focus of LREs (grammar,
lexis, etc.). Second, the analysis of the nature of the problem-solving behavior writers engaged in,
which, in turn, entailed three steps: (a) identification of the writing strategies (or cluster of strategies)
used; (b) analysis of the orientation of the strategies used in each problem-solving unit (as in Man-
chón et al., 2009, they distinguished between a “compensatory” and an “upgrading” orientation),
and (c) analysis of the depth of processing of problem-solving behavior.
407
Rosa M. Manchón
Based on their analyses, they provided the following working hypothesis for the exploration of
writing as a site for language learning:
LREs in self-sustained writing tasks will hypothetically involve more language learning
potential if (a) the complexity of the linguistic unit attended to is in consonance with the
proficiency level of the writer; (b) the use of writing strategies is fexible and upgrading
in orientation; and (c) the production of the structure concerned involves high depth of
processing and repeated engagement in the subprocesses of writing, with special emphasis
on the stages of translating and monitoring, which allow learners to search the available
knowledge (explicit and implicit) they may possess.
(López-Serrano et al., 2019, p. 523)
This study constitutes the most comprehensive account of the linguistic processing dimension of individual
writing to date, and it makes a theoretical and methodological contribution for future studies intended to
zoom into linguistic processing while writing and the resulting learning outcomes. Importantly, as noted in
an earlier section, it would be relevant to test the predictions made about learning through writing in stud-
ies with a wider range of L2 users in an attempt to cover a fuller spectrum of, among others, educational
backgrounds, degrees of writing expertise, motivation and goals for writing, and proficiency in the L2.
Studies of written corrective feedback processing

Feedback processing studies provide empirical evidence for the theoretical accounts about the connec-
tion between feedback processing, in particular how deeply L2 writers process the feedback they are
provided with, and potential language learning gains (Bitchener, 2019; Leow, 2020). In the context of
SLA approaches to attention and noticing, a growing research interest concerns the manner in which
diferent types of feedback (more/less comprehensive, more/less explicit) foster depth of processing
(DoP; e.g., Cerezo et al., 2019; Kim & Bowles, 2019). DoP is often operationalized as “the relative
amount of cognitive efort, level of analysis, elaboration of intake together with the usage of prior
knowledge, hypothesis testing and rule formation employed in decoding and encoding some gram-
matical or lexical item in the input” (Leow, 2015, p. 204), in this case the WCF. An additional research
focus concerns the potential correlations between DoP and the nature of the revisions incorporated.
It is important to note some characteristics of this body of work, namely (i) a focus on younger and
older L2 users; (ii) the inclusion of L2 users with or without background in language and linguistics
(an incipient line of research), and in in- or out-of-school contexts; (iii) the L2s investigated (which so
far include only English and Spanish), and, very importantly, (iv) the fact that, in some studies, WCF
provision, processing, and use was embedded in the instructional sequence (see Leow & Manchón,
2022; Manchón & Leow, 2020, for further analysis). This research has contributed relevant insights
into the processing dimension of input in the form of WCF, especially for how input processing in
writing is linked to output production (see discussion in Leow & Manchón, 2022; Roca de Larios
& Coyle, 2022). Some studies have also tried to establish diferent levels of processing. For instance,
Cerezo et al. (2019) studied the written refections of L2 writers while they were comparing their
original texts and the feedback provided on their writing. To this end, they set up a coding scheme
intended to grasp degrees of cognitive efort put into the processing of feedback, as well as levels of
analysis and elaboration. These levels were established on the basis of the presence or absence in the
written verbalizations of (a) the errors in the original texts and corrections provided in the feedback,
(b) the error category (grammar, lexis, etc.), and (c) (meta)linguistic explanations on the nature of the
error, regardless of whether or not these were correct. They hypothesized that these levels of elabora-
tion refected diferent levels of awareness, ranging from “awareness at the level of noticing” (the low-
est level of cognitive engagement, Level 1), “awareness at the level of reporting” (a medium level of
processing), and awareness at the level of “understanding” (the highest level of processing).
408
Although it is possible to capture DoP in this way, it is important to note that very few studies
have so far found positive correlations between levels of processing and improvements in subsequent
revisions. In this respect, it has been speculated that variations in the level of engagement with the
WCF received might be a function of the degree of explicitness of the feedback, and that this degree
of explicitness may, in turn, interact with writing conditions. For instance, Manchón et al. (2020)
speculated that, in collaborative writing conditions, deeper engagement with the feedback could be
linked to the greater need for negotiation that would arise when learners have to process less explicit
types of feedback. They also pointed to key considerations regarding the methodological procedures
used to tap into feedback processing. In this respect, McBride and Manchón (2023) looked into the
afordances of diverse data collection procedures (namely, think-aloud, written languaging, and a
combination of both) in terms of the light they could shed on how L2 writers engage with and pro-
cess the feedback provided on their writing. They also investigated the extent to which any observed
efects were mediated by writing conditions (pen and paper vs. digital writing) and learner-related
characteristics, especially their previous backgrounds (or lack thereof) in language and linguistics. An
important finding of the study was that, while think-aloud protocols alone enabled a deeper insight
into the levels of the participants’ processing when compared to written languaging, the simultane-
ous action of both oral and written processing shed a much stronger light on the cognitive processes
participants underwent while completing the WCF processing stage. It follows that methodological
consideration should be central in future feedback processing studies.

Despite the progress made at the levels of theory and research, future advances in the psycholinguis-
tics of writing as a site for language learning require further substantive and methodological develop-
ments (see Textbox 32.2).
First, research on writing processes should be central in future SLA-oriented research agendas
given that this research connects directly with central empirical questions on how and why L2 writ-
ing may be a site for language learning through opportunities for monitoring, restructuring, elabora-
tion, consolidation, and/or refinement of the language used (see Leow & Manchón, 2022; Manchón
& Leow, 2020; Michel et al., 2022, for further elaboration). Additionally, research on writing pro-
cesses needs to expand both its scope—so that digital writing features more prominently in research
agendas—and the temporal dimension of writing in an attempt to account for the sequential, time-
distributed nature of writing events. As mentioned at diferent points in the chapter, future work also
needs to provide empirical evidence for the proposed WCF processing frameworks.
Second, it is still an open question what kind of learning accrues in writing, for instance, expan-
sion/consolidation/restructuring of explicit/implicit grammatical/lexical knowledge. Future work
should investigate potential learning outcomes across tasks and writing environments, crucially
including the comparison of paper-based and screen-based writing.
Third, more research is needed on individual diferences (IDs). A key avenue for future research
would be to examine the interaction of IDs at play in writing and in the use of WCF, as well as
whether or not efects of IDs vary as a function of the writing environment (paper-based vs. screen-
based writing) and writing conditions. Here, it is an open question whether research findings on
timed-constrained writing conditions also apply to those conditions that mirror the time-extended
nature of most forms of writing in academic and workplace settings.
Fourth, future work must go beyond one-shot studies and instead investigate writing and WCF
processing and use as part of the instructional sequence, which means adopting a curricular perspec-
tive (see also Byrnes, 2020; Leow, 2020, 2023 [this volume]).
In order to progress in these research domains, several methodological considerations will be cru-
cial. Future work should encompass lab-based studies as well as classroom-based studies. Some of the
questions awaiting an answer would require the adoption of a longitudinal, time-extended perspective,
409
Rosa M. Manchón
for instance when investigating learning through and by writing in CLIL contexts, or the implementa-
tion of writing processes in the time-extended conditions of real writing. It is also relevant to expand
the populations: similar to the arguments by Andringa and Godfroid (2020) for SLA studies globally,
our insights on the connection between writing and language learning will be limited if target popu-
lations continue to be primarily university students with a background in language and linguistics.
Similarly, studies with younger populations are still under-represented. Future researchers should also
strive for data triangulation, and employ methodological procedures that allow them to capture writ-
ing processes both at and beyond the point of inscription.
In which way do studies of writing processes that focus on online writing phenomena (such as pausing)
and those that target macro-writing processes (such as formulation or depth of processing of written
corrective feedback) diferentially contribute to our understanding of the psycholinguistics of writing
as a site for L2 learning?
How can L2 writing lead to L2 grammar and vocabulary learning? What kind of learning would result and
under what conditions? Are potential learning outcomes moderated by IDs, task-related factors, writ-
ing conditions (individual vs. collaborative writing; time-compressed vs. time-extended writing; with/
without access to external sources) or writing environments (paper-based vs. screen-based writing)?
What is the infuence of theoretically and pedagogically relevant task complexity factors on writing pro-
cesses? How do these processes correlate with writing products and longer-term learning?
When and how does explicit knowledge gained via feedback processing become available for later use
in subsequent written performance, and when does L2 knowledge gained via feedback processing
become implicit knowledge?
How does depth of processing of written corrective feedback relate to short-term (uptake) and longer-
term (retention) learning efects? What are the factors that may moderate potential efects?
Acknowledgements
This chapter is an expanded version of the author’s plenary address “L2 writing as a site for language
learning: A missing piece in the SLA puzzle,” delivered at the 2021 EUROSLA Conference.
Further reading
Manchón, R. M., & Polio, C. (Eds.). (2022). The Routledge handbook of second language acquisition and writing.
Routledge.
This handbook is a compendium of theoretical perspectives and empirical developments on the connection
between writing and language learning. Chapters in the volume provide in-depth critical analyses of past achieve-
ments and needed future developments, many of which are closely related to the psycholinguistics of L2 writing.
Roca de Larios, J. (2013). Second language writing as a psycholinguistic locus for L2 production and learning.
This paper presents a discussion of key theoretical and methodological considerations in research exploring
writing as a site for language learning from a psycholinguistic perspective.
Williams, J. (2012). The potential role(s) of writing in second language development. Journal of Second Language
Writing, 21, 321–331.
This article ofers a detailed SLA-oriented discussion of the language learning potential of L2 writing.
410
References
Ahmadian, M., & Vasylets, O. (2022). The role of cognitive individual diferences in L2 writing performance
and written corrective feedback processing and use. In R. M. Manchón & C. Polio (Eds.), The Routledge
handbook of second language acquisition and writing (pp. 139–151). Routledge.
Andringa, S., & Godfroid, A. (2020). Sampling bias and the problem of generalizability in applied linguistics.
Annual Review of Applied Linguistics, 40, 134–142.
Barkaoui, K. (2019). What can L2 writers’ pausing behavior tell us about their L2 writing processes? Studies in
Bialystok, E. (1994). Representation and ways of knowing: Three issues in second language acquisition. In
N. Ellis (Ed.), Implicit and explicit factors in second language learning: Interdisciplinary perspectives (pp. 549–569).
Academic Press.
Bitchener, J. (2019). The intersection between SLA and feedback research. In K. Hyland & F. Hyland (Eds.),
Feedback in second language writing. Contexts and issues (pp. 85–105). Cambridge University Press.
Byrnes, H. (2020). Toward an agenda for researching L2 writing and language learning. The educational context
of development. In R. M. Manchón (Ed.), The language learning potential of L2 writing: Moving forward in theory
and research (pp. 73–94). Benjamins.
Cerezo, L., Manchón, R. M., & Nicolás-Conesa, F. (2019). What do learners notice while processing written
corrective feedback? A look at depth of processing via written languaging. In R. P. Leow (Ed.), The Routledge
handbook of second language research in classroom learning (pp. 173–187). Routledge.
Chukharev-Hudilainen, E., Saricaoglu, A., Torrance, M., & Feng, H. H. (2019). Combined deployable keystroke log-
ging and eyetracking for investigating L2 writing fuency. Studies in Second Language Acquisition, 41(3), 583–604.
Cumming, A. (1989). Writing expertise and second language proficiency. Language Learning, 39, 81–141.
Cumming, A. (1990). Metalinguistic and ideational thinking in second language composing. Written Commu-
nication, 7, 482–511.
Cumming, A. (2020). L2 writing and L2 learning. Transfer, self-regulation, and identities. In R. M. Manchón
(Ed.), Writing and language learning. Advancing research agendas (pp. 29–48). Benjamins.
DeKeyser, R. (2007). Introduction: Situating the context of practice. In R. DeKeyser (Ed.), Practicing in a second
language: Perspectives from applied linguistics and cognitive psychology (pp. 1–18). Cambridge University Press.
Gánem-Gutiérrez, A., & Gilmore, A. (2018). Tracking the real-time evolution of a writing event: Second lan-
guage writers at diferent proficiency levels. Language Learning, 68(2), 469–506.
Gass, S. (1997). Input, interaction, and the second language learner. Lawrence Erlbaum.
Harklau, L. (2002). The role of writing in classroom second language acquisition. Journal of Second Language
Writing, 11, 329–350.
Johnson, M. (2022). Task complexity studies. In R. M. Manchón & C. Polio (Eds.), The Routledge handbook of
second language acquisition and writing (pp. 42–64). Routledge.
Kim, H. R., & Bowles, M. (2019). How deeply do second language learners process written corrective feed-
back? Insights gained from think-alouds. TESOL Quarterly, 4, 913–938.
Leow, R. P. (2020). L2 writing-to-learn: Theory, research, and a curricular approach. In R. M. Manchón (Ed.),
Writing and language learning. Advancing research agendas (pp. 95–117). John Benjamins.
Leow, R. P. (2023). The infuence of learning contexts on the psycholinguistics of second language learning.
In A. Godfroid & H. Hopp (Eds.), The Routledge handbook of second language acquisition and psycholinguistics
Leow, R. P., & Manchón, R. M. (2022). Directions for future research agendas on L2 writing and feedback as
language learning from an ISLA perspective. In R. M. Manchón & C. Polio (Eds.), The Routledge handbook
of second language acquisition and writing (pp. 299–311). Routledge.
Leow, R. P., & Su, B.-R. (2022). Theoretical perspectives on L2 writing, written corrective feedback, and
language learning in individual writing conditions. In R. M. Manchón & C. Polio (Eds.), The Routledge
López-Serrano, S., Roca de Larios, J., & Manchón, R. M. (2019). Language refection fostered by individual L2
writing tasks: Developing a theoretically-motivated and empirically-based coding system. Studies in Second
López-Serrano, S., Roca de Larios, J., & Manchón, R. M. (2020). Reprocessing output during L2 individual
writing tasks: An exploration of depth of processing and the efects of proficiency. In R. M. Manchón (Ed.),
Writing and language learning. Advancing research agendas (pp. 231–253). John Benjamins.
Manchón, R. M. (2020). Writing and language learning. Looking back and moving forward. In R. M. Man-
chón (Ed.), Writing and language learning. Advancing research agendas (pp. 3–26). John Benjamins.
Manchón, R. M., & Cerezo, L. (2018). Writing as language learning. In J. Liontas (Ed.), The TESOL encyclope-
dia of English language teaching. John Wiley & Sons, Inc.
411
Rosa M. Manchón
Manchón, R. M., & Leow, R. P. (2020). Investigating the language learning potential of L2 writing: Method-
ological considerations for future research agendas. In R. M. Manchón (Ed.), Writing and language learning.
Manchón, R. M., Nicolás-Conesa, F., Cerezo, L., & Criado, R. (2020). L2 writers’ processing of written cor-
rective feedback: Depth of processing via written languaging. In W. Suzuki & N. Storch (Eds.), Languaging
in language learning and teaching (pp. 241–265). John Benjamins.
Manchón, R. M., & Roca de Larios, J. (2007). On the temporal nature of planning in L1 and L2 writing. Lan-
guage Learning, 57(4), 549–593.
Manchón, R. M., & Roca de Larios, J. (2008). Writing-to-learn in instructed language learning contexts. In E.
A. Soler & M. P. S. Jordá (Eds.), Intercultural language use and language learning (pp. 101–121). Springer.
Manchón, R. M., Roca de Larios, J., & Murphy, L. (2009). The temporal dimension and problem-solving
nature of foreign language composing: Implications for theory. In R. M. Manchón (Ed.), Writing in foreign
language contexts: Learning, teaching, and research (pp. 102–129). Multilingual Matters.
Manchón, R. M., & Vasylets, O. (2019). Language learning through writing: Theoretical perspectives and
empirical evidence. In J. B. Schwieter & A. Benati (Eds.), The Cambridge handbook of language learning (pp.
341–362). Cambridge University Press.
Manchón, R. M., & Williams, J. (2016). L2 writing and SLA studies. In R. M. Manchón & P. K. Matsuda
(Eds.), The handbook of second and foreign language writing (pp. 567–586). Mouton De Gruyter.
McBride, S., & Manchón, R. M. (2023). Analysing L2 writers’ processing of written corrective feedback via
written languaging and think-aloud protocols. Methodological considerations. In R.M. Manchón & J. Roca
de Larios (eds.), Research methods in the study of writing processes. John Benjamins.
Michel, M., Révész, A., Lu, X., Kourtali, N. E., Lee, M., & Borges, L. (2020). Investigating L2 writing processes
across independent and integrated tasks: A mixed-methods study. Second Language Research, 36(3), 277–304.
Michel, M., Stiefenhöfer, L., Verspoor, M., & Manchón, R. M. (2022). L2 writing processes of language learn-
ers in individual and collaborative writing conditions. In R. M. Manchón & C. Polio (Eds.), The Routledge
Ortega, L. (2012). Epilogue: Exploring L2 writing—SLA interfaces. Journal of Second Language Writing, 21(4), 404–415.
Papi, M. (2022). The role of motivational and afective factors in L2 writing performance and written corrective
feedback processing and use. In R. M. Manchón & C. Polio (Eds.), The Routledge handbook of second language
acquisition and writing (pp. 152–165). Routledge.
Révész, A., Kourtali, N. E., & Mazgutova, D. (2017). Efects of task complexity on L2 writing behaviors and
linguistic complexity. Language Learning, 67, 208–241.
Révész, A., Michel, M., & Lee, M. (2019). Exploring second language writers’ pausing and revision behaviors:
A mixed-methods study. Studies in Second Language Acquisition, 41(3), 605–631.
Roca de Larios, J. (2013). Second language writing as a psycholinguistic locus for L2 production and learning.
Roca de Larios, J., & Coyle, Y. (2022). Learners’ engagement with written corrective feedback in individual and
collaborative L2 writing conditions. In R. M. Manchón & C. Polio (Eds.), The Routledge handbook of second
language acquisition and writing (pp. 81–93). Routledge.
Storch, N. (2022). Theoretical perspectives on L2 writing and language learning in collaborative writing and the
collaborative processing of written corrective feedback. In R. M. Manchón & C. Polio (Eds.), The Routledge
Swain, M. (1985). Communicative competence: Some roles of comprehensible input and comprehensible output in
its development. In S. Gass & C. Madden (Eds.), Input in second language acquisition (pp. 235–253). Newbury House.
Tavakoli, P. (2014). Storyline complexity and syntactic complexity in writing and speaking tasks. In H. Byrnes &
R. M. Manchón (Eds.), Task-based language learning—Insights from and for L2 writing (pp. 217–236). Benjamins.
Vasylets, O., & Gilabert, R. (2022). Task efects across modalities. In R. M. Manchón & C. Polio (Eds.), The
Routledge handbook of second language acquisition and writing (pp. 39–51). Routledge.
Vasylets, O., Gilabert, R., & Manchon, R. M. (2017). The efects of mode and task complexity on second lan-
guage production. Language Learning, 67(2), 394–430.
Vasylets, O., Gilabert, R., & Manchón, R. M. (2020). Task modality, communicative adequacy and CAF mea-
sures. The moderating efect of task complexity. In R. M. Manchón (Ed.), Writing and language learning.
Williams, J. (2012). The potential role(s) of writing in second language development. Journal of Second Language
Writing, 21, 321–331.
Zalbidea, J. (2017). “One task fits all”? The roles of task complexity, modality, and working memory capacity in
L2 performance. The Modern Language Journal, 101(2), 335–352.
Zalbidea, J. (2021). On the scope of output in SLA: Task modality, salience, L2 grammar noticing, and develop-
ment. Studies in Second Language Acquisition, 43, 50–82.
412
33
SECOND LANGUAGE ASSESSMENT
Rob Schoonen
Introduction
Psycholinguists who investigate the cognitive underpinnings of second language (L2) learning or
multilingualism often want to, or need to, monitor the development of the second or additional
language(s) being learned by the individuals, just as applied linguists and educators need to, albeit
for potentially diferent purposes. If language researchers or educators indeed want to know an L2
learner’s level of proficiency in the language of interest, they will have to measure it. However, most
readers will realize that this is easier said than done.
Language proficiency (see Textbox 33.1) is generally considered a psychological ability or char-
acteristic of language users and learners (henceforth, language learners). Although some researchers
distinguish between “ability” and “proficiency,” here the two terms will be used interchangeably.
This attribute of language proficiency is not directly observable nor obvious. Language researchers
must then elicit a language performance to infer a learner’s level. This language performance will
most likely form the basis of their evaluation of proficiency. This description is, however, deceivingly
simplistic. Questions that immediately arise are: What is the task the language learner has to perform,
and what are the conditions for task performance? What are the knowledge and skills the language
learner has to employ for the task performance? Is there only one optimal way to perform the task,
or are there more sequences of cognitive processes that lead to an optimal task performance? And
last but not least: What are the criteria for evaluating the performance? It will be clear from this set
of questions that “Pandora’s box,” in the words of Tim McNamara (1996), has indeed been opened.
A related question is why researchers or educators want to know the language learner’s level of
language proficiency. Do they just want to monitor development, or is there other inference- or
decision-making involved, based on the measurements (Davies, 1990; Hamp-Lyons, 2016)? If educa-
tors need to decide whether the language learners have reached a pre-set level (for example, in order
to be admitted to certain academic programs) other requirements apply to the assessment than for
diagnostic purposes targeting remediation or training (Alderson, 2005; Bachman & Palmer, 1996).
Establishing a language learner’s level is one possible target of assessment, but at least equally impor-
tant is assessing the potential someone has for learning a new language; that is, the language learning
ability or language aptitude (cf. MacWhinney, 1994; Poehner & Infante, 2016). In the context of L2
research, applied linguists and educators are especially interested in individual diferences between
language learners, both in terms of their proficiency level and their language learning potential. Valid
tools and procedures to discriminate reliably between language learners are paramount in such cases.
DOI: 10.4324/9781003018872-38 413

Rob Schoonen
Language proficiency: The capacity to use and process language in language use tasks
Language assessment: The evaluation of a language user’s level of language proficiency
Cognitive processes: The mental operations a language user has to go through to perform (language use) tasks
Language performance: The actual language use in language use tasks and its results, such as language
comprehension and language production
Individual diferences: The variation among individuals in the level and quality of language use and
knowledge, and/or related variables such as motivation and language exposure
Psycholinguistics and language assessment have diferent histories. Levelt (2013) painted the rich his-
tory of psycholinguistics, with two pioneering studies in the nineteenth century that have left their
traces in current experimental and observational L2 research, respectively, i.e., a study using mental
chronometry by Donders (1865) and a study of speech errors by Meringer and Mayer (1895).
The beginning of the academic study of language testing and assessment is often attributed to Rob-
ert Lado’s (1961) Language Testing: The Construction and Use of Foreign Language Tests. Lado focused on
foreign language learning and stated that the primary goal of testing is to give feedback a role in the
language learning process. He distinguished between the “elements of language” (i.e., pronunciation,
intonation, stress, grammatical structure, and vocabulary) and “integrated language skills” (i.e., auditory
comprehension, speaking, reading, writing, and translation). This distinction suggested two approaches
to language assessment, one that is based on a componential analysis of language proficiency and one
that focuses on language performance holistically. Lado also recognized that there is more to language
learning than just learning language (i.e., “beyond language,” p. 275), such as cross-cultural understand-
ing and “other higher values” (p. 290). Historically, Lado had his precursors in testers of mental abilities
at the beginning of the twentieth century. Still relevant to current language research is one of the mental
abilities from military contexts, that is, language learning aptitude. For an overview of the history of
language assessment, see Spolsky (1995) and in relation to testing purposes, Hamp-Lyons (2016).
While language testers aim to measure language proficiency, psycholinguists aim to understand
the processing of language. However, it is difcult to understand the processing of language without
measuring language comprehension or production, either cross-sectionally or longitudinally. Con-
versely, researchers may find themselves unable to measure language proficiency validly without a
proper understanding of language processing. Psycholinguists and language testers share their wish
to unravel language comprehension and production, to report language development and language
learning progress, and to diagnose issues in language processing. In language assessment, and increas-
ingly in psycholinguistics, there is ample attention to the constructs tested, questioning not only “how
they are tested”, but also “what it is that is tested” (Alderson et al., 2017). Therefore, psycholinguistics
and L2 assessment or language testing may (and should) benefit from each other’s research traditions
and accumulated knowledge. Transdisciplinary exchange can support researchers from both fields in
their attempts to assess, understand, and further language learners’ development.
Critical issues and approaches

The purpose of assessment is a critical issue that afects the design and use of L2 assessment
instruments. Such instruments have to meet the psychometric requirements of good test-
ing. Most importantly, interpretations of assessment scores need to be valid and meaningful
414
The psycholinguistics of second language assessment
(Borsboom et al., 2004; Chapelle, 2021; Messick, 1989). Some validity theorists focus on the
meaning and explanation of the test scores, whereas others extend the concept of validity to
include the consequences of test use (see Markus & Borsboom, 2013, and Chapelle, 2021, for
corresponding discussions). In this chapter, the focus will be on the interpretation of score
meaning and whether there is evidence that the intended construct is measured in a given assess-
ment. Obviously, using a test in other assessment contexts or with other target populations may
require new validity evidence.
Borsboom and colleagues (2004) presented two simple conditions for the construct validity
of the measurement of an attribute such as L2 proficiency. A measurement is valid “if and only
if (a) the attribute exists and (b) variations in the attribute causally produce variations in the
outcomes of the measurement procedure” (Borsboom et al., 2004, p. 1061). Although these
conditions are simple, they are incredibly hard to verify. Can researchers or test-score users
assume that hypothesized psycholinguistic constructs or attributes of a language learner exist?
For instance, do language learners have an ability to produce writing that is “lexically diverse”?
And moreover, can researchers or test-score users assume a causal relationship between a lan-
guage learner’s “lexical diversity” ability and the scores on a type-token or Voc D measurement
(Richards & Malvern, 2007)? If not, what then causes the measurement scores? It is critical that
psycholinguistic research plays a role here by providing or informing substantive theories of
response behavior that explain what happens between item administration and item response
(see Borsboom et al., 2004).
From a psycholinguistic perspective, learners’ language use is largely determined by their
language knowledge and the cognitive processes involved in applying the knowledge (Hulstijn,
2015). Psycholinguists try to identify and measure these processes by administering simple and
more complex tasks or by creating two or more experimental conditions and then comparing the
responses. For instance, learners’ response times can be measured by subtracting the simple from
the more complex task or condition. This subtraction method focuses on the processes involved
in each condition; it generalizes away from individual diferences in executing these cognitive
processes. In language testing and educational assessment, on the other hand, individual difer-
ences in performance are the primary target for measurement: Why is language learner A more
proficient or a more successful learner than language learner B? Integrating these two diferent
research traditions is a second critical issue for validity research. Although Cronbach (1957) tried
to reconcile these two paradigms of research, the study of cognitive processing and assessment of
individual diferences have not been connected explicitly until recently (Draheim et al., 2019;
Kidd et al., 2018).
A third critical issue concerns the level of granularity of assessment. In L2 learning and assess-
ment, the construct of language proficiency can be approached as a (single) complex integrated
skill (Oller, 1983) or as a set of a few or many subskills or “language elements” (cf. Lado, 1961).
Over time, with a greater understanding of the various components and processes of L2 use, lan-
guage learners and other stakeholders might want more detailed information about language per-
formances. Is poor speaking performance due to slow word retrieval processes, unfamiliarity with
unusual prosodic patterns, poor articulatory control, or something else? This kind of diagnostic
information may be needed to develop educational interventions. The level of detail or granular-
ity in the assessment may also depend on the possibilities that exist to remedy potential causes of
poor speaking performances. Psycholinguistics has added to the large and growing collection of
fine-grained procedures to measure language processing. However, in the educational setting, per-
formance feedback often needs to be provided at the level of the individual instead of the group
level. The psychometric quality of psycholinguistic measurements at the level of the individual,
and thus the usefulness for assessments, is often an open question (cf. Lee, 2015), and represents an
opportunity for future development.
415
Rob Schoonen
Theoretical perspectives
Language profciency and its components

The diagnosis of language proficiency is one of the many purposes (Hamp-Lyons, 2016) a language
assessment can serve, and is a growing strand of assessment research (Alderson, 2005; Alderson &
Huhta, 2011; Lee, 2015). Diagnosis can unravel the diferent components of language proficiency, as
is done in psycholinguistics, but more often the motive is to identify strengths and weaknesses at the
individual level to inform learning and teaching.
One approach to diagnosis could be to develop discrete-point tests that target language elements,
which may be as specific as the individual phonemes of the foreign language (Lado, 1961, p. 369).
Additionally, the speed with which cognitive processes are executed (e.g., lexical retrieval in speak-
ing) could be a focus of diagnostic testing. These diagnostic assessments can be scored objectively
without involving human raters, and responses can be collected independently of the language per-
formance. The assumption is that the test scores for components or subskills are predictive of the
overall performance because the constituent components are presumed to be causally related to the
overall performance. These tests for components or subskills (e.g., an oral productive vocabulary
test, a sentence production test, and a pronunciation test) are more fine-grained than (speaking) per-
formance assessments and therefore may have more diagnostic potential than general performance
assessments (see below for an approach to derive diagnostic information from a performance assess-
ment). In the development of discrete-point tests, it is essential to have a valid psycholinguistic model
outlining the cognitive steps language learners take to carry out language performance tasks. These
models can provide a framework for analyzing problems in language performance and creating test
specifications that target subskills for diagnostic discrete-point tests (e.g., word retrieval or word rec-
ognition). As a case in point, psycholinguistic models for speaking and writing, but also for reading
and listening, have (to varying degrees) informed various subskill testing (e.g., De Jong et al., 2012;
Van Gelderen et al., 2007; Vafaee & Suzuki, 2020; Van Moere, 2012).
The issue of granularity, however, is key. Individual diferences in very minute subskills or cog-
nitive processes may not always manifest in the overall language performance, or at all levels of
proficiency. For example, at higher levels of language proficiency, diferences in speed of word recog-
nition, an established psycholinguistic subprocess of reading, may no longer afect the general reading
comprehension of the language learner (Stanovich, 1991; Van Gelderen et al., 2007).
For authenticity and validity reasons, many language testers prefer performance assessments to
assess language learners’ (productive) language proficiency. However, holistic ratings of these perfor-
mances may ofer few leads for diagnosis. Analytic ratings in which trained raters provide subscores
for various aspects of writing or speech (e.g., fuency, pronunciation) are an attempt to provide more
granular information about language proficiency (cf. Weigle, 2002, and Fulcher, 2003, for examples
of rating scales for writing and speaking, respectively).
Second language acquisition (SLA) researchers, on the other hand, often replace human ratings by
analyses of complexity, accuracy, and fuency (Housen et al., 2012), which may be indicators of lan-
guage development. A large number of studies have used this triplet of variables to describe language
development or the efects of interventions. The more accurate a language performance is, for example,
the more advanced the language learner is presumed to be (cf. the study of speech errors, “Historical
perspectives”). A similar interpretation of fuency and complexity, however, is not as straightforward.
For example, fuency often refers to ease and smoothness of language use, but it is also operationalized
as speed of language production (see De Jong, 2018, for further discussion). The question then arises
whether faster is always better, or what the limit for speed is, given the learner’s personal (L1) speaking
style (De Jong, 2018; De Jong et al., 2015) and what a listener can process. This interpretation problem
is similar for complexity. One cannot claim that more complex written or spoken products necessarily
416
indicate higher proficiency. Thus, there may be optimal levels of fuency and complexity, which are
difcult to quantify, which undermines validity arguments for using complexity and fuency measures
in high-stakes assessments.
Other studies have shown that it is not easy to diagnose strong and weak performances in terms
of objective linguistic features, and thus to provide useful feedback for language teaching. Iwashita et
al. (2008) analyzed 200 transcribed English as a foreign language (EFL) speaking performances, rep-
resenting five levels of overall language proficiency. Over 20 lexical, grammatical, and phonetic-pho-
nological variables were evaluated for their discriminatory power of the proficiency levels. Although
a number of these variables showed an efect of level, the efect sizes were generally small to moder-
ate, and often adjacent proficiency levels could not be distinguished based on the linguistic features,
which makes these features less suited for diagnostics. With increased technological developments,
larger numbers of variables can be derived from written and spoken language performances automat-
ically (Kyle & Crossley, 2018; Van Moere & Downey, 2016), and major language testing organiza-
tions use these scored variables for automatic assessment of performances. These assessments function
reasonably well for the purposes for which they are used, such as admission decisions. However, one
may question to what extent information on these large numbers of features is helpful for feedback
in language learning or teaching (Chapelle et al., 2015; Van Moere & Downey, 2016).
Another reason why the use of linguistic features in assessment may be problematic is that linguis-
tic analyses work fairly well in unobtrusive research contexts when learners are unaware of the assess-
ment criteria. Once the analytic metrics are known, however, test takers may feel the need to use
infrequent words and/or complex morphological or syntactic structures at high speed (cf. the “had
I known” efect, Schoonen, 2011), which may raise test scores, but not necessarily refect improved
language proficiency.
In sum, to address the question of which features psycholinguists should measure as aspects of
language proficiency, three criteria seem to be applicable. Firstly, from a language learning perspec-
tive, it is important to establish that language learners show individual diferences in the execution
of subprocesses, either in quality or in speed. Secondly, individual diferences in the execution of the
subcomponent should be causally related to diferences in the overall language performance. If that
can be established, a third criterion is whether the subcomponent is trainable. If a specific language
processing component causes the language learner problems, targeted practice might help overcome
the hindrances and improve language performance. Some language learners will be more sensitive to
training than others, which in itself is an important attribute of the language learner.
Language learning potential and its components

Assessment of language learning potential can be considered a special strand of research, building
on insights of the psycholinguistics of (instructed) second language acquisition. Language learning
potential relates to the ability of a language learner to acquire new linguistic knowledge and skills in
the language learning process. MacWhinney (1994) classified foreign languages according to specific
features that may be new to the language learner and that add to the L2 learning task, ranging from
orthographic learning, phonological and phonetic learning, and lexical learning, to morphosyntactic
learning and syntactic processing and learning. The extent to which these learning tasks cause dif-
ficulties depends on the learners’ L1s, but also on the learners’ (basic) language learning skills, typi-
cally assessed with tests for basic cognitive skills such as paired-associate learning (for lexical learning)
and category inference (for morphosyntactic and syntactic learning). This research strongly relates to
language aptitude, which often is included in language learning studies.
The first modern aptitude studies of Carroll date back to the late 1950s. Basic but (presumed)
essential skills are operationalized in language aptitude tests, and the scores on these tests are consid-
ered to be predictive of a test taker’s language learning potential. Carroll’s modern language aptitude
417
Rob Schoonen
test (MLAT) focused on phonetic coding ability, grammatical sensitivity, rote-learning, and memory
of foreign language materials (Carroll, 1990; Carroll & Sapon, 1959). A more recent language apti-
tude test is Meara’s LLAMA test (Meara, 2005), which includes measures for vocabulary learning,
sound recognition, sound-symbol correspondence, and grammatical inferencing (LLAMA B, D, E,
and F, respectively). These measures suggest that language aptitude is multi-faceted (Li, 2016; Rog-
ers et al., 2017). Most of these measures tap into explicit language aptitude, which is particularly
relevant to instructed language learning. Li and DeKeyser (2021) make a clear distinction between
implicit and explicit language aptitude, both of which are multi-componential and predict diferent
aspects of SLA, such as learning rate and ultimate attainment. However, the validity of the measure-
ment of implicit language aptitude has yet to be established. The assessment of aptitude requires an
extensive test battery to diagnose language learners’ strengths and weaknesses in language learning
and should take into account the stage of the learners’ development (see Li & DeKeyser, 2021, for a
more extensive discussion).
Another way to assess language learning potential is through dynamic assessment (Poehner &
Infante, 2016). This approach builds on Vygotskian socio-cultural theory and the concept of zones
of proximal development. Dynamic assessment expands a “static” assessment of a learner’s knowledge
and skills with information about the learners’ ability to take the next step in their language develop-
ment. This potential can be assessed by ofering a test-taker help in accomplishing a difcult task.
The question then becomes how much help the learner needs, and whether the support leads to
learning and success in new tasks. Learning potential scores are established by comparing an actual
(unmediated) score with a mediated score that is achieved with support (Poehner & Lantolf, 2013).
Designing such dynamic assessments presumes explicit insight in the stages of development (i.e., the
zones of proximal development). In a successful dynamic assessment, it is key to develop steps for
a meaningful scafolding in complex task performance (Gonulal & Loewen, 2018). The amount of
controlled empirical research is still very limited, but Poehner and Lantolf (2013) provided an illustra-
tive example with a study of the efect of standardized computerized dynamic assessments of reading
and listening of French and Chinese as foreign language for English L1 students.

As was already indicated, psycholinguists have developed various measurement procedures and tools
to conduct research into cognitive language processing (see Jiang, this volume). A number of these
measurements and techniques have slowly spread into the field of language assessment. Constructs
from psycholinguistics have also emerged in language assessment, most notably processing speed (in
terms of reaction times, RT) (see Hamrick, 2023 [this volume], and Jiang, 2012, for an extensive
overview), but also the concept of automatization or automaticity (DeKeyser, 1996; Segalowitz,
2003, 2010; Suzuki, this volume). Automaticity, the seemingly efortless processing of language, is
seen as a high level of proficiency. Because automatic language processing does not require much
cognitive efort or deliberate decision making, it tends to coincide with stable processing. Segalowitz
operationalized processing stability in terms of the coefcient of variation (CV; Segalowitz & Sega-
lowitz, 1993), a measure that captures variability in repeated reaction times and that is now being
used in research contexts (e.g., Hulstijn et al., 2009; Hui & Godfroid, 2021; Yu et al., 2021).
Knowing a word
Lexical research is one of the areas where psycholinguistics has particularly infuenced language assess-
ment. The network model of the mental lexicon is a case in point as it underlies the development
of word association tasks and related measures, tapping into semantic, phonological, orthographic,
and collocational knowledge, and multi-faceted word relations. Aspects of lexical knowledge may be
418
easily accessible for advanced language learners and not accessible or much more difcult to access for
beginners. Read (1993, 2000) introduced the assessment of “deep” word knowledge via a structured,
multiple choice word association task for students. Schoonen and Verhallen (2008) adapted the for-
mat into word webs to test semantic word knowledge in primary school students. Fitzpatrick (Fitz-
patrick, 2003; Meara & Fitzpatrick, 2000) developed a productive counterpart for word associations
with the Lex30. These word association measures are intended to advance researchers’ understanding
of diferent facets of lexical knowledge. Fitzpatrick added a speed component to the access of word
associations (Fitzpatrick & Izura, 2011), as did Cremer and Schoonen (2013) to the assessment of
deep word knowledge in fifth graders. These reaction time measures distinguished between native
and non-native language users. The fifth graders’ lexical fuency was related to independent reading
comprehension scores. However, the assessment of automatic activation via priming scores did not
relate to children’s reading comprehension (Spätgens & Schoonen, 2019). Measures that discriminate
between language and age groups or experimental conditions may not be sensitive enough to be used
in individual diferences (correlational) studies, especially when the predictor variable is a derived
measure (a diference score), like a priming score (cf. Draheim et al., 2019).
Reaction time measures have also been used to explore and validate more straightforward lexi-
cal measures, such as the well-known and often used Yes/No vocabulary test. To establish a learner’s
potential overestimation of their own vocabulary knowledge in this self-report test, test developers
typically include a number of pseudo-words, which complicates scoring adjustment substantially.
Pellicer-Sánchez and Schmitt (2012) investigated whether response times in this Yes/No procedure
could reveal test-takers’ confidence in word recognition and thus contribute to valid score interpre-
tation. Although the RTs were helpful in some cases, they were not informative for all learners and
contexts.
Eyes paying attention

The work by Pellicer-Sánchez and Schmitt (2012) illustrates that psycholinguistic methods can be
useful in exploring the cognitive processes that lead to responses in tests. Following Borsboom et
al. (2004) and others, the cognitive processes involved in test taking should be similar to the ones
involved in actual target language use. A more recent development, compared to RT measurement,
is the application of eye-tracking in test validation. Eye-tracking is well-known from applications
in reading research (Just & Carpenter, 1980), and also in (first) language acquisition research with
visual world scenes (Trueswell & Gleitman, 2004; see Godfroid, 2020, for an overview). As a direct
measure of focus of attention, the technique has become a tool in analyzing test-takers’ behaviors
and cognitive processes.
In modern listening assessments, supporting visuals are used to increase authenticity and to invite
language learners to integrate visual and auditory information. In a study of the role of visuals in
academic EFL listening assessment, Suvorov (2015) asked language learners to watch six videoclips
of lectures. Some of the videos contained content information and others only provided contex-
tual information. Suvorov showed that test-takers looked at the content videos longer compared
to the context videos, suggesting that test-takers adjusted their attention when content informa-
tion was present. In a similar vein, Batty (2020) showed that in video-mediated L2 listening assess-
ment, test-takers paid attention to the face of the speaker far more than the gestures. In Suvorov’s
study, participants showed relatively large individual diferences on the psycholinguistic measures, yet
these diferences did not correlate with test scores. A recurring pattern is thus: Diferences between
(experimental) conditions can be reliably established with psycholinguistic measures, but individual
diferences are more difcult to establish. As a result, these psycholinguistic measures may not always
be suitable for correlational analyses with, for example, language proficiency measures (Draheim et
al., 2019).
419
Rob Schoonen
In a somewhat diferent context, Lee and Winke (2018) investigated young English language
learners’ (ELL) test performances on a computerized speaking test and their resulting test anxiety.
An analysis of eye gaze patterns showed that the young ELLs were more distracted by a timer in the
computerized test context than their native peers. This caused them to devote less time to visual cues
that were meant to help with their speaking performance.
Cognitive diagnostics
Cognitive diagnostic assessment (CDA) aims to infer information about cognitive skills from test or general
language performances. For example, in a CDA, researchers may try to relate (fine-grained) analytic scores
of written compositions or test-takers’ response patterns in a reading test to presumed underlying cogni-
tive components of language proficiency. CDA approaches are theory-driven in that experts analyze the
components or subskills that need to be mastered to pass the successive test items of a comprehension test,
or to score on diferent linguistic analytic criteria in language production. In a Q-matrix, the relationship
between the set of test items or criteria, and the presumed component skills is recorded (see Table 33.1).
Through cognitive diagnostic modeling of the test takers’ response patterns, their level of mastery
of the subskills is then estimated. The overall aim of CDA is to increase the diagnostic value of an
assessment in terms of the underlying cognitive skills (Jang, 2009; Lee & Sawaki, 2009b). This retro-
fitting of a test performance to more diagnostic subskills presumes that the item responses contain the
relevant diagnostic information and that the items cover all relevant subskills. This may not always be
the case, at least to the extent that is required for a useful analysis (Alderson et al., 2015; Jang, 2009).
Lee and Sawaki (2009a) applied this cognitive diagnostic approach to an integrated test of reading
to infer diagnostic information. Their analysis highlighted the importance of finding the right level
of granularity when designing a Q-matrix, given that for a viable diagnosis, the listed cognitive sub-
skills should each contribute to multiple, but not all test items. The results of the analysis consisted
of cognitive profiles of the language learners; for example, some were more successful with items
where Subskill 1 was involved and others with items that called upon Subskill 2, or other (combina-
tions of) subskills. The CDA showed that equal overall reading comprehension test scores can hide
diferent profiles of cognitive component skills, which may be highly relevant for planning teaching
interventions. Jang (2009) related test items to nine reading subskills (e.g., “Determine word meaning
out of context with recourse to background knowledge”), but despite the number of subskills, the
number of diferent profiles that emerged was rather limited (as in Lee & Sawaki, 2009a). In conclu-
sion, overall scores may indeed hide diferent subskill profiles, but the number of profiles and their
distinctiveness is still limited and depends very much on test design.
Table 33.1 An example Q-matrix for a seven-item test and three presumed subskills (after Lee & Sawaki, 2009b).
Item number Subskill 1: Understanding details Subskill 2: Connecting ideas Subskill 3: Inferencing
1 1 0 0
2 0 1 0
3 0 1 0
4 1 0 0
5 0 1 1
6 1 1 0
7 0 1 0
420
Xie (2017) applied cognitive diagnostic modeling to the L2 writings of 472 first-year under-
graduate students. Xie drew up a list of about 30 textual features, as counterparts of reading test
items, and related these features in a Q-matrix to five “component skills of writing”: content, orga-
nization, grammar, vocabulary, and mechanics. The Q-matrix again described which combination
of the five skills had to be mastered to score (high) on certain textual features. For example, an
essay that is scored as “concisely written and contains few redundant linguistic expressions” suggests
that the writer is good at organization, grammar, and vocabulary. The cognitive diagnostic analysis
of the writing scores led to various score profiles regarding the mastery of the five writing subskills.
Again, students with equal overall scores were shown to have diferent profiles and thus diferent
strengths and weaknesses (Xie, 2017).

Insights and techniques from psycholinguistic research permeate language assessments in applied lin-
guistic and language learning research, as well as research validating language assessments. Language
processing models are useful frameworks to identify cognitive processes and linguistic knowledge as
component skills of language use and thus are vital for diagnosis in language learning (see open ques-
tions and issues in Textbox 33.2). The explicit connection between processing and knowledge com-
ponents and language use, however, is not very frequently made, despite its importance (Alderson,
2005; Alderson et al., 2015; Lee, 2015; Van Moere, 2012). Conversely, diagnostic language testing
can and does make a valuable contribution to language learning research, but often lacks a psycho-
linguistic underpinning. Xie (2019), for example, developed a systematic inventory of linguistic
errors in EFL writing as a diagnostic instrument for writing instruction; while undoubtedly helpful
for teachers, the inventory lacks a psycholinguistic rationale other than a suggestion that the errors
refer to writers’ misrepresentations of the target language. The same applies to automatic evaluation
of speech and writing. Automated scores often are very successful in predicting general language
performance, but to what extent they are cognitively grounded and can contribute to useful feedback
remains to be confirmed.
Cognitive diagnostic assessment is just emerging in the field of language testing (Lee, 2015;
Lee & Sawaki, 2009b; Xie, 2017). Although this development is important, it is not without crit-
ics, because language performance assessments are usually not designed to cover and diagnose all
components skills. Discrete-point tests seem to be more appropriate for diagnosis of language pro-
ficiency or language learning potential, while psycholinguistic techniques and measures provide
many opportunities for in-depth analyses of language processing. In combination with appropriate
technology, online processing can be unraveled, and this can be done with timed task admin-
istration, controlled support, or feedback for dynamic assessment tasks (Mehri Kamrood et al.,
2019), or the registration of fuency in online response behavior via keystroke logging (Van Waes
& Leijten, 2015), eye-tracking, or response latency or stability. These measures are well estab-
lished in psycholinguistics and reliably display diferences between groups of language learners or
conditions of language learning. Developing these measures so they can reliably reveal individual
diferences in language proficiency still requires further investigation and validation. Research on
the appropriate and meaningful level of granularity will certainly help solve this problem, since
more fine-grained components tend to be measured more reliably, but also have less impact on the
overall performance. It is important for researchers and educators to be confident that individual
diferences or variability in the psycholinguistic component skills are causally related to the vari-
ability in language proficiency (cf. Borsboom et al., 2004), such that testing can support meaning-
ful teaching and learning.
421
Rob Schoonen
How far do you go? Psycholinguistic measurements allow for very detailed diagnosis of language perfor-
mance, but to what level of detail are these psycholinguistic measures predictive of the quality of the
general language performance?
Cause or efect? How do component scores relate to general language performance, that is, can we
assume that these processing and knowledge components relate causally to language performance?
Proficiency or behavior? To what extent do language measures indicate language proficiency or just typi-
cal behavior?
The more, the better? This motto does not apply to all (linguistic) features of written or spoken language
production, but then, what is optimal for the reader or listener?
Acknowledgements
The author wishes to thank the editors and an anonymous reviewer for their helpful comments.
Further reading
Bachman, L. F., & Palmer, A. S. (1996). Language testing in practice: Designing and developing useful language tests.
Bachman and Palmer (1996) is a classical source for the basic principles of test development that is still very
valuable and instructive. The book also provides informative examples.
De Jong, N. H. (2018). Fluency in second language testing: Insights from diferent disciplines. Language Assess-
ment Quarterly, 15(3), 237–254.
De Jong’s paper provides an overview of the construct of fuency from various perspectives and how these
perspectives are relevant to language assessment.
Ercikan, K., & Pellegrino, J. W. (Eds.). (2017). Validation of score meaning for the next generation of assessments: The
use of response processes. Taylor & Francis.
This is an interesting collection of papers that focus on cognitive processes in test-taking, as are measured
by modern data collection procedures such as logging and eye-tracking, in order to validate score meaning.
Winke, P., & Brunfaut, T. (Eds.). (2021). The Routledge handbook of second language acquisition and language testing.
Routledge.
Language assessment is a growing field that is interconnected with other domains of applied linguistics, such
as second language acquisition, as is evidenced by a number of recent handbooks on language assessment
and these related domains, including Winke and Brunfaut’s handbook as one of the most recent examples.
References
Alderson, J. C. (2005). Diagnosing foreign language profciency: The interface between learning and assessment.
Continuum.
Alderson, J. C., Brunfaut, T., & Harding, L. (2015). Towards a theory of diagnosis in second and foreign lan-
guage assessment: Insights from professional practice across diverse fields. Applied Linguistics, 36(2), 236–260.
Alderson, J. C., Brunfaut, T., & Harding, L. (2017). Bridging assessment and learning: A view from second and
foreign language assessment. Assessment in Education: Principles, Policy & Practice, 24(3), 379–387.
Alderson, J. C., & Huhta, A. (2011). Can research into the diagnostic testing of reading in a second or foreign
language contribute to SLA research? EUROSLA Yearbook, 11(1), 30–52.
Bachman, L. F., & Palmer, A. S. (1996). Language testing in practice: Designing and developing useful language tests.
422
Batty, A. O. (2020). An eye-tracking study of attention to visual cues in L2 listening tests. Language Testing, 38(4),
511–535.
Borsboom, D., Mellenbergh, G. J., & van Heerden, J. (2004). The concept of validity. Psychological Review,
111(4), 1061–1071.
Carroll, J. B. (1990). Cognitive abilities in foreign language aptitude: Then and now. In T. Parry & C. Stansfield
(Eds.), Language aptitude reconsidered (pp. 11–29). Prentice Hall.
Carroll, J. B., & Sapon, S. M. (1959). Modern language aptitude test. Psychological Corporation.
Chapelle, C. A. (2021). Validity in language assessment. In P. Winke & T. Brunfaut (Eds.), The Routledge hand-
book of SLA and language testing (pp. 11–20). Routledge.
Chapelle, C. A., Cotos, E., & Lee, J. (2015). Validity arguments for diagnostic assessment using automated writ-
ing evaluation. Language Testing, 32(3), 385–405.
Cremer, M., & Schoonen, R. (2013). The role of accessibility of semantic word knowledge in monolingual and
bilingual fifth-grade reading. Applied Psycholinguistics, 34(6), 1195–1217.
Cronbach, L. J. (1957). The two disciplines of scientific psychology. American Psychologist, 12(11), 671–684.
Davies, A. (1990). Principles of language testing. Wiley-Blackwell.
De Jong, N. H. (2018). Fluency in second language testing: Insights from diferent disciplines. Language Assess-
ment Quarterly, 15(3), 237–254.
De Jong, N. H., Groenhout, R., Schoonen, R., & Hulstijn, J. H. (2015). L2 fuency: Speaking style or profi-
ciency? Correcting measures of L2 fuency for L1 behavior. Applied Psycholinguistics, 36(2), 223–243.
De Jong, N. H., Steinel, M. P., Florijn, A., Schoonen, R., & Hulstijn, J. H. (2012). Facets of speaking profi-
ciency. Studies in Second Language Acquisition, 34(1), 5–34.
DeKeyser, R. M. (1996). Exploring automatization processes. TESOL Quarterly, 30(2), 349–357.
Donders, F. C. (1865). Over de snelheid der gedachte en der wilsbepaling: Voorloopige mededeeling [On the
speed of thought and decision making: Preliminary communication]. Nederlandsch Archief voor Genees- en
Natuurkunde, 1, 518–521.
Draheim, C., Mashburn, C. A., Martin, J. D., & Engle, R. W. (2019). Reaction time in diferential and devel-
opmental research: A review and commentary on the problems and alternatives. Psychological Bulletin, 145(5),
508–535.
Fitzpatrick, T. (2003). Eliciting and measuring productive vocabulary using word association techniques and frequency bands
[Unpublished doctoral dissertation]. Swansea University.
Fitzpatrick, T., & Izura, C. (2011). Word association in L1 and L2: An exploratory study of response types,
response times, and interlingual mediation. Studies in Second Language Acquisition, 33(3), 373–398.
Fulcher, G. (2003). Testing second language speaking. Pearson Education.
guide. Routledge.
Gonulal, T., & Loewen, S. (2018). Scafolding technique. In J. I. Liontas, TESOL International Association, &
M. DelliCarpini (Eds.), The TESOL encyclopedia of English language teaching. John Wiley & Sons.
Routledge.
Hamp-Lyons, L. (2016). Purposes of assessment. In D. Tsagari & J. Banerjee (Eds.), Handbook of language assess-
ment (pp. 13–28). De Gruyter Mouton.
Housen, A., Kuiken, F., & Vedder, I. (2012). Complexity, accuracy and fuency. In A. Housen, F. Kuiken, & I.
Vedder (Eds.), Dimensions of L2 performance and profciency: Complexity, accuracy and fuency in SLA (pp. 1–20).
Benjamins.
Hui, B., & Godfroid, A. (2021). Testing the role of processing speed and automaticity in second language listen-
ing. Applied Psycholinguistics, 42(5), 1089–1115.
Hulstijn, J. H. (2015). Language profciency in native and non-native speakers: Theory and research (Vol. 41). Benjamins.
Hulstijn, J. H., Van Gelderen, A., & Schoonen, R. (2009). Automatization in second language acquisition: What
does the coefcient of variation tell us? Applied Psycholinguistics, 30(4), 555–582.
Iwashita, N., Brown, A., McNamara, T., & O’Hagan, S. (2008). Assessed levels of second language speaking
proficiency: How distinct? Applied Linguistics, 29(1), 24–49.
Jang, E. E. (2009). Cognitive diagnostic assessment of L2 reading comprehension ability: Validity arguments for
Fusion Model application to LanguEdge assessment. Language Testing, 26(1), 31–73.
Just, M. A., & Carpenter, P. A. (1980). A theory of reading: From eye fixations to comprehension. Psychological
Review, 87(4), 329–354.
Kidd, E., Donnelly, S., & Christiansen, M. H. (2018). Individual diferences in language acquisition and process-
ing. Trends in Cognitive Sciences, 22(2), 154–169.
423
Rob Schoonen
Kyle, K., & Crossley, S. A. (2018). Measuring syntactic complexity in L2 writing using fine-grained clausal and
phrasal indices. The Modern Language Journal, 102(2), 333–349.
Lado, R. (1961). Language testing: The construction and use of foreign language tests. A teacher’s book. Longman.
Lee, S., & Winke, P. (2018). Young learners’ response processes when taking computerized tasks for speaking
assessment. Language Testing, 35(2), 239–269.
Lee, Y.-W. (2015). Diagnosing diagnostic language assessment. Language Testing, 32(3), 299–316.
Lee, Y.-W., & Sawaki, Y. (2009a). Application of three cognitive diagnosis models to ESL reading and listening
assessments. Language Assessment Quarterly, 6(3), 239–263.
Lee, Y.-W., & Sawaki, Y. (2009b). Cognitive diagnosis approaches to language assessment: An overview. Language
Assessment Quarterly, 6(3), 172–189.
Levelt, W. J. M. (2013). A history of psycholinguistics: The pre-Chomskyan era. Oxford University Press.
Li, S. (2016). The construct validity of language aptitude: A meta-analysis. Studies in Second Language Acquisition,
38(4), 801–842.
Li, S., & DeKeyser, R. (2021). Implicit language aptitude: Conceptualizing the construct, validating the mea-
sures, and examining the evidence. Introduction to the special issue. Studies in Second Language Acquisition,
43(3), 473–497.
MacWhinney, B. (1994). Psycholinguistic issues in the assessment of the sub-components of language abilities. In
J. Thornton (Ed.), Language aptitude invitational symposium program proceedings. ERIC ED 461 264.
Markus, K. A., & Borsboom, D. (2013). Frontiers of test validity theory: Measurement, causation, and meaning.
Routledge.
McNamara, T. F. (1996). Measuring second language performance. Longman.
Meara, P. M. (2005). Llama language aptitude tests. Lognostics.
Meara, P. M., & Fitzpatrick, T. (2000). Lex30: An improved method of assessing productive vocabulary in an
L2. System, 28(1), 19–30.
Mehri Kamrood, A., Davoudi, M., Ghaniabadi, S., & Amirian, S. M. R. (2019). Diagnosing L2 learners’ develop-
ment through online computerized dynamic assessment. Computer Assisted Language Learning, 34(7), 868–897.
Meringer, R., & Mayer, C. (1895). Versprechen und Verlesen: Eine psychologisch-linguistische Studie [To mis-speak and
to misread: A psychological-linguistic study]. Göschen.
Messick, S. (1989). Validity. In R. L. Linn (Ed.), Educational measurement (3rd ed, pp. 13–103). American Council
on Education/Macmillan Pub. Co.
Oller, J. W., Jr., (1983). Evidence for a general language proficiency factor: An expectancy grammar. In Oller, J.
W., Jr., (Ed.), Issues in Language Testing Research (pp. 3–10). Newbury House.
Pellicer-Sánchez, A., & Schmitt, N. (2012). Scoring Yes—No vocabulary tests: Reaction time vs. nonword
approaches. Language Testing, 29(4), 489–509.
Poehner, M. E., & Infante, P. (2016). Dynamic assessment in the language classroom. In D. Tsagari & J. Banerjee
(Eds.), Handbook of Language Assessment (pp. 275–290). De Gruyter Mouton.
Poehner, M. E., & Lantolf, J. P. (2013). Bringing the ZPD into the equation: Capturing L2 development during
Computerized Dynamic Assessment (C-DA). Language Teaching Research, 17(3), 323–342.
Read, J. (1993). The development of a new measure of L2 vocabulary knowledge. Language Testing, 10(3),
355–371.
Read, J. (2000). Assessing vocabulary. Cambridge University Press.
Richards, B., & Malvern, D. (2007). Validity and threats to the validity of vocabulary measurement. In H. Daller,
J. Milton, & J. Trefers-Daller (Eds.), Modelling and Assessing Vocabulary Knowledge (pp. 79–92). Cambridge
University Press.
Rogers, V., Meara, P., Barnett-Legh, T., Curry, C., & Davie, E. (2017). Examining the LLAMA aptitude tests.
Journal of the European Second Language Association, 1(1), 49–60.
Schoonen, R. (2011). How language ability is assessed. In E. Hinkel (Ed.), Handbook of research in second language
teaching and learning (pp. 701–716). Routledge.
Schoonen, R., & Verhallen, M. (2008). The assessment of deep word knowledge in young first and second
language learners. Language Testing, 25(2), 211–236.
Segalowitz, N. S. (2003). Automaticity and second languages. In C. J. Doughty & M. H. Long (Eds.), The hand-
book of second language acquisition (pp. 382–408). Blackwell.
Segalowitz, N. S. (2010). Cognitive bases of second language fuency. Routledge.
Segalowitz, N. S., & Segalowitz, S. J. (1993). Skilled performance, practice, and the diferentiation of speed-
up from automatization efects: Evidence from second language word recognition. Applied Psycholinguistics,
14(3), 369–385.
Spätgens, T., & Schoonen, R. (2019). Individual diferences in reading comprehension in monolingual and
bilingual children: The infuence of semantic priming during sentence reading. Learning and Individual Dif-
ferences, 76, 101777.
424
Spolsky, B. (1995). Measured words: The development of objective language testing. Oxford University Press.
Stanovich, K. E. (1991). Word recognition: Changing perspectives. In Handbook of reading research, Vol. 2 (pp.
418–452). Lawrence Erlbaum Associates, Inc.
Suvorov, R. (2015). The use of eye tracking in research on video-based second language (L2) listening assess-
ment: A comparison of context videos and content videos. Language Testing, 32(4), 463–483.
Trueswell, J., & Gleitman, L. (2004). Children’s eye movements during listening: Developmental evidence for
a constraint-based theory of sentence processing. In J. M. Henderson & F. Ferreira (Eds.), The interface of
language, vision, and action: Eye movements and the visual world (pp. 319–346). Psychology Press.
Vafaee, P., & Suzuki, Y. (2020). The relative significance of syntactic knowledge and vocabulary knowledge in
second language listening ability. Studies in Second Language Acquisition, 42, 383–410.
Van Gelderen, A., Schoonen, R., Stoel, R. D., De Glopper, K., & Hulstijn, J. (2007). Development of adoles-
cent reading comprehension in language 1 and language 2: A longitudinal analysis of constituent compo-
nents. Journal of Educational Psychology, 99(3), 477–491.
Van Moere, A. (2012). A psycholinguistic approach to oral language assessment. Language Testing, 29(3), 325–344.
Van Moere, A., & Downey, R. (2016). Technology and artificial intelligence in language assessment. In D.
Tsagari & J. Banerjee (Eds.), Handbook of second language assessment (pp. 342–357). De Gruyter Mouton.
Van Waes, L., & Leijten, M. (2015). Fluency in writing: A multidimensional perspective on writing fuency
applied to L1 and L2. Computers and Composition, 38(A), 79–95.
Weigle, S. C. (2002). Assessing writing. Cambridge University Press.
Xie, Q. (2017). Diagnosing university students’ academic writing in English: Is cognitive diagnostic modelling
the way forward? Educational Psychology, 37(1), 26–47.
Xie, Q. (2019). Error analysis and diagnosis of ESL linguistic accuracy: Construct specification and empirical
validation. Assessing Writing, 41, 47–62.
Yu, X., Janse, E., & Schoonen, R. (2021). The efect of learning context on L2 listening development: Knowl-
edge and processing. Studies in Second Language Acquisition, 43, 329–354.
425
34
COGNITIVE EFFECTS OF
BILINGUALISM
Gregory J. Poarch
Introduction
In recent years, there has been heightened interest as to whether navigating several languages in daily
life has an efect on cognitive development and cognition. Such a question arises from the cognitive
efort involved in juggling two or more languages in the mind (Kroll et al., 2015). While there was
some initial research in this field in the early 1900s (see “Historical perspectives”, which follows),
the notion that bilingualism afects cognition has been under closer scrutiny since the 1990s (Kroll &
Bialystok, 2013), stemming also from psycholinguistic findings that language processing in bilinguals
difers substantially from that of monolinguals. For example, bilinguals show slower lexical processing
and retrieval during picture naming than monolinguals (for children: Poarch & Van Hell, 2012; for
adults: Sullivan et al., 2018). Such findings have been attributed to both languages in bilinguals always
being active, irrespective of which language is in use (Thierry & Wu, 2007). Hence, when word
candidates in one language are processed and retrieved, there is likely competition from not only
similar word candidates in the same language (within-language competition) but also from translation
equivalents in the other language(s) (across-language competition). Controlling this competition and
interference from the language not-in-use requires a cognitive mechanism that inhibits one or several
word candidates in favor of the target word for successful retrieval (Green, 1998). It is hypothesized
that this process of controlling language co-activation is subserved by a mechanism that humans also
rely on when performing non-verbal cognitive processing, the so-called executive function, and
particularly one of its sub-components, inhibitory control (Bialystok, 2017, Green, 1998). Inhibition
is also implicated as the cognitive mechanism drawn on during bilingual language switching, thus
when switching between a dominant language A into a non-dominant language B. The inhibition
exerted to switch from language A to language B persists when switching back from language B to
language A, thus yielding asymmetrical switching costs that are larger in the switch back to the domi-
nant language than in the switch to the non-dominant language (Meuter & Allport, 1999; Philipp
& Koch, 2009). Bilinguals thus incur substantial cognitive control training by repeatedly drawing on
cognitive resources during language processing. Critically, the increased and regular cognitive efort
needed to control several languages may thus also spill-over to non-language cognitive processing
and thus have an efect on domain-general cognition (Adesope et al., 2010; Bialystok, 2017). Recent
supporting evidence for such common underlying inhibition during language production and non-
verbal task performance comes from research on language and task switching in trilinguals (Declerck
et al., 2020).
426 DOI: 10.4324/9781003018872-39

Cognitive effects of bilingualism
Bilingualism: In the psychological literature, as opposed to that used in second language acquisition (SLA)
and applied linguistics, bilingualism is defined as the knowledge and use of more than one language
or dialect by an individual in everyday life (Grosjean, 2013). However, not all bilinguals are the same:
They will difer to various degrees based on their experiences with their languages, at which age they
began using each language, the contexts in which the languages were learned, how proficient they
are in each of their languages, how often the languages are currently used (Surrain & Luk, 2019), and
in which interactional context (Gullifer & Titone, 2020). Hence, the term bilingualism cannot be
considered a homogeneous construct because it characterizes a very broad spectrum of dual-language
use. Finally, bilingualism should not be viewed as a categorical variable given the difculty in clearly
delineating bilingual from monolingual experiences (Luk & Bialystok, 2013).
Executive function: Executive function (EF) is an umbrella term for cognitive processes situated in the
pre-frontal cortex of the human brain. EF is generally described as including attention, updating,
shifting, and inhibition (Miyake & Friedman, 2012) or, more elaborately, selective attention, task
shifting and updating, inhibitory control and response inhibition, and working memory (Botvinick
et al., 2001; Diamond, 2013). Because our cognitive system needs to regularly and repeatedly choose
between several alternative and competing responses, it relies on a monitoring mechanism that detects
confict and subsequently attempts to resolve it (Keye et al., 2009). EF has also been described as the
“adaptive goal-directed behaviors that enable individuals to override more automatic or established
thoughts and responses” (Garon et al., 2008, p. 31). The EF components develop throughout child-
hood but show somewhat varied developmental trajectories, with inhibitory control approaching
maturity before or during early adolescence (Best & Miller, 2010), while task shifting and updat-
ing continue maturing through adolescence and early adulthood (Huizinga et al., 2006); EF then
decreases in older adults (Wilson et al., 2002).
Inhibitory control: Inhibitory control (IC) is considered to be a sub-component of executive function
and it is drawn on for cognitive, behavioral, and emotional inhibition of dominant or so-called prepo-
tent responses in favor of alternative responses that are more appropriate in a given situation (Miyake
& Friedman, 2012; Munakata et al., 2011).
Cognitive reserve: Following Stern (2002), cognitive reserve can be described as the individual capacity to
tolerate age-related changes in the brain and the disjoint between the degree of pathological changes
and the clinical outcomes such as dementia. It thus has implications for healthy aging in that some
individuals are better able to cope with brain damage than others by recruiting alternative brain net-
works in a compensatory manner. There are several life experiences that have been found to modu-
late cognitive reserve such as level of education, active social networks, and intellectually stimulating
activities (Del Maschio et al., 2018).
The effects of bilingualism on cognition

Early research on the efects of bilingualism on cognition, conducted at the end of the nineteenth
and in the first half of the twentieth century, predominantly conceptualized bilingualism as a problem
and a debilitating factor on intelligence. Saer (1923), for example, reported that monoglot children
displayed considerably superior performance over bilingual children on an intelligence test and con-
cluded that mental confusion existed in bilingual children more so than in monolingual children in
427
Gregory J. Poarch
tests for dextrality and rhythm. Similarly, but formulated more severely, Goodenough (1926) stated
that “the use of a foreign language in the house is one of the chief factors in producing mental retar-
dation” (p. 393). Darcy (1953) identified two main confounds in this early body of research: 1) Often
information was missing on the socio-economic backgrounds of the tested participants and 2) “there
was only a very general and subjective determination of bilingualism” (p. 32). Hence, any significant
diferences found between groups of monolinguals and bilinguals on, for example, an intelligence test
could just as well be attributed to other factors than the number of languages spoken by the individu-
als. Note that such confounds were alluded to even at that time (Mitchell, 1937). Hence, drawing
conclusions on the intelligence of bilingual children solely from verbal intelligence tests, which may
disadvantage bilingual children particularly if administered in the weaker of their two languages,
should be done with caution (Pintner, 1923).
However, these voices remained unheard until the second half of the twentieth century, when Peal
and Lambert (1962) studied bilingual children in Canada, and an alternative view on bilingualism and
its cognitive efects began to emerge. Peal and Lambert found that balanced French-English bilingual
children outperformed French monolingual children on 15 of 18 variables measuring intelligence.
Critically, unlike in much of the previous research, the groups had been matched on socio-economic
background. The authors concluded that experience with two language systems seems to have left
bilingual children “with a mental fexibility, a superiority in concept formation, a more diversified
set of mental abilities” (p. 20). Starting from this seminal study that shed a new light on the efects of
bilingualism on cognitive development, a stream of research arose that further explored the impact of
everyday use of multiple languages on cognition (Bialystok, 2009). In particular, studies from the late
1980s onwards compared bilingual and monolingual children in non-verbal selective attention and
metalinguistic development (Bialystok, 1988; Galambos & Hakuta, 1988), yielding further results
showing that bilinguals outperformed monolinguals (pioneered by Bialystok & Majumder, 1998).
As Poarch and Krott (2019) point out, this turn from mostly detrimental cognitive consequences of
bilingualism towards a more positive view may also be traced to the prevalent language policy at the
time (see also Hakuta, 1986).

The view that bilingualism afects cognition rests on the assumption that a key aspect of the bilingual
experience is language processing under conditions of constant language co-activation (Shook &
Marian, 2019; Thierry & Wu, 2007). Consequently, for any activity in which language is relevant,
bilinguals need to draw on a cognitive control mechanism that enables selecting the appropriate lan-
guage by resolving the between-language competition. The assumed resolution mechanism involves
inhibiting the irrelevant language, for which the executive function system (EF) is a likely candidate
(see Textbox 34.1). The development of EF has been found to be modulated by such diverse factors
as socio-economic status (SES; Noble et al., 2007), circadian rhythm (Hahn et al., 2012), diet (Kim
& Wang, 2017), musical expertise (Zuk et al., 2014), physical activity (Best, 2010), and culture (Tran
et al., 2019).
A hypothesis linking the regular use of multiple languages to the development of EF was put
forth by Green (1998) in his inhibitory control (IC) model (see also Bak & Robertson, 2017). The
IC model posits the cognitive mechanism underlying the efects of bilingualism mainly in the EF
sub-component inhibitory control since this is drawn on to resolve competition between languages
during language processing. As lexical representations of both languages become activated, a cogni-
tive advisory system assesses which language is relevant for a given situation, after which inhibition
is applied to suppress the non-intended language. The stronger and more dominant the language
(usually the L1), the stronger the required inhibitory processes. Furthermore, given that not all
bilinguals use their languages in identical contexts in daily life, Green and Abutalebi (2013) proposed
428
the adaptive control hypothesis, which distinguishes between three language switching contexts in
daily life: (1) single language interactional contexts in which only one language is used in each con-
text (e.g., work vs. home); (2) dual language contexts in which two languages are used in the same
context but not with the same interlocutor, which poses higher inhibitory demands given that the
correct language needs to be chosen for each interlocutor, and (3) dense code-switching contexts in
which both languages are used freely and interchangeably with interlocutors and inhibition may play
less of a role. Depending on the predominant language switching contexts that bilingual individuals
find themselves in, they may also accrue varying levels of language control training.
There is ample evidence that speakers of multiple languages rely on the EF system during lan-
guage use (Filippi et al., 2015) and when switching from one language to another (Anderson et al.,
2018). The network in the brain subserving EF is thus utilized repeatedly to successfully regulate
cross-language competition and choose the appropriate language in any given context (Marian et al.,
2017). Such linguistic domain-specific extensive training of EF through bilingual language use may
over time make processing in this neural network more efcient (Calabria et al., 2018), which may
also have a domain-general efect on the performance on cognitive tasks that are non-verbal (Green
& Abutalebi, 2013, but see Branzi et al., 2016).
Several experimental approaches have attempted to explore possible diferences between mono-
lingual and bilingual populations in the field of bilingualism and EF, some of which use behavioral,
others neurophysiological measures. Some of the behavioral experimental tasks that have been cre-
ated to tap into EF and non-verbal cognitive processing are the color-shape switching task (Zelazo,
2006), the Stroop color-word task (Stroop, 1935), the Simon task (Simon & Rudell, 1967), the
Flanker task (Eriksen & Eriksen, 1974), and a more elaborate version of the Flanker task, the atten-
tion network task (ANT; Fan et al., 2002), to name just a few. All of these tasks, in one way or
another, require selective attention in switching between rules, updating information, and inhibiting
irrelevant information. In a variant of the color-shape switching task, the dimensional change card
sort (DCCS; Zelazo, 2006), participants are required to first use one rule, namely sorting cards by
one dimension (color), and later suppress the first rule to apply a second rule, sorting cards by another
dimension (shape). The Simon and the Flanker tasks, presented here in more detail (see Figure 34.1),
have been and are presently still used pervasively in the research field. Both are assumed to induce
cognitive confict and require the subsequent engagement of inhibitory control. While the tasks have
often been used interchangeably in research on EF, they induce confict in diferent ways (see sec-
tion “Critical issues”): In the Simon task, colored squares are presented on a computer screen and,
through the absence or presence of a spatial stimulus-response mismatch, they induce cognitive con-
fict on incongruent vs. congruent trials. In the Flanker task, in contrast, strings of arrows are either
incongruent or congruent (see Figure 34.1A), using fanking distractor arrows to the left and right
of a middle arrow to induce interference that needs to be resolved.
In both tasks, overall reaction times for congruent and incongruent conditions represent the speed
of how quickly participants monitor and selectively attend to changing trial types. Additionally, an
inhibitory control score can be calculated using the diference score between the two conditions,
the magnitude of which is an indicator of how strongly individuals are infuenced by the mismatch
(Simon task) or distractors (Flanker task) in incongruent vs. congruent conditions. Smaller magni-
tudes index better interference control (Keye et al., 2009).

While there is ample evidence documenting cognitive advantages for bilinguals over monolinguals in
classical EF tasks such as the Simon and the Flanker task, not all studies have documented such dif-
ferences (for a review, see Van den Noort et al., 2019). This section ofers a non-exhaustive overview
of the more recent research pertinent to the topic of cognitive efects of bilingualism, starting with
429
Gregory J. Poarch
Figure 34.1 Flanker task (panel A) and Simon task (panel B) congruent and incongruent conditions. Stimulus
displayed on screen requires indicated button press.
430
behavioral research with children and moving via young adults to older adults, thus covering the life
span. Most of the studies presented in this section have compared groups of monolinguals and bilin-
guals. Thus, participant groups will be described in more detail only if there were deviations from
this grouping pattern. The section ends with a brief overview of recent neurophysiological research
(for extensive reviews, see Antoniou, 2019; Bialystok, 2017).
To begin with research on children, De Cat et al. (2018) tested five- to seven-year-olds in the UK
and, after controlling for such critical background variables as age, socioeconomic status, and sex,
found that bilinguals outperformed monolinguals in the Simon task. Similarly, using a battery of EF
tasks, Sorge and colleagues (2017) tested eight- to eleven-year-old children in Canada and reported
bilingualism to be the best predictor for Flanker task performance (see also Chung-Fat-Yim et al.,
2020). Furthermore, in longitudinal research spanning the course of 12 months, Tran and colleagues
(2019) tested three-year-old monolinguals and bilinguals from Argentina, Vietnam, and the USA and
found that bilingualism modulated children’s performance on a color-shape switching task. Corrobo-
rating longitudinal evidence comes from Park et al. (2018) who used the Flanker task and a color-
shape switching task to test eight- to twelve-year-old monolingual and bilingual children in the USA
over the same time span. Their findings showed a strong improvement of inhibitory control only
in the bilingual children, but not in the monolingual children. For task switching, however, group
diferences did not emerge. The authors thus propose that bilingualism modulates EF component
development diferentially at various ages (see also Textbox 34.1), which in turn may only yield dif-
ferences between bilinguals and monolinguals in performing specific EF tasks at specific time points
during development. Support for this interpretation comes from a study performed in Germany by
Poarch (2018), who reported inhibitory control diferences between monolingual second language
(L2) learners and bilingual L3 learners aged 5–13 in the Flanker task, but no such diferences in
the Simon task. Finally, Simonis et al. (2020) compared 10-year-olds and 16-year-olds enrolled in
bilingual immersion programs to age-matched, non-immersed children in the French-speaking part
of Belgium. The children were administered the Simon, ANT, and DCCS tasks, yet no significant
diferences were obtained between the groups at any age level.
Turning to recent research with young adults, there is a somewhat more mixed picture. For
example, no diferences were found between groups of UK-based 18- to 30-year-olds in Simon task
performance (Naeem et al., 2018). The diferences that were found between participants were not
modulated by bilingualism but by socioeconomic status (SES) instead. While Grundy and colleagues
(2017) found similar Flanker task performances across groups of young adults, they did report smaller
sequential congruency efects for bilinguals, which was interpreted as bilinguals commanding more
efcient attentional disengagement from preceding trials compared to monolinguals. However, this
interpretation was challenged by Goldsmith and Morton (2018), who argued that smaller sequential
congruency efects represented a disadvantage in terms of learning and memory. Finally, Desideri
and Bonifacci (2018) reported advantages for bilinguals over monolinguals on ANT performance as
did Xie (2018) for high-proficient compared to low-proficiency L2 learners on Flanker task per-
formance. Note that an absence of diferences between groups of young adults in EF tasks may also
stem from the fact that in early adulthood cognitive processing capacity reaches its highest efciency,
in turn inducing ceiling performances on EF tasks (Antoniou, 2019). Inadvertently, performance
variability is limited and may obscure possible systematic diferences across groups in this age range.
In research with older adults, there is more consistent evidence that bilinguals outperform mono-
linguals in EF task performance (Goral et al., 2015; for review, see Bialystok et al., 2016). In this line
of research, there has been particular interest in the neuroprotective potential of bilingualism, specifi-
cally testing if the lifelong exercise of language control afects the brain networks in the prefrontal
cortices that show vulnerability to the aging process (Bialystok & Sullivan, 2017). Research findings
indicate that bilingualism delays the onset of dementia (Alladi et al., 2017; Woumans et al., 2015) and
mild cognitive impairment (Wilson et al., 2015) and that bilingualism leads to less severe cognitive
431
Gregory J. Poarch
outcomes after stroke (Alladi et al., 2016; Paplikar et al., 2019). There is thus converging evidence that
bilingualism contributes to cognitive reserve (see Textbox 34.1), which in turn postpones the impact
of neurodegenerative disease in older adults (though see Del Maschio et al., 2018, for critical review).
As the research particularly on older adults indicates, the bilingual experience has also been shown
to afect brain structure, function, and connectivity (DeLuca et al., 2019), which is particularly rel-
evant when exploring the efects of bilingualism on cognition since significant functional overlap has
been documented for brain regions subserving language control and domain-general EF (Abutalebi
& Green, 2016; Sulzipio et al., 2020). While the neuroscientific findings have also been somewhat
mixed in terms of positive efects of bilingualism vs. null results, in his review, Pliatsikas (2019) mod-
els variability between studies and individuals as dependent on the quantity and quality of bilingual
exposure and use. Critically, EF task performance has also been shown to correlate with structural
changes in the brain, observed in functional magnetic resonance imaging (fMRI; Olsen et al., 2015)
and functional diferences measured in electroencephalography (EEG; Grundy et al., 2017), suggest-
ing diferences between monolingual and bilingual brains when performing EF tasks.
In summary, the research reviewed above indicates converging evidence for cognitive diferences
between monolinguals and bilinguals in classical EF tasks such as the Simon and the Flanker task.
While not all studies document such diferences, when they do appear, they are in favor of bilinguals
(Grundy, 2020; Van den Noort et al., 2019).

The number of studies exploring the cognitive aspects of bilingualism has been exponentially grow-
ing since the turn of the century. While earlier work reported consistent advantages for bilinguals
over monolinguals in non-verbal processing tasks, in the past ten years, several studies have also failed
to report any significant between-group diferences (Paap & Greenberg, 2013). Hence, the notion
that bilingual language co-activation and the subsequently required and repetitive cognitive control
demands can have a beneficial efect on the development of EF has become highly disputed (cf. Bak,
2016). There is a vast research body attesting to consistent cognitive control advantages of bilinguals
over monolinguals in performing cognitive tasks for children (Blom et al., 2017; De Cat et al., 2018;
Poarch, 2018; Poarch & Bialystok, 2015), young adults (Blumenfeld & Marian, 2014), and older
adults (Bialystok et al., 2014). At the same time, a number of studies have reported no across-group
diferences, particularly in children (Antón et al., 2014; Duñabeitia et al., 2014). Thus, an ongoing
debate ensued concerning the viability and robustness of the reported cognitive efects of bilingual-
ism. The debate centers on whether various confounding factors that are germane to experimental
research (such as, e.g., SES and relative language proficiencies) have been taken into account suf-
ciently when interpreting research results. Furthermore, studies that show no between-group difer-
ences “do not invalidate all prior work indicating such diferences” (Antoniou, 2019, p. 408; see also
Grundy, 2020). At the same time, studies reporting null results emphasize the necessity for replication
research to unambiguously and consistently define and follow research procedures and design (see
also Poarch & Krott, 2019). Ultimately, given that not all bilinguals are the same in terms of individ-
ual language experience and exposure (De Bruin, 2019; Surrain & Luk, 2019), “advantages found for
bilinguals are unlikely to extend to all bilinguals under all circumstances” (Antoniou, 2019, p. 408).
There have also been concerns in the research community concerning the EF tasks used and the
manner in which collected data is processed further. Not only do such experimental tasks, admin-
istered in the lab, lack ecological validity, they also show inconsistent convergent validity (Poarch,
2018; Valian, 2015), elicit confict diferently (Poarch & Van Hell, 2019), and induce varying cogni-
tive loads (Qu et al., 2016). Hence, researchers should not use available EF tasks indiscriminately
and instead strive to create tasks with greater ecological validity. Furthermore, Poarch and Krott
(2019) note that studies are not always comparable since diferent choices in designing experiments,
432
selecting stimuli, as well as in testing procedures, and in processing and analyzing data may inevitably
have added variability to past research and thus contributed to the mixed results. Hence, there is a
need for implementing EF tasks in a (more) standardized fashion, handling and pruning data using
consistent outlier identification and exclusion criteria (De Cat et al., 2018; Zhou & Krott, 2016), and
choosing appropriate statistical analyses (Poarch et al., 2019; Zhou & Krott, 2018). More generally,
the authors also suggested running more longitudinal studies to track EF development over time (see,
e.g., Park et al., 2018; Tran et al., 2019), pre-registering studies to address the reproducibility crisis
in research (Open Science Collaboration, 2015), choosing within-group over across-group designs
to avoid having to exhaustively match groups (Baum & Titone, 2014; Van Hell & Poarch, 2014),
heeding individual diferences to a greater extent (DeLuca et al., 2019; Yamasaki et al., 2018), mov-
ing away from null-hypothesis significance testing (NHST) and p-values with the arbitrary cut-of
point of .05 to determine significance (McShane et al., 2017), and embracing reliance on efect sizes
or alternative statistical approaches such as Bayesian statistics (Wagenmakers et al., 2016) to determine
whether diferences found are dependable and robust.

Given the ongoing discussion in the research field highlighted earlier, Poarch and Krott (2019) have
recently suggested ways to move forward. For one, while the finding of subtle cognitive diferences in
EF task performance between groups may be interesting in itself, it may be of even greater relevance
to investigate how bilingualism impacts real life. Enhanced skills for bilinguals have thus also been
documented for perspective taking (for review, see Schroeder, 2018), creativity and divergent thinking
(Kharkhurin, 2009), and tolerance of ambiguity (Dewaele & Wei, 2013). Such performance difer-
ences may indicate more relevant and thus important diferences in real life. Furthermore, there have
also been attempts to move away from focusing solely on the development of EF towards exploring
the reverse direction, namely whether better EF skills are also linked to improved language processing.
Indications of such links have been reported for lexical processing (Gangopadhyay et al., 2019), sen-
tence comprehension (Navarro-Torres et al., 2019), sentence processing (Filippi et al., 2015), and syn-
tactic ambiguity resolution (Teubner-Rhodes et al., 2016). For example, Filippi and colleagues (2015)
found that bilingual children outperformed monolingual children in syntactically complex sentence
comprehension under linguistic noise, corroborating earlier research with adults (Filippi et al., 2012).
Furthermore, future research will need to take into account that maintaining a dichotomy between
monolinguals and bilinguals is problematic since individuals from both these distinct groups may have
a possibly greater overlap in language experience than previously assumed (see, e.g., Blanco-Elorrieta
& Pylkkänen, 2018; Hulstijn, 2019; Poarch & Krott, 2019; Poarch & Van Hell, 2017). Such individual
diferences within groups of participants may be better captured by describing specific language usage
patterns, as, for example, captured in the adaptive control hypothesis (Green & Abutalebi, 2013) and
the characterization of social diversity using language entropy (Gullifer & Titone, 2020).
Which specific bilingual experiences induce changes in executive function (EF)?

Which EF sub-components are infuenced by which bilingual experiences?
Which bilingual experiences induce neural re-organization, structural changes, and changes in plasticity
and connectivity?
How does domain-general EF afect language processing?
To what extent does domain-general EF afect switching between languages?
433
Gregory J. Poarch
Further reading
Antoniou, M. (2019). The advantages of bilingualism. Annual Review of Linguistics, 5, 395–415.
This paper examines evidence for and against the beneficial efects of bilingualism on executive functions,
cognitive aging, and brain plasticity.
Baum, S., & Titone, D. (2014). Moving toward a neuroplasticity view of bilingualism, executive control, and
aging. Applied Psycholinguistics, 35, 857–894.
A review of cognitive aging and the efects of life-long bilingualism on neuroplasticity and age-related
declines in executive control functions.
Bialystok, E. (2017). The bilingual adaptation: How minds accommodate experience. Psychological Bulletin, 143,
233–262.
The paper evaluates evidence for systematic modifications of brain and cognitive systems induced by bilin-
gualism against the backdrop of experience-dependent plasticity.
Valian, V. (2015). Bilingualism and cognition. Bilingualism: Language & Cognition, 18, 3–24.
An overview of diferent conceptualizations of executive function and cognitive reserve and the range of
factors that are known to improve executive function.
References
Abutalebi, J., & Green, D. W. (2016). Neuroimaging of language control in bilinguals: Neural adaptation and
reserve. Bilingualism: Language and Cognition, 19, 689–698.
Adesope, O. O., Lavin, T., Thompson, T., & Ungerleider, C. (2010). A systematic review and meta-analysis of
the cognitive correlates of bilingualism. Review of Educational Research, 80, 207–245.
Alladi, S., Bak, T. H., Mekala, S., Rajan, A., Chaudhuri, J. R., Mioshi, E., Krovvidi, R., Surampudi, B., Dug-
girala, V., & Kaul, S. (2016). Impact of bilingualism on cognitive outcome after stroke. Stroke, 47, 258–261.
Alladi, S., Bak, T. H., Shailaja, M., Gollahalli, D., Rajan, A., Surampudi, B., Hornberger, M., Duggirala, V.,
Chaudhuri, J. R., & Kaul, S. (2017). Bilingualism delays the onset of behavioral but not aphasic forms of
frontotemporal dementia. Neuropsychologia, 99, 207–212.
Anderson, J. A. E., Chung-Fat-Yim, A., Bellana, B., Luk, G., & Bialystok, E. (2018). Language and cognitive
control networks in bilinguals and monolinguals. Neuropsychologica, 117, 352–363.
Antón, E., Duñabeitia, J. A., Estévez, A., Hernández, J. A., Castillo, A., Fuentes, L., Davidson, D. J., & Car-
reiras, M. (2014). Is there a bilingual advantage in the ANT task? Evidence from children. Frontiers in Psy-
chology, 5, 398.
Bak, T. (2016). Cooking pasta in La Paz. Linguistic Approaches to Bilingualism, 6, 699–717.
Bak, T., & Robertson, I. (2017). Biology enters the scene. A new perspective on bilingualism, cognition, and
dementia. Neurobiology of Aging, 50, iii–iv.
Best, J. R. (2010). Efects of physical activity on children’s executive function: Contributions of experimental
research on aerobic exercise. Developmental Review, 30, 331–551.
Best, J. R., & Miller, P. H. (2010). A developmental perspective on executive function. Child Development, 81,
1641–1660.
Bialystok, E. (1988). Levels of bilingualism and levels of linguistic awareness. Developmental Psychology, 24, 560–567.
Bialystok, E. (2009). Bilingualism: The good, the bad, and the indiferent. Bilingualism: Language and Cognition,
12, 3–11.
Bialystok, E., Abutalebi, J., Bak, T. H., Burke, D. M., & Kroll, J. F. (2016). Aging in two languages: Implications
for public health. Ageing Research Review, 27, 56–60.
Bialystok, E., & Majumder, S. (1998). The relationship between bilingualism and the development of cognitive
processes in problem-solving. Applied Psycholinguistics, 19, 69–85.
Bialystok, E., Poarch, G. J., Luo, L., & Craik, F. I. M. (2014). Efects of bilingualism and aging on executive
function and working memory. Psychology and Aging, 29, 696–705.
Bialystok, E., & Sullivan, M. D. (Eds.). (2017). Growing old with two languages: Efects of bilingualism on cognitive
aging. John Benjamins.
Blanco-Elorrieta, E., & Pylkkänen, L. (2018). Ecological validity in bilingualism research and the bilingual advan-
tage. Trends in Cognitive Sciences, 22, 1117–1126.
434
Blom, E., Boerma, T., Bosma, E, Cornips, L., & Everaert, E. (2017). Cognitive advantages of bilingual children
in diferent sociolinguistic contexts. Frontiers in Psychology, 8, 552.
Blumenfeld, H. K., & Marian, V. (2014). Cognitive control in bilinguals: Advantages in stimulus-stimulus inhi-
bition. Bilingualism: Language and Cognition, 17, 610–629.
Botvinick, M. M., Braver, T. S., Barch, D., Carter, C. S., & Cohen, J. D. (2001). Confict monitoring and cogni-
tive control. Psychological Review, 108, 625–652.
Branzi, F. M., Calabria, M., Boscarino, M. L., & Costa, A. (2016). On the overlap between bilingual language
control and domain-general executive control. Acta Psychologica, 166, 21–30.
Calabria, M., Costa, A., Green, D. W., & Abutalebi, J. (2018). Neural basis of bilingual language control. Annals
of the New York Academy of Sciences, 1426, 221–235.
Chung-Fat-Yim, A., Sorge, G. B., & Bialystok, E. (2020). Continuous efects of bilingualism and attention on
Flanker task performance. Bilingualism: Language and Cognition, 1–6.
Darcy, N. T. (1953). A review of the literature on the efects of bilingualism upon the measurement of intel-
ligence. Journal of Genetic Psychology, 82, 21–57.
De Bruin, A. (2019). Not all bilinguals are the same: A call for more detailed assessments and descriptions of
bilingual experiences. Behavioral Sciences, 9, 33.
De Cat, C., Gusnanto, A., & Serratrice, L. (2018). Identifying a threshold for the executive function advantage
in bilingual children. Studies in Second Language Acquisition, 40, 119–151.
Declerck, M., Schuch, S., & Phillip, A. M. (2020). Confict adaptation during multilingual language production
as evidenced by the n-3 efect. Bilingualism: Language and Cognition, 1–4.
Del Maschio, N., Fedeli, D., & Abutalebi, J. (2018). Bilingualism and aging: Why research should continue.
Linguistic Approaches to Bilingualism. Advance online publication.
DeLuca, V., Rothman, J., Bialystok, E., & Pliatsikas, C. (2019). Redefining bilingualism as a spectrum of experiences
that diferentially afects brain structure and function. Proceedings of the National Academy of Sciences, 116, 7565–7574.
Desideri, L., & Bonifacci, P. (2018). Verbal and nonverbal anticipatory mechanisms in bilinguals. Journal of Psy-
cholinguistic Research, 47, 719–739.
Dewaele, J.-M., & Wei, L. (2013). Is multilingualism linked to a higher tolerance of ambiguity? Bilingualism:
Language and Cognition, 16, 231–240.
Diamond, A. (2013). Executive functions. Annual Review of Psychology, 64, 135–168.
Duñabeitia, J. A., Hernandez, J. A., Antón, E., Macizo, P., Estevez, A., Fuentes, L. J., & Carreiras, M. (2014). The
inhibitory advantage in bilingual children revisited: Myth or reality? Experimental Psychology, 61, 234–251.
Eriksen, B. A., & Eriksen, C. W. (1974). Efects of noise letters upon the identification of a target letter in a
nonsearch task. Perception Psychophysics, 16, 143–149.
Fan, J., McCandliss, B. D., Sommer, T., Raz, A., & Posner, M. I. (2002). Testing the efciency and indepen-
dence of attentional networks. Journal of Cognitive Neuroscience. 14, 340–347.
Filippi, R., Leech, R., Thomas, M. S., Green, D. W., & Dick, F. (2012). A bilingual advantage in controlling
language interference during sentence comprehension. Bilingualism: Language and Cognition, 15, 858–872.
Filippi, R., Morris, J., Richardson, F. M., Bright, P., Thomas, M. S. C., Karmilof-Smith, A., & Marian, V.
(2015). Bilingual children show an advantage in controlling verbal interference during spoken language
comprehension. Bilingualism: Language and Cognition, 18, 490–501.
Galambos, S. J., & Hakuta, K. (1988). Subject-specific and task-specific characteristics of metalinguistic aware-
ness in bilingual children. Applied Psycholinguistics, 9, 141–162.
Gangopadhyay, I., Ellis Weismer, S., & Kaushanskaya, M. (2019). Domain-general inhibition and lexical pro-
cessing in monolingual and bilingual children: A longitudinal approach. Cognitive Development, 49, 68–80.
Garon, N., Bryson, S. E., & Smith, I. M. (2008). Executive function in preschoolers: A review using an integra-
tive framework. Psychological Bulletin, 134, 31–60.
Goldsmith, S. F., & Morton, J. B. (2018). Sequential congruency efects in monolingual and bilingual adults: A
failure to replicate Grundy et al. (2017). Frontiers in Psychology, 9, 2476.
Goodenough, L. (1926). Racial diferences in the intelligence of school children. Journal of Experimental Psychol-
ogy, 9, 358–397.
Goral, M., Campanelli, L., & Spiro, A. (2015). Language dominance and inhibition abilities in bilingual older
adults. Bilingualism: Language and Cognition, 18, 79–89.
Green, D. W. (1998). Mental control of the bilingual lexico-semantic system. Bilingualism: Language and Cogni-
tion, 1, 67–81.
Green, D. W., & Abutalebi, J. (2013). Language control in bilinguals: The adaptive control hypothesis. Journal of
Cognitive Psychology, 25, 515–530.
Grosjean, F. (2013). Bilingualism: A short introduction. In F. Grosjean & P. Li (Eds.), The psycholinguistics of
bilingualism (pp. 5–25). Wiley-Blackwell.
435
Gregory J. Poarch
Grundy, J. G. (2020). The efects of bilingualism on executive functions: An updated quantitative analysis. Journal
of Cultural Cognitive Science, 4, 177–199.
Grundy, J. G., Anderson, J. A., & Bialystok, E. (2017). Neural correlates of cognitive processing in monolinguals
and bilinguals. Annals of the New York Academy of Sciences, 1396, 183–201.
Hahn, C., Cowell, J. M., Wiprzycka, U. J., Goldstein, D., Ralph, M., Hasher, L., & Zelazo, P. D. (2012). Cir-
cadian rhythms in executive function during the transition to adolescence: The efect of synchrony between
chronotype and time of day. Developmental Science, 15, 408–416.
Hakuta, K. (1986). Mirror of language: The debate of bilingualism. Basic Books.
Huizinga, M., Dolan, C. V., & van der Molen, M. W. (2006). Age-related change in executive function: Devel-
opmental trends and a latent variable analysis. Neuropsychologia, 44, 2017–2036.
Hulstijn, J. H. (2019). An individual-diferences framework for comparing nonnative with native speakers: Per-
spectives from BLC theory. Language Learning, 69(1), 158–183.
Keye, D., Wilhelm, O., Oberauer, K., & van Ravenzwaaij, D. (2009). Individual diferences in confict-moni-
toring: Testing means and covariance hypothesis about the Simon and the Eriksen Flanker task. Psychological
Research, 73, 762–776.
Kharkhurin, A. V. (2009). The role of bilingualism in creative performance on divergent thinking and invented
alien creatures tests. Journal of Creative Behavior, 43, 59–71.
Kim, J. Y., & Wang, S. W. (2017). Relationships between dietary intake and cognitive function in healthy Korean
children and adolescents. Journal of Lifestyle Medicine, 7, 10–17.
Kroll, J. F., & Bialystok, E. (2013). Understanding the consequences of bilingualism for language processing and
cognition. Journal of Cognitive Psychology, 25, 497–514.
Kroll, J. F., Dussias, P. E., Bice, K., & Perrotti, L. (2015). Bilingualism, mind, and brain. Annual Review of Lin-
guistics, 1, 377–394.
Luk, G., & Bialystok, E. (2013). Bilingualism is not a categorical variable: Interaction between language profi-
ciency and usage. Journal of Cognitive Psychology, 25, 605–621.
Marian, V., Bartolotti, J., Rochanavibhata, S., Bradley, K., & Hernandez, A. E. (2017). Bilingual cortical control
of between- and within-language competition. Scientifc Reports, 7, 11763.
McShane, B. B., Gal, D., Gelman, A., Robert, C, & Tackett, J. L. (2017). Abandon statistical significance. arXiv.
https://arxiv.org/abs/1709.07588
Meuter, R. F., & Allport, A. (1999). Bilingual language switching in naming: Asymmetrical costs of language
selection. Journal of Memory and Language, 40, 25–40.
Mitchell, A. J. (1937). The efect of bilingualism in the measurement of intelligence. The Elementary School
Journal, 38, 29–37.
Miyake, A., & Friedman, N. P. (2012). The nature and organization of individual diferences in executive func-
tions: Four general conclusions. Current Directions in Psychological Science, 21, 8–14.
Munakata, Y., Herd, S., Chatham, C., Depue, B. E., Banich, M. T., & O’Reilly, R. C. (2011). A unified frame-
work for inhibitory control. Trends in Cognitive Sciences, 15, 453–459.
Naeem, K., Filippi, R., Periche-Tomas, E., Papageorgiou, A., & Bright, P. (2018). The importance of socio-
economic status as a modulator of the bilingual advantage in cognitive ability. Frontiers in Psychology, 9, 1818.
Navarro-Torres, C., Lopez Garcia, D., Chidambaram, V., & Kroll, J. F. (2019). Cognitive control facilitates atten-
tional disengagement during second language comprehension. Brain Sciences, 9, 95.
Noble, K. G., McCandliss, B. D., & Farah, M. J. (2007). Socioeconomic gradients predict individual diferences
in neurocognitive abilities. Developmental Science, 10, 464–480.
Olsen, R. K., Pangelinan, M. M., Bogulski, C., Chakravarty, M. M., Luk, G., Grady, C. L., & Bialystok, E. (2015).
The efect of lifelong bilingualism on regional grey and white matter volume. Brain Research, 1612, 128–139.
Open Science Collaboration (2015). Estimating the reproducibility of psychological science. Science, 349, aac4716.
Paap, K. R., & Greenberg, Z. I. (2013). There is no coherent evidence for a bilingual advantage in executive
processing. Journal of Cognitive Psychology, 66, 232–258.
Paplikar, A., Mekala, S., Bak, T. H., Dharamkar, S., Alladi, S., & Kaul, S. (2019). Bilingualism and the severity
of poststroke aphasia. Aphasiology, 33, 58–72.
Park, J., Ellis Weismer, S., & Kaushanskaya, M. (2018). Changes in executive function over time in bilingual and
monolingual school-aged children. Developmental Psychology, 54, 1842–1853.
Peal, E., & Lambert, W. E. (1962). The relation of bilingualism to intelligence. Psychological Monographs: General
and Applied, 76, 1–23.
Philipp, A. M., & Koch, I. (2009). Inhibition in language switching: What is inhibited when switching between
languages in naming tasks? Journal of Experimental Psychology: Learning, Memory, and Cognition, 35, 1187–1195.
436
Pintner, R. (1923). Comparison of American and foreign children on intelligence tests. Journal of Educational
Psychology, 14, 292–295.
Pliatsikas, C. (2019). Understanding structural plasticity in the bilingual brain: The Dynamic Restructuring
Model. Bilingualism: Language and Cognition, 23, 459–471.
Poarch, G. J. (2018). Multilingual language control and executive function: A replication study. Frontiers in Com-
munication, 3, 46.
Poarch, G. J., & Bialystok, E. (2015). Bilingualism as a model for multitasking. Developmental Review, 35,
113–124.
Poarch, G. J., & Krott, A. (2019). A bilingual advantage? An appeal for a change in perspective and recommen-
dations for future research. Behavioral Sciences, 9, 95.
Poarch, G. J., & Van Hell, J. G. (2012). Cross-language activation in children’s speech production: Evidence from
second language learners, bilinguals, and trilinguals. Journal of Experimental Child Psychology, 111, 419–438.
Poarch, G. J., & Van Hell, J. G. (2017). Bilingualism and the development of executive function in children: The
interplay of languages and cognition. In S. Wiebe & J. Karbach (Eds.), Executive function: Development across
the life span (pp. 172–187). Taylor & Francis.
Poarch, G. J., & Van Hell, J. G. (2019). Does performance on executive function tasks correlate? Evidence from
child trilinguals, bilinguals, and second language learners. In I. Sekerina, V. Valian, & L. Spradlin (Eds.), Bi-
lingualism, executive function, and beyond: Questions and insights (pp. 223–236). John Benjamins.
Poarch, G. J., Vanhove, J., & Berthele, R. (2019). The efect of bidialectalism on executive function. International
Journal of Bilingualism, 23, 612–628.
Qu, L., Low, J. J. W., Zhang, T., Li, H., & Zelazo, P. D. (2016). Bilingual advantage in executive control when
task demands are considered. Bilingualism: Language and Cognition, 19, 277–293.
Saer, D. J. (1923). The efect of bilingualism on intelligence. British Journal of Psychology, 14, 25–38.
Schroeder, S. R. (2018). Do bilinguals have an advantage in theory of mind? A meta-analysis. Frontiers in Com-
munication, 3, 36.
Shook, A., & Marian, V. (2019). Covert co-activation of bilinguals’ non-target language: Phonological competi-
tion from translations. Linguistic Approaches to Bilingualism, 9, 228–252.
Simon, J. R., & Rudell, A. P. (1967). Auditory S—R compatibility: The efect of an irrelevant cue on informa-
tion processing. Journal of Applied Psychology, 51, 300–304.
Simonis, M., Van der Linden, L., Galand, B., Hiligsmann, P., & Szmalec, A. (2020). Executive control perfor-
mance and foreign-language proficiency associated with immersion education in French-speaking Belgium.
Sorge, G. B., Toplak, M. E., & Bialystok, E. (2017). Interactions between levels of attention and levels of bilin-
gualism in children’s executive functioning. Developmental Science, 20, e12408.
Stern, Y. (2002). What is cognitive reserve? Theory and research application of the reserve concept. Journal of the
International Neuropsychological Society, 8, 448–460.
Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18,
643–662.
Sullivan, M. D., Poarch, G. J., & Bialystok, E. (2018). Why is lexical retrieval slower for bilinguals? Evidence
from picture naming. Bilingualism: Language and Cognition, 21, 479–488.
Sulzipio, S., Del Maschio, N., Del Mauro, G., Fedeli, D., & Abutalebi, J. (2020). Bilingualism as a gradient mea-
sure modulates functional connectivity of language and control networks. Neuroimage, 205, 116306.
Surrain, S., & Luk, G. (2019). Describing bilinguals: A systematic review of labels and descriptions used in the
literature between 2005–2015. Bilingualism: Language and Cognition, 22, 401–415.
Teubner-Rhodes, S. E., Misher, A., Corbett, R., Andreu, L., Sanz-Torrent, M., Trueswell, J. C., & Novick, J.
M. (2016). The efects of bilingualism on confict monitoring, cognitive control, and garden-path recovery.
Cognition, 150, 213–231.
Thierry, G., & Wu, Y. J. (2007). Brain potentials reveal unconscious translation during foreign-language com-
prehension. Proceedings of the National Academy of Sciences, 104, 12530–12535.
Tran, C. D., Arredondo, M. M., & Yoshida, H. (2019). Early executive function: The infuence of culture and
bilingualism. Bilingualism: Language and Cognition, 22, 714–732.
Van den Noort, M., Struys, E., Bosch, P., Jaswetz, L., Perriard, B., Yeo, S., Barisch, P., Vermeire, K., Lee, S.-H.,
& Lim, S. (2019). Does the bilingual advantage in cognitive control exist and if so, what are its modulating
factors? A systematic review. Behavioral Sciences, 9, 27.
Van Hell, J. G., & Poarch, G. J. (2014). How much bilingual experience is needed to afect executive control?
Wagenmakers, E. M., Morey, R. D., & Lee, M. D. (2016). Bayesian benefits for the pragmatic researcher. Current
Directions in Psychological Science, 11, 19–23.
437
Gregory J. Poarch
Wilson, R. S., Beckett, L. A., Barnes, L. L., Schneider, J. A., Bach, J., Evans, D. A., & Bennett, D. A. (2002).
Individual diferences in rates of change in cognitive abilities of older persons. Psychology and Aging, 17,
179–193.
Wilson, R. S., Boyle, P. A., Yang, J., James, B. D., & Bennett, D. A. (2015). Early life instruction in foreign
language and music and incidence of mild cognitive impairment. Neuropsychology, 29, 292–302.
Woumans, E., Santens, P., Sieben, A., Versijpt, J., Stevens, M., & Duyck, W. (2015). Bilingualism delays clinical
manifestation of Alzheimer’s disease. Bilingualism: Language and Cognition, 18, 568–574.
Xie, Z. (2018). The infuence of second language (L2) proficiency on cognitive control among young adult
unbalanced Chinese-English bilinguals. Frontiers in Psychology, 9, 412.
Yamasaki, B. L., Stocco, A., & Prat, C. S. (2018). Relating individual diferences in bilingual language experi-
ences to executive attention. Language, Cognition, and Neuroscience, 33, 1128–1151.
Zelazo, P. D. (2006). The Dimensional Change Card Sort (DCCS): A method of assessing executive function in
children. Nature Protocols, 1, 297–301.
Zhou, B., & Krott, A. (2016). Data trimming procedure can eliminate bilingual cognitive advantage. Psychonomic
Bulletin & Review, 23, 1221–1230.
Zhou, B., & Krott, A. (2018). Bilingualism enhances attentional control in non-verbal confict tasks—evidence
from ex-Gaussian analyses. Bilingualism: Language and Cognition, 21, 162–180.
Zuk, J., Benjamin, C., Kenyon, A., & Gaab, N. (2014). Behavioral and neural correlates of executive functioning
in musicians and non-musicians. PLoS ONE, 9, e99868.
438
35
SYNTHESIS
Transdisciplinary perspectives on second
language acquisition: exploration
versus explanation
Jan Hulstijn
Introduction
The central theme in this chapter concerns the potential of a transdisciplinary perspective on second
language acquisition (SLA). Is it the case that, by incorporating views from ever more other disciplines
(themselves also changing rapidly, infuenced by other fields), we might get a better chance of increas-
ing our understanding of second language (L2) acquisition, perhaps even attain (or at least approach)
an all-inclusive theory of L2 acquisition? Or is this goal unattainable and the ambition to strive for
it simply naïve? This chapter seeks to answer these questions. The answers are colored (critics might
say “biased”) by the author’s stance, rooted in Popper’s (1959) critical rationalism. Critical rationalism
gives priority to theory construction and theory testing over attempts to define a discipline in terms
of its objects of investigation, and over worries of the danger that the discipline might disintegrate.
The chapter is structured in the following way. For a proper treatment of the issues just mentioned
it is mandatory to first consider what scientific inquiry is and in what ways psycholinguistics and
SLA became multidisciplinary fields. The other chapters of this section of the handbook are then
discussed with a view on transdisciplinarity and exploration versus explanation. Holbrook (2013;
cited in Douglas Fir Group, 2016) assigns diferent meanings to the adjectives transdisciplinary, cross-
disciplinary, interdisciplinary, and multidisciplinary from her philosophical perspective. However, for the
topic of the present chapter, Holbrook’s fine distinctions are not needed and the terms will be used,
where appropriate, without principled diferences in meaning. The question of SLA’s scope was
pushed very strongly by the “Transdisciplinary framework for SLA in a multilingual world,” presented
by the Douglas Fir Group (DFG, 2016). This framework is presented and discussed with a view on
its potential for exploration and explanation. The chapter ends with directions for future research.
A theory (also called local theory) is an explanation of observations or a solution to a problem.

A theoretical framework (also called metatheory) ofers answers to wider, ultimate questions (e.g., the evolu-
tion of language and the human species).
DOI: 10.4324/9781003018872-40 439

Jan Hulstijn
From exploration to explanation (and back)

Scientific inquiry, one might think, is the business of exploration: collecting data, making observa-
tions, searching for facts. SLA researchers want to know how control of an L2 develops over time.
They want to know in what ways L2 learners difer in their learning and in the proficiency they
attain. They want to know which curricular settings or pedagogical practices are best for L2 learn-
ing. And they want to know dozens of other things. But scientific inquiry is more than exploration.
Observations of developmental patterns, observations of individual diferences, and observations of
the efects of instructional practices may be puzzling, calling for explanation. Theories are attempts to
provide explanations for non-trivial, puzzling observations. Empirical testing of a hypothesis derived
from a theory may produce new observations, throwing new light on the puzzling observations
which the theory sought to explain. This may necessitate the rejection or adaptation of the theory.
Ideally, a theory forms the heart (point 3) of a transparent argument of the following type:
1. The goal is to explain the following non-trivial observations: …

2. The explanation (theory) rests on the following assumptions: … Because of their conditional
status, assumptions are not proposed as testable claims.
3. The explanation (theory) runs as follows: … Typically, a theory introduces some constructs and
specifies the nature of their relationships.
4. From the theory, the following testable (falsifiable) claims (hypotheses) are derived: …
5. From the theory, the following as yet speculative claims are derived: … Future developments or
innovations in … will hopefully allow their empirical testing at some point in the future.
According to Muthukrishna and Henrich (2019), it makes good sense to distinguish between theories
and metatheories, also called theoretical frameworks. While theories address what the ecologist Tinber-
gen (1963) called proximate questions (about causation and development), metatheories (theoretical
frameworks) address ultimate questions (concerned with evolution and function).1 Darwin’s (1859)
views on evolution began as a theory but they have, meanwhile, developed into a metatheory, con-
ceiving nature as a complex adaptive system (Van den Bergh, 2018). In linguistics and psycholin-
guistics, the ultimate puzzle to be solved is how and why (typical) members of the human species are
capable of acquiring language(s) while members of other species are incapable of doing so. Genera-
tive linguists, along with some cognitive scientists and biologists, have provided their metatheory for
this feat (Berwick et al., 2013; Chomsky, 2005). Usage-based linguists, along with other cognitive
scientists, have provided a radically diferent metatheory, viewing language as a complex adaptive
system (the Five Graces Group, 2009). Like theories, theoretical frameworks should be falsifiable
(Hulstijn, 2020).
In sum, scientific inquiry is a matter of exploration and explanation (proximate and ultimate),
in a cyclical manner. As will become clear in the remainder of this chapter, the distinction between
exploration and explanation should be kept in mind when discussing possibilities for multidisci-
plinary work on L2 acquisition.
Psycholinguistics and SLA in historical perspective

SLA and psycholinguistics, the two disciplines forming the backbones of this handbook, go back a
long time. Psycholinguistics has its origins in the eighteenth century (Levelt, 2013) and SLA, the
study of people learning additional languages, has its roots in “25 centuries of language teaching”
and the book by the same name by Kelly (1969). However, the cognitive revolution in the twentieth
century gave an unprecedented boost (qualitatively and quantitatively) to both SLA and psycholin-
guistics (Table 35.1).
440
Synthesis
Table 35.1 Timeline and schools (paradigms) in linguistics and psychology.
Timeline Linguistics Psychology
1910 - Structuralism Behaviorism

Cognitive Revolution2
1960 - Generative linguistics Boxes-and-arrows cognitive psychology,
deterministic perspectives
1985 - Usage-based linguistics Neural-network psychology,
probabilistic perspectives
From around 1960 to 2000, psycholinguistics was typically concerned with the study of processes
at the lower levels (sounds, phonemes, morphemes, words, syntactic patterns), involved in speech
production, speech perception/comprehension and in reading. The table of contents of textbooks
and handbooks confirm this characterization (e.g., Traxler & Gernsbacher, 2006). Research tools ini-
tially allowed the measurement of overt behavior. Later, with the advance of new technology, covert
behavior (through EEG, fMRI, eye-tracking) could be observed as well (De Groot & Hagoort,
2018). Until recently, research was almost exclusively conducted in “the lab”—a physical place that
people had to visit in order to participate.
The study of writing processes has hardly been reported in textbooks of psycholinguistics for
many years. For example, the seminal article of Flower and Hayes (1981) was not published in a
psycholinguistic journal. By its nature, the study of information processing at higher linguistic levels
(discourse, paragraph, text) includes the role of metacognitive skills (e.g., metalinguistic knowledge,
reading and writing strategies), leading to a further broadening of psycholinguistics. Consequently,
psycholinguistics after approximately 2000 must be characterized as a much wider field than psycho-
linguistics in the 1960–2000 period, currently including the acquisition of sign languages, L2 acquisi-
tion, bilingualism, language processing in special populations (children, adults, and aging people with
atypical profiles), and the role of person attributes such as working-memory capacity and attentional
skills (executive functions; see Poarch, 2023 [this volume]).
Like psycholinguistics, SLA went through a period of enormous growth since the cognitive revo-
lution, both in volume (number of journals and publications) and scope. Most SLA researchers who
published their first papers in the 1970s had received their academic training in a language (e.g.,
English in the U.S. and the U.K.), not in psychology or another discipline. Many of them began their
career as an L2 instructor. At the time of their training, in the 1960s, the prevalent ideas about teach-
ing and learning came from structural linguistics and behavioral psychology (Gagné, 1965; Hilgard &
Bower, 1975; Skinner, 1957), materialized in the audiolingual method (Rivers, 1964).
After approximately 1970, infuences from cognitive psychology were numerous. Infuential con-
structs in the SLA literature such as automatization and skill acquisition (DeKeyser, 1997), intake
(Gass, 1988), input processing (VanPatten, 1996), focus on form (Long & Robinson, 1998), atten-
tion, awareness, and noticing (Schmidt, 1995), task-based language teaching (Robinson, 2002; Ske-
han, 1998), and involvement load (Laufer & Hulstijn, 2001) all refect (albeit in diferent degrees)
information-processing work in cognitive psychology. Levelt’s (1989) psycholinguistic model of
speaking infuenced Pienemann’s (1998) processability theory of L2 acquisition and De Bot’s (1992)
model of L2 speaking. A huge body of research on explicit and implicit L2 learning (Ellis, 1994;
Rebuschat, 2015) was (and still is) based on implicit and statistical learning in cognitive psychology
(Williams & Rebuschat, this volume) and on procedural and declarative memory in neuropsychol-
ogy (Morgan-Short & Ullman, this volume). The examples given so far illustrate the infuence of
cognitive psychology and cognitive neuroscience on SLA. But there is more. Lantolf (2000) based his
441
Jan Hulstijn
sociocultural theory of L2 acquisition on the work of the Russian psychologist Vygotsky. Almost all
textbooks of SLA contain sections devoted to constructs from other fields in the social sciences, such
as learning styles, personality, motivation, anxiety, willingness to communicate, language-learning
aptitude, attitudes (towards an L2, towards speakers of the L2, towards learning the L2), and accul-
turation. Only a handful of SLA researchers have tried to explain phenomena of L2 acquisition with
notions from outside the social sciences. Schumann (1997) sought to explain afectual aspects of L2
acquisition from a neurobiological perspective, arguing that emotion and cognition are tightly con-
nected in the brain through a long evolutionary process. A perspective from physics on L2 acquisi-
tion is ofered by Han et al. (2017), who argue that the law of energy conservation in physics can
be used to account for ultimate attainment of an L2. The authors ofer mathematical formulae for
the dynamic interplay of motivation, aptitude, L1-L2 distance, and L2 input. Finally, with respect
to methods of empirical inquiry, SLA has been afected by experimental psychology, neuroscience
(neuroimaging), psychometrics, and (frequentist and Bayesian) statistics (Norouzian et al., 2018).
Despite many infuences from other disciplines, much current SLA work still is conducted by
researchers trained in linguistics. Both generative and usage-based linguistics have put their stamp
on large amounts of theoretical and empirical work in SLA. The poverty-of-the stimulus argu-
ment (Chomsky, 1980, p. 3) and universal grammar, consisting of “principles” and “parameters”
(Chomsky, 1981, p. 3–4) had (and still have) a huge impact on SLA research. Usage-based linguistics
(insights of scholars like Langacker, Croft, Bybee, Hopper, Goldberg, and others), originating some
25 years later than generative linguistics, began to impact SLA researchers in the 1990s, notably
through the work of Nick Ellis.
In sum, psycholinguistics and SLA, until approximately the year 2000, can be characterized as
largely separate disciplines, as observed (more generally, for linguistics and psychology) already twenty
years ago by Segalowitz (2001). However, the second wave of the cognitive revolution, characterized
by thinking of the mind as a neural network, along with technological innovation, and probabilistic
usage-based linguistics made the two disciplines partly overlap in scope. Fortunately, there is no place
in academia for an authority dictating boundaries of, and between, disciplines. The current partial
overlap of psycholinguistics and SLA is the result of natural progress in scientific inquiry, driven by
researchers’ curiosity and desire to deepen their insights by taking account of ever more factors, as
the contributions to this handbook illustrate.
The chapters in this section

The authors of the five individual chapters in this section of the handbook review the study of L2
learning and use in various learning contexts, from various disciplines (such as education and devel-
opmental psychology) “neighboring” SLA, but also partially overlapping with it. Although explora-
tion and explanation must stand in a cyclical relationship to one another (as explained previously),
exploration arguably tends to dominate explanation in much (but not all) educational research,
given the need to find solutions to practical, educational problems. Thus, in their chapter, Sachs,
Baralt, and Gurzynski-Weiss (2023 [this volume]) address possible educationally favorable interac-
tions between learner attributes (in the cognitive realm) and pedagogical arrangements (in particular,
pedagogical tasks of diferent cognitive complexity). This work belongs to the tradition of aptitude-
treatment interaction research, which is motivated by the idea that it is, potentially, educationally
more fruitful to investigate pedagogical tasks and learner attributes in combination with one another
than separately. Providing a related educational perspective, Leow (2023 [this volume]) addresses
the question of how local contexts (e.g., diferent programs within a curriculum) or global contexts
(e.g., L2 learning in school or elsewhere) might afect the way in which the L2 is processed (i.e.,
psycholinguistically) when people are learning an L2. Leow notices a tremendous paucity of research
on this question, proposing that it “provides a worthy direction for future studies” (p. 382). Thus,
442
Synthesis
the work reviewed in this chapter is based on the premise that, albeit in ways poorly understood
so far, naturalistic and instructional environments (global contexts) and pedagogical procedures,
such as learner feedback (local contexts) may indeed afect L2 input processing in educationally
meaningful ways. Exploration characterizes the work reviewed in the chapters of Sachs et al.
and Leow more than explanation and hypothesis testing. The few hypotheses that do form the
starting point of empirical work reviewed in these chapters (associated with names such as Ske-
han, Robinson, and Krashen) are local theories (in the sense defined above, as an explanation of
observations). This observation is not a criticism. The distance between global or local contexts
of L2 learning and the way in which L2 information is being processed by L2 learners (from
millisecond to millisecond) is indisputably huge and it would thus be unfair to criticize brave
attempts to explore possible efects of learning contexts on what goes on in the mind/brain (see
“Open questions”).
Manchón (2023 [this volume]) addresses the question of how various cognitive processes involved
in writing (conceptualized as a kind of conscious, intentional information processing) may lead to
L2 learning. As in Leow’s contribution, Manchón’s contribution adopts an educational, instru-
mental perspective, guided by the question of which task-mediated processes causally lead to L2
learning. The “theoretical perspectives” on L2 writing as a site for L2 learning (section “Theoretical
perspectives and approaches”) are derived from local theories, which originated in the deterministic
school of cognitive psychology rather than in the probabilistic, neural-network school (see Table
35.1). Regarding the primary processes of writing (planning and formulation), these theoretical
perspectives pertain to the distinction between implicit (direct) learning, including automatization,
and explicit (indirect) learning. Regarding the processes involved in revision, including feedback
processing, the theoretical perspectives pertain to task-mediated depth of processing, monitor-
ing, activation of prior knowledge, and hypothesis testing, based on Gass’s (1997) model of input
processing.
Schoonen (2023 [this volume]) shows how the understanding of language proficiency, the cen-
tral construct in language testing, has changed substantially since the days of Lado (1961) as a
result of work in psycholinguistics. Lado distinguished between elements of language use and inte-
grated skills, refecting insights from structural linguistics and behaviorist psychology (Table 35.1).
In contrast, current work on language assessment (as reviewed by Schoonen) takes psycholinguistic
processing models into account. This change refects the paradigm shift, caused by the cognitive
revolution, calling for investigating and explaining covert cognitive processing in relation to overt
behavior. Schoonen points out that researchers of language assessment have to weigh the relevance
(or lack thereof) of psycholinguistic research on language processing at diferent levels of granularity
for diferent types of language assessment with diferent purposes (diagnostic testing as well as test-
ing of level of proficiency). Referring to Borsboom et al. (2004), Schoonen addresses the need of a
causal theory of construct validity, which turns out to be a tall order. The need of such a theory of
response behavior and the weighing of findings and insights from various disciplines makes language
assessment a fascinating field of inquiry, as illustrated by “Open questions and issues” in Schoonen’s
chapter.
Poarch (2023 [this volume]) addresses a psycholinguistic question, not connected to practi-
cal matters of language education and testing. The question is a typical and fascinating example
of the succession of exploration and explanation in a cyclical manner. It starts with observations
(obtained in some studies) of a possible association between control of various types of executive
functions and (degree of) bilingualism. Poarch shows that researchers have tried to investigate the
generalizability of such an association (i.e., between cognitive control and bilingualism) and, at the
same time, have tried to explain the existence (or lack) of patterns of association between cognitive
control and bilingualism (modulated by other factors, in diferent populations, involving difer-
ent tasks). Work in this field nicely illustrates how the role of explanans (cause) and explanandum
443
Jan Hulstijn
(efect) might change place. Although the jury may still be out with respect to the existence and
explanation of an association between cognitive control and bilingualism, both exploration and
explanation in this field—conducted along the conventions of critical rationalism—have deep-
ened (according to some), or not (according to others) our understanding of the control of more
than one language. As Berthele (2021, p. 110) remarks, “Multilingualism is large and old enough
to teach us modesty.”
A transdisciplinary framework for SLA (the Douglas Fir group, 2016)

In this section, we return to the topic of this chapter: transdisciplinarity in SLA. In 2016, a seminal
paper on the topic of transdisciplinarity was published in the Modern Language Journal that deserves to
be discussed here because it argues for transdisciplinarity from a wider perspective than discussed in
the previous sections. The paper, entitled “A transdisciplinary framework for SLA in a multilingual
world,” was authored by a collective of 15 scholars, united under the name of the Douglas Fir Group
(DFG, 2016). These scholars have “diferent theoretical roots, including in no particular order: socio-
cultural theory (Johnson, Lantolf, Negueruela, Swain), language socialization theory (Duf), social
identity theory (Norton), complexity and dynamic systems theory (Larsen-Freeman), usage-based
approaches (Ellis, Ortega), the biocultural perspective (Schumann), ecological and sociocognitive
approaches (Atkinson), variationist sociolinguistics (Tarone), systemic functional linguistics (Byrnes,
Doran), and conversation analysis (Hall)” (DFG, 2016, p. 20). The paper addresses historical, philo-
sophical, educational, and ethical issues of SLA as well as the demarcation lines with other disciplines.
The DFG framework “assumes the embedding, at all levels, of social, sociocultural, sociocognitive,
sociomaterial, ecosocial, ideological, and emotional dimensions” (p. 24). It pursues “an integrative
consideration of learners’ mental and neurobiological processing, remember-language, and moment-
to-moment language use” (p. 24).
The framework renders L2 learning as a process beginning “at the micro level of social activity.”
The context of these activities is situated and shaped at a meso level by “sociocultural institutions and
communities (e.g., neighborhood, work, leisure-time activities), characterized by pervasive social
conditions (e.g., economic, cultural, religious, political),” afecting the creation of people’s social
identity. At the macro level there are “society-wide ideological structures” and belief systems with
particular orientations toward the use and learning of (additional) languages (p. 24). Each of the three
levels “exists only through constant interaction with the others” (p. 25).
After the presentation of the framework, the DFG authors devote ten sections to “fundamental
themes” (p. 26), formulated as propositional statements.
1. Language competences are complex, dynamic, and holistic (p. 26).

2. Language learning is semiotic learning (p. 27).
3. Language learning is situated and attentionally and socially gated (p. 27).
4. Language learning is multimodal, embodied, and mediated (p. 29).
5. Variability and change are at the heart of language learning (p. 29).
6. Literacy and instruction mediate language learning (p. 30).
7. Language learning is identity work (p. 31).
8. Agency and transformative power are means and goals for language learning (p. 33).
9. Ideologies permeate at all levels (p. 33).
10. Emotion and afect matter at all levels (p. 36).
Three years later, the Modern Language Journal (volume 103 (S1), 2019) published another supple-
mentary issue, in which several individual members of the DFG were given room to set their indi-
vidual stamps on the framework.
444
Synthesis
Appraisal of the DFG framework

Google Scholar (accessed 22 March 2021) lists 176 citations of the 2016 DFG paper. Interest-
ingly, almost all papers citing the DFG paper are concerned with applied linguistics in the wide
sense, including language teaching and language and ideology. Only two of the citations provide an
appraisal by researchers of SLA in the narrow sense (linguists): Han (2016) and Slabakova (2019). Of
these, Han’s rejoinder gives the most critical appraisal of the DFG article. Han (2016) noted that the
ten themes, “with apparent lack of epistemological coherence,” do not make up a theory, although
Han acknowledged that the DFG authors did not claim the framework to be a theory (p. 738). As a
“way forward” for SLA, Han proposed to distinguish three subfields: (i) basic or fundamental SLA,
(ii) instructed SLA (ISLA), and (iii) applied SLA (ASLA). While basic/fundamental SLA and ISLA
have established themselves already (see Leow, 2023 [this volume]) the DFG framework brought
ASLA to the fore “as a potential area” (Han, 2016, p. 739). In her invited peer commentary, Slaba-
kova (2019) argued that the approach of generative SLA (GenSLA) in its more recent appearance
(Rothman & Slabakova, 2018) is partially commensurable with usage-based approaches to SLA and
the DFG’s framework. Leow (2023, pp. 373–374) notes that it is “challenging to address the infu-
ence of this transdisciplinary context on the psycholinguistics of L2 learning, when viewed from a
process-oriented perspective. . ., especially given the many variables postulated to play a role in L2
learning.”
By bringing in notions from the meso and macro levels, the DFG as a collective as well as indi-
vidual members of it are clear in their call for broadening the object of SLA, thereby redefining
SLA. With their proposal, the DFG authors “call to SLA researchers to expand their analytic gaze
to diferent dimensions of social activity and . . . to think integratively” (p. 38). Many researchers
(including myself) will think positively about parts or all of the contents of sections headed by the ten
statements listed earlier. But what kind of wisdom do these propositions refect? Are they generaliza-
tions supported by robust evidence? Are they assumptions, hypotheses, or speculative views? What
the DFG authors failed to do is point out that there is no certainty in scientific knowledge, that gen-
eralizations, assumptions, claims, and speculative views have to be translated into scientific inquiry of
the exploratory and explanatory kind, in a cyclical process. In the absence of such a warning, SLA
researchers, in particular novice researchers, may be led to “accept” the content of the propositions
as rendering the truth.
Whether the ideas proposed by the DFG will turn out to have a major (transdisciplinary) impact
on SLA as a discipline remains to be seen. At the time of writing this chapter (2020–2021) it is
probably too early to tell. One can sympathize with, support, or advocate the ethical and political
ideology of the proponents of a social and ecosocial turn in SLA as a desired goal for SLA. Novel and
“weird” ideas are needed, scientific inquiry is essentially anarchistic,3 but the real future of a scientific
discipline is eventually determined by its capability of solving fundamental puzzles through explora-
tion and explanation in a cyclical manner.
For SLA researchers sympathizing with the DFG framework and the social turn, the challenge,
therefore, is (i) to translate the framework’s main ideas into fundamental issues (puzzling phenomena
and problems in need of explanation and solution), (ii) find explanations and solutions, and (iii)
empirically test hypotheses derived from these explanations and solutions. A discipline may have
as its main goal to improve the world (create better ecosystems, prevent natural disasters, address
migration challenges, battle poverty, improve health care, improve education, improve L2 instruc-
tion) but the bottom line is that, if it loses its brief of being critical of virtually every statement,
examining its empirical evidence and finding explanations for crucial observations, it runs the
risk of becoming marginalized. More importantly, it may lose its capacity to produce cumulative,
robust, evidence-based knowledge, albeit even evidence-based knowledge will not give us absolute
certainty. As Popper said, there is no place for certainty in scientific inquiry but there are rational
445
Jan Hulstijn
and critical ways to reduce uncertainty maximally through theory construction and testing. Or, in
the succinct statement of Gass et al. (2021, p. 245), “Scientific rigor is the sine qua non in all areas
of scientific inquiry.”
Directions for future research

This chapter of the handbook addressed the degree of transdisciplinarity in current SLA, appraising
the preceding five individual chapters of this section and the DFG’s “Transdisciplinary framework for
SLA,” from a perspective on scientific inquiry as a cyclical process of exploration and explanation.
SLA’s history shows that the first phase of predominantly linguistic studies of “interlanguage” soon
evolved into a thriving phase, characterized by numerous infuences from various other disciplines
in the social sciences.
Transdisciplinary research and interdisciplinary collaboration are likely to be fruitful when
the researchers involved (with roots in diferent disciplines) share (i) a theoretical perspective
(preferably even an overarching theoretical framework [metatheory], addressing so-called ulti-
mate questions of evolution and function), and (ii) views on what should be accepted as robust
knowledge. Because of its nature, a theoretical framework provides converging perspectives from
diferent disciplines (linguistics, psychology, sociology, economy, brain sciences, physics, biology).
Because these perspectives converge, the framework forms the basis of potentially fruitful multi-
disciplinary work. Language as a complex adaptive system (CAS) is such a theoretical framework
(De Bot et al., 2007; Ellis, 2019; Han, 2019; Larsen-Freeman, 2018, 2019; and others). Language
as CAS is multidisciplinary as of necessity, with a potential for SLA, provided that its claims are
falsifiable (Hulstijn, 2020).
If researchers of diferent schools, embracing diferent epistemological stances, attempt to collabo-
rate, no fruitful work can be expected. To my knowledge, no important work has been conducted
by structuralists collaborating with generativists, by generativists collaborating with usage-based lin-
guists, or, in psychology, by cognitive psychologists with behaviorists. Equally important, collabora-
tion is not likely to be fruitful when researchers do not agree on the weights given to qualitative and
quantitative research methods.
The DFG framework should be taken seriously because it forces SLA researchers to refect on
what their goals are and how they might achieve them. In concrete terms, the first positive potential
of the DFG framework is that it encourages SLA researchers to replicate existing empirical work with
more varied samples of participants. Hitherto, most empirical work in SLA has been conducted
with rather homogeneous samples of L2 learners (often university students) in restricted arenas of L2
acquisition. Replication studies should include participants (i) with other backgrounds, (ii) exposed
to the L2 in other situations, as suggested by Andringa and Godfroid (2020; see also Bolibaugh et al.,
2021). A point not raised in the DFG article, which I propose to add, is that (iii) more typologically
diferent second languages should be included, acquired by people living in more parts of the world
than has been done in SLA hitherto. Empirical work along these three lines will surely produce
surprising findings which in turn will shed new light on some fundamental issues, supporting some
theoretical claims but challenging others. The bottom line is that transdisciplinarity is not a goal in
itself.
In conclusion, SLA is still in a healthy state of Darwinian variability: It is developing naturally and
exploring potentially promising new avenues. An equally natural process of Darwinian selection will
surely take place, with some theories surviving and others not. There is thus much work to do for
(new generations of) researchers, to make this selection process happen, through rigorous cycles of
exploration, explanation, and hypothesis testing.
446
Synthesis
Is it desirable (and is it indeed possible) to somehow unite (i) a theory of L2 acquisition as a mental,
cognitive phenomenon (the processing and representation of linguistic information in the mind/
brain) with (ii) a theory of learning and use of additional languages in diferent socio-economic or
ideological contexts?
What might the phenomena at the meso and macro levels of the DFG’s framework be that need a theo-
retical explanation?
Which theoretical framework (metatheory) would provide the best umbrella for the explanation of phe-
nomena of L2 acquisition and use at all three levels (micro, meso, macro)?
Is it necessary that every “local” question or claim be in line with an overarching theoretical framework?
For instance, does research on instructed L2 acquisition really need a theoretical framework (metathe-
ory) for it to produce robust, evidence-based knowledge, fruitful for L2 instruction?
Acknowledgments
I would like to thank William Grabe (Northern Arizona University), Robert DeKeyser (University
of Maryland), ZaoHong Han (Columbia University), and in particular Aline Godfroid (Michigan
State University) for their extremely helpful feedback on the first draft of this chapter.
Notes
1 Konrad Lorentz and Niko Tinbergen were awarded the Nobel Prize in Physiology/Medicine in 1973 “for
their discoveries concerning organization and elicitation of individual and social behavior patterns in animals.”
(https://en.wikipedia.org/wiki/Nikolaas_Tinbergen, archived 18 May 2020)
2 According to Miller (2003, p. 142), one can assign a real date to the cognitive revolution. The date is Sep-
tember 11, 1956. when Noam Chomsky, Paul Newman, Herb Simon, George Miller, and others met at a
symposium at M.I.T. For linguistics, Chomsky’s seminal books appeared in 1957 and 1965. In psychology, the
seminal publication was Miller et al. (1960).
3 “Scientific teamwork is essentially anarchistic: the best ideas should govern the process, not scientific author-
ity,” says Willem (Pim) Levelt, one of the founders of the Max Planck Institute for Psycholinguistics in
Nijmegen, in an interview in 2020 on the occasion of the institute’s fortieth anniversary (www.mpi.nl/40th-
anniversary, accessed 20 March 2021).
Further reading
The question of whether SLA has been, can, or should (not) be split up in various subfields, such as
instructed second language acquisition (Leow, 2023 [this volume]) and applied SLA (Han, 2016) has
been discussed by, among others, Ellis (2021), Long (2007), DeKeyser (2010), Hulstijn (2013), and Gass,
Loewen, and Plonsky (2021).
The question of whether SLA researchers, working in diferent cognitive and non-cognitive branches of SLA,
might be able to bridge their diferences has ramifications in ideology (the goals of scientific inquiry), philosophy
of science (epistemology and ontology in critical rationalism and relativism), sociology (the role of communities
of researchers in a scientific discipline), and research methodology (emic and etic perspectives; quantitative vs.
qualitative methods). These matters have been addressed by (in alphabetical order) Martha Bigelow, Nick Ellis,
Robert DeKeyser, Jan Hulstijn, James Lantolf, Alison Mackey, Lourdes Ortega, Steven Talmy, and Richard
Young in an article published in Studies in Second Language Acquisition (Hulstijn et al., 2014). See also Ortega
(2019).
447
Jan Hulstijn
References
Andringa, S. J., & Godfroid, A. (2020). Sampling bias and the problem of generalizability in applied linguistics.
Annual Review of Applied Linguistics, 40, 134–142.
Berthele, R. (2021). The extraordinary ordinary: Re-engineering multilingualism as a natural category. Language
Learning, 71(S1), 80–120.
Berwick, R. C., Friederici, A. D., Chomsky, N., & Bolhuis, J. J. (2013). Evolution, brain, and the nature of
language. Trends in Cognitive Sciences, 17(2), 89–98.
Bolibaugh, C., Vanek, N., & Marsden, E. J. (2021). Towards a credibility revolution in bilingualism research:
Open data and materials as stepping stones to more reproducible and replicable research. Bilingualism: Lan-
guage and Cognition, 24, 801–806.
Borsboom, D., Mellenbergh, G. J., & van Heerden, J. (2004). The concept of validity. Psychological Review,
111(4), 1061–1071.
Chomsky, N. (1980). Rules and representations. Basil Blackwell.
Chomsky, N. (1981). Lectures on government and binding. Foris Publications.
Chomsky, N. (2005). Three factors in language design. Linguistic Inquiry, 36(1), 1–22.
Darwin, C. (1859). On the origin of species by means of natural selection, or the preservation of favoured races in the struggle
for life. John Murray.
De Bot, K. (1992). A bilingual production model: Levelt’s “speaking” model adapted. Applied Linguistics, 13(1),
1–24.
De Bot, K., Lowie, W., & Verspoor, M. (2007). A dynamic systems theory approach to second language acquisi-
tion. Bilingualism: Language and Cognition, 10(1), 7–21.
De Groot, A. M. B., & Hagoort, P. (Eds.). (2018). Research methods in psycholinguistics and the neurobiology of lan-
guage: A practical guide. Wiley Blackwell.
DeKeyser, R. M. (1997). Beyond explicit rule learning. Automatizing second language morphosyntax. Studies
DeKeyser, R. M. (2010). Where is our field going? The Modern Language Journal, 94(4), 646–647.
Journal, 100(S1), 19–47.
Ellis, N. C. (Ed.). (1994). Implicit and explicit learning of languages. Academic Press.
Ellis, N. C. (2019). Essentials of a theory of language cognition. The Modern Language Journal, 103(S1), 39–60.
Ellis, R. (2021). A short history of SLA: Where have we come from and where are we going? Language Teaching.,
54, 190–205.
The Five Graces Group (Beckner, C., Blythe, R., Bybee, J., Christiansen, M. H., Croft, W., Ellis, N. C., Hol-
land, J., Ke, J., Larsen-Freeman, D., Schoenemann, T.) (2009). Language is a complex adaptive system.
Language Learning, 59(S1), 1–26.
Flower, L., & Hayes, J. R. (1981). A cognitive process theory of writing. College Composition and Communication,
32(4), 365–387.
Gagné, R. M. (1965). The conditions of learning. Holt, Rinehart & Winston.
Gass, S. M. (1988). Integrating research areas: A framework for second language studies. Applied Linguistics, 9(2),
198–217.
Gass, S. M. (1997). Input, interaction, and the second language learner. Lawrence Erlbaum.
Gass, S. M., Loewen, S., & Plonsky, L. (2021). Coming of age: The past, present, and future of quantitative SLA
research. Language Teaching, 54, 245–258.
Han, Z.-H. (2016). A “reimagined SLA” or an expanded SLA? A rejoinder to the Douglas Fir Group (2016).
The Modern Language Journal, 100(4), 736–740.
Han, Z.-H. (2019). Researching CDST: Promises and pitfalls. In Z.-H. Han (Ed.), Profling learner language as a
dynamic system (pp. 1–13). Multilingual Matters.
Han, Z.-H., Bao, G., & Wiita, P. (2017). Energy conservation: A theory of L2 ultimate attainment. International
Review of Applied Linguistics (IRAL), 50(2), 133–164.
Hilgard, E., & Bower, G. (1975). Theories of learning. Prentice-Hall.
Holbrook, J. B. (2013). What is interdisciplinary communication? Refections on the very idea of disciplinary
integration. Synthese, 190(11), 1865–1879.
Hulstijn, J. H. (2013). Is the second language acquisition discipline disintegrating? Language Teaching, 46,
511–517.
Hulstijn, J. H. (2020). Proximate and ultimate explanations of individual diferences in language use and lan-
guage acquisition. Dutch Journal of Applied Linguistics, 9(1/2), 21–27.
448
Synthesis
Hulstijn, J. H., Young, R. F., Ortega, L., Bigelow, M., DeKeyser, R., Ellis, N. C., Lantolf, J., Mackey, A., &
Talmy, S. (2014). Bridging the gap: Cognitive and social approaches to research in second language learning
and teaching. Studies in Second Language Acquisition, 36(3), 361–421.
Kelly, L. G. (1969). 25 centuries of language teaching. Newbury House.
Lado, R. (1961). Language Testing: The construction and use of foreign language tests. A teacher’s book. Longman.
Lantolf, J. (Ed.). (2000). Sociocultural theory and second language learning. Oxford University Press.
Larsen-Freeman, D. (2018). Looking ahead: Future directions in, and future research into, second language
acquisition. Foreign Language Annals, 51(1), 55–72.
Larsen-Freeman, D. (2019). On language learner agency: A complex dynamic systems theory perspective. The
Modern Language Journal, 103(S1), 61–79.
Laufer, B., & Hulstijn, J. (2001). Incidental vocabulary acquisition in a second language: The construct of task-
induced involvement. Applied Linguistics, 22(1), 1–26.
Leow, R. P. (2023). The infuence of learning contexts on the psycholinguistics of second language learning.
In A. Godfroid & H. Hopp (Eds.), The Routledge handbook of second language acquisition and psycholinguistics
Levelt, W. J. M. (1989). Speaking: From intention to articulation. MIT Press.
Levelt, W. J. M. (2013). A history of psycholinguistics: The pre-Chomskyan era. Oxford University Press.
Long, M. H. (2007). Problems in SLA. Erlbaum.
Long, M. H., & Robinson, P. (1998). Focus-on-form: Theory, research and practice. In J. Doughty & J. Wil-
liams (Eds.), Focus-on-form in classroom second language acquisition (pp 15–41). Cambridge University Press.
Miller, G. A. (2003). The cognitive revolution: A historical perspective. Trends in Cognitive Science, 7(2), 141–144.
Miller, G. A., Galanter, E., & Pribram, K. H. (1960). Plans and the structure of behavior. Holt Rinehart and
Winston.
Muthukrishna, M., & Henrich, J. (2019). A problem in theory. Nature Human Behaviour, 3(3), 221–229.
Norouzian, R., de Miranda, M., & Plonsky, L. (2018). The Bayesian revolution in second language research: An
applied approach. Language Learning, 68(4), 1032–1075.
Ortega, L. (2019). SLA and the study of equitable multilingualism. The Modern Language Journal, 103(S1), 23–38.
Pienemann, M. (1998). Language processing and second language development: Processability Theory. John Benjamins.
Poarch. G.J. (2023). Cognitive efects of bilingualism. In A. Godfroid & H. Hopp (Eds.), The Routledge handbook
Popper, K. R. (1959). The logic of scientifc discovery. Hutchinson.
Rebuschat, P. (Ed.). (2015). Implicit and explicit learning of languages. John Benjamins.
Rivers, W. (1964). The psychologist and the foreign-language teacher. The University of Chicago Press.
Robinson, P. (2002). Learning conditions, aptitude complexes and SLA: A framework for research and peda-
gogy. In P. Robinson (Ed.), Individual diferences and instructed language learning (pp. 95–112). John Benjamins.
Rothman, J., & Slabakova, R. (2018). The generative approach to SLA and its place in modern second language
studies. Studies in Second Language Acquisition, 40(2), 417–442.
Sachs, R., Baralt, M., & Gurzynski-Weiss, L. (2023). Psycholinguistics for the language classroom. In A. God-
froid & H. Hopp (Eds.), The Routledge handbook of second language acquisition and psycholinguistics (pp. 387–399).
Routledge.
Schmidt, R. (1995). Consciousness and foreign language learning: A tutorial in the role of attention and aware-
ness in learning. In R. Schmidt (Ed.), Attention and awareness in foreign language learning (pp 1–63). University
of Hawai’i Press.
Schoonen, R. (2023). The psycholinguistics of second language assessment. In A. Godfroid & H. Hopp (Eds.),
Schumann, J. H. (1997). The neurobiology of afect in language. Blackwell.
Segalowitz, N. (2001). On the evolving connections between psychology and linguistics. Annual Review of
Skehan, P. (1998). A cognitive approach to language learning. Oxford University Press.
Skinner, B. (1957). Verbal behavior. Appleton-Century-Crofts.
Slabakova, R. (2019). “L” stands for language. The Modern Language Journal, 103(S1), 152–160.
Tinbergen, N. (1963). On aims and methods of ethology. Zeitschrift für Tierpsychologie, 20, 410–433.
Traxler, M. J., & Gernsbacher, M. A. (Eds.). (2006). Handbook of psycholinguistics (2nd ed.). Academic Press.
Van den Bergh, J. C. J. M. (2018). Human evolution beyond biology and culture. Cambridge University Press.
VanPatten, B. (1996). Input processing and grammar instruction in second language acquisition. Ablex.
449
INDEX
Note: Page numbers in italics and bold refer to figures and tables respectively.
Abrahamsson, N. 30 Andringa, S. 300, 302, 410

accented speech 15 anecdotal observations 268
accuracy 43, 77, 112–113, 142, 160 animacy 222
acquisition 66, 317; see also adult second language ANT performance 431
acquisition; child L2 acquisition; child second anxiety: cognition in language classroom 387;
language acquisition, psycholinguistics; heritage cognitive processing 392–393
language acquisition; language acquisition; second applied SLA (ASLA) 378, 445; see also second
language acquisition (SLA) language acquisition (SLA)
across-language competition 426 aptitude-treatment interactions (ATI) 387–389
activation threshold 67, 204 Arechabaleta, B. 79
active gap-filling strategy 219 Arslan, S. 80
ACT-R architecture 323 articulation 353–354
Adamou, E. 262, 272 artificial grammar 283
adaptive control hypothesis 273 artificial grammar learning (AGL) 282, 285,
adaptive control of thought-rational (ACT-R) 289, 361
theory 310 artificial languages 282, 299; (semi-)artificial language
Adelman, J. 167 studies 299; sentences 283
adult second language acquisition: classroom-based artificial/mini-grammars 56
learning contexts 41; current trends and future artificial speech segmentation 283
45; grammatical vs. ungrammatical L2 sentences association measures (AMs) 169–170
41; immersion-based L2 learning 41; L1/L2 Associative Deficit Hypothesis 90
structural similarity and learning context 41; L2 associative learning 230–231, 233
processing, developmental stages 41; L2 syntactic asymmetrical switching cost 127
priming 43–45; learning to parse and parsing to attention 17–18
learn 39–40; lexical-semantic information 42; attention-demanding task 285
prediction-based learning accounts 42; predictive attitude 139
processing 42–43 attrition 62, 66
affective dimension 274 Au, T. 76
age at migration 64 auditory presentation, second language
age of acquisition (AoA) 77, 118, 325 psycholinguistics 153–154
age of onset 29 auditory statistical learning 285
age-related declines 88 autocorrelation 168
Aleman Banon, J. 223 automaticity 308–309; construct of 310;
Allaw, E. 391 measurements 312
Allen, L. Q. 143 automatization 308–309, 317, 365; blocked
Allport, A. 127 and interleaved practice 314–315; declarative
alternative learning-based account 44 knowledge in grammar practice for automatization
450
Index
313; distributed practice 313–314; explicit Blanco-Elorrieta, E. 260

knowledge 301; issues and topics 311; knowledge blocking 230–231
315; L2 practice in L2 education 317–318; Bloom, P. 257, 366
measurements and nature of automatization in L2 Bonifacci, P. 431
learning 316; speed and stability of L2 processing Booth, B. 315–316
312; theoretical perspectives and approaches Bordag, D. 196
309–311; transfer of learning 315–316; vocabulary Borro I. 343
learning for automatization 312 Borsboom, D. 415
awareness 295 brain: activity during language processing 128;
AX continuous performance task (AX-CPT) based response data 151; electrophysiological
271, 273 responses 142
Aydın, O. 53–54 British National Corpus 172
Ayneto, A. 18 Broersma, M. 258, 260
Brysbaert, M. 103
Baayen, H. 167 Bullemer, P. 282
Baddeley, A. 349–350, 363 Bultena, S. 259–260
balanced bilinguals 125 Burt, M. 27
Balota, D. A. 126 Byers-Heinlein, K. 20
Baralt, M. 376, 390 Bygate, M. 337
basal ganglia 325
basic memory process 286 Callies, M. 169
Batterink, J. L. 283 Carrol, G. 208, 211
Batty, A. O. 419 case marking 79
Beatty-Martinez, A. L. 262–263, 273 Catalan-Spanish bilinguals 127, 129
Bell, P. K. 299 Cedden, G. 53–54
Belletti, A. 245 Cedrus RB: Response Pads 155; series of response
benchmarking 151; and calibration, second language boxes 154
psycholinguistics 159; independent 154 CEM see cumulative enhancement model (CEM)
Bernolet, S. 44 central executive 349
bilinguals 426; advantage 18; children 15; Chang, C. B. 76
environments, natural experiment 21; experience CHILDES database 367
18–19; interlocutors 128; L1 learners 28; language childhood bilingualism 100; evaluation 21. see also
development 19, 61; languages 114; learning new bilingualism 21
words, strategies for 16; lexicons 130–131; mental child-internal and child-external variables 26, 28
lexicon 30; speakers 56, 127; vocabulary 13; child L2 learners: of English as adults 27; L1 effects
vocabulary growth 15. see also bilingualism 32; syntactic processing 32
18–19 children: dominant vs. non-dominant languages 17;
bilingual IA (BIA) model 209 knowledge of language 16; language acquisition
bilingual infants: auditory processing skills 139; language processing 17; skills across languages
14; bilingual and monolingual children 16; 17; vocabulary 16
bilingualism 17–18; cognitive benefits 18; familiar- child second language acquisition 26–27,
sounding speakers 14–15; independent languages psycholinguistics: age of onset 27–28; child-
processing 17; input and vocabulary in each internal and child-external variables 28;
language 16; language skills 13; language switching current trends and future 34; explicit and
17; learning problem 13; monolinguals and 14; implicit knowledge, differences between 29;
native languages sounds 14; non-native contrasts historical and theoretical perspectives 27; L1
14; vocabularies 15 and relation between two languages 28; lexical
bilingualism 69, 296, 426–427; bilinguals’ degree of and morphological processing 30–31; linguistic
125; consequences 17–18; critical issues and topics domains, differences between 28–29; outcome-
432–433; current contributions and research oriented studies 28; phonological production and
429–431; current trends and future directions perception 29–30; process-oriented studies 28;
433; historical perspectives 427–428; older age, profile effects 33–34; sentence- and discourse-level
new language language in 85–86; theoretical processing 31–33
perspectives and approaches 428–429; Chinese-English bilinguals 116–118
unbalanced 74 Chondrogianni, V. 32
bimodals picture-sentence matching chronometry 151–152
task 272 Chronos (Psychology Software Tools) 155
Birdsong, D. 250 CHRONSET 130
Bitchener, J. 405 Cintrón-Valentín, M. C. 300
451
Index
Clahsen, H. 40, 75, 77, 115, 219, 223 context-based implicit learning 167
classroom-based priming tasks 44 context-dependent memory 167
CLI see cross-linguistic influence (CLI) contingency 165, 169–170
Clyne, M. 257–258, 260 control process model (CPM) 257–258
CMP see cross-modal priming (CMP) conversational scaffolding 337
co-activation 426 Conway, C. M. 363
code-switching 14, 17, 255–256; cognate triggering Cook, S. 77–78
259; control process model 262–263; factors coordinate control 262
modulating cost 261–262; historical perspectives coreferential antecedent 221
257–258; interlocutor characteristics 260–261; corpora, in second language psycholinguistics
psycholinguistic research 257; sociolinguistic research: contingency 169–170; data for L2
perspectives 257; structural linguistic approaches psycholinguistics 172–174; entrenchment
257; triggering hypothesis 258–260 166–167; and experimental design 171; as primary
coefficient of variance (CV) 312 data 172; productivity 167; prototypicality and
cognate facilitation effect 127 salience 170; recency 167–169; surprisal 170–171;
cognates 51, 112, 127 triangulating corpus-based and experimental
cognition 426 approaches 171–172
cognitive and psycholinguistic processes 13, 16 corpus-based research 168–169
cognitive associations 67 corpus linguistics (CL) 164, 167
cognitive development 29, 426 Corpus of Contemporary American English
cognitive diagnostic assessment (CDA) 420–421 (COCA) 171
cognitive ecosystem 90 corrective feedback 308
cognitive/educational psychology 308 co-speech gestures 139
cognitive effects of bilingualism 426–433 Costa, A. 127, 129
cognitive flexibility and control 18 co-varying collexeme analysis 170
cognitive-interactionist theory 341 Cox, J. G. 92
cognitive-linguistic frameworks 164 Cremer, M. 419
cognitive models 231 critical foreign language geragogy 93
cognitive processes 111–112, 400, 414 critical period/sensitive period 26
cognitive-psychological aspects of language critical rationalism 439
learning 294 Cronbach, L. J. 415
cognitive psychology 90, 102, 139138 Crookes, G. 375
cognitive reserve 87, 427 cross-language: activation 31, 115, 116–117, 126,
cognitive resource limitations 40 270; identity effect 129; synonyms 15
cognitive skills, development 18 cross-linguistic influence (CLI) 28, 54, 61, 65
cognitive tasks 18; complexity 387 cross-linguistic: competition 67; differences 39;
cohesive devices 243 similarity 54, 65–66, 68; variation 138
collocational knowledge 169 cross-modal priming (CMP) 117–118
collostructional analysis 170 cross-situational learning 283–284
color naming 129 cue: availability 230; reliability 230; strength 88;
color-shape switching task 429 validity 230
communicative/content-based activities 315 Cumming, A. 401–403, 406
compensation strategies 87 cumulative enhancement model (CEM) 54–55
competence vs. performance 63 cumulative priming 44
complex, dynamic, and adaptive knowledge Cunnings, I. 221, 224
networks, and dialectic (CREED) 230 Curcic, M. 300
complex adaptive system (CAS) 446
complex dynamic systems theory (CDST) 89 Darcy, I. 195
compositional multiword sequences 208 Darcy, N. T. 428
comprehension 17; of sentences 117–118 Darwin, C. 440
computational models 208 De Bot, K. 50, 130–131, 133, 258, 260
computer-assisted language learning (CALL) 376 De Cat, C. 431
computer-mediated feedback 389 declarative and procedural memory in second
concordances 167 language learning 322; implications 331;
concreteness 76–77, 112 330–327
Conklin, K. 208, 211 declarative grammar test 315
connectives 33 declarative knowledge 309–310, 314, 351
Contemori, C. 250 declarative memory 67, 87, 324, 326–328; in lexicon
content-based curriculum 375 65–66; system 322
452
Index
declarative/procedural (DP) model of language Ellis, N. C. 234, 300, 339

322, 324, 366 Ellis, R. 296, 301
decoupling 68, 88 emblems 143
Degand, L. 245 English as a foreign language (EFL) 417
De Jong, N. H. 298–299, 316 English as second language (ESL) 181, 295
DeKeyser, R. 75, 184, 299, 308, 312, 323, 365 English language learners (ELL) test performances 420
Dekydtspotter, L. 40 enhanced incidental learning 343
de Larios, R. 402 entrenchment 88, 165, 166–167
deliberate vocabulary practice 312 episodic buffer 349
Depth of processing (DoP) 400, 405 episodic representations 192–193
derivational morphology 30–31 E-Prime (Psychology Software Tools) 153,
Desideri, L. 431 156–157, 159
Destruel, E. 249 Erlam, R. 389
detection 335, 342 error-based learning 43, 44
developmental language disorders (DLD) 32 Etherton, J. L. 285
developmental phenomena 67 Etxebarria, E. 77
developmental sharpening 341 eurocentrism 104–105
Dewaele, J. -M. 392–393 event conceptualization 65
Dewey, J. 373 event-related-potentials (ERP) 41, 111, 116–118,
Dienes, Z 299 128, 142–143, 179, 223–224, 234–235, 262, 270
Differential Object Marking (DOM) 79 executive function (EF) 426–428
digit naming 127 executive working memory 349, 354, 356
director-matcher tasks 140 experiment-generation software manufacturers 153
discourse 242; coherence markers 33; contributions explicit instruction on online processing 300–301
and research 249–251; Interface Hypothesis 247; explicit knowledge 26, 284, 294–295, 299
inferences in 243; information structure 246–247; explicit language aptitude (ELA) 344
prominent antecedents 221; RAGE hypothesis explicit learning 281, 295
248; referential coherence 244–245; relational externally induced switches 256
coherence 245–246; salient subjects 221; Shallow eye movements 115, 179; research 115–116; and
Structure Hypothesis 248–249 sentence processing 118; studies 111
dispersion 167, 169 eye-tracking 115, 118, 142, 151, 381; methodology
distractor 126 300; paradigm 170; during reading 79
DLD see developmental language disorders (DLD)
DMDX 156, 157 face-to-face (FTF) instruction 354, 376
Doherty, S. 300 Feature Reassembly Hypothesis (FRH) 68
Dohn, N. B. 373 feedback processing 400, 408
domain-by-age model 31 Felser, C. 40, 118, 219, 221, 223
domain-general learning 323 Fenn, K. M. 299
dominant language 20, 77 Ferreira, V. S. 127
Dominguez, L. 68 Filiaci, F. 244
Donaldson, B. 249 filler-gap dependencies 216, 219–220
Doughty, C. J. 317 finite-state grammar 361
Douglas Fir Group (DFG) 373 first language (L1) attrition 26, 67; cross-linguistic
d-pronouns 247 influence (CLI) 61; extralinguistic factors 65–66;
drift-diffusion model 366 historical perspectives 62–63; Interface Hypothesis
Drijvers, L. 142 and Feature Reassembly Hypothesis 68; language
dual-language environments 17–18 development 62; linguistic features, selective
dual-route model of lexical formation 366 vulnerability of 64–65; neurolinguistic perspective
Dulay, H. 27 66–67; reversibility of attrition 62, 64; scope of
Dussias, P. E. 250, 262–263 attrition 63–64; Unified Competition Model
(UCM) 67–68; usage-based approaches 67
early bilingualism see bilingual infants first-pass hypothesis 51–52
economy-driven parser 40 Fitzpatrick, T. 419
Egi, T. 337 Flanker task performances 263, 430, 431
electroencephalography (EEG) 151 Fletcher, P. C. 287
electrophysiological brain responses 111 Flower, L. 441
electrophysiological research 262 focal attention, awareness, and depth of processing 387
Elgort, I. 116, 179, 312 Fodor, J. 39–40, 363
elicited production tasks 139 Foote, R. 78
453
Index
foreign language instruction 380 Granena, G. 344

form-focused instruction (FFI) 300, 315 Gray, S. 350
form-meaning-function associations 343 Green, D. W. 262–263, 428
formulaic sequences 198 Grey, S. 56
formulation 400 Gries, St. 168
Foucart, A. 250 Grosjean, F. 66, 260
Franco, A. 283 Grundy, J. G. 431
Frenck-Mestre, C. 118 Gruter, T. 248, 250
frequency 67; and multiword units 207–208; and Gumperz, J. 257
single-word recognition 205–207 Gupta, P 363
frequency-as-repetition 167 Gurzynski-Weiss, L. 394
Fricke, M. 262 Gygax, P. M. 246
Friesen, D. C. 126
Fuchs, Z 78 HALA instrument 81
Fukkink, R. G. 315 Hamp-Lyons, L. 414
Fulga, A. 45 Hamrick, P. 284
functional magnetic resonance imaging (fMRI) 142 Han, Z. 441
fuzzy form-meaning mappings 195 Hardison, D. M. 142
fuzzy representations hypothesis 193 Harklau, L. 401
fuzzy semantic representations 196 Harrington, M. 117, 348
Hartsuiker, R. J. 44
Gánem-Gutiérrez, A. 406 Hatch, E. M. 336, 340
garden-path sentences 216–219 Haugen, E. 62
Gardner, M. K. 211 Hayes, J. R. 441
Gass, S. M. 404, 446 Henrich, J. 440
Gathercole, S. 356 heritage language acquisition 73–74; child and adult
Geeslin, K. L. 394 heritage speakers 75; current trends and future
gender: agreement 57, 78; -marked determiner 80–81; development 75; diverse languages 75;
43; marking in nouns 79; -matching object 32; dominance shifts 80; high-proficiency heritage
gender-case-number (GCN) marking 363 speakers 80; historical perspectives 74–75; home
Gernsbacher, M. A. 211 language 74; human language, theories of 74;
gesture 139; coding 140; comprehensibility 143; rates lexical access and morphological processing 76–78;
141; strokes 140 literacy role in 75; morphosyntactic processing
Gibson, E. 219 78–80; phonological perception and production
Gilmore, A. 406 76; speakers 73; syntax, semantics, and discourse
Godfroid, A. 167, 298–300, 362, 403, 410 80; theoretical perspectives and approaches 75
Goldsmith, S. F. 431 heritage languages 68
Golestani, N. 132 heritage speakers 34
Gollan, T. H. 127 heterogeneity among bilinguals 15
go/no-go paradigm 116 Hicks, G. 68
González Alonso, J. 53, 56–57, 98 higher-order cognitive processes 406
Goo, J. 341, 344 higher-order writing processes 406
Goodenough, L. 427 high-frequency words 203, 206–207, 209
Gor, K. 76–79, 186 hippocampalcortical declarative system 366
Gorilla 158 hippocampus 192
grammar-based processing 216 Holbrook, J. B. 439
grammar learning 356 Holmqvist, K. 159
grammatical case and gender concord 65 home language 74
grammatical core vs. periphery features 63 homophones 112–113, 204
grammatical encoding 130–131 Hopp, H. 32, 42, 223, 247, 268
grammatical islands 220 Howard, D. V. 287
grammaticality judgment test (GJT) 27, 180–181, Howard, J. H. 287
184, 288, 299; see also tasks Hoyer, W. J. 90
grammatical knowledge 28–29 Hsieh, H.-C. 391
grammatical morphemes 198 Hsu, N. 271
grammatical noun phrases 78–79 Hyltenstam, K. 30, 99
grammatical sensitivity (GS) 389, 395n1, 418 hypnosis 288
grammatical/syntactic representations 39 hypothesis linking 428
454
Index
idioms 198, 199, 207 interference tasks 125

IELTS 185 interlanguage grammars 222
imageability 76–77, 112 interlingual homographs 112–113, 126, 204;
implicit-explicit interface 296 interference 115
implicit knowledge 26, 294–296, 297 internally induced switches 256
implicit learning 44, 281, 295; attention, role of internal message 353
285–286; consolidation 289; cross-situational International Corpus of Learner English (ICLE) 167
learning 289; experimental paradigms 282–283; interrater reliability 140
grammar aspect 282; implicitness of implicit intra-sentential code-switching 255–256
statistical learning 283–285; individual differences Issa, B. I. 300
289; input statistics 282; interaction 287–289; Ito, K 300
mechanism 286; paradigms 282; regularities Iwashita, N. 417
286–287; statistical learning paradigms 282–283
implicit linguistic competence 323 Jackson, C. N. 41
incidental learning 282 Jacob, G. 77
incrementality 216 Jaeger, T. F. 171
individual communicative style 141 Jaensch, C. 50
individual differences 327, 414 Jágrová, K. 170
inflectional morphology 31, 78 Jang, E. E. 420
inflection vs. derivation 198 JavaScript: 159, RTs 158
information packaging 243 Jegerski, J. 79–80
information structure (IS) 243, 246 Jia, R. 31
inhibitory control (IC) 427–428 Jiang, N. 184
innate language processing principles 27–28 jitter (precision) 151, 158–159
input-driven factors 274 Johns, M. A. 273
input elaboration 335 Johnson, J. 27, 32
instance learning 343 jsPsych 158
instructed language learning (ILL) 378 judgment and rating tasks 139
instructed second language acquisition (ISLA) 373, Juffs, A. 117
445; applied ISLA vs. ISLA applied 378; curricular
context 375; delivery mode of instruction 376–377; Kaan, E. 261, 274
educational environment 380; elementary vs. high Kelly, S. D. 143, 440
school vs. tertiary levels 380–381; empirical research Kerr, R. 315–316
378–380; feedback 377; foreign language vs. second Keyboards 155
language environments 381; learning context from Kim, K. M. 299, 302
a local perspective 375–378; learning context Kirsner, K. 92
from global perspective 379–381; methodological Kırkıcı, B. 77
context 377; naturalistic vs. instructed environment knowledge/skill transfer 309
379; pedagogical context 376; programmatic levels Kohnert, K. 31
375–376; research setting and design 378; study Kootstra, G. J. 259
abroad/immersion vs. at home 379–380; theoretical Köpke, B. 64
psycholinguistic perspectives 374–381; theory 401; Kotz, S. A. 116
trends and future directions 381–382 Krashen, S. 295
integrated language skills 414 Krott, A. 428
interactional contexts 272–274 Kuhl, P. 367
interactional modifications 336
interactional structure of conversation 336 L1 grammar specifications 40
interactive activation (IA) models 204, 208–209 L1–L2 lexical distinctions 62
interactive gestures 139 L1 morphosyntactic processing 87
Interface Hypothesis (IH) 68, 224, 247, 296 L1 processing 39–40
interface hypothesis, explicit and implicit knowledge L2 automatization 316
294; disentangling automatized explicit and L2 formulaic sequences 343
implicit knowledge 301–302; explicit instruction L2 gender agreement processing 236
on online processing 300–301; issues and topics L2 grammar practice 313
297–298; reverse interface hypothesis 296–297, L2 knowledge and skill development 311
302–303; simultaneous acquisition of explicit and L2 lexical encoding 199
implicit knowledge from instruction 298–300; L2 lexical learning 364
theoretical perspectives 295 L2 metalinguistic knowledge 355
455
Index
L2 phonolexical encoding 195–196 lexical and morphological processing 30–31

L2 phonological contrasts 194–195 lexical decision 112–113
L2 proceduralization 316 lexical decision task (LDT) 112, 178–179, 180
L2 psycholinguistics: corpora data for 172–174; L1 lexical-developmental trajectory 312
and L2 speakers 182; reaction time research in lexical entrenchment 206
second language psycholinguistics 155 lexical incorporation 53
L2 sentence processing 118 lexical inhibition 114
L3/Ln acquisition: age-matched bilingual 52; lexical knowledge 28–29, 312
cross-language effects in, theoretical perspectives lexical knowledge quality 114
54–57; cross-linguistic influence (CLI) 54; current lexical memory 326
trends and future 57; dedicated investigations 57; lexical processing 426
language-specific exposure 52; morphosyntax Lexical Quality Hypothesis 199
54; non-native languages 50, 52; sequential lexical representations 191
multilingualism, nonlinear consequences 51–54 lexical segmentation 286
LaBrozzi, R. 234 lexical-thematic bias 219
Lado, R. 414 lexicon 204
lag (accuracy) 151, 158–159 lexicon-syntax interface 224
Lago, S. 56 LexTALE 185
Lakoff ’s Cognitive Commitment 164 Li, C. 185
Lambert, W. E. 428 Li, S. 344, 418
language-ambiguous words 125 Linck, J. A. 352
language analytic ability (LAA) 389 linguistic behavior 33
language assessment 414 linguistic competence 117
language background dimension 274 linguistic conceptualization 140
language-based curriculum 375 linguistic domains, differences between 28–29
language brokers 212n.2 linguistic hierarchy 178
language classroom 387; anxiety and cognitive linguistic markedness considerations 39
processing 392–393; aptitude-treatment linguistic meaning-making activity 401
interactions 388–390; attention, awareness, and linguistics and psychology 441
depth of processing 391–392; cognitive task loanwords 212n.1
complexity 390–391; issues and topics 388; local theory 439
teacher/interlocutor individual differences 393–394 Loewen, S. 337
language co-activation 426 Logan, G. D. 285
language curriculum 375, 381, language curriculum Long, M. H. 375, 394
(writing–to–learn); writing-to-learn 375 long-distance dependencies 219
Language Experience and Proficiency Questionnaire long-term migrants 62
(LEAPQ) 185 long-term priming 44
Language History Questionnaire 185 Lopez, B. 212n.2
language learning affordances of writing 405 López-Serrano, S. 406
language patterns 363 lower-order cognitive processes 406
language processing activity 407 low-frequency words 193, 197, 205–207, 209
language-related episodes (LRE) 407–408 Lozano-Arguelles, C. 234
language switching: bilingual infants’ processing 17; Lukyanchenko, A. 76
experiments 127; paradigm 127; tasks 125
language use dimension 274 MacIntyre, P. D. 394
LDT see lexical decision task (LDT) Mackey, A. 339, 341, 344, 348, 393
learned attention 230–231, 341 macro- and micro-writing processes 405
learner corpus research (LCR) 167 MacSwan, J. 257
learning contexts 374; on psycholinguistics 373 MacWhinney, B. 363, 417
learning new words, strategies for 16 magneto-encephalography 260
learning under distraction 285 Maie, R. 299
Lee, J. F. 300 mainstream psycholinguistics 98
Lee, Y.-W. 420 majority languages 21, 74
Lemmerth, N. 32 Manchón, R. M. 401, 403–404, 409
Lenet, A. E. 90 Mandarin–English bilinguals 56
Leow, R. P. 374, 378, 391, 404 Mani, N. 31
Levelt, W. J. M. 353 Marinis, T. 32, 40, 220
Lew-Williams, C. 100 markedness of forms 67
lexical access: 39; and morphological processing 76–78 Marsden, E. 300, 313
456
Index
masked orthographic form priming experiment 193 Multifactorial Prediction and Deviation Analysis
masked priming 114 using Regression (MuPDAR) approach 172
Mason, S. 77–78 multilingualism 26
Mathews, R. C. 288 multilinguals 51; speakers 28; vocabulary acquisition
maturational constraints 341 literature 54
McBride, S. 409 Multilink simulations 209–210
McCarthy, C. 30 multimodal L2 research 139
McClelland, J. L. 208–209 multimodal language processing: congruent
McDonough, K. 45, 168, 313, 391 gesture 144; cross-modal integration 138;
McManus, K. 300, 313 cross-modal interdependence 138; evidence for
McNamara, T. 413 137–138; gestures facilitate lexical ambiguity
McNeill, D. 140 resolution 138; historical perspectives 138–139;
Meara’s LLAMA test 418 incongruent gesture 144; innovations and
memory 17–18 future directions 145; language-specific gestural
memory-based episodic L2 hypothesis 192–193 repertoires 138; methodological reflections
memory mechanisms 323 144–145; methods and paradigms 139–144;
memory of foreign language materials 418 multimodal comprehension processing 141–143;
memory systems 324–325 multimodal L2 learning 143–144; multimodal
mental lexicon 62, 180, 191, 418 production processing 140–141; pre-test/post-test
mental model 242–243 design 143; representational gestures 138
mental retardation 428 MultiPic database 126
meta-analyses on corrective feedback 381 multiple-word expressions 180
metalanguage 294 multiword sequences 204, 207
metalinguistic awareness 29, 87 multi-word units: morphologically complex words
metalinguistic knowledge 29, 296, 323 197–198; processing 198–199
metalinguistic reflection 89 MuPDAR approach 172
metatheory 439 Muthukrishna, M. 440
Meuter, R. F. 127 mutual exclusivity 13, 16
Michigan Test of English Language Proficiency 185
Midford, R. 92 Nakatsukasa, K. 391
Miller, Z. F. 299 naming latencies 129–130
miniature linguistic system (MLS) 365 naming tasks 125, 181–182
mini-grammars based on natural languages 91 native language processing, dual mechanism
minority languages 21, 74 model 75
mirror-image design 51 native-like linguistic knowledge 29, 56, 98
mismatch paradigms 139 native speakers (NSs) 217, 335
missing surface inflection hypothesis 32 natural language paradigms 330
mixed-language sentences 17 Navarra, J. 30
Miyake, A. 356 Navarro-Torres, C. 271
modern language aptitude test (MLAT) 417–418 Naveh-Benjamin, M. 90
Monaghan, P. 283–284 negative feedback (NF) 338
monolinguals 13; -bilingual differences 18; negative transfer 88
children 16–17; environments 19–20; infants 15; negotiation for meaning 335–336, 340
interlocutors 128; learning new words, strategies neurocognitive techniques 263
for 16; speakers 20, 33; vocabulary development neurogenetics 87
17; see also bilingual infants neurolinguistic theory of bilingualism 66–67, 323
Montrul, S. 78–79 neurological activity 67
Morgan-Short, K. 300, 365–367 Neville, H. J. 118
morphological masked priming 114 Newport, E. 27, 32
morphosyntactic agreement 222 Nissen, M. J. 282
morphosyntactic agreement violations 222 non-adjacent dependencies 286
morphosyntactic associations 229 non-canonical (passive and object-initial) sentences 33
morphosyntactic information 42 non-code-switchers 272
morphosyntactic knowledge 34 non-interface position 295
morphosyntactic processing 42, 232; heritage non-pathological L1 attrition 62
language acquisition 78–80 non-propositional meaning 139
Morton, J. B. 431 non-stereotypical local antecedents 221
motivation 67, 89 non-verbal cognitive processing 426
multi-componential model 349, 351 non-verbal learning 327
457
Index
non-verbal processing tasks, 432 Petitto, L. A. 19

Northbrook, J. 211 Pham, G. 31
noticing 335, 342 phonemic coding ability 389
Novick, J. M. 271 phonetic coding ability 418
null-hypothesis significance testing (NHST) 433 phonetic plan 353
null pronoun 245 phonolexical representations 191
null vs. overt pronouns 64 phonological confusion 194
phonological encoding 191
object-subject ambiguity 218 phonological loop 349
Obler, L. K. 99 phonological neighborhood effects 329
obligatory encoding assumption 285 phonological neighbors 191, 194, 204
O’Brien, I. 354 phonological priming 194
offline and online measures 30 phonological processing 30
offline behavioral studies 55–57 phonological production and perception 29–30
offline methods 229 phonological short-term memory (PSTM) 349–389
offline tasks and knowledge 34 phono-translation distractors 129
Oh, J. S. 76 phono-translation interference 129
older age, new language language in: aging 85; Piaget, J. 373
bilingualism 85–86; cognitive decline 86; Pica, T. 339
contributions and research 90–92; critical issues picture description 132–133; tasks 125
and topics 89–90; current trends and future 92–93; picture naming 126–129
declarative/procedural model (DP) model 87; picture-picture interference 128
multiple aging symptoms 85; nature and nurture picture-word interference paradigm 126, 128, 184
interactions 86; second language acquisition (SLA) planning 400
86; theoretical perspectives and approaches 87–89; Plant, R. 153
unified competition model (UCM) 88; working Poarch, G. J. 31, 53, 428
memory and information processing mechanisms 85 Polio, C. 393
Omaki, A. 220 Popper, K. R. 439
one-shot design 378 postpuberty L2 learners 75
online sensitivity to L2 grammatical information 41 post-task grammar instruction 356
Open Science Framework (OSF) 119, 160 Potter, C. E. 100
OpenSesame 157, 158 power law 311
oral productive vocabulary test 416 practice 308–309
Orquin, J. L. 159 predicate proximity 219
orthographic neighbors 114 predicted behavior/inference tasks 142
Oxford Placement Test 185 predictive processing 39
Ozyurek, A. 142 Presentation (NBS) 157
pre-task grammar instruction 356
Pacton, S. 362 preverbal infants 18
Pan, H. Y. 249 priming 113–114, 168–169, 179; cumulative 44
Paradis, J. 31 priming paradigms 111
Paradis, M. 66, 323, 363 probabilistic associations 231
Parkinson’s disease 329 problem-solving behavior 404
parsing 216 problem-solving strategies 402
parsing-to-learn hypotheses 39, 41 proceduralization 310, 313, 361, 363, 365; and
passive exposure to speech 335 automatization 311; of L2 skills 313
patterning: consonants 286; vowels 286 proceduralized knowledge 351
pattern recognition 389 procedural knowledge 309
Peabody Picture Vocabulary Test-IV 185 procedural learning 325
Peal, E. 428 procedural memory 65, 67, 87, 322, 325
pedagogically oriented psycholinguistics 394 processability theory of L2 acquisition 441
Pellicer-Sanchez, A. 169 processing-based methods 56
Pellicer-Sánchez, A. 419 processing instruction (PI) 300, 363, 367
perceptual narrowing 13 Processing Rich Information from Multidimensional
perceptual salience 230 Interactive Representations (PRIMIR) model 19
perceptual speed 389 process-oriented perspective 382
Perfetti, C. A. 316 process-oriented studies 30
permastore 87 production compilation 323
Perruchet, P. 362 Productive Vocabulary Levels Test 185
458
Index
productivity, psycholinguistic notion 165 Reber, A. S. 282, 289

proficiency 413–414 Reber, P. J. 282
property-by-property transfer 55 Rebuschat, P. 284, 299, 361
prototypicality and salience 165, 170 recast 335; contingency of recasts 339
pseudocognates 53 recency 165, 167–169
pseudowords 282–284 receptive task 184
PsychoJS 158 recognition and comprehension of words 112–117;
psycholinguistic research, benchmarked timing event-related brain potentials 116–117; eye-
latencies 154 movement research 115–116; lexical decision
psycholinguistics and cognitive science 101–103 112–113; priming 113–114; semantic decision 113
psycholinguistics of L2 interaction 335; cognitive- reduced relative clause ambiguity 217
interactionist theory of ISLA 341; individual re-exposure 64
differences 344; input and interaction 336–337; referential dependencies 216, 221–222
input and interactional modifications on refresh rate 151
comprehension and acquisition 339–340; linguistic Reichle, R. 250
target 343; methodology 343–344; negative Reifegerste, J. 88
feedback and recasts 338–339, 343; output: relative clause attachment ambiguity 217, 219
scaffolding and pushed output 337–338; recasts relative clause island 220
343; relationships between interaction 343; Renaud, C. 40
salience, noticing, and detection 342–343 repetition 308
psycholinguistics of L2 learning 374 representational deficit models 222
psycholinguistics of writing 409 representational gestures 143
psychological ability 413 representational redescription 296
Psychology Software Tools’ E-Prime and Chronos resonance 67–68
package 154, 159 resource-dispersing variables 390
Psychophysics Toolbox 157 resting level activation (RLA) 204
PsychoPy 156, 157 reverse interface hypothesis 296–298, 302–303
Puig-Mayenco, E. 55 Roberts, L. 244
Pylkkanen, L. 128, 260 Robinson, P. 354, 389
Python 159 Roca de Larios, J. 401, 403
rote-learning 418
Q-matrix 420, 420, 421 rote memory (RM) 389
Rothman, J. 56
radical plasticity thesis 296 Roux, F. 130
RAGE hypothesis 248 rule-governed constraints 65–66
reaction time (RT) 111, 112, 203, 312 rule-governed grammatical knowledge 326
reaction time research in second language Rumelhart, D. E. 208–209
psycholinguistics: auditory presentation
153–154; benchmarking and calibration 159; saccades 115
brain-based response data 151; cathode ray Saer, D. J. 427
tube (CRT) monitors 153; chronometry in Saffran, J. R. 283
151–152; data collection on single system 158; Sagarra, N. 80, 234
E-Prime 159; L2 psycholinguistics research 155; salience 67, 230, 342
liquid crystal display (LCD) 152–153; practices Sanchez, D. J. 288
reporting and replicability 160; psycholinguistic Sanders, T. 245
research, benchmarked timing latencies 154; Santesteban, M. 127
recommendations 161; response devices 154–155; Sanz, C. 92
software 156–158; solutions and future directions Sato, M. 313
158–161; stimulus presentation 151–152; timing Sawaki, Y. 420
errors illustration 152; visual language stimuli Sawyer, M. 348
158–159; visual presentation 152–153; web-based Schmid, M. S. 64, 66
chronometric research 158 Schmidt, R. W. 285
Read, J. 419 Schmitt, N. 419
reading-aloud data 126 Schoonen, R. 419
reading process 245 Schulz, B. 220
real-life L2 use 317 Schumann, J. H. 441
real-time comprehension 111; processes 115 Schwartz, B. D. 31
real-time language processing 40, 41 Scontras, G. 78
real-word prime 193 Sebastian-Galles, N. 18
459
Index
second language assessment, psycholinguistics of incorporation 53; pseudocognates 53; self-paced

413; cognitive diagnostics 420–421; contributions reading task 53–54
and research 418–421; current trends and future serial reaction time (SRT) task 282
directions 421; eyes paying attention 419–420; Shallow Structure Hypothesis (SSH) 223, 248–249
historical perspectives 413–414; issues and Shehadeh, A. 337
approaches 414–415; knowing a word 418–419; Shen, X. R. 262, 272
language learning potential and components short-term memory 354, 356
417–418; language proficiency and components short-term priming 43
415–417 signal conversion 296
second language comprehension: cognitive processes Simonis, M. 431
111–112; comprehension of sentences 117–118; Simon task 429
innovations and future directions 118–119; single-and mixed-language sentences 20
methods and paradigms 111–112; reaction single-language task 115
time (RT) 112; real-time comprehension 111; single-language utterances 20
recognition and comprehension of words 112–117 situation model 242
second language production: color naming 129; Siyanova-Chanturia, A. 169
picture description 132–133; picture naming Skehan, P. 389
126–129; sentence completion 131–132; sentence skill acquisition theory 310, 323
production 130–131, 133–134; word naming SLA-oriented L2 writing 400
125–126; word production tasks 129–130 Slevc, L. R. 356
second language writing 400; historical perspectives SLM see speech learning model (SLM)
401–402; studies of writing processes 406–408; Smith, L. 284, 394
task-related studies 405–406; theoretical Snider, E. 171
perspectives and approaches 402–405; trend and social isolation 88
future directions 409–410; written corrective social learning 17–18
feedback processing 408–409 society-wide ideological structures 444
self-paced reading (SPR) 53–54, 78–79, 117, 179, socio-economic status (SES) 73, 428, 431
181, 182, 220, 316 Sokolov, J. L. 367
self-reported proficiency 185 Solovyeva, K. 312
semantic decision 113 Sorace, A. 244–245, 247
semantic knowledge 325 Sorge G. B. 431
semantic lures 193 speaking: articulation 353–354; conceptualization
semantic overlap, degree 140 353; formulation 353
semantic priming 114, 117 speech and gesture 140
semantics-based generalizations 286 speech-gesture congruity 143
semi-artificial languages 299, 317; learning 286 speech learning model (SLM) 65
sentence- and discourse-level processing 31–33 speech-specific patterns 169
sentence completion tasks 125, 131–132 speech stream 283
sentence processing 216; cognitive approaches speed-accuracy trade-off 113
229–230; cognitive characteristics 232–233; spoken language processing 78–79
experiential characteristics 231; gender 236; Spolsky, B. 414
historical perspectives 217; L1 transfer 232; SPR see self-paced reading (SPR)
L2 proficiency 232; limitations 237; linguistic SSARC (Simplify-Stabilize, Automatize,
approaches 223–224; morphosyntactic Restructure-Complexify) model 390
information, processing 222–223; number statistical learning 20, 281, 283
235–236; physical characteristics 230–231; statistical learning (SL) paradigm 282–283, 362
psychological characteristics 231; referential stereotype threat 87
dependencies 221–222; research 224–225; Steuck, J. 273
resolving syntactic ambiguities 217–219; syntactic stimulated recall (SR) 391
constraints, processing 219–223; tense 233–234 stimulus encoding 285
sentence production 130–131, 133–134 stimulus presentation, reaction time research 151–152
sentences comprehension: cross-modal priming striatal-cerebellar-thalamic procedural system 366
(CMP) 117–118; eye movements and sentence Stroop color-word task 429
processing 118; self-paced reading (SPR) 117 Stroop paradigm 129
sequence learning 282 Stroop-related conflict 271
sequential multilingualism, nonlinear consequences structural priming 39, 43–44; longer-term learning
51–54; cognate effects 53; cross-linguistic via 45; paradigm 132; technique 259
similarity 54; first-pass hypothesis 51–52; lexical studying abroad (SA) 380
460
Index
subject-first preference 42 UCM see unified competition model (UCM)

Sueyoshi, A. 142 Uggen, S. 167
Sullivan, M. 52 Ullman, M. T. 75, 324, 365–367
SuperLab (Cedrus) 157 unbalanced bilingualism 74
SuperLab 6 (RB-740 Response Pad) 155 unbalanced bilinguals 206
suprasegmentals 65 unified competition model (UCM) 67–68, 88, 367
surprisal 165, 170–171 universal grammar (UG) 217
Suvorov, R. 419 upper-level writing course (learning-to-write) 375
Suzuki, Y. 308, 313, 316–317, 365 usage-based approaches 88, 168, 310; language
switches spanning sentence boundaries 17 acquisition 27, 67; literature 165
switching cost 127 USB Game Pads 155
syllabus on L2 learning 375
synchronous computer-mediated communication VAC-verb contingency 172
(SCMC) 376, 388 Vaid, J. 212n.2
syntactic associations 231 van Hell, J. G. 31, 53
syntactic disambiguation 218 VanPatten, B. 300
synthetic syllabus 375 variability in bilingual processing 268–270; cognitive
control 270–272; interactional contexts 272–274;
Tagarelli, K. M. 88, 299 language regulation 270–272
target-language interactions 89 Vasylets, O. 406
task-based language teaching (TBLT) 375, 390 verb argument constructions (VACs) 170
task-modality studies 406 verb conjugation 300
task-related studies 405 verb frequency 172
tasks: accuracy in stimulus display and behavior verb-object-particle (VOP) constructions 172
179; and cognitive process 179; language use and verb particle construction (VPC) 172
acquisition 178–179; real-life language behavior verb-particle-object (VPO) 172
179; sensitivity 182–183; technical aspects of Verhaeghen, P. 90
181–182; theoretical aspects of 180–181; variable Verhallen, M. 419
manipulation and control 179; word association visual language stimuli, reaction time research 158–159
task 179 visual presentation, reaction time research 152–153
task-switching 263 visual-spatial short-term memory 349
teacher and interlocutor individual differences 387 visual-spatial sketchpad 349
theoretical/generative linguistics 63 Visual World eye-tracking paradigm 80
Thierry, G. 116–117 vocabulary learning 53, 143, 196, 286
think-aloud protocols 381 Voc D measurement 415
threshold hypothesis 353 voice onset time (VOT) 65
tick 151 von Holzen, K. 31
timing errors illustration, reaction time VOT see voice onset time (VOT)
research 152 vowel formants 65
Tinbergen, N. 440 VPC see verb particle construction (VPC)
tip-of-the-tongue states 140 VPO see verb-particle-object (VPO)
TOEFL 185 Vygotskian socio-cultural theory 418
TOEIC 185 Vygotsky, L. S. 373
token frequencies 164–167
Tomasello, M. 366 Wang, S. W. 350
Toro, J. M. 285 Warren, P. 179, 312
total conceptual vocabulary 13, 15 web-based chronometric research 158
Tran, C. D. 431 web-based functionality 158
transdisciplinary research 275 webDMDX 186
transfer of learning 315–316 Weber-Fox, C. 118
transitional probabilities 282 Wei, L. 262–263
translation equivalents 13 weighted-attribute approach 170
triadic componential framework (TCF) 390 Weinreich, U. 257
triangulating corpus-based and experimental Williams, J. N. 299, 361, 403
approaches 171–172 Wilson, F. 247
triggering hypothesis 258–260 Wong, W. 300
Trofimovich, P. 168 word and multi-word processing 191, 203–204;
Turner, G. 153 frequency and multiword units 207–208;
461
Index
frequency and single-word recognition models 348–349; overview 352–357; reading

205–207; fuzzy representations hypothesis 352–353; SLA theories 351; speaking 353–355;
193; fuzzy semantic representations 196; vocabulary 356
L2 phonolexical encoding, problem with Wray, A. 169
195–196; L2 phonological contrasts 194–195; writing processes 400
memory-based episodic L2 hypothesis written corrective feedback (WCF) 377, 404
192–193; morphologically complex words written naming, naming latencies in 130
197–198; multiword units processing 198–199; Wu, Y. J. 116–117
phonolexical encoding 194–195; RLA vs. RT Wulff, S. 168, 171
distribution 210; theoretical perspectives and
approaches 191–193, 208–210 Xie, Z. 421
word-fragment completion 168
word monitoring 180, 316 Yilmaz, Y. 338–339, 344
word naming 125–126, 130 Yılmaz, G. 66
word-object mappings 283 Yu, C. 284
word production: cognates 124; interlingual
homographs 124; interlingual homophones Zamora, C. C. 391
124; in L2 124; L2 naming tasks 124; language- zero anaphora 141
ambiguous words 124; tasks 129-130 Ziegler, N. 394
working memory (WM) 89, 229–230, 344, Zirnstein, M. 270
348, 364, 389; critical issues and topics zones of proximal development 418
352; grammar 355–356; measures 349–351; Zufferey, S. 245–246
462

The Routledge Handbook of Second Language Acquisition and Psycholinguistics

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Routledge Handbook of Second Language Acquisition and Psycholinguistics

Uploaded by

Copyright:

Available Formats

THE ROUTLEDGE HANDBOOK OF

SECOND LANGUAGE ACQUISITION AND

Holger Hopp is a Professor of English Linguistics at Technische Universität Braunschweig (Institute

The Routledge Handbooks in Second Language Acquisition are a comprehensive, must-have

The Routledge Handbook of Second Language Acquisition and Pragmatics

The Routledge Handbook of Second Language Acquisition and Corpora

The Routledge Handbook of Second Language Acquisition and Language Testing

The Routledge Handbook of Second Language Acquisition and Technology

The Routledge Handbook of Second Language Acquisition and Writing

The Routledge Handbook of Second Language Acquisition and Speaking

The Routledge Handbook of Second Language Acquisition and Sociolinguistics

The Routledge Handbook of Second Language Acquisition and Individual

The Routledge Handbook of Second Language Acquisition and Psycholinguistics

For more information about this series, please visit: www.routledge.com/The-Routledge-Handbooks-

Edited by Aline Godfroid and Holger Hopp

1 Introduction: second language acquisition and psycholinguistics 1

2 The psycholinguistics of early bilingualism 13

3 The psycholinguistics of child second language acquisition 26

4 The psycholinguistics of adult second language acquisition 38

5 The psycholinguistics of L3/Ln acquisition 50

6 The psycholinguistics of frst language attrition 61

7 The psycholinguistics of heritage language acquisition 73

8 Learning a new language in older age: psycholinguistic and cross-

9 Synthesis: psycholinguistics across the lifespan 97

10 Studying second language comprehension 111

11 Studying second language production 124

12 Studying multimodal language processing: the integration of speech

13 Conducting reaction time research in second language psycholinguistics 150

14 Using corpora in research on second language psycholinguistics 164

15 Synthesis: research methods in second language psycholinguistics 178

16 Word and multi-word processing: memory-based and linguistic approaches 191

17 Word and multiword processing: cognitive approaches 203

18 Sentence processing in a second language: linguistic approaches 216

19 Sentence processing: cognitive approaches to second language

20 Discourse processing 242

22 Synthesis: transdisciplinary approaches to understanding variability in

23 Implicit learning and second language acquisition: a cognitive

24 Hypotheses about the interface between explicit and implicit

25 Automatization and practice 308

26 Declarative and procedural memory in second language learning:

27 The psycholinguistics of second language interaction 335

28 Working memory and second language learning: a critical and

29 Synthesis: the psycholinguistics of learning 361

30 The infuence of learning contexts on the psycholinguistics of second

31 Psycholinguistics for the language classroom 387

32 The psycholinguistics of second language writing 400

33 The psycholinguistics of second language assessment 413

34 Cognitive efects of bilingualism 426

35 Synthesis: transdisciplinary perspectives on second language

11.1 Examples of possible stimuli in (a) regular picture naming experiments,

Casey Lew-Williams is a professor in the Department of Psychology at Princeton University

Christine E. Potter is an assistant professor of psychology at the University of Texas at El Paso

Nuria Sagarra (PhD, University of Illinois at Urbana-Champaign, USA) is an associate professor

Sarah Schimke is a professor of German linguistics at Ludwig-Maximilians-Universität in Munich (Ger-

Michael T. Ullman is a professor in the Department of Neuroscience at Georgetown University

Shifting themes and perspectives within psycholinguistic research on SLA

Topics and themes of the handbook

I. Psycholinguistics Across the Lifespan

Notes for readers and instructors

1. Introduction with defnitions

4. Theoretical perspectives and approaches

Textbox 2.1 Key terms and concepts

Critical issues and topics

Bilingual infants can successfully discriminate between their two languages