The Acquisition of Spanish in Understudied Language Pairings

Issues in Hispanic and Lusophone Linguistics (IHLL)

issn 2213-3887
IHLL aims to provide a single home for the highest quality monographs and edited volumes
pertaining to Hispanic and Lusophone linguistics. In an effort to be as inclusive as possible, the
series includes volumes that represent the many sub-fields and paradigms of linguistics that
do high quality research targeting Iberian Romance languages. IHLL considers proposals that
focus on formal syntax, semantics, morphology, phonetics/phonology, pragmatics from any
established research paradigm, as well as psycholinguistics, language acquisition, historical
linguistics, applied linguistics and sociolinguistics. The editorial board is comprised of experts
in all of the aforementioned fields.
For an overview of all books published in this series, please see
http://benjamins.com/catalog/ihll

Editors
Jason Rothman Jennifer Cabrelli Amaro
University of Reading University of Illinois at Chicago

Editorial Board
Sonia Colina
University of Arizona
João Costa
Universidade Nova de Lisboa
Inês Duarte
Universidade de Lisboa
Sónia Frota
Universidade de Lisboa
Ángel J. Gallego
University of Barcelona
María del Pilar García Mayo
Universidad del País Vasco
Anna Gavarró
Universitat Autònoma de Barcelona
Kimberly L. Geeslin
Indiana University
Michael Iverson Liliana Sánchez
Macquarie University Rutgers University
Paula Kempchinsky Ana Lúcia Santos
University of Iowa Universidade de Lisboa
Juana M. Liceras Carmen Silva-Corvalán
University of Ottawa University of Southern
John M. Lipski California
Pennsylvania State University Juan Uriagereka
Gillian Lord University of Maryland
University of Florida Elena Valenzuela
Jairo Nunes University of Ottawa
Universidade de São Paulo Bill VanPatten
Acrisio Pires Michigan State University
University of Michigan, Ann Arbor
Pilar Prieto
Universitat Pampeu Fabra

Volume 3
The Acquisition of Spanish in Understudied Language Pairings
Edited by Tiffany Judy and Silvia Perpiñán
The Acquisition of Spanish
in Understudied Language Pairings

Edited by

Tiffany Judy
Wake Forest University

Silvia Perpiñán
The University of Western Ontario

John Benjamins Publishing Company

Amsterdam / Philadelphia
 TM
The paper used in this publication meets the minimum requirements of
8

the American National Standard for Information Sciences – Permanence

of Paper for Printed Library Materials, ansi z39.48-1984.

doi 10.1075/ihll.3
Cataloging-in-Publication Data available from Library of Congress:
lccn 2014038339 (print) / 2014047123 (e-book)
isbn 978 90 272 5802 1 (Hb)
isbn 978 90 272 6908 9 (e-book)

© 2015 – John Benjamins B.V.

No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any
other means, without written permission from the publisher.
John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands
John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa
 Table of contents

Introduction
The importance of crosslinguistic comparison in the study
of the acquisition of Spanish 1
Tiffany Judy & Silvia Perpiñán

part i.  Spanish as the L2 in a bilingual society

Crosslinguistic influences in the mapping of functional features
in Quechua-Spanish Bilingualism 21
Liliana Sánchez
Verbal agreement in the L2 Spanish of speakers of Nahuatl 49
Alma P. Ramírez-Trujillo & Joyce Bruhn de Garavito
Early coda production in bilingual Spanish and Basque 75
Maria-José Ezeizabarrena & Alaitz Alegria
The locative paradigm in the L2 Spanish of Catalan native speakers 105
Silvia Perpiñán

part ii.  Spanish as an L2 in a non-bilingual society

The acquisition of Spanish in a bilingual and a trilingual L1 setting:
Combining Spanish with German, French and Catalan 135
Laia Arnaus Gil & Natascha Müller
Knowledge and processing of subject-related discourse properties
in L2 near-native speakers of Spanish, L1 Farsi 169
Tiffany Judy
Subject pronouns in the L2 Spanish of Moroccan Arabic speakers:
Evidence from bilingual and second language learners 201
Aurora Bel & Estela García-Alcaraz
The construal of goal-oriented motion events by Swedish speakers
of L2 Spanish: Encoding of motion endpoints and Manner of motion 233
Alejandra Donoso & Emanuel Bylund
 The Acquisition of Spanish in Understudied Language Pairings

part iii.  Spanish as an L2 in an instructional context

Object drop in L2 Spanish, (complex) feature reassembly, and
L1 pre-emption: Comparing English, Chinese, European and
Brazilian Portuguese learners 257
Michael Iverson & Jason Rothman
The acquisition of differential object marking in Spanish
by Turkish speakers 281
Silvina Montrul & Ayʂe Gürel
Copula choice in adjectival constructions in Dutch L1 Spanish L2 309
Manuela Pinto & Alexia Guerra Rivera
Typological proximity in L2 acquisition: The Spanish non-native grammar
of French speakers 329
Juana M. Liceras & Anahí Alba de la Fuente
Index 359
 The importance of crosslinguistic comparison
in the study of the acquisition of Spanish

Tiffany Judy & Silvia Perpiñán

Wake Forest University / The University of Western Ontario

The aim of this volume is to present to both the reader and the field a collection
of empirical studies examining the acquisition of Spanish, currently one of the
world’s most spoken and studied languages, in combination with languages other
than English. Despite the multitude of speakers of varied L1s that acquire Spanish,
no collection of understudied language combinations, such as those that comprise
this volume, has thus far been compiled. The research that appears herein includes
a variety of acquisition scenarios (child and adult), learning contexts (classroom,
naturalistic immersion), societal contexts (bilingual and non-bilingual societies),
language pairings (typologically similar and typologically dissimilar) and linguistic
properties (discourse constraints, adjectival constructions and coda production to
name a few), yet holds constant the acquisition of Spanish. This variety and breadth
of coverage is the base of what we consider a well-rounded survey of Spanish in
the context of bi/multilingualism and second language acquisition (SLA), but also
makes possible several important theoretical contributions.
For example, a major theoretical question examined in acquisition studies,
whether early bilingualism or adult SLA, regards the nature and representation
of the speaker’s developing linguistic system. While it is assumed in this volume
and the chapters that comprise it that language acquisition is guided by Universal
Grammar (UG), previous linguistic knowledge has observable effects on bilingual
development and SLA. It is only through a systematic comparison of different lan-
guage combinations that we can tease apart what comes from first language trans-
fer (whether facilitative or non-facilitative) and what comes from other sources
such as Universal Grammar, frequency of input, or general cognition. One per-
tinent example of this type of research is Iverson and Rothman’ chapter, where
object drop in Spanish is examined in two Romance languages (Brazilian and
European Portuguese) in conjunction with two non-Romance languages (­English
and Mandarin Chinese). The purpose of this chapter is to account for some
observed differences in the developmental sequences and ultimate attainment in
the L2 acquisition of the same target property from various L1 backgrounds. The
chapter contributes significantly to SLA theory and data analysis conventions in

doi 10.1075/ihll.3.01jud
© 2015 John Benjamins Publishing Company
 Tiffany Judy & Silvia Perpiñán

that it demonstrates, through comparison of the four aforementioned language

pairings, that a simple view of L1 transfer alone does not adequately explain the
different behavior from the L1 groups. Instead, Iverson and Rothman propose that
the theoretical constructs of L1 pre-emption (Trahey & White 1993) and feature
reassembly (Lardiere 2008, 2009) are essential in L2 theorizing. This and other
crosslinguistic descriptions, many of which are original, add to the field’s general
knowledge of linguistic systems and, more importantly, to the description of uni-
versal principles and parameterized properties.
Furthermore, systematic comparisons of this sort allow researchers to inves-
tigate how Spanish may develop at different rates or through different transitional
stages depending on the first or other language(s). An example from the literature
is VanPatten (1987), where it was proposed that native (L1) English speakers prog-
ress through a five-stage development in their acquisition of ser/estar in Spanish.
Taking into account the syntactic structure and the semantic features involved
in the different ser/estar constructions, VanPatten (2010) simplified this proposal
and concluded that the actual learning problem is estar, given its marked aspec-
tual nature in comparison to the unmarked aspectual feature of ser (Schmitt 1997,
2005). However, this proposal can only be generalized if corroborating evidence
from other language combinations obtains. In this volume, three chapters inves-
tigate the acquisition of Spanish ser/estar in combination with Catalan, Dutch,
German and French, and in different linguistic contexts. For example, Pinto and
Guerra Rivera examine the acquisition of the copular verbs with adjectives by
adult classroom L2 learners whose native language is Dutch, another language
with a binary copular system. Pinto and Guerra Rivera argue that acquisition of
ser/estar + adjective, is not problematic across the board, and that only a sub-
set of semantic properties denoted by estar + adjective are problematic for their
Dutch speakers. Perpiñán’s chapter explores the expression of location with ser/
estar (and also haber) in Spanish, and demonstrates that some semantic proper-
ties encoded with the expression of location can still somehow be problematic
for very advanced bilingual speakers whose L1 is Catalan. These two chapters
highlight the difficulties posed by the acquisition of estar and the importance of
understanding the underlying syntax and semantics of the ser/estar structures,
as proposed by VanPatten (2010). Arnaus Gil and Müller, on the other hand, in
their study of bilingual and trilingual children, observe that ser is more problem-
atic than estar for children, unlike what has been previously reported for adult L2
acquirers. These new data on the acquisition of ser/estar complement, redefine
and corroborate some proposals put forward by VanPatten (1985, 1987, 2010) for
English-speaking learners of Spanish, showing that it is only through examination
of multiple language combinations that we can gain a more accurate and complete
picture of the acquisition stages.
 The importance of crosslinguistic comparison in the study of the acquisition of Spanish 

Still other pairings from the same language family, namely Romance lan-
guages, make it possible to investigate microparametric differences and their
potential effects on SLA. Liceras and Alba de la Fuente’s overview chapter on the
interlanguage of French speakers learning Spanish, as well as Perpiñán’s study on
the Spanish grammar of Catalan-dominant Spanish-Catalan bilinguals demon-
strate that only by exploring subtle differences (i.e. microparametric differences)
between typologically closely-related languages can we properly describe the
implicit grammars of these speakers. Specifically, Liceras and Alba de la Fuente’s
chapter explores microparametric differences between French and Spanish on
noun compounds, plural realization, clitic clusters, plural and case marking in
quantifiers, passives and subject realization and conclude that, even when the lan-
guages are typologically proximate to one another, they might not be typologically
similar. This may be a cause of difficulties comparable to those that L2 speakers
encounter when faced with strong parametric differences or surface structure fil-
ters. These authors, among others, provide data on plural and case marking in
quantifiers, and demonstrate that the advanced French-Spanish speakers show
evidence of neither the L1 nor the L2 setting, but rather that of another typological
proximate Romance language, in this case Catalan. This type of research not only
contributes to the current debate regarding microparametric variation in mor-
phosyntax and SLA theory, but also makes contributions to the field of linguistics
as it provides a more fine-grained analysis of the linguistic variation evidenced in
the world’s languages as well as potentially providing insight into the L2 develop-
mental process.
Lastly, some language combinations provide a piece of the puzzle regarding
examination of certain hypotheses that are not fully testable in the absence of the
specific language pairing. For example, the Interface Hypothesis (Sorace 2011,
2012) claims that language combination is not a deterministic factor in native-like
processing of external interface-conditioned properties. That is, the facilitative or
non-facilitative nature of the L1 and L2 is not predicted to determine native-like
processing. To test this prediction, though, both facilitative and non-facilitative
language pairings must be examined. Judy’s chapter provides one necessary piece
of the puzzle for falsifying this claim in that it examines a language pairing and
property for which L1 transfer is facilitative. Thus, holding methodology and pro-
ficiency of the participants constant, if speakers of facilitative language pairings
such as Farsi-Spanish outperform speakers of non-facilitative pairings such as
English-Spanish, counterevidence is found for this specific claim.
An ongoing debate exists in the field regarding the role of input in language
development and language acquisition. In generative linguistics, however, the
role of other linguistic factors, usually those external to the speaker’s mind such
as quantity and quality of input or societal context, have been downplayed to a
 Tiffany Judy & Silvia Perpiñán

degree since the emphasis has largely been placed on describing implicit gram-
mars and linguistic competence. While describing language acquirers’ implicit
grammars remains a goal of this volume, we aim to do so while not obviating
the possible influence of external factors in the acquisition process. For this rea-
son, the volume is organized in three sections based on the type of contact the
speakers had with the Spanish language. Section 1 of the volume presents four
chapters that examine acquisition of Spanish in four distinct bilingual society con-
texts: ­Quechua-Spanish in Peru; Nahuatl-Spanish in Mexico; Basque-Spanish in
the Basque Country; and Catalan-Spanish in Catalonia. Section 2 contains four
chapters that present research conducted on naturalistic aquirers of Spanish, be
they children or adults, in non-bilingual societies: German-Spanish, French-
Spanish and German-Catalan-Spanish child speakers in Europe; Farsi-Spanish
adult speakers in Argentina; Moroccan Arabic-Spanish child sequential speakers
tested as teenagers in Spain as well as typical classroom L2 learners of Spanish; and
Swedish-Spanish speakers in Chile. Lastly, Section 3 is largely comprised of typical
adult classroom learners of Spanish: an overview chapter that includes B ­ razilian
­Portuguese, ­European Portuguese and English classroom learners of Spanish as
well as Mandarin Chinese-Spanish child, teenage and adult naturalistically acquir-
ing bilinguals in Peru; Turkish-Spanish learners in Turkey; Dutch-Spanish learners
tested in the Netherlands; and an overview chapter that examines French-Spanish
learners. The following sections provide a brief introduction to the acquisitional
setting, merits and description of each chapter.

1.  Spanish as the L2 in a bilingual society

The collection of studies included in this section describes the grammar of bilin-
guals living in a bilingual society and whose non-dominant or second language is
Spanish. In a bilingual society, both languages generally have considerable presence
in the life of the community and are both frequently activated in the bilingual’s
mind. Still, in most bilingual situations, speakers tend to identify themselves, be
dominant or be raised in one language, which naturally creates an asymmetry in
the speakers’ bilingual competence. Nonetheless, the linguistic dominance in one
language over the other can be dynamic and might change over the life span of the
speaker depending on the context or the speaker’s needs, as seen in Ramírez Trujillo
and Bruhn de Garavito’s chapter. In the studies included in this section, Spanish is
the language learnt later in life, as the majority language, usually in a naturalistic
environment, although sometimes also in instructional contexts since Spanish is
taught in the classroom. Our aim in distinguishing between Spanish acquisition in
a bilingual context versus a non-bilingual context (Section 2), is to properly define
 The importance of crosslinguistic comparison in the study of the acquisition of Spanish 

the unique linguistic reality of places such as the Basque Country and Peru, for
example, where the speakers included in the four chapters outlined below live in
communities where two languages enjoy varying levels of presence and prestige in
the realms of government, education, politics, society and family life. Two of the
studies were carried out in bilingual communities of Latin A ­ merica (Sánchez in
Quechua-Spanish bilingual speech communities in the Southern Andes of Peru and
Bolivia and from the region of Lamas in the Amazonian region of Peru; Ramírez
Trujillo and Bruhn de Garavito in a predominantly Nahuatl-speaking village in the
state of Tlaxcala, Mexico). The other two studies took place in bilingual regions
of Spain in which both languages have official status and the minority languages
have comparable prestige – albeit not presence – to that of the majority language
(Ezeizabarrena and Alegria in Basque-Spanish bilinguals in the Basque Country,
and Perpiñán in Catalan-Spanish bilinguals in Catalonia).
The existence of the two languages in the society and in the speaker’s mind
makes crosslinguistic interference a typical outcome of bilingual competence.
A case in point is Sánchez’s overview chapter, which goes through previously-­
published studies on Spanish-Quechua bilingualism and demonstrates the
­existence of crosslinguistic influence in the Spanish spoken by Quechua speakers.
Furthermore, she provides an explanation for the difficulties that these bilinguals
experience in terms of problems with the mapping of functional features onto
morphology. Building from the Feature Reassembly Hypothesis (Lardiere 1998,
2003, 2005), Sánchez proposes a fine-grained account to explain crosslinguistic
influence in Quechua-Spanish speakers by defining and teasing apart functional
interference from functional convergence. Sánchez surveys a series of studies
and linguistic phenomena, paying special attention to the linguistic phenomena
placed at the lexicon/morphology/syntax interface. She documents innovative
mappings between functional features and morphology such as the emergence of
non-argumental clitics (syntactic level), different uses of the tense, aspect and evi-
dentiality systems (syntax-morphology interface), changes in argument structure
(syntax-lexicon interface), and pervasive use of null objects with definite anteced-
ents (syntax-discourse interface). Sánchez concludes that these new patterns of
feature-morphology mapping force us to have a more nuanced view of language
modularity in bilinguals.
Next, Ramírez Trujillo and Bruhn de Garavito’s chapter investigates two types
of bilinguals, a group that has been exposed to both Nahuatl and Spanish since
birth (simultaneous bilinguals), and another group exposed to Spanish later in life,
between 6 and 20 years of age (mean 11.7 years old). Both groups were exposed to
the majority language naturalistically; nonetheless, their performance in Spanish
is not defined by the onset of acquisition, but rather by the degree of contact with
the majority language. Ramírez Trujillo and Bruhn de Garavito’s study showed no
 Tiffany Judy & Silvia Perpiñán

differences in subject-verb agreement or tense forms between the two groups of

bilinguals classified by their onset of acquisition of Spanish. However, the authors
note that the speakers’ performance is significantly different depending on the
actual contact with Spanish. Thus, if the speakers frequently travel or work outside
the village and, therefore, have extensive contact with Spanish, regardless of the
age they started being exposed to it, they demonstrate a more native-like accu-
racy than those that have less frequent contact with Spanish. Crucially, Ramírez
­Trujillo and Bruhn de Garavito argue that this difference is not only due to quan-
tity of input, but also to quality of input, since the speakers that work or travel
outside the village are exposed to the standard variety of Spanish. Another impor-
tant finding in this chapter regards the error patterns found in the speakers: both
bilingual groups produced the same type of errors, not explicable by L1 transfer,
but rather by the use of default forms such as non-finite verbal forms. Moreover,
the speakers did not make mistakes with clitic placement, showing full knowledge
of syntax at the same time that morphology was shown to remain incomplete.
The authors interpret these results as supporting evidence of the Missing Surface
Inflection Hypothesis (Prévost & White 2000). Both Sánchez’s, as well as Ramírez
Trujillo and Bruhn de Garavito’s studies highlight the importance of the strength
of the associations between functional features and morphological forms in the
bilinguals’ speech, particularly for the acquisition of morphology, the bottleneck
of L2 acquisition (Slabakova 2009).
With respect to the acquisition of phonology in a bilingual context,
­Ezeizabarrena and Alegria’s chapter investigates the emergence of codas, a well-
known milestone in phonological development, in the spontaneous speech of a
Basque-Spanish bilingual child (1;09–2;01). This study is of great import because
the authors show that the two phonological systems of the early bilingual, despite
the punctual presence of crosslinguistic interference, develop separately and
according to the developmental stages of each language. They also show that the
bilingual child produced more codas in the language with more frequent and/or
varied types of codas (i.e. Basque) compared to his production in Spanish, a lan-
guage with less frequent and fewer types of codas, corroborating previous findings
(­Borràs-Comes & Prieto 2013; Polo 2011). The type and frequency of segments
produced in each language were also shown to correlate with the frequency of
those segments in the target language (except for the production of sibilants in
Spanish), further demonstrating the independence of the two systems in the bilin-
gual mind. Finally, Ezeizabarrena and Alegria provide evidence towards the delay
of morphological codas in the children’s production, as previously observed (Lleó
2003). Overall, this study argues that the bilingual speech follows language specific
strategies and that bilingual children have high sensitivity to inter-linguistic dis-
tinctions, keeping separate representations for each phonological system.
 The importance of crosslinguistic comparison in the study of the acquisition of Spanish 

The last chapter of this section is devoted to the study of the Spanish grammar
of Catalan-dominant sequential bilinguals. By means of experimental method-
ology typically used for SLA research, Perpiñán’s chapter explores three micro-
parametric differences between Catalan and Spanish regarding the expression of
location and existential constructions. The speakers investigated therein iden-
tify themselves with, prefer and mostly use Catalan in their daily lives; still, their
­Spanish knowlege, as measured by a standard proficiency test, is indistinguish-
able from that of Spanish native speakers. However, when these Catalan-dominant
bilinguals are thoroughly tested in their production and grammatical intuitions
in Spanish, they display significant differences in their acceptance of the definite
effect in existential constructions, underuse of estar to express location, and over-
acceptance of ser to locate objects, as employed in Catalan grammar. These results
suggest that, even in the case of highly functional bilinguals, we may still find
crosslinguistic interference in certain linguistic domains, and significant differ-
ences in the implicit grammars of the bilinguals compared to that of monolinguals.
Furthermore, this chapter demonstrates that experimental SLA methodology is an
appropriate and fruitful approach to the study of bilingualism.
These studies on the acquisition of Spanish in a bilingual context show us,
on the one hand, that early bilinguals soon distinguish between the two linguistic
systems that they are acquiring simultaneously. On the other hand, they show that
an early onset of acquisition does not guarantee a native-like ultimate attainment,
downplaying the always-questioned factor of onset of acquisition and age of acqui-
sition. Finally, we would like to point out that the issue of the emergence of contact
varities in bilingual societies, and the type/quality of input that these bilinguals are
exposed to are two topics that are not discussed in detail here. Although it is an
extremely interesting topic and very relevant to the discussion and description of
bilingual grammars, it does not speak to the main goals of this volume.

2.  Spanish as an L2 in a non-bilingual society

Acquiring an L2 in a non-bilingual society can have a variety of different modali-

ties. For example, a child can acquire two or more languages at the same time
because her parents speak to her in two different languages (i.e. the one parent/
one language strategy), or her parents may have immigrated to a different country
but still speak the family language at home, while the societal, or majority lan-
guage is mostly used outside the home. This latter scenario is depicted in Arnaus
Gil and Müller in the first chapter of this section, which investigates the grammar
of bilingual and trilingual children with different language combinations and in
different countries. Another way of acquiring an L2 in a non-bilingual society is
 Tiffany Judy & Silvia Perpiñán

to study abroad for a certain period of time, but in reality, if the study abroad
session is not very long, this may more accurately be considered an instructional
environment. And finally, a more radical way in which a language can be acquired
is via rather continuous exposure to it through immigration to a country in which
that language is spoken, as in the cases that Judy, Bel and García Alcaraz, and
Donoso and Bylund portray. These cases share the characteristic that the speak-
ers were first exposed to the L2 in adulthood, which might render very dissimi-
lar outcomes, from fossilization to full convergence depending on the linguistic
property and the language pairing. Differently from the studies reported on in
Section 3, it is widely assumed, given certain conditions like integration with the
larger society, that speakers in a naturalistic context such as those described in
Judy, Bel and García Alcaraz and Donoso and Bylund, are exposed to L2 input
in high quantities, which is likely to be of high quality, too. Thus, research on
this acquisitional context is highly valuable since it allows researchers to explore if
and how L2 speakers’ acquisition potential is limited regardless of age and in the
absence of restrictions often present in instructional contexts (i.e. reduced quan-
tity and quality of L2 input). The following section explores some of these contexts
and presents a description of the variety of outcomes and situations that we find
in child bilingualism in societies where Spanish is not the predominant language,
and situations of adult SLA in Spanish-speaking societies.
The first chapter in this section describes the acquisition of two and three
languages very early in life. Arnaus Gil and Müller investigate the language of
simultaneous bilinguals and trilinguals in Germany, France, and Catalonia, Spain.
Specifically, they examine the acquisition of copular verbs combined with adjec-
tives in the verbal domain, and the position of adjectives (pre or post-nominal)
in the nominal domain, in bilingual (Spanish-German/ Spanish-French) and
trilingual children (Spanish-Catalan-German). Generally speaking, the authors
show that the acquisition of the copulas can also be a problematic area for bilin-
gual children, as it has been shown for L2 learners, especially with adjectives that
accept both copulas. Furthermore, Arnaus and Müller demonstrate crosslinguistic
­influence in the bilingual Spanish-German children, who have more difficulties
assigning permanent properties (ser) to the predicate than temporal properties
(estar). These and other similar findings lead the authors to propose that if the
target language provides the children with a complex and a less complex syn-
tactic derivation, and the other language the bi/trilingual is exposed to presents
only the less complex derivation, then the bi/trilingual children’s tendency is to
use the less complex syntactic derivation. If, on the other hand, the bi/trilingual
children is exposed to the two types of derivations (complex and simple) in the
two languages, then the children will be more likely to use the complex syn-
tactic derivation. Arnaus Gil and Müller’s chapter is quite relevant to the study
 The importance of crosslinguistic comparison in the study of the acquisition of Spanish 

of  ­multilingualism because it shows that the acquisition of complex derivations

can be expedited if children are exposed to several languages with these complex
constructions, and that trilingual speakers can be more successful than bilingual
speakers in the acquisition process.
Next, Judy’s chapter examines convergence on the discourse-constrained dis-
tribution of null and overt referential subject pronouns in Spanish by near-native,
L1 Farsi speakers. Data from both offline and online tasks examining Contrastive
Focus, Topic Shift and Topic Maintenance were collected and reported on in order
to test the Interface Hypothesis’ prediction (Sorace 2011, 2012) that even near-
native speakers of facilitative language pairings will evidence processing differ-
ences on external interface-conditioned properties. While evidence of native-like
knowledge and processing was found for some conditions, differences were also
found highlighting the need for further empirical investigation of the tenability of
these claims. Importantly, and in line with the goals of this volume, Judy’s chapter
contributes to the debate surrounding the IH’s claims by examining a facilitative
language pairing. That is, much work conducted on the distribution of subject pro-
nouns has focused on non-facilitative language pairings such as English-­Spanish
or French-Spanish, where transfer of L1 values does not immediately result in con-
vergence. Thus, if divergence in non-facilitative language pairings is found, it is
not possible to tease apart the effects of L1 transfer from the increased processing
burden experienced purportedly by bilinguals (IH). However, finding divergence
in speakers for whom L1 transfer is facilitative lends credence to the IH’s claims.
Without amassing research on both types of language combinations, it is not pos-
sible to comment on this prediction.
In a similar vein, Bel and García-Alcaraz examine interpretation of null and
overt third person subject pronouns in Spanish by early sequential and typical L2
classroom Moroccan Arabic speakers. The inclusion of a pre-test for anaphora res-
olution strategies in the participants’ L1 not only aided in the analysis of the results,
but also contributes new empirical data regarding the Position of ­Antecedent
Hypothesis (PAH, Carminati 2002) and microvariation. Bel and García-Alcaraz
found that the PAH is only partly applicable in Moroccan Arabic and that while
native speakers’ preferences with null subjects patterned with results found for
Spanish, Italian, and Catalan, preferences for overt subjects differed from ­Italian
and Catalan, instead patterning with Spanish preferences. Even so, a further dif-
ference, based on clause order, was found between Spanish and Moroccan ­Arabic
preferences, indicating a further level of variation. The lack of significant differ-
ences between group (early bilinguals, L2 speakers and native Spanish speakers)
on the Spanish experiments suggests that, in general, both bilingual groups have
converged on native-like anaphora resolution strategies in Spanish. However, a
difference in condition (subordinate-main order) where transfer of L1 anaphora
 Tiffany Judy & Silvia Perpiñán

resolution strategies would have proved facilitative but was not evidenced and
points to something highlighted in Iverson and Rothman: the roles of the L1 and
of L1 transfer is not yet clear and may likely function in conjunction with other
processes (e.g. feature rebundling).
Finally, Donoso and Bylund’s chapter investigates construal of goal-oriented
motion events in Swedish speakers of L2 Spanish. While 7 of the 17 experimen-
tal participants reported having received some instruction in Spanish, all par-
ticipants were living in Chile at the time of testing with a mean length of stay of 8
years. Thus, due to the length of time spent in a Spanish-speaking environment,
we have grouped this chapter in the non-bilingual society section. Regarding
the property, motion events are described by the authors as events that involve
“physical displacement whereby an entity occupies different spatial positions
at different temporal intervals” (p. 233). Languages differ with respect to how
they encode manner, path and endpoint, and Swedish and Spanish are no excep-
tion. While the bilinguals of Donoso and Bylund’s study used the same amount
of manner verbs as the Spanish control group, they tended to provide supple-
mentary manner information via the use of periphrastic constructions and they
produced more endpoints than the controls, which are both indicative of L1
transfer. These differences are in line with previous research showing difficulties
in L2 acquisition of such structures, even at advanced proficiency levels. Since
no research currently exists on this linguistic property for Swedish-­Spanish
bilinguals, this chapter contributes new data on a completely novel language
pairing. Additionally, Donoso and Bylund’s chapter is unique to the volume
since, unlike the majority of the chapters, it examines speakers’ knowledge of
information structure, which, when combined with research on purely formal
features, enriches the whole of our understanding about what L2 speakers must
master during acquisition.
As mentioned above, an advantage of studying naturalistic acquirers is
that one of the variables commonly thought to hinder L2 classroom learners’
­acquisition – the quantity and quality of the input – is eliminated, or, at min-
imum, reduced, thus providing researchers a window into the upper limits of
non-native language acquisition. For example, in Judy’s study, the adult near-
native Farsi-Spanish speakers demonstrated native-like processing of Topic Shift
tokens in a Self-Paced Reading task, a property for which other researchers have
found divergence in offline methods in classroom learners (see Sorace 2011,
2012 and references therein). Similarly, the studies by Bel and García-Alcaraz
and Donoso and Bylund respectively found high levels of convergence in their
naturalistic speakers. However, some (subtle) differences as compared to native
Spanish speakers were found in each of these three studies. The fact that length
and amount of exposure, which is quite difficult to accurately measure, vary from
 The importance of crosslinguistic comparison in the study of the acquisition of Spanish 

study to study may have some bearing on the results. This may especially be true
of Arnaus Gil and M ­ üller’s study on bi/trilingual children who had certainly
received less exposure to the target language at the time of testing. Nonetheless,
these studies provide a point of comparison that allows researchers to control for
variables that may negatively affect language acquisition by classroom learners,
the focus of the next section.

3.  Spanish as an L2 in an instructional context

The final section of this volume is devoted to learners exposed to explicit L2

instruction. While classroom learners are likely the most commonly studied
L2 population in the United States, this acquisition context has not consistently
proved itself to be neither maximally effective nor efficient in terms of learner
outcomes, perhaps as a result of the quantity and quality of the primary linguistic
data available to the learners. In fact, it has been shown, in a variety of languages
and linguistic properties, that classroom learners are not universally successful
in their convergence on the L2 (DeKeyser 2005; Long 2007). Instead, substan-
tially varying outcomes obtain within analogous or, at least, similar settings. This
is not to say, however, that no classroom learner is successful as much empirical
research in the generative paradigm has shown that even adult classroom learn-
ers can acquire sophisticated knowledge of the target language (see White 2003a,
2003b, 2008, 2011 for overview). This is true of some of the properties tested in the
following chapters. For example, Montrul and Gürel found that, even at low and
intermediate proficiency levels, learners demonstrated knowledge of the distribu-
tion of the differential object marker a in Spanish despite the fact that it is not very
noticeable in the input due to lack of phonological saliency and its polyfunctional
use. Still, the very lowest proficiency learners showed some divergence that was
evident of L1 transfer. Pinto and Guerra Rivera also showed that Dutch-Spanish
classroom learners make use of the appropriate copula verb with non-scalar grad-
able adjectives, but that their performance with a subset of the scalar gradable
adjectives, the irreversible ones that denote a property that cannot be reverted,
was target-deviant. These results mirror previous findings on the general difficulty
of acquiring the ser/estar distinction with adjectives (VanPatten 1985, 1987), par-
ticularly the acquisition of estar (VanPatten 2010), but at the same time diverge
from those previous findings in that this new language combination forces us to
be more precise in our linguistic predictions and in describing the selective role
of L1 transfer. We believe that the data showing convergence and the explanations
for any lack of convergence found in the classroom learners examined in the fol-
lowing four chapters contribute to the field’s understanding of the developmental
 Tiffany Judy & Silvia Perpiñán

stages of language acquisition and can serve as a point of comparison for studies
conducted on speakers that are exposed to the L2 in naturalistic environments like
bilingual Spanish-speaking societies as discussed in Section 1 or in immigration
and/or home language scenarios as described in Section 2.
The first of the chapters included in the classroom context section, that of
Iverson and Rothman, actually contains both naturalistic speakers in a non-­
bilingual society and typical classroom speakers as it is an overview of the avail-
able literature on object drop in Spanish in four language pairings mentioned
above: ­ Brazilian Portuguese-Spanish, European Portuguese-Spanish, English-
Spanish and Mandarin Chinese-Spanish. Since three of the four datasets com-
ment on classroom learners, we have grouped this chapter in the instructional
context section, but there is some crossover between this and the previous section.
In addition to bringing together four datasets, an important contribution in its
own right, ­Iverson and Rothman’s chapter speaks to one of the main goals of the
volume: through a careful analysis of datasets from various language pairings, it
demonstrates that considering data from only one language pairing reduces the
overall explanatory adequacy in that an explanation that may suffice for a particu-
lar language combination may not be generalizable to another. As a major aim of
modern generative linguistics is to provide universal (i.e. not language-specific)
theories of acquisition, this approach is highly welcomed. Although it may be diffi-
cult for researchers to examine all available datasets for other properties, especially
those that have received more empirical attention (e.g. null/overt subjects), this
chapter clearly demonstrates the value in doing so.
Similar to Iverson and Rothman’s chapter, the chapter by Montrul and Gürel
examines an object-related property, specifically knowledge and production of
Differential Object Marking (DOM) in native Turkish-speaking classroom learn-
ers of L3 Spanish (L2 English). DOM is evidenced in both Turkish and Spanish,
although the features implicated in each language differ from one another: ani-
macy and specificity are relevant for Spanish, while only specificity is relevant for
Turkish for the contexts tested for in the study. While they do not analyze data from
English-Spanish bilinguals, the results, overall, demonstrate that the developmen-
tal delays reported in the literature on English-Spanish learners are not evidenced
to the same degree in Turkish-Spanish bilinguals, even at low and intermediate
proficiency levels. Still, some of the lowest proficiency participants showed a more
Turkish-like constrained distribution of DOM in that some allowed DOM with
inanimate, specific direct objects. The authors interpret the whole of the results as
providing evidence for L1 transfer (Schwartz & Sprouse 1996) as well as for featur-
ing rebundling (Lardiere 2008, 2009). In addition to these theoretical contribu-
tions, this study is unique in that it is of the first to examine convergence on DOM
in Spanish by speakers of a language that also instantiates DOM.
 The importance of crosslinguistic comparison in the study of the acquisition of Spanish 

A classic topic investigated in instructional SLA is the role of the L1. As dis-
cussed above, the L1 can be both facilitative and non-facilitative, and its effects
can be long-lasting in the development of the L2. Pinto and Guerra Rivera’s
chapter highlights the importance of analyzing the two languages at hand in
depth so specific linguistic properties that subtly differ between the languages
and that can ultimately determine the successful acquisition process can be
localized and explained. Once these linguistic differences are understood, our
knowledge of the L2 acquisition process will increase since they may explain
apparent random mistakes. In particular, Pinto and Guerra Rivera explore the
acquisition of the ser/estar contrast with different types of adjectives, paying
special attention to the inherent semantic properties of the Spanish copular
verbs in comparison to the properties of Dutch copular verbs (zijn and worder).
Their study shows that the Spanish learners have problems with estar, but not
so much with ser, and that estar was not problematic with all types of adjectives,
only with those that denote scalar gradable properties, such as flaco ‘thin’ or
gordo ‘fat’. A further examination of the data offers an even more fine-grained
classification of the adjectives, which in turn reflects the copula choice in
Dutch. Dutch-speaking learners of Spanish encounter problems selecting ser/
estar with irreversible scalar gradable adjectives, such as viejo ‘old’, not revers-
ible, but not with an adjective such as gordo, which is reversible. The analysis of
errors demonstrates that the learners do not make random choices and do not
commit superficial mistakes. Instead, learners have access to inherent semantic
features in both languages and difficulties may arise due to some sort of mis-
match in very subtle semantic properties at the lexical level.
Like Sánchez’s chapter and that of Iverson and Rothman, the final chapter of
the volume is an overview by Liceras and Alba de la Fuente that explores bilinguals
of two typologically related languages – French and Spanish. Although typologi-
cally related, one aim of this chapter is to explore microparametric variation and
elucidate the difference between typological similarity and typological proximity,
which is sometimes conflated in the literature. The authors describe the differences
and similarities between French and Spanish for seven linguistic properties: noun
compounds, plural realization, clitic clusters, plural and case marking in quanti-
fiers, passives and subject realization. A common assumption in the field has been
that typological proximity, usually taken to be membership to the same language
family, is advantageous in language acquisition. Nonetheless, through analysis
of existing data on the seven linguistic properties mentioned above, ­Liceras and
Alba de la Fuente demonstrate that proximity does not equal similarity and that
speakers of typologically proximate yet typologically dissimilar languages may
experience acquisition difficulties akin to those experienced by L2 learners whose
L1 differs parametrically from the L2. A further contribution of the chapter is
 Tiffany Judy & Silvia Perpiñán

that, by systematically examining L1 French and L2 Spanish, the authors provide

microparametric-based explanations for structures that surface in L2 Spanish that
are not part of the L1 or the L2 grammar as well as L1-based explanations for
non-Spanish-like structures. As mentioned above, research of this type informs
our understanding of microparametric variation in conjunction with bilingualism
and SLA theory.
While not all instructed learners come to enjoy high levels of L2 proficiency,
their time in the classroom is quite well-spent. From the learners’ perspective,
knowledge of subtle grammatical properties can be obtained even in the begin-
ning stages of acquisition. For example, Montrul and Gürel showed that Turkish
learners of Spanish DOM demonstrated knowledge of the phonologically discreet
and polyfunctional marker a even at low proficiency levels. The same is true of the
English-speaking Spanish learners described in Iverson and Rothman’s chapter
(data from Bruhn de Garavito and Guijarro-Fuentes (2002)): even at intermediate
proficiency level, these classroom learners demonstrated knowledge of semantic
restrictions on definiteness and subjacency constraints in complex DP and sen-
tential CP islands. From the researchers’ perspective, studies of this acquisition
context can provide much insight into the stages and developmental timeline of
instructed learners, with the benefit that instructed learners usually form a rela-
tively homogenous group, an advantage when searching for generalizations.

4.  Concluding remarks

We expect that the collection of studies that this book offers, together in one vol-
ume, will be of future reference for the field of Spanish acquisition as a second lan-
guage and bi/multilingualism from a linguistic perspective and that it will inspire
more research of the kind exemplified herein. While the common denominator
of these studies is the acquisition of Spanish, the merit of the chapters and the
findings detailed therein can, and should, be applied to other language pairings
and acquisition contexts, such as L3 and Ln acquisition. The overview chapters
as well as those that advance original data underline the importance of accurate
descriptions of the languages under investigation, emphasizing the interconnec-
tion between linguistic theory and SLA theorizing. We believe that this useful
linguistic information can only be broadened by crosslinguistic comparisons, the
heart and soul of this book. Crosslinguistic comparisons allow us to define general
tendencies that go beyond transfer or superficial similarities. As we have seen in
several studies, a detailed and precise understanding of the L1 and the proper-
ties that must be acquired in the L2 help us to understand not only the learnabil-
ity tasks in each situation, but more generally speaking, they help us to describe
 The importance of crosslinguistic comparison in the study of the acquisition of Spanish 

the nature of interlanguage grammars, one of the main purposes of SLA research.
These fruitful and novel crosslinguistic comparisons have contributed to a better
understanding of the role of L1 transfer, the impact of typological (dis)similarity,
the effect of age of onset of acquisition, type of linguistic environment and the
function of the quantity and quality of input.
At the same time, several of these studies have called attention to the need
and value in looking beyond the traditional idea of parameter (re)setting in SLA,
advocating for a more nuanced view of the linguistic system and the modular-
ity of language, particularly in L2 learners and bilingual speakers. Whereas the
idea of principles and parameters (Chomsky 1981) revolutionized the field of
contemporary linguistics, and tremendously advanced the description of gen-
eralizations among the languages of the world, several studies in this collection
have shown that an explanation based solely on parameters might not be suf-
ficiently adequate to capture the subtle difficulties that L2 speakers display when
acquiring a language. Therefore, the study of interlanguage grammars pushes us
to resort to the description of microparameters and more particularly, to a more
atomic view of the system in which features and feature bundles mapped onto
lexical items make up the skeleton of language, much in line with the new tenets
of the Minimalist linguistic framework (Chomsky 1995, 2004). In this sense, the
study of L2 acquisition and bilingualism also serves to move general linguis-
tic theory forward. The danger here is getting lost in itemized, specific descrip-
tions hardly generalizable that lead us to inconclusive and irrelevant deductions;
hence, we need to find the middle point between explanatory adequacy and
­generalizable findings.
The collection of studies included here has allowed us to gain ground on our
understanding of L1 transfer in the sense that several studies have pointed out that
it is selective and by no means random. L2 speakers create coherent interlanguage
grammars that are not always copies of the L1 grammar, although they might
not be target-like yet; moreover, these developing grammars are possible, natural
grammars constrained by Universal Grammar. Another matter that this compen-
dium of articles has made clearer is the effect of linguistic typology in the acquisi-
tion process. We have seen that proximity between languages does not necessarily
imply ease of acquisition or convergence, and that, in fact, typological proximity
with minimal differences between the languages may delay development. Finally,
some studies have addressed the issue of onset of acquisition, particularly the stud-
ies included in Section 1, which focus on the acquisition of Spanish in a bilingual
society, providing evidence that a very early age of onset of acquisition does not
guarantee native-like ultimate attainment; but at the same time, other studies have
demonstrated that convergence on difficult linguistic aspects is still possible in
adult L2 speakers, likely owing to constant exposure to the target language. Thus,
 Tiffany Judy & Silvia Perpiñán

variability in acquisitional outcomes remains a core characteristic of SLA and bi/

multilingualism.
In summary, this volume took as its central goal the expansion of the lan-
guage pairings combined with Spanish in generative language acquisition
research as we believe much is to be learned from a wider and more representa-
tive body of research. The results of the 12 contributing chapters, individually
and collectively, speak to the usefulness of examining less commonly-studied
language pairings in a variety of contexts and we hope that this volume will
serve as a reference and motivating force for continued empirical investigations
of this sort.

Acknowledgments

Of course, this volume would not have been possible without the contribution
of the reviewers, listed in alphabetical order: Mohammad Alhawary, Laia Arnaus
Gil, Anna Babel, Aurora Bel, Joan Borràs, Josep Maria Brucart, Joyce Bruhn de
­Garavito, Alejandro Cuza, María del Pilar García Mayo, Pedro Guijarro-­Fuentes,
Alberto Hijazo-Gascón, Pilar Larrañaga, Conxita Lleó, Karen Miller, Antje
Muntemdam, Öner Özçelik, Maria Carme Parafito Couto, Claire Renaud, Ana
Lucia Santos, Cristina Schmitt, Ludovica Serratrice, Roumyana Slabakova, Julio
Villa-García, Christiane von Stutterheim, and two reviewers who wish to remain
anonymous. Their comments and insights have greatly shaped each of the chapters
and we are very appreciative of their efforts. We are also indebted to our editorial
assistant, Ansley Byers, for her hard work and precision, especially in the final
moments of editing. Lastly, we are grateful to Jason Rothman and Cornelis H. J.
Vaes for their expert guidance and support.

References

Borràs-Comes, J., & Prieto, P. (2013). The acquisition of coda consonants by Catalan and S­ panish
children: Effects of prominence and frequency of exposure. Probus, 25(1), 1–24.
DOI: 10.1515/probus-2013-0001
Bruhn de Garavito, J., & Guijarro-Fuentes, P. (2002). L2 acquisition of indefinite object drop
in Spanish. In J. Costa & M.J. Freitas (Eds.), GALA 2001 Proceedings (pp. 60–67). Lisbon:
Associação Portuguesa de Linguística.
Carminati, M.N. (2002). The Processing of Italian Subject Pronouns. Unpublished Ph.D. dis-
sertation. University of Massachusetts at Amherst.
Chomsky, N. (1981). Lectures on Government and Binding. Dordrecht: Foris.
DOI: 10.2307/2273965
 The importance of crosslinguistic comparison in the study of the acquisition of Spanish 

Chomsky, N. (1995). Minimalist inquiries: The framework. MIT Occasional Papers in Linguis-
tics, 15, 1–56.
Chomsky, N. (2004). Beyond explanatory adequacy. In A. Belletti (Ed.), Structures and Beyond
(pp. 104–131). Oxford: OUP.
DeKeyser, R. (2005). What makes second-language grammar difficult? A review of issues. Lan-
guage Learning, 55(1), 1–25. DOI: 10.1111/j.0023-8333.2005.00294.x
Lardiere, D. (1998). Dissociating syntax from morphology in a divergent L2 end-state grammar.
Second Language Research, 14, 359–375. DOI: 10.1191/026765898672500216
Lardiere, D. (2003). Second language knowledge of [±Past] vs. [±Finite]. In J. M. Liceras,
H. Zobl, & H. Goodluck. (Eds.), Proceedings of the 6th Generative Approaches to Second
Language Acquisition Conference (GASLA 2002) (pp. 176–189). Somerville, MA: Cascadilla
Proceedings Project.
Lardiere, D. (2005). On morphological competence. In L. Dekydtspotter, R. Sprouse, &
A.  Liljestrand (Eds.), Proceedings of the 7th Generative Approaches to Second Language
Acquisition Conference (GASLA 2004), (pp. 178–192). Somerville, MA: Cascadilla Pro-
ceedings Project.
Lardiere, D. (2008). Feature-assembly in second language acquisition. In J. Liceras, H. Zobl, &
H.  Goodluck (Eds.), The Role of Formal Features in Second Language Acquisition
(pp. 106–140). Mahwah, NJ: Lawrence Erlbaum Associates.
Lardiere, D. (2009). Some thoughts on the contrastive analysis of features in second language
acquisition. Second Language Research, 25, 173–227. DOI: 10.1177/0267658308100283
Lleó, C. (2003). Prosodic licensing of coda in the acquisition of Spanish. Probus, 15, 257–281.
DOI: 10.1515/prbs.2003.010
Long, M.H. (2007). Problems in SLA. Mahwah, NJ: Lawrence Erlbaum Associates.
DOI: 10.1080/09500780802152622
Polo, N. (2011). Notas sobre la adquisición fonológica bilingüe del español: El desarrollo de las
codas en español como lengua materna en hablantes bilingües del País Vasco. Oihenart,
26, 399–421.
Prévost, P., & White, L. (2000). Missing surface inflection or impairment in second language
acquisition? Evidence from tense and agreement. Second Language Research, 16, 103–133.
Schmitt, C.J. (1997). Aspect and the Syntax of Noun Phrases. Unpublished Ph.D. dissertation.
University of Maryland, College Park.
Schmitt, C.J. (2005). Semi-copulas: Event and aspectual composition. In P. Kempchinsky
& R.  ­Slabakova (Eds.), Syntax, Semantics and the Acquisition of Aspect (pp. 121–145).
New York, NY: Springer.
Schwartz, B.D., & Sprouse, R.A. (1996). L2 cognitive states and the full transfer/full access
model. Second Language Research, 12, 40–72. DOI: 10.1177/026765839601200103
Slabakova, R. (2009). What is easy and what is hard to acquire in a second language? In
M. Bowles, T. Ionin, S. Montrul, & A. Tremblay (Eds.), Proceedings of the 10th Generative
Approaches to Second Language Acquisition Conference (GASLA 2009) (pp. 280–294).
Sorace, A. (2011). Pinning down the concept of “interface” in bilingualism. Linguistic Approaches
to Bilingualism, 1(1), 1–33. DOI: 10.1075/lab.1.1.01sor
Sorace, A. (2012). Pinning down the concept of “interface” in bilingualism: A reply to peer com-
mentaries. Linguistic Approaches to Bilingualism, 2, 209–216.
DOI: 10.1075/lab.2.2.04sor
 Tiffany Judy & Silvia Perpiñán

Trahey, M., & White, L. (1993). Positive evidence and preemption in the second language class-
room. Studies in Second Language Acquisition, 15, 181–204.
VanPatten, B. (1985). The acquisition of ser and estar by adult learners of Spanish: A preliminary
investigation of transitional stages of competence. Hispania, 68, 399–406.
DOI: 10.2307/342218
VanPatten, B. (1987). Classroom learners’ acquisition of ser and estar: Accounting for develop-
mental patterns. In B. VanPatten, T.R. Dvorak, & J.F. Lee (Eds.), Foreign Language ­Teaching:
A Research Perspective (pp. 61–75). Cambridge, MA: Newbury House.
VanPatten, B. (2010). Some verbs are more perfect than others: Why learners have difficulty
with ser and estar and what it means for instruction. Hispania, 93(1), 29–38.
White, L. (2003a). Second Language Acquisition and Universal Grammar. Cambridge: CUP.
DOI: 10.1017/cbo9780511815065.002
White, L. (2003b). On the nature of interlanguage representation: Universal Grammar in second
language. In C. Doughty & M. Long (Eds.), The Handbook of Second Language Acquisition
(pp. 19–42). Oxford: Blackwell. DOI: 10.1002/9780470756492.ch2
White, L. (2008). Different? Yes. Fundamentally? No. Definiteness effects in the L2 English of
Mandarin speakers. In R. Slabakova, J. Rothman, P. Kempchinsky, & E. Gavruseva (Eds.),
Proceedings of the 9th Generative Approaches to Second Language Acquisition Conference
(pp. 251–261). Somerville, MA: Cascadilla Proceedings Project.
White, L. (2011). Second language acquisition at the interfaces. Lingua, 121, 577–590.
DOI: 10.1016/j.lingua.2010.05.005
part i

Spanish as the L2 in a bilingual society

 Crosslinguistic influences in the mapping
of functional features in Quechua-Spanish
Bilingualism

Liliana Sánchez
Rutgers University

This paper discusses evidence of crosslinguistic influence that involves the

emergence of new patterns of feature-morphology mapping in Spanish as spoken
by native speakers of Quechua. It builds on the notions of functional interference
and functional convergence (Sánchez 2003, 2004) and of feature reassembly
(Lardiere 2003, 2005) in order to account for crosslinguistic influence at: (a) the
syntactic level (feature reassembly and the emergence of non-argumental clitics)
(b) the syntax/morphology interface (feature reassembly) (c) syntax/lexicon
interface (feature reassembly and changes in argument structure), and (d) the
syntax/pragmatics interface (licensing of null objects and focus fronting). The
evidence shows that activation of features, feature reassembly and mapping onto
morphology is an important source of crosslinguistic influence.

1.  Introduction

The understanding of second language acquisition and bilingualism from a

generative perspective has been characterized in the last decades by multiple
approaches. They have ranged from earlier studies that provided strong empiri-
cal support to the availability of universal language acquisition mechanisms and
the autonomy of syntax (Flynn 1987; Liceras 2010; Schwartz & Sprouse 1996;
White 1989, 2003) to studies that emphasize the role of the interfaces between
language components (Montrul 2010; Sorace 2000, 2005; Sorace & Serratrice
2009; White 2011). These interfaces have become increasingly acknowleged as
crucially involved in the development of the grammatical representation of adult
and child sequential bilinguals (Pladevall 2010; Serratrice, Sorace & Paoli 2004;
Sorace & Serratrice 2009).
In the study of bilingual first language acquisition there have also been devel-
opments that have moved research beyond the autonomy of syntax (Meisel 1997;

doi 10.1075/ihll.3.02san
© 2015 John Benjamins Publishing Company
 Liliana Sánchez

Paradis & Genesee 1996) and have emphasized the role of crosslinguistic influ-
ence at the interfaces between syntax and other components such as the pragmatic
component (Hopp 2009; Hulk & Müller 2000; Müller & Hulk 2001).
In that respect, there has been a confluence of evidence coming from adult
and child L2 language acquisition as well as from bilingual first acquisition data
towards a more complex view of the relationship between the computational and
the interpretive components (Chomsky 2000). An example of this is the evidence
of the acquisition of the syntactic licensing of null subjects by L2 learners that may
show some residual instability in the distribution of null vs. overt subjects in dis-
course (Sorace 2000). Furthermore, recent developments in L2 processing studies
have also highlighted the need to understand the relationship between activation
of the lexicon and syntactic development. For instance, Hopp (2012) has found
evidence that mastery of lexical gender, namely the correct assignment of gender
values to nouns, is correlated with mastery of gender agreement.
These findings point out at the relationship between syntax, discourse-level
properties, and the lexicon. They bring us closer to a more articulated view of
how different types of linguistic knowledge are stored and accessed in the bilin-
gual mind. This view is compatible with a more nuanced approach to modularity
of language (Jackendoff 1997, 2002). It is also compatible with the exploration
of the interaction across components as part of the study of bilingual develop-
ment (­Ullman 2005). This perspective takes into account the separate and dis-
tinct nature of lexical and syntactic knowledge. At the same time, it recognizes
that interactions across components (Jackendoff 1997; Slabakova 2008) take place
when language representations and lexical items are activated for production and
comprehension purposes. The evidence from bilingual data represents an invalu-
able window into how these interactions take place. Such evidence can not be eas-
ily obtained from monolingual data.
In this context, the need to study bilingual and second language acquisition
in naturalistic contexts (Meisel 2008; Serratrice, Sorace & Paoli 2004) has become
more pressing since they are a source of evidence of the interactions between the
narrow syntax and other language modules such as the lexicon and the interpre-
tive component. These interactions cannot be easily observed in monolinguals
because they do not exhibit differences in proficiency across components. They are
also difficult to observe in early acquirers in instructional contexts because activa-
tion of some components does not reach enough frequency to observe such inter-
actions. The study of bilingual and second language acquisition in children and
adults who live in language contact situations allows us to better observe the pro-
cesses involved in the development of a grammatical representation in ­situations
in which input and processing for comprehension and production are more fre-
quent than in instructional contexts.
 Crosslinguistic influences in Quechua-Spanish bilingualism 

In this paper, I will focus on some of the major findings and contributions
that the study of Quechua-Spanish bilingualism from a generative perspective
has brought to the field of second language acquisition studies. In this paper,
I adopt a perspective that assumes modularity and interactions across components
(­Jackendoff 1997). The contributions include evidence of crosslinguistic influence
in the Spanish of Quechua speakers: (a) at the core syntax in the projection of
non-argumental clitics (Camacho, Paredes & Sánchez 1995; Escobar 1991; Kalt
2000, 2002, 2012b; Sánchez 2003) (b) at the lexico-morphosyntactic interface in
the mapping of functional features onto overt morphemes or independent words
(Kalt 2009; Sánchez 2004); and (c) at the syntax-pragmatics interface in the inter-
action between syntactic position, morphology and the interpretation of topic and
focus (Camacho, Paredes & Sánchez 1997; Muntendam 2013).
The overall picture that emerges from the study of the Spanish acquired by
Quechua speakers is one in which divergences from the Spanish input have as
their basic source the activation of Quechua functional features in association
with Spanish lexical items (Sánchez 2003, 2004). For instance, as we will see in the
paper in more detail, new associations between Spanish past tenses and reportative
evidentiality emerge in Spanish verbs such as the use of the pluperfect past tense
to convey hearsay information (Sánchez 2004). Cases such as this one involve a
process of reassembly (Lardiere 1998, 2003, 2005) in which Quechua evidentiality
feaures are mapped onto the Spanish pluperfect morphology.
The paper is organized as follows: Section 2 presents an overview of the main
concepts and of the perspective adopted in this paper; Section 3 presents the rel-
evant properties of Quechua and Spanish languages; and Section 4 presents previ-
ous studies that show evidence of crosslinguistic influence related to functional
features, involving new patterns of morphological mapping of non-argumental
clitics, functional interference and reassembly of aspectual features, and new pat-
terns of mapping onto L2 morphology. It also presents evidence of convergence at
the syntax/discourse interface.

2.  C
 rosslinguistic influence in Quechua-Spanish Bilingualism:
Feature reassembly, functional convergence, and the lexicon

In parallel fashion to a better understanding of the role that interfaces play in second
language acquisition and bilingual development, our knowledge of the relationship
between declarative and procedural memory and its role in second language acqui-
sition has also developed (Gürel 2004; Paradis 1993, 2009; Ullman 2001, 2005).
In particular, we now know that the lexicon is stored in declarative memory and
that grammatical aspects of language are stored in procedural memory. We also
 Liliana Sánchez

know that both memory systems interact (Ullman 2005). The same is true of the
role of access to the L1 lexicon by L2 learners (Kroll & Stewart 1994; Schwartz &
Kroll 2006). We know that access to items in the L2 lexicon can be characterized
by suppression of their corresponding lexical equivalents in the L1. There is also
current research that assumes non-selective access to the lexicon in each of the lan-
guages spoken by a bilingual individual but focuses on syntactic cues as predictors
of crosslinguistic activation of the lexicon, namely, the possibility of higher activa-
tion of the L1 lexicon where some level of syntactic equivalence exists between the
L1 and the L2 (Marull 2013). In this respect, constant activation of lexical items in
a language contact situation allows us to observe phenomena not easily observed in
instructional contexts, among them: the emergence of new mappings of syntactic
features onto morphology (Putnam & Sánchez 2013) as well as the mapping of
syntactic operations triggered by those features that are grammaticalized in one
language but not in the other and are accessible to the interpretive component such
as evidentiality or discourse topic. This is so because language contact situations
involve a wider range of interactive contexts.
I assume that in addition to the basic syntactic operations (Merge, Agree, and
Move), the syntactic component requires access to functional features which are
syntactic minimal elements stored in the speaker’s mind in association with some
areas of the lexicon (Chomsky 2000). These features may or may not be mapped
onto specific overt morphological units in a one-to-one relationship or in a one-
to-many syncretic one (Giorgi & Pianesi 1997). Some of these features have a role
in the interface between the lexicon and syntax (aspectual features for example)
whereas other do not (EPP features). In this paper, I focus on the former.
In terms of how crosslinguistic influence reflects the complex relationship
between language representation and access to different language components in
second language acquisition, studies on Quechua-Spanish bilingualism have pro-
vided evidence of long-term sharing of feature mapping onto morphology between
the two languages. The acquisition of Spanish as a second language by Quechua
speakers usually takes place in naturalistic contexts as well as in instructional con-
texts.1 It supposes frequent activation of the L2 lexicon for c­ omprehension and
production purposes. The nature of L2 acquisition in language contact situations

.  The examples presented in this article come from previous research on bilingualism in
Quechua and Spanish among bilingual speech communities in the Southern Andes of Peru
and Bolivia and from the region of Lamas in the Amazonian region of Peru. Most of the
Quechua examples belong to Southern Quechua varieties especially from Cuzco Quechua as
described in Cerrón-Palomino (1987) and Sánchez (2010). The examples in Spanish are from
non-contact General Spanish unless they come from a specific study in Quechua-Spanish
bilingualism.
 Crosslinguistic influences in Quechua-Spanish bilingualism 

allows us then to address issues such as the activation of lexical and functional
elements in less asymmetrical conditions than instructional contexts (Kroll &
Bogulski 2013; Kroll & Stewart 1994; Kroll, Van Hell, Tokowicz & Green 2010).
In this perspective, previously proposed notions such as the need for reassembly
of functional features (Lardiere 2005) and the development of new patterns of
feature-morphology mapping due to crosslinguistic influence (Sánchez 2003) can
be viewed as stemming from the interaction in the activation of functional fea-
tures and their morphological mapping. The difficulties found in L2 feature reas-
sembly, expressed in the lack of target-like acquisition of morphology (­Lardiere
1998, 2003, 2005), may stem from weak mappings between L2 functional features
and morphological forms. Strong associations between L1 and L2 features and
their mapping onto L2 morphology may also result in grammars divergent from
the input.
In fact, difficulties in feature-morphology mapping in bilinguals have been
noticed by Sánchez (2003). According to Sánchez (2003), there are two processes
involving functional feature activation in bilinguals: functional interference and
functional convergence. Functional interference takes place when: “the activation
of functional features in one language, triggered by input in the other language,
generates syntactic changes in the bilingual grammars.” (Sánchez 2003, p. 13). One
case is the activation of evidentiality features in the use of the pluperfect tense
in order to convey hearsay information in Spanish (Sánchez 2004).2 In this case
rather than lacking the mapping of a particular feature onto an overt morphologi-
cal form, what we see is the mapping of a feature grammaticalized in one language
onto the morphological patterns of the other language.
Functional convergence, on the other hand, is defined as the specification of
a common set of features shared by the equivalent functional categories in the
two languages spoken by a bilingual (Sánchez 2003, p. 15). According to Sánchez
(2003), functional convergence takes place when a set of features not activated in
language A is frequently activated by input in language B in the bilingual mind.
This can result in a fusion of functional features from the two languages. Both
Lardiere’s (2003) and Sánchez’s (2003) proposals hinge on the acquisition of new
patterns of morphological competence by the bilingual learner.
In the next sections, I will assess the evidence for crosslinguistic influence
coming from the acquisition of Spanish by Quechua speakers, in order to show
how these different notions (the availability of feature reassembly, functional
interference and functional convergence) can be better understood as phenomena
related to the activation of lexical items and functional features and to the ­mapping

.  This particular phenomenon will be discused in detail in the next section of the paper.
 Liliana Sánchez

of these features onto morphology. In doing so my purpose is to highlight the need

for more detailed study of the lexicon/morphology/syntax interface, namely the
activation of lexical items and their morphological make-up in connection with
functional features relevant to the lexicon and to some syntactic operations such
as Merge or Agree. This is especially needed for language pairs with significant
differences in morphological make-up since some of the new patterns that emerge
have important consequences for how syntax develops and evolves in the bilingual
mind, as we will see in Section 3.
I will focus especially on studies that have looked at L2 characteristics stem-
ming from differences in the association of lexical items, functional features and
morphological mapping between the two languages. I will pay special attention to
those cases in which functional interference triggers feature reassembly and may
result in convergence, understood as the emergence of new mappings between
features and morphemes that are shared by both languages.
One basic tenet of minimalism is that at the level of the core syntax, defined
as the basic operational system, operations such as Merge, Agree and Move are
universally available (Chomsky 2000).3 In that respect these operations should be
available to all Quechua-Spanish bilinguals. However, the way in which Merge
generates basic constituents according to the argument structure of a verb differs
in both language families, especially in regard to the mapping of non-subject per-
son and number features onto morphology. For instance, one of the main charac-
teristics of second language Spanish as spoken by Quechua speakers has been the
great variety of phenomena that affect the distribution of direct, indirect and non-
argument clitics (Camacho & Sánchez 2002; Camacho, Paredes & Sánchez 1995,
1997; Escobar 1991; Kalt 2009, a.o.; Sánchez 2003). These and other phenomena
will be presented in more detail in the next section.

3.  M
 ain morphosyntactic and syntactic characteristics
of Quechua and Spanish

Some of the main syntactic and morphological characteristics of Quechua and

Spanish are partially shared. Both Quechua and Spanish are null subject languages
with subject agreement marking on the verb, as exemplified in the following
sentences:

.  The discussion of the data in this paper will not center around constraints on structures
or on general principles of economy, as they are not relevant to the data.
 Crosslinguistic influences in Quechua-Spanish bilingualism 

Quechua
(1) Papa-ta mikhu-n-mi.
Potato-acc eat-prog-3.s
‘S/He is eating (a) potato(-es).’
(Sánchez 2010, p. 24)
Spanish
(2) Com-e papa-s.
Eat-3.sg potato-pl
‘S/He eats potatoes.’

As these examples show, both languages share morphological marking of subject

person and number on the verb. This means that in both languages acquisition
of subject morphology is relevant to the interpretation of the null subject and is
salient because it is not simply a phonological form (PF) phenomenon but part of
the numeration (Liceras, Fernández & Alba 2011). This commonality implies that
for Quechua-Spanish bilinguals the mapping of features such as subject, person
and number onto morphology would not represent an area in which a reassembly
of functional features would have to take place.
There are however important differences in the way in which direct and indi-
rect objects are mapped onto morphemes in both languages. Quechua languages
are agglutinative SOV languages characterized by roots and suffixes and with
a neutral word order in which indirect objects precede direct objects (Cerrón-­
Palomino 1987). Spanish varieties are SVO languages with affixes and phono-
logically independent ‘functional’ words and with a neutral word order in which
direct objects may or may not precede indirect objects. Some ditransitive verbs
are usually marked with a third person dative clitic that is also marked for features
such as definiteness or specificity. This is shown in:
Quechua
(3) Tayta churi-n-man wasi-ta qu-n.
Father son-3.sg-dat house-acc give-3.sg
‘The father gives the house to his son.’
(Sánchez 2010, p. 13)

Spanish
(4) a. El padre le da la casa a su hijo.
The father cl give the house to his son
b. El padre le da a su hijo la casa.
The father cl give to his son the house
‘The father gives his son the house.’
 Liliana Sánchez

As (4a) and (4b) show, Spanish marks indirect objects with the phonologically
independent preposition a ‘to’ that has been argued to mark definiteness and/
or specificity similar to the clitic specification (Bleam 1999; Leonetti 2004). In
terms of morphological marking of arguments on the verb, Quechua varieties and
­Spanish also differ greatly. Whereas in Quechua third person direct objects are not
marked as morphemes on transitive verbs and they may be null, in Spanish they
are marked as clitics if their referent is definite and/or specific:
Cuzco Quechua
(5) a. Huwan-ta riku-rqa-nki-chu?
Huwan-acc see-pst.att-2.sg-int
‘Did you see Huwan?’
b. Arí, riku-rqa-ni.
Yes, see-pst.att-1.sg
‘Yes, I saw (him).’
Spanish
(6) a. ¿Vi-ste a Juan?
   See-pst.2.sg.perf to Juan
‘Did you see Juan?’
b. Sí, lo- v–i.
Yes, 3.sg.acc see-pst.1.sg.perf
‘Yes, I saw him.’

There is, however, some direct, indirect and oblique object marking on the verb in
Quechua involving first and second persons that does not involve features such as
definiteness or specificity. Quechua also allows for overt pronouns as independent
words marked with case:
(7) Pusa-wa-sqa
bring-1.sg obj- past.rep evid
‘S/he brought me.’
(Cusihuamán 2001, p. 163)
(8) Mariya-m (pay-ta) riku-n.
Mariya-foc/evid (s/he-acc) see-3.sg
‘Mariya sees (him/her).’
(Sánchez 2010, p. 27)
Notice also that there are significant differences in the way in which tense and
person features are mapped onto morphemes in both languages. In Examples (5a)
and (5b) the morpheme -rqa encodes past tense and attested evidentiality and the
morphemes -nki and -ni encode person and number whereas in Spanish the mor-
pheme -ste encodes tense, aspect, person, and number.
 Crosslinguistic influences in Quechua-Spanish bilingualism 

In addition to differences with respect to argumental morphological marking

on the verb, Quechua and Spanish differ with respect to the relevance of deri-
vational morphology to syntax. As noted by Cerrón-Palomino (1987), Quechua
languages have a complex morphological system that includes derivational and
inflectional morphemes. The basic sequence of affixation proposed by Cerrón-
Palomino (1987) in Quechua languages is:
(9) Mikhu-naya-n-mi
Eat-des-3.sg-foc.evid
‘S/he is about to eat.’
(10) Root + Derivational morpheme+Inflectional morpheme+Sentence level
morpheme
(Cerrón-Palomino 1987, p. 267)

While it is traditionally the case to think of derivational morphemes as involv-

ing only the generation of new lexical items and of inflectional morphemes as
being related to the syntax, in Quechua the separation is not as clear cut. Van
de Kerke (1996) notes that in Quechua, there are at least six derivational suffixes
that change the case or theta grid of the verb and therefore have syntactic con-
sequences. These are: causative -chi, assistive -ysi, reciprocal -na, reflexive -ku,
bi-locational -mu, benefactive -pu, and desiderative -naya. An example of how
derivational structure may affect argument structure is shown in the following
sentences by the addition of the desiderative/imminent suffix -naya in (12) to the
root mikhu- ‘eat’ in (11):
(11) Mikhu-n.
Eat-3.sg
‘(S/he) eats.’
(12) Mikhu-naya-n.
Eat-des-3.sg
‘(S/he) wants/is about to eat.’
 (Based on Cerrón-Palomino 1987)

In its desiderative meaning, the suffix -naya introduces the agent of the desire
as an addition to the agent of the eating but in its imminent meaning it does not
introduce a new agent. The addition of an agent in the theta grid is not reflected
at the phonological word level in Quechua given that (12) is a single word. The
equivalent expression in Spanish has two independent verbs (each one an inde-
pendent phonological word):
(13) María quie-re com-er.
Maria want-3.sg eat-inf
‘Maria wants to eat.’
 Liliana Sánchez

As will be shown in the next section, this difference is relevant to the understand-
ing of crosslinguistic influence in feature-morpheme mapping and the projection
of argument structure in Quechua-Spanish bilinguals.
Another important area in which the boundaries between derivational mor-
phemes restricted to the lexicon and inflectional morphemes restricted to syntax
are blurred in Quechua is aspect. As noted by Hintz (2011), there is a continuum
in Quechua of aspectual derivational suffixes that modify the meaning of a verbal
root and of aspectual inflectional suffixes. Both systems interact in many Quechua
varieties. An example of such interaction can be found in the derivational mor-
pheme -ru that indicates perfectivity as part of the lexical meaning of the verb but
doubles in this expression as a past tense marker:
(14) Chay allqu-cha-kuna-pis phiña-ru-ku-n-ña
That dog-dim-pl-add upset-perf-refl-3.sg-dis
chay chiku-cha-n-ta.4


that boy-dim-3.sg.poss-acc
‘Those dogs too had already upset that young boy.’

A third type of important crosslinguistic differences in the mapping of features

onto morphemes comes from left dislocated elements. In this case, the informa-
tional structure of the sentence is also involved. Both Quechua and Spanish variet-
ies exhibit topic and focus fronting in sentences with non-canonical word orders
(answers to questions about objects or sentences with a reintroduction of a previ-
ously mentioned topic). In Quechua, the dislocated element is morphologically
marked (Cerrón-Palomino 1987; Muysken 1995; Sánchez 2010), but there is no
morphological marking on the verb. In Spanish, left dislocated definite topics are
not marked themselves but there are clitics marked for case, person and number
on the verb as shown in (16):
Quechua
(15) [Wasi-ta-qa] Pirdu-m ruwa-rqa-n.
   House-acc-top Pirdu-direv.foc build-pst-3.sg
‘The house, Pirdu built (it).’
(Sánchez 2010, p. 31)
Spanish
(16) [La casa], Pedro la- hizo.
   The house Pedro acc.3.sg made
‘As for the house, Pedro made it.’

.  This example comes from a study for which data were collected using a picture-based
story task in Cuzco in 2005 currently in preparation. The speaker is an adult female.
 Crosslinguistic influences in Quechua-Spanish bilingualism 

With respect to left dislocated focalized constituents, in Quechua they are mor-
phologically marked, but they are not in Spanish and it is possible that they may
occur without a sentence-internal clitic.
Quechua
(17) [Kustal-ta-m] qunqa-rqu-n chay chufir.
  Package-acc-foc/evid forget-perf-3S that driver
‘It was the package that that driver forgot.”
(Sánchez 2010, p. 30)
Spanish
(18) El paquete olvidó el chofer.
The package forgot the driver
‘It was the package that the driver forgot.’

As we will see in the next section, crosslinguistic influence at the syntax/informa-

tional structure interface might affect some aspects of Merge but it may not neces-
sarily affect structural restrictions.

4.  F
 unctional interference, feature re-assembly and functional
convergence: Evidence of crosslinguistic influence
in Quechua-Spanish bilingualism studies

4.1  C
 rosslinguistic influence at the syntactic level: Functional interference,
feature re-assembly and the emergence of non-argumental clitics
One of the sources of evidence for functional interference that results in new pat-
terns of L2 mapping onto morphology is the emergence of new non-argumental
clitics in Quechua-Spanish bilinguals. As noted by Camacho, Paredes and S­ ánchez
(1995) and Kalt (2009), a possessor clitic has been found at different stages of
acquisition of Spanish by Quechua speakers from Southern Peru, as shown in
Example (19):
(19) [Lo]i- amarran su pata [del condor]i
   3.sg tie his leg    of-the condor
como si estuviera montando.


as if was riding’
‘They tie the condor’s leg as if it was riding.’
(Camacho, Paredes & Sánchez 1995, p. 135)

In this sentence the clitic is coindexed with the genitive expression del condor ‘of
the condor’ and not with the whole direct object su pata del condor ‘the condor’s
leg’ as it would be expected if this were a case of direct object clitic doubling. This
 Liliana Sánchez

possessor clitic has no corresponding clitic in general or in monolingual varieties

of Spanish:
(20) *Lo-/*le- toqué la mano del niño.
acc.3.sg/dat.3.sg touched the hand of-the boy
‘I touched the hand of the boy.’
(Camacho, Paredes & Sánchez 1995, p. 135)

Camacho, Paredes and Sánchez (1995) show evidence from wh-extraction that the
clitic lo is not associated with the full direct object DP but only with the genitive
expression del condor ‘of the condor’ as it can coexist with the wh-extraction of the
possesed constituent, as shown by the question in (21). This is consistent with a
more general pattern of clitic doubling found in L2 Spanish:
(21) ¿Qué loi- amarran ti [del cóndor]?
   what acc.3.sg tie t [of-the condor]
‘What (part) of the condor do they tie?’
(Camacho, Paredes & Sánchez 1995, p. 136)

The emergence of this possessor clitic is indicative of a process of activation of

L1 genitive features and of their association with L2 pronominal clitics. What
makes the reconfiguration or feature reassembly possible is the fact that a simi-
lar structure can be found with non-direct objects in General Spanish (Camacho,
Paredes & Sánchez 1995; Kalt 2012b). In fact, Kalt (2012b) notes that, in general
Spanish, structures with oblique or applicative clitics are characterized by a chain
in which the clitic is related to a possessor but is preceded by the prepositon a ‘to’
and not by the preposition de ‘of ’, as shown by the grammaticality of (22) and the
ungrammaticality of (23):
(22) María lei- robó el dinero a Juani.
Maria dat.3.sg stole the money to Juan
‘Maria stole the money from Juan.’
(23) *María lei- robó el dinero de Juani.
  Maria dat.3.sg stole the money of Juan
‘Maria stole Juan’s money.’
(Kalt 2012b, p. 174)

Following Masullo (1992), Kalt proposes that, in General Spanish, some oblique
objects incorporate into the theta-grid of the verb and can be clitic doubled, if
they are introduced by the preposition a. The process of incorporation is blocked
by the preposition de because it assigns structural case to its complement result-
ing in the ungrammaticality of (23). Camacho, Paredes and Sánchez (1995) and
Kalt (2012b) propose some form of reanalysis of the features of the clitic in order
to account for sentences such as (16) in the L2 Spanish of Quechua speakers.
 Crosslinguistic influences in Quechua-Spanish bilingualism 

Camacho, Paredes and Sánchez (1995) propose that the doubling in (19) and (21)
is possible because the clitic is marked in the L2 for genitive features. The crucial
aspect of this type of proposal is that in L2 Spanish the preposition de is not a case
assigner but the spell out of genitive features. The reassembly of features in this
case involves the mapping of the Quechua case marking suffix -pa shown in (24)
onto the Spanish preposition de deprived of its case assigning features.
(24) Kuntur-pa chaki-n.
condor-gen leg-3.sg
‘The condor’s leg.’
(Camacho, Paredes & Sánchez 1995, p. 142)

This possibility combined with the availability of preverbal dative clitics in dou-
bling structures in the L2 allows for the mapping of genitive features onto an dative
clitic, a feature not associated with clitics in general Spanish.
The process of functional interference and differences between the L1 and
the L2 in the mapping of features onto morphology can also result in a higher
level of underspecification for non-argumental clitics in L2 Spanish. Kalt (2002)
found evidence of differences in the interpretation of dative clitics as oblique clit-
ics between 84 Southern Quechua-Spanish bilingual children (ages 6–14) and
16 monolingual Spanish-speaking children from Bolivia (ages 6–10). Whereas in
General Spanish, dative clitics are coindexed with indirect objects rather than with
oblique benefactives (see contrast between (25) and (26)), this is not the case for
first person object markers in Quechua, as shown by (27) where the first person
object marker in combination with -pu (-puwa) results in the oblique benefactive.
(25) Mi madre lek- dar-á el libro (a Anak)
My mom dat.3.sg give-3.sg.fut the book (to Ana)
‘My mom will give the book to Ana.’
(26) *Mi madre mek- dar-á el libro a Ana (para mík)
  My mom ben-1.sg give-3.sg.fut the book to Ana (for mek)
‘My mom will give Ana the book for me.’
(27) Mama-y Ana-man libru-ta haywa-pu-wa-nqa.
Mom-1pos Ana-dat book-ac give-ben-1obj-3.sg.fut
‘My mom gave Ana the book for me.’
(Kalt 2012b, p. 175)

Example (27) shows that the aspectual suffix -pu introduces a benefactive mean-
ing and coexists with the marking of first person object features on the verb (-wa).
These features are not associated with an indirect object but with an oblique object
with a benefactive interpretation, as the English translation ‘for me’ indicates.
Kalt (2002) tested the bilingual children’s knowledge of Spanish sentences with
 Liliana Sánchez

i­ ntransitive verbs and transitive verbs with reflexive and oblique objects expressed
by clitics using a picture selection task in which the children listened to a sentence
and then were asked to choose from several pictures the one that reflected the
meaning of the sentence. She also asked the children to describe pictures. She
tested sentences with and without clitics. She found that Quechua-Spanish bilin-
gual children had their lowest levels of accuracy in the strict dative interpretation
of sentences involving an indirect object clitic le of the type shown in (28) due to
high levels of variability in the interpretation of the clitic as an indirect object, an
oblique or a locative expression, the latter a possible interpretation for the null
marker available in Quechua. Their results were significantly different from those
of the Spanish monolingual control group.
(28) Anai lej pone la chompa.
Ana dat.3.sg puts the sweater
‘Ana puts his sweater on.’

Kalt (2012b) attributes the variability in interpretation to a pattern of reassocia-

tion of the aspectual benefactive features of the suffix -pu, usually projected along
with some form of direct object marking, to the Spanish clitic le due to the fact
that le- has dative case features. Notice also the viability of mapping other fea-
tures such as locative onto the clitic. These results are consistent with a a reas-
signment of features associated with the clitic le that allows for a wider range of
non-argumental cases.
Kalt (2012b) sought further confirmation that the wider range of options
comes from the L1. She reports on a subsequent study conducted in Quechua
with Quechua L1 children. The results of that study show that for the third person
indirect object null morpheme with ditransitive verbs, shown in sentence (29),
several interpretations are possible. The sentence can be interpreted as involving
an action performed on an individual other than the subject, an action performed
on the subject itself and an action performed towards a location.
(29) Ana chumpa-ta chura-0-n.
Ana sweater-acc put-3.sg obj-3.sg subj
‘Ana put the sweater (on somebody specified in discourse/ on somebody
unspecified/ there/ somewhere).’
(Kalt 2012b, p. 183)

These data point in the direction of a process of functional interference under-

stood as the activation of features from the L1 such as oblique or locative case and
their mapping onto L2 morphology even in contexts in which the clitic le is only
the spell out of dative features in general Spanish.
The emergence of non-argumental clitics such as genitive clitics, oblique and
locative clitics in the L2 Spanish of Quechua speakers are processes that highlight
 Crosslinguistic influences in Quechua-Spanish bilingualism 

the relevance that activation of L1 features and their mapping onto L2 morphology
has in the development of L2 syntax.

4.2  C
 rosslinguistic influence at the syntax-morphology interface:
Functional interference, feature reassembly and functional
convergence in tense, aspect and evidentiality
Functional interference and feature reassembly can also be found in the tense,
aspect and evidentiality systems. Sánchez (2004) presents evidence of func-
tional interference and convergence in a study of oral production of 38 Southern
­Quechua-Spanish bilingual children (ages 10–16) for whom Spanish is a second
language. In the Spanish spoken by the children there is evidence of the mapping
of reported evidentiality and past tense features onto the Spanish morphological
forms that spell out aspect and tense features. In Quechua, past tense features
are strongly linked to evidentiality features and mapped onto syncretic suffixes
(Cerrón-Palomino 1987; Cusihuamán 2001). The following example shows how
the suffix -sqa spells out past tense and reportative evidential features. This suf-
fix is used when referring to past events for which the speaker has no direct
evidence:
(30) Manku Qhapaq-qa Titiqaqa qucha-manta-s lluqsimu-sqa.
Manku Qhapaq-top Titikaka Lake-abl-indir.ev emerge-3.sg.pst.rep
‘Manku Qhapaq emerged from the Titicaca Lake.’
(Cusihuamán 2001, p. 161)

The suffix -sqa contrasts with the syncretic suffix -rqa, which spells out past and
attested evidential features and is used when referring to a past event for which the
speaker has evidence:
(31) Huwan-mi Mariya-ta qhawa-rqa-n.
Huwan-dir.ev Mariya-acc see-past-3sg
‘Huwan saw Mariya.’
(Sánchez 2004, p. 149)

In Spanish, on the other hand, past tense and aspectual features are mapped onto
single morphemes as shown in (32–33).
(32) Compr-é.
Buy-1.sg.pst.perf
‘I bought.’ (perfective)

(33) Compr-aba.
Buy-1.sg.pst.imperf
‘I used to buy.’ (imperfective)
 Liliana Sánchez

Sánchez (2004) notes that the imperfective forms of Spanish share some meanings
with the reportative past forms of Quechua such as a mirative interpretation that
indicates surprise:
(34) Anda, ¡sabía nadar!
Hey, knew to swim
‘Hey, (s/he) knew how to swim!’
(Sánchez 2004, p. 149)

Sánchez (2004) proposes that this partial similarity serves as a trigger to generate
new associations between grammaticalized evidentiality features in Quechua and
past morphology in Spanish. Using a story-retelling task in which the children
listened to a story with perfective and imperfective forms and then were asked to
retell the story in the past, Sánchez (2004) found evidence of strong differences
in the use of Spanish verb forms between Quechua-Spanish bilingual children
and Spanish monolingual children. Whereas Spanish monolingual children had
higher frequencies of preterite past perfective forms, Quechua-Spanish bilingual
children used almost exclusively imperfective and pluperfect forms in Spanish
to convey hearsay or reportative evidential past. Their Quechua narratives of
the same stories also reflected high frequencies of reportative past verb forms.
The type of narratives Sánchez (2004) found are exemplified by the following
sentences:
(35) Había una viejita, dice.
Have-pst.imperf an old woman, say
‘(There) was an old woman, (they) say.’
(36) Había sembr-ado maíz.
Have-pst.imperf.3.sg sow-part corn
‘(S/he) had sowed corn.’
(37) Ella sembr-aba maíz.
She sow-pst.imperf.3.sg corn
‘She sowed corn.’
(38) Le- hab-ía encontr-ado un pajarito amarillo.
dat.3.sg have-pst.imperf.3.sg find-part a birdy yellow
‘She found a yellow bird.’
(39) Y le- había llev-ado a su casa.
and dat.3.sg have-pst.imperf.3.sg take-part to her home
‘And (she) took him home.’
(40) Y le había h-echo
And dat.3.sg have-pst.imperf.3.sg make-part
 Crosslinguistic influences in Quechua-Spanish bilingualism 

comer trigo, agua, pan.

eat wheat, water, bread
‘And (she) made him eat wheat, water and bread.’
(Sánchez 2004, p. 157–158)

This sequence of sentences is not typical of a Spanish narrative. Firstly, it lacks

preterite forms and secondly, sentences (36), (38), (39) and (40) show consistent
use of the pluperfect and sentences (35) and (37) show imperfective forms. Both
pluperfect and imperfective forms are used to convey hearsay information in a
manner that ressembles the distribution of reportative past forms in Quechua.
Sánchez (2004) proposes an extension of her previously proposed Functional
Convergence Hypothesis (FCH) (Sánchez 2003) as follows:
“Convergence, the common specification for equivalent functional features in the
two languages spoken by the bilingual in a language contact situation, takes place
when the languages have partially similar matrices of features associated with
the same functional category. Frequent activation of the two matrices triggers
convergence in features.” (Sánchez 2004, p. 150)

The FHC predicts that among Quechua-Spanish bilinguals the activation of evi-
dentiality features in Quechua affects bilingual Spanish representations and activa-
tion of aspectual features in Spanish affects the bilingual Quechua representations.
Evidence for the mapping of evidentiality onto Spanish morphology was found in
the consistent use of pluperfect and imperfective morphemes in the Spanish data.
The bilingual Quechua results also exhibited convergence because rather than
showing a consistent use of reportative past forms, attested past morphemes were
also found in the narratives. Their distribution was similar to the distribution of
perfective and imperfective forms in Spanish.
Like the emergence of oblique clitics, the emergence of reportative evidential-
ity in the Spanish of Quechua speakers is also evidence in favor of a view of cross-
linguistic influence that is triggered by the activation of functional features in both
languages and a common pattern of feature-morphology mapping.

4.3  C
 rosslinguistic influence at the syntax-lexicon interface: Functional
interference, functional convergence, aspect and argument structure
Another piece of evidence in favor of functional interference and convergence can
also be found in the mapping of some grammaticalized Quechua aspectual fea-
tures associated with syncretic derivational morphology onto independent verbs
in Spanish. Sánchez (2006) presents evidence of a complex type of convergence
in which the syncretic derivational morpheme -naya associated with v­ olitive and
 Liliana Sánchez

imminent modal/aspectual features is mapped onto the Spanish modal verb q­ uerer
‘to want’. The study of picture-based narratives of 30 Lamas Kechwa-Spanish bilin-
gual children (ages 9–13) shows evidence of crosslinguistic influence between
Lamas Kechwa desiderative progressive forms (41) and Spanish modal periphras-
tic verb forms (42):
(41) Miku-naya-yka-n.
Eat-des-prog-3.sg
‘(S/he) wants to/is about to eat.’
(Sánchez 2006, p. 540)
(42) Está queriendo comer.
Is want-ger eat-inf
‘(S/he) wants to eat/She is about to eat.’

The latter form conveys a desiderative/imminent aspectual meaning absent in the

narratives of a comparison group of 25 Spanish-speaking children. The results of
the study showed evidence of functional interference. The narratives showed that
bilingual children used the desiderative form in Quechua:
(43) Sap-itu urma-naya-n yaku-pi.
Toad-dim fall-des-3.sg water-loc
‘The toad wants/is about to fall into the water.’
(Sánchez 2006, p. 544)

They also used it with progressive forms as in:

(44) Kay achku muku-chi-naya-yka-n kay sap-itu-ta.
This dog bit-caus-des-prog-3.sg toad-dim-acc
‘This dog is wanting to have this toad bit.’
(Sánchez 2006, p. 544)

In the Spanish narratives, modal progressive forms with a desiderative/imminent

reading were found:
(45) Un wamr-illu (e)stá queriendo agarrar su sapo.
A boy-dim (i)s wanting (to) grab his toad
‘A boy wants/is about to grab his toad.’
(Sánchez 2006, p. 545)

In these cases, the progressive marker contributes to the imminent reading of

the verbal periphrasis, an interpretation not found in monolingual varieties of
­Spanish. Notice that, in Quechua, the grammaticalized aspectual imminent fea-
ture and the modal desiderative feature are syncretically mapped onto the mor-
pheme -naya. Spanish has no corresponding bound morpheme. In the Spanish
of these ­Quechua-speaking children there is mapping of both features onto an
 Crosslinguistic influences in Quechua-Spanish bilingualism 

independent modal verb in Spanish. This type of mapping in its imminent inter-
pretation has an important syntactic consequence in that it forces the elimination
of the volitional agent of the verb querer ‘to want’ and bleaches the argumental
structure of the verb in its volitional intrepretation leaving only the imminent
interpretation.
The evidence from these emergent verbal periphrasis supports the idea that
there is convergence in features and in morphological mapping. It also supports
the need for a more nuanced view of the relationship between features and their
mapping onto morphology in areas of the lexicon related to syntax such as argu-
ment structure, one that takes into account the differences across languages
between mapping features onto suffixes or onto verbs and verbal periphrasis.This
suggests that activation of some grammaticalized features does indeed affect the
configuration of some aspects of the lexicon that have consequences for syntactic
development.

4.4  Evidence of crosslinguistic influence at the syntax/pragmatics interface

In addition to new patterns of feature-morphology mapping, the study of the L2
Spanish of Quechua speakers has also shown evidence of transfer of mapping
strategies at the syntax/pragmatics interface involving the spell out of functional
features. Camacho, Paredes and Sánchez (1997) have shown that the Spanish
of Southern Quechua speakers is characterized by the pervasive nature of null
objects with definite antecedents. In their study of 33 adult L1 Southern Quechua-
L2 Spanish speakers, they found evidence of null objects with definite antecedents
in the speech of all proficiency subgroups. The following are examples of the types
of null objects found in the dialogues between an interviewer and the participants
in the study:
(46) a. ¿Y Uds. preparaban el desayunoi para los pensionistas?
‘And did you prepare breakfast for the guests?’
b. Sí, mis hermanas ei preparaban.
‘Yes, my sisters prepared it.’
(47) a. ¿Extrañas mucho a tu papai?
‘Do you miss your father?’
b. Sí, sí ei extraño.
‘Yes, yes I miss him.’
(48) a. ¿Qué pasó?- me dijo- Loi había atropellado una moto
‘What happened? -he told me-. A motorcycle has run her over.’
le dije, entonces:
‘I told him, then:’
 Liliana Sánchez

b. Hemos llevado ei ya al Hospital del Niño, ya ei

‘Then we took her to the hospital;’
hemos llevado al Emergencia.
‘We took her to the Emergency room.’

(49) a. ¿Qué hace, la mata (el lobo) o no la mata a la ovejai?

‘What does the wolf do? Does it kill the sheep?’
b. Sí mata ei, sí mata ei
‘Yes, it kills (it), yes it kills it.’
(Camacho, Paredes & Sánchez 1997, p. 59)

Notice that, in all these cases, the antecedents of null objects were definite.
­Camacho, Paredes, and Sánchez (1997) note that the pervasive nature of null
objects can be accounted for by a topic operator that licenses the null object. This
type of licensing takes place despite the availability of clitics in the L2 Spanish of
these speakers as exemplified in sentences (48a) and (48b). In sentence (48a), the
discourse topic is introduced as a clitic, but afterwards, a null object is used to refer
to that topic (48b). This type of licensing is also found in Quechua as illustrated in
sentences (5a) and (5b). Notice that the null object cannot be accounted for by a
general absence of clitics in L2 Spanish. I would like to propose that, in these cases,
there is a mapping strategy that is transferred from one language to the other.
There is no overt object third person verbal morphology in Quechua and a null
object appears in contexts in which there is a third person topic object as in (5b).
It is the mapping strategy of Quechua, namely the lack of third person object overt
morphology that is at the source of the definite null objects in L2 Spanish. Inde-
pendent evidence for such a mapping has also been found in the L2 Spanish of
Chinese speakers (Cuza, Pérez-Leroux & Sánchez 2013). In this case, crosslinguis-
tic influence is shown by the absence of third person object morphology in the L2
verb following the L1 patterns in null topic contexts.
Evidence of null objects with definite antecedents in the Spanish of two
groups of Central Quechua-Spanish and Southern Quechua-Spanish bilingual
children (ages 9–13) was also found by Sánchez (2003). The picture-based nar-
ratives of both groups of bilingual children showed that the second most fre-
quent type of direct object in Quechua was a null object, the most frequent type
being an overt DP. This was not the case in the Spanish narratives. These had a
higher frequency of clitics and structures with clitic doubling than of null objects.
Null objects, however, had overt definite DPs previously mentioned in the text as
their antecedents and also antecedents not present in the text but shown in the
pictures, a pattern also found in the Quechua narratives. This pattern is consis-
tent with the idea that the null objects correspond to a discourse salient element,
namely, a topic.
 Crosslinguistic influences in Quechua-Spanish bilingualism 

I would like to discuss a final piece of evidence of crosslinguistic influence

at the interface of syntax and the pragmatic component reported in a study on
word order, focus, and weak crossover in the Spanish of adult Quechua speak-
ers (Muntendam 2013). This evidence will be helpful in distinguishing between
crosslinguistic influence at the interface of the lexicon and syntax involving feature
reassembly and morphological mapping, and crosslinguistic influence in restric-
tions to syntactic operations such as weak crossover effects. Muntendam (2013)
collected naturalistic data and conducted a grammaticality judgment test on weak
crossover with 15 adult Quechua-Spanish bilinguals in Bolivia.5 Her study aimed
at finding evidence on whether previous findings of the prevalence of OV word
orders in the Spanish of Quechua speakers, analyzed as focus fronting (Camacho
1999), had a correlate in a restriction on focus fronting widely documented for
General Spanish: weak crossover effects. Muntendam (2013) notes that Quechua
focus fronting lacks weak crossover effects but Spanish focus fronting involves
such effects. If OV word orders are evidence of transfer of focus fronting strategies
from Quechua into Spanish, then Quechua influenced Spanish (which she labels
Andean Spanish) should also lack weak crossover effects. The contrast in weak
crossover effects is found in the following examples:
(50) ¿Mayqen wawa-ta-taq mama-n
   which child-ac-q mother-3pos
apa-mu-chka-n
yachay wasi-man?

bring-dir-progR-3sg school-dir
‘Which child does his mother bring to school?’
(51) [F Sapa wawa-ta] mama-n
 each child-acc mother-3pos
apa-mu-chka-n
yachay wasi-man.

bring-dir-progr-3sg school-dir
‘His mother brings each child to school.’
(52) ??¿[A qué niñoi] trajo sui madre a la escuela?
   To which boy brought his mother to the school
‘Which child did his mother bring to school?’
(53) ??[FA cada niñoi] trajo sui madre a la escuela.
   To each child brought his mother to the school
‘His mother brought each child to school.”
(Muntendam 2013, p. 119)

.  Muntendam (2013) also reports results of other tasks but I will only discuss the ones rel-
evant to the argument presented.
 Liliana Sánchez

The results of the grammaticality judgment task showed that no bilingual speak-
ers showed sensitivity to sentences with focus fronting (48) and questions (47) in
Quechua revealing no weak cross overeffects. However, in Spanish, 10 out of the
15 speakers showed sensitivity to weak crossover effects with questions, and 11 out
of 15 showed sensitivity with focus fronting sentences in a proportion almost as
high as that of the control groups of speakers of non-Andean Spanish. At the same
time, the naturalistic data examined showed 18.5 % of OV structures, a percentage
similar to those found in other varieties of Spanish in contact with Quechua and
much higher than the 7.9% found in naturalistic data from Argentinian Spanish, a
variety not in contact with Quechua. From these data, Muntendam concludes that,
while the preference for OV could be attributed to some form of focus fronting for
pragmatic purposes, it does not correlate with the main syntactic characteristic of
focus fronting in Quechua.
In view of the previous evidence of functional interference, convergence, and
differences in feature assembly and mapping onto morphology from a variety of
studies on Quechua-Spanish bilingualism, Muntendam’s work suggests that it is
mostly those features that are at the interface of the lexicon, syntax and morphol-
ogy that are affected by crosslinguistic influence, whereas restrictions on syntactic
operations such as movement are not thus affected.
Overall, the evidence presented in this paper points in the direction of a more
nuanced view of the interface of the lexicon, feature specification and morphologi-
cal mapping that is at the basis of crosslinguistic influence. It also highlights the
need for more specific studies of the relationship between activation of the lexicon
and of related functional features such as aspectual or discoursive features and
how it affects morphological mapping in the bilingual mind.

5.  Concluding remarks

In this paper, I have discussed evidence of crosslinguistic influence that involves

the activation of functional features that generates new patterns of feature-­
morphology mapping across languages that includes crosslinguistic influence at:
(a) the syntactic level (feature reassembly and the emergence of non-argumental
clitics) (b) the syntax/lexicon interface (feature reassembly and changes in aspec-
tual configurations and argument structure), and (c) the syntax/pragmatics inter-
face (licensing of null objects and focus fronting but no weak crossover effects). At
the syntactic level, feature reassembly is represented in the emergence of new non-
argumental clitics in Quechua-Spanish bilinguals because Quechua oblique case
marking is mapped onto the Spanish dative clitic le. This new mapping allows for
an extension of verbal agreement from arguments to non-argumental ­constituents
 Crosslinguistic influences in Quechua-Spanish bilingualism 

such as benefactives. This in turn generates a new syntactic configuration in which

rather than being merged as prepositional phrases such as para/por mi ‘for mi’ as
in Example (26), benefactives are directly merged as morphological markers on
the verb. Thus the interaction between morphology and syntax in the bilingual
grammar has had an effect on the way in which non-arguments are merged in one
of the languages.
At the syntax/lexicon interface, the mapping of the Quechua imminent suffix
-naya onto a modal verb such as querer ‘to want’ involves a change in the way in
which lexical roots are activated and merged in the syntax of bilingual Spanish.
While a single lexical root is activated in Quechua, in Spanish two verbal roots are
activated. This opens up new possibilities for the projection of external arguments
of the first and the second root and has consequences for the syntax.
Finally, at the syntax/pragmatice interface, convergence with the Quechua
mapping of continuing topics in bilingual Spanish shows evidence that sensitivity
to pragmatic conditions can alter the morphosyntactic representation of objects.
On the other hand, the availability of focus fronting as a Quechua-influenced
mapping strategy in bilingual Spanish is not accompanied by the lack of weak
cross over effects that characterizes Quechua. This shows that while focus fronting
for pragmatic purposes might be prevalent, actual movement is available in bilin-
gual Spanish in right weak cross over contexts.
Overall, this evidence has been used to show that activation of features, feature
reasembly, and mapping onto morphology force a more nuanced view of language
modularity in bilinguals. The data discussed support the view that multiple lan-
guage components interact in the representation of languages (Jackendoff 1997)
and especially in the bilingual mind (Slabakova 2008; Sorace 2000) in ways that are
innovative and may lead to language change. The availability of multiple compo-
nents in two languages allow multiple interactions between the lexicon of the two
languages, the inventory of functional features, their morphological expressions,
and the pragmatic knowledge associated with the interpretation of some features.

References

Bleam, T. (1999). Object bare plurals in Spanish and the semantics of personal A. In J. M. ­Authier,
B. Bullock, & L. Reed (Eds.), Formal Perspectives on Romance Linguistics (pp.  21–37).
Amsterdam: John Benjamins. DOI: 10.1075/cilt.185.04ble
Camacho, J. (1999). From SOV to SVO: The grammar of interlanguage word order. Second Lan-
guage Research, 15(2), 115–132. DOI: 10.1191/026765899673532714
Camacho, J., & Sánchez, L. (2002). Explaining clitic variation in Spanish. In M. Amberber &
P. Collins (Eds.), Formal Approaches to Language Universals and Language Variation
(pp. 21–40). Norwood, CT: Ablex.
 Liliana Sánchez

Camacho, J., Paredes, L., & Sánchez, L. (1995). The genitive clitic and the genitive construction
in Andean Spanish. Probus, 7(2), 133–146. DOI: 10.1515/prbs.1995.7.2.133
Camacho, J., Paredes, L., & Sánchez, L. (1997). Null objects in Bilingual Andean Spanish. In
E. Hughes, M. Hughes, & A. Greenhill (Eds.), Proceedings of the 21st annual Boston Univer-
sity Conference on Language Development, Vol 1. (pp. 55–66). Somerville: MA: Cascadilla
Press.
Cerrón-Palomino, R. (1987). Lingüística Quechua. Cusco, Peru: Centro de Estudios Rurales
Andinos Bartolomé de las Casas.
Chomsky, N. (2000). Minimalist inquiries: The framework. In R. Martin, D. Michaels, &
J. Uriagereka (Eds.), Step by Step: Essays on Minimalist Syntax in Honor of Howard Lasnik
(pp. 89–155). Cambridge, MA: The MIT Press. DOI: 10.1017/s0022226702271625
Cusihuamán, A. (2001). Gramática Quechua Cuzco-Collao. Lima: Instituto de Estudios Peruanos.
Cuza, A., Pérez-Leroux, A.T., & Sánchez, L. (2013). The role of semantic transfer in clitic-drop
among Chinese L1-Spanish L2 bilinguals. Studies in Second Language Acquisition, 35(1),
93–125. DOI: 10.1017/s0272263112000691
Escobar, A.M. (1991). Los bilingües en el Perú. Lima, Peru: Instituto de Estudios Peruanos.
DOI: 10.1017/s0022216x00000882
Flynn, S. (1987). L2 acquisition of pronoun anaphora: Resetting the parameter. In B. Lust
(Ed.), Studies in the Acquisition of Anaphora: Defining the Constraints, 2, (pp. 227–243).
­Dordrecht: Reidel. DOI: 10.1007/978-94-009-3387-3_10
Giorgi, A., & Pianesi, F. (1997). Tense and Aspect. Oxford: OUP. DOI: 10.1515/ling.2003.019
Gürel, A. (2004). Selectivity in L2-induced L1 attrition: A psycholinguistic account. Journal of
Neurolinguistics, 17, 53–78. DOI: 10.1016/s0911-6044%2803%2900054-x
Hintz, D. (2011). Crossing Aspectual Frontiers. Los Angeles, CA: University of California Press.
Hopp, H. (2009). The syntax-discourse interface in near-native L2 acquisition: Off-line and on-
line performance. Bilingualism, 12, 463–483. DOI: 10.1017/s1366728909990253
Hopp, H. (2012). Grammatical gender in adult L2 acquisition: Relations between lexical and
syntactic variability. Second Language Research, 29(1), 33–56.
DOI: 10.1177/0267658312461803
Hulk, A., & Muller, N. (2000). Bilingual first language acquisition at the interface between ­syntax
and pragmatics. Bilingualism: Language and Cognition, 3, 227–244.
DOI: 10.1017/s1366728900000353
Jackendoff, R. (1997). The Architecture of the Language Faculty. Cambridge, MA: The MIT Press.
Jackendoff, R. (2002). Foundations of Language: Brain, Meaning, Grammar, Evolution. Oxford:
OUP. DOI: 10.2298/jfi0561213i
Kalt, S. (2000). Non-direct object agreement in the second language Spanish of Southern
­Quechua speakers. In B. Swierzbin, F. Morris, M.E. Anderson, C.A. Klee, & E. Tarone
(Eds.), Social and Cognitive Factors in Second Language Acquisition: Selected Proceedings
of the 1999. Second Language Research Forum (pp. 222–242). Somerville, MA: Cascadilla
Press.
Kalt, S. (2002). Second Language Acquisition of Spanish Morpho-Syntax by Quechua-Speaking
Children. Unpublished Ph.D. dissertation. University of Southern California at Los Angeles.
Kalt, S. (2009). Bilingual children’s object agreement and case marking in Cusco Quechua. In
H. Bliss & E. Gerard. (Eds.), University of British Columbia Working Papers in Linguistics,
26, (pp. 126–142). Vancouver, BC: University of British Columbia.
Kalt, S. (2012b). Cambios morfosintácticos en castellano impulsados porel quechua hablante.  In
P. Dankel, V.F. Mallat, J.C. Godenzzi, & S. Pfänder (Eds.), El español andino. Espacios comu-
nicativos y cambios gramaticales, 41, (pp. 165–192). Neue Romania.
 Crosslinguistic influences in Quechua-Spanish bilingualism 

Kerke, S.C., van de. (1996).
Affix Order and Interpretation in Bolivian Quechua. Leiden: CNWS.
Kroll, J.F., & Bogulski, C.A. (2013). Cognitive second language acquisition: Organization of the
second language lexicon. In C.A. Chapelle (Ed.), The Encylopedia of Applied Linguistics.
(pp. 4322–4330). Oxford: Blackwell. DOI: 10.1002/9781405198431.wbeal0886
Kroll, J.F., & Stewart, E. (1994). Category interference in translation and picture naming:
­Evidence for asymmetric connections between bilingual memory representations. Journal
of Memory and Language, 33, 149–174. DOI: 10.1006/jmla.1994.1008
Kroll, J.F., Van Hell, J.G., Tokowicz, N., & Green, D.W. (2010). The Revised Hierarchical Model:
A critical review and assessment. Bilingualism: Language and Cognition, 13, 373–381.
DOI: 10.1017/s136672891000009x
Lardiere, D. (1998). Dissociating syntax from morphology in a divergent L2 end-state grammar.
Second Language Research, 14, 359–375. DOI: 10.1191/026765898672500216
Lardiere, D. (2003). Second language knowledge of [±Past] vs. [±Finite]. In J.M. Liceras,
H. Zobl, & H. Goodluck. (Eds.), Proceedings of the 6th Generative Approaches to Second
Language Acquisition Conference (GASLA 2002) (pp. 176–189). Somerville, MA: Casca-
dilla Proceedings Project.
Lardiere, D. (2005). On morphological competence. In L. Dekydtspotter, R. Sprouse, &
A.  Liljestrand (Eds.), Proceedings of the 7th Generative Approaches to Second Language
Acquisition Conference (GASLA 2004), (pp. 178–192). Somerville, MA: Cascadilla
­Proceedings Project.
Leonetti, M. (2004). Specificity and differential object marking in Spanish. Catalan Journal of
Linguistics, 3, 75–114.
Liceras, J. (2010). Second language acquisition and syntactic theory in the 21st century. Annual
Review of Applied Linguistics, 30, 248–269. DOI: 10.1017/s0267190510000097
Liceras, J., Fernández, R., & Alba, A. (2011). Overt subjects and copula omission in the Spanish
and the English grammar of English–Spanish bilinguals: On the locus and directionality of
interlinguistic influence. First Language, 32(1–2), 88–115.
DOI: 10.1177/0142723711403980
Marull, C. (2013, April). Language-specific syntax constrains cross-linguistic activation. Paper
presented at the 43rd Linguistic Symposium on Romance Languages, CUNY Graduate
Center, New York.
Masullo, P. (1992). Incorporation and Case Theory in Spanish: A Crosslinguistic Perspective.
Unpublished Ph.D. dissertation. University of Washington, WA.
Meisel, J.M. (1997). The acquisition of the syntax of negation in French and German: Contrast-
ing first and second language development. Second Language Research, 13(3), 227–263.
DOI: 10.1191/026765897666180760
Meisel, J. (2008). Child second language acquisition or successive first language acquisition? In
B. Haznedar & E. Gavruseva (Eds.), Current Trends in Child Second Language Acquisition:
A Generative Perspective (pp. 55–80). Amsterdam: John Benjamins.
DOI: 10.1017/s0272263109990076
Montrul, S. (2010). Dominant language transfer in Spanish L2 learners and heritage speakers.
Second Language Research, 26(3), 293–327. DOI: 10.1177/0267658310365768
Müller, N., & Hulk, A. (2001). Crosslinguistic influence in bilingual language acquisi-
tion: Italian and French as recipient languages. Bilingualism: Language and Cognition,
4(1), 1–21. DOI: 10.1017/s1366728901000116
Muntendam, A. (2013). On the nature of cross-linguistic transfer: A case study of Andean Span-
ish. Bilingualism: Language and Cognition, 16(1), 111–113.
DOI: 10.1017/s1366728912000247
 Liliana Sánchez

Muysken, P. (1995). Focus in Quechua. In K. Kiss (Ed.), Discourse Configurational Languages

(pp. 375–393). Oxford: OUP.
Paradis, M. (1993). Linguistic, psycholinguistic, and neurolinguistic aspects of “interference” in
bilingual speakers: The activation threshold hypothesis. International Journal of Psycholin-
guistics, 9(2), 133–145.
Paradis, M. (2009). Declarative and Procedural Determinants of Second Languages. Amsterdam:
John Benjamins. DOI: 10.1017/s0272263110000306
Paradis, J., & Genesee, F. (1996). Syntactic acquisition in bilingual children. Studies in Second
Language Acquisition, 18(1), 1–25. DOI: 10.1017/s0272263100014662
Pladevall, E. (2010). Child L2 development of syntactic and discourse properties of Spanish
subjects. Bilingualism: Language and Cognition, 13(2), 185–216.
DOI: 10.1017/s1366728909990447
Putnam, M., & Sánchez, L. (2013). What’s so incomplete about incomplete acquisition? –
A ­prolegomenon to modeling heritage language grammars. Linguistic Approaches to Bilin-
gualism, 3(4), 478–504. DOI: 10.1075/lab.3.4.04put
Sánchez, L. (2003). Quechua-Spanish Bilingualism: Functional Interference and Convergence.
Amsterdam: John Benjamins. DOI: 10.1075/lald.35
Sánchez, L. (2004). Functional convergence in the tense, evidentiality and aspectual systems of
Quechua-Spanish bilinguals. Bilingualism: Language and Cognition, 7, 147–162.
DOI: 10.1017/s136672890400149x
Sánchez, L. (2006). Kechwa and Spanish bilingual grammars: Testing hypotheses on functional
interference and convergence. International Journal of Bilingual Education and Bilingual-
ism, 9, 535–556. DOI: 10.2167/beb379.0
Sánchez, L. (2010). The Morphology and Syntax of Topic and Focus: Minimalist Inquiries in the
Quechua Periphery. Amsterdam: John Benjamins. DOI: 10.1075/la.169
Schwartz, A., & Kroll, J. (2006). Bilingual lexical activation in sentence context. Journal of Mem-
ory and Language, 55, 197–212. DOI: 10.1016/j.jml.2006.03.004
Schwartz, B.D., & Sprouse, R.A. (1996). L2 cognitive states and the “full transfer/full access”
model. Second Language Research, 12, 40–72. DOI: 10.1177/026765839601200103
Serratrice, L., Sorace, A., & Paoli, S. (2004). Crosslinguistic influence at the syntax-pragmatics
interface: Subjects and objects in Italian-English bilingual and monolingual acquisition.
Bilingualism: Language and Cognition, 7, 183–206.
DOI: 10.1017/s1366728904001610
Slabakova, R. (2008). Meaning in the Second Language. Berlin: Mouton de Gruyter.
DOI: 10.1017/s0272263110000124
Sorace, A. (2000). Syntactic optionality in L2 acquisition. Second Language Research, 6, 93–102.
DOI: 10.1191/026765800670666032
Sorace, A. (2005). Syntactic optionality at interfaces. In L. Cornips & K. Corrigan (Eds.), Syntax
and Variation: Reconciling the Biological and the Social (pp. 46–111). Amsterdam: John
Benjamin. DOI: 10.1017/s1360674307002316
Sorace, A., & Serratrice, L. (2009). Internal and external interfaces in bilingual language devel-
opment: Beyond structural overlap. International Journal of Bilingualism, 13(2), 195–210.
DOI: 10.1177/1367006909339810
Ullman, M. (2001). The neural basis of lexicon and grammar in first and second language: The
declarative/procedural model. Bilingualism: Language and Cognition, 4, 105–122.
DOI: 10.1017/s1366728901000220
 Crosslinguistic influences in Quechua-Spanish bilingualism 

Ullman, M. (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). Washington, DC:
­Georgetown University Press.
White, L. (1989). Second Language Acquisition and Universal Grammar. Amsterdam: John
­Benjamins. DOI: 10.1177/026765839100700306
White, L. (2003). Universal Grammar and Second Language Acquisition. Cambridge: CUP.
DOI: 10.1017/cbo9780511815065.002
White, L. (2011). Second language acquisition at the interfaces. Lingua,121, 577–590.
DOI: 10.1016/j.lingua.2010.05.005
 Verbal agreement in the L2 Spanish
of speakers of Nahuatl

Alma P. Ramírez-Trujillo1 & Joyce Bruhn de Garavito1, 2

1Emory and Henry College & 2The University of Western Ontario

This paper examines the production of subject/verb agreement in the L2

Spanish of Nahuatl L1 speakers in light of the debate on whether problems
with morphology are evidence for a deficit at the functional level in second
language acquisition. In particular, it focuses on whether age of acquisition is
the determining factor in the production of agreement, or whether quantity and
quality of input are. Results show that there was no difference in the rate or type
of error according to age of acquisition, but there was an important difference
in rate between those participants who had little or no contact with standard
Spanish and those that had a great deal due to the fact that they had to leave the
village in order to find work. These results are taken as evidence against a deficit
in adult second language acquisition.

1.  Introduction1

This paper describes the interlanguage grammar of second language (L2) learn-
ers of Spanish whose first language (L1) is Nahuatl, an indigenous language spo-
ken in Mexico. It focuses on the production of agreement between the subject
and the verb in Spanish. A great deal of literature on the Spanish spoken by
indigenous peoples in Mexico and other parts of Latin America either e­ xplicitly
or implicitly assumes it is a variety that differs substantially from the norm.

.  We wish to thank The University of Western Ontario for an International Research Award
to Joyce Bruhn de Garavito and a New Research and Scholarly Initiative Award (ADF) to
Joyce Bruhn de Garavito and Silvia Perpiñán. The data reported on here form part of Alma P.
Ramírez-Trujillo’s Ph.D. thesis.
We also wish to express gratitude to the people of San Isidro Buensuceso for their patient
help and hospitality, particularly Don Fidencio and his extended family. Finally, thanks are
due to Jeremías Cabrera who produced the translations to and from Nahuatl.

doi 10.1075/ihll.3.03ram
© 2015 John Benjamins Publishing Company
 Alma P. Ramírez-Trujillo & Joyce Bruhn de Garavito

­ astra (1992), for example, in a paper that is frequently cited, asks whether there
L
is an Indian Spanish. She analyzes the production of several groups of indig-
enous speakers, including Otomí and Nahuatl, pointing out the ways in which
they diverge from the standard. However, in most of this research, it is not clear
whether the participants are first or second language speakers, simultaneous or
sequential bilinguals, whether they learned Spanish from other L2 speakers or
whether they have any contact with the ‘standard’ Spanish of the region. The
present study remedies these shortcomings by focusing explicitly on bilinguals.
We address two main issues: first, whether there are differences between early
and late bilinguals, that is, whether age is a determining factor in the produc-
tion of the standard agreement morphology; second, whether amount and
type of contact with the norm may be the best explanation for the properties
of the Spanish of these speakers. We also examine the type of error commit-
ted in light of current predictions that the least specified form will be used as a
default (McCarthy 2006, 2007). Although we focus on one particular group of
speakers in one particular village, we will show that the kind of input received
and not age of acquisition is the main factor in the consistency of production
of agreement. Optionality is associated with lack of contact with the generally
monolingual speakers of the regional norm, while second language learners of
Spanish who have frequent outside contact do not exhibit variability in agree-
ment between the subject and the verb. Besides the difference in quantity of
input received by each group, there is a difference in quality given that those
with significant amounts of input from outside the village are in contact with
the standard ­Spanish of the region, while those that do not have this contact will
receive input from speakers whose variety exhibits the same properties as their
own. We argue that there is no particular deficit in second language acquisition
that necessarily leads to optionality in the production of morphology.
In the following section we will briefly give an overview of the context in
which Nahuatl is spoken. This is important for several reasons. In the first place,
it is difficult to understand the findings if one does not have a general idea regard-
ing the sociolinguistic circumstances of native languages in Latin America. This
impinges directly on our approach to these learners who are in great part illiterate.
It is also important to note that, because Nahuatl is not a frequently studied lan-
guage, there is disagreement as to its structure or, depending on how we view the
different dialects, its possible structures.
The paper is organized as follows: Section 2 will consist of a short descrip-
tion of the sociolinguistic background; in Section 3 we will describe the proper-
ties of agreement in Spanish and Nahuatl; in Section 4 we will comment on the
relevant theoretical approaches to the acquisition of agreement, turning then in
 Verbal agreement in the L2 Spanish of speakers of Nahuatl 

Section 5 to the methodology followed in Section 6 by the results, with a discus-

sion and conclusion in 7.

2.  Sociolinguistic background

Nahuatl or Mexicano, as it is sometimes called, is the language spoken by the

descendants of the Aztecs. There are over one million speakers so at first sight it
would not seem to be an endangered language. However, it is generally spoken
in small villages in different regions, and this isolation has reinforced not only
important dialectal variations but also notable differences in attitude towards
the language and its future (Hill & Hill 1986). For example, the village where the
present research took place is situated among a cluster of villages that lie on the
slopes of a mountain, La Malinche, in the state of Tlaxcala. These villages were
most notably studied by Hill and Hill (1986) in their seminal work on the use of
­Mexicano in this region. When we visited, around 10 years after the publication
of Hill and Hill’s book and perhaps 20 years after the original research was carried
out, we found villages where adults spoke to each other in Nahuatl and to their
children in Spanish, a sure sign that the language is endangered. In contrast, in
San Isidro Buensuceso, where our participants live, Nahuatl is the language of the
village, heard everywhere, and the speakers are proud of their language, although
they do not seem to have any idea that it was once the language of an empire.
The economic situation described by Hill and Hill (1986) has not improved.
It is not possible for people to continue to live off the land as they had in the past,
and most younger members of the village now travel to the large city of Puebla
to work. This situation of economically-forced migration is reflected across the
country; in fact, one of the greatest concentrations of Nahuatl speakers is found
in Mexico City. This means that Spanish is becoming more and more important,
leading to a generational divide. In San Isidro, the older members of the commu-
nity are either monolingual in Nahuatl or second language learners of Spanish.
Middle-aged people may also be second language learners of Spanish, but they will
have a great deal more contact with the language. The younger generation is often
bilingual (See Ramírez-Trujillo 2013).
The situation regarding schooling is also complicated, as our research
has shown. Many of the participants in our study are illiterate or semi-literate.
­Elementary school gives children the opportunity to learn to read and write
Nahuatl, although not in all classes. High school is generally in Spanish. However,
few members of the community, particularly girls, finish high school, and of the
older generations few even finished elementary school.
 Alma P. Ramírez-Trujillo & Joyce Bruhn de Garavito

It is clear from the above that these speakers are not used to being asked about
their language or to being tested (see Rice, Libben, & Derwing (2002) for a dis-
cussion of testing illiterate speakers). Within the generative approach, besides
the authors of the present paper, only work by MacSwan (1998, 1999, 2000), who
looked at code switching, has considered the structure of Nahuatl. However, he
did not conduct an experiment; rather Nahuatl speakers served as informants
regarding code-switching intuitions. Given the population studied and the first
language of the speakers, the present paper is clearly breaking new ground, not
only because of the language combination but also because this paper is one of the
few that examines illiterate speakers, using a novel oral production methodology.
As linguists we often take for granted that our participants will know what we are
asking them to do because they have experience with testing, but in this case a
certain amount of latitude must be accepted, particularly in the placement test, as
we shall see. We note, however, that, in spite of the lack of familiarity with what
we were trying to accomplish, responses were as expected and consistent with the
aims of the tasks. The people of this town were always very willing to help, and we
must express deep gratitude for their kindness.

3.  Agreement in Spanish and Nahuatl

3.1  S panish
Spanish is a pro-drop language that exhibits rich subject agreement on the verb.
In fact, all verbs in Spanish are bound roots that must attach to either a non-finite
ending or to person and tense suffixes. The infinitive, for example, is made up of
the bound root (e.g. compr- ‘buy’), the thematic vowel (compr-a), and the infini-
tival ending -r (compr-a-r ‘to buy’). Tense suffixes generally do not attach directly
to the verb but rather to the thematic vowel that is used by linguists to categorize
verbs into first, second or third class. The most common class has -a as the the-
matic vowel, with around 80% of verbs belonging to it. The second class thematic
vowel is -e, and the third is -i. In these two classes the thematic vowel is not as con-
sistent as in the first class, as you see in Table 1. Table 1 illustrates the infinitive and
the conjugation in the present tense indicative of the three classes, Table 2 provides
the simple past. In each case we find the root, followed by the thematic vowel,
with the tense, aspect and mood (TAM) morphology following, and the person/­
number agreement morphemes in final position. We have excluded the second
person plural as this form is not used in Latin America. It is important to note
that the third person singular of the present tense ends only in the thematic vowel,
with no overt person agreement. It is for this reason that it has been argued that
the third person is the default (Bruhn de Garavito 2003; McCarthy 2006, 2007).
 Verbal agreement in the L2 Spanish of speakers of Nahuatl 

Table 1.  Present tense of three classes of verbs

Infinitive (stem + theme compr-a-r (to buy) com-e-r (to eat) viv-i-r (to live)
vowel + infinitival ending)

Person (pronoun) Stem + theme vowel + person/number

1sg (yo) compr-o com-o viv-o

2sg (tú) compr-a-s com-e-s viv-e-s
3sg (él/ella)/2sg compr-a com-e viv-e
formal (usted)
1pl (nosotros/as) compr-a-mos com-e-mos viv-i-mos
3pl (ellos/ellas)/2pl compr-a-n com-e-n viv-e-n
formal and informal (Uds.)

The simple past or preterite is not as transparent as the other tenses, in par-
ticular the second person is not marked with -s which is the case in every other
tense. However, in informal speech many speakers in all dialects generalize and
add a final -s, producing non-standard forms such as compr-a-ste-s. The speakers
in the region we studied did this often and we did not code this as an error.

Table 2.  Preterite tense of three classes of verbs

Infinitive (stem + theme compr-a-r com-e-r viv-i-r
vowel + infinitival ending)

Person (pronoun) Stem + theme vowel + TAM + Person/Number

1sg (yo) compr-é com-í viv-í

2sg (tú) compr-a-ste com-i-ste viv-i-ste
3sg (él/ella)/2sg compr-ó com-i-ó viv-i-ó
formal (usted)
1pl (nosotros/as) compr-a-mos com-i-mos viv-i-mos
3pl (ellos/ellas)/2pl compr-a-r-on com-ie-r-on viv-ie-r-on
formal and informal (Uds.)

Although the present tense may be used in Spanish to refer to the present time
(Miro la tele ‘I am watching TV’) in many places the present progressive is used
instead (Estoy mirando la tele). This tense, which does not have identical uses to
the English progressive but is somewhat similar, is formed with the auxiliary estar
‘to be’ and the gerund of the main verb. We will see below that the present progres-
sive was often used by our participants instead of the simple present.
Within the Principles and Parameters framework (Chomsky 1981) it was
assumed the null subject parameter was defined by the presence of a small pro
 Alma P. Ramírez-Trujillo & Joyce Bruhn de Garavito

in subject position with which the verb agreed. Recent reformulations within the
Minimalist Program (Chomsky 1995) argue that pro is an unnecessary category.
Rather, verbal agreement suffixes are considered pronominal elements that include
the feature [+D] which checks the EPP2 feature by merging with the root of the
verb (Speas 1994; Alexiadou & Anagnostopoulou 1998). This is illustrated in (1).
It is interesting to note that this brings Spanish more in line with some analyses
regarding Nahuatl, as we will see below.
(1) a. comemos
b. [IP comei-mosj [VP [DP mosj] [V come]]]

In those cases in which the subject pronoun is expressed, it is assumed that it is in

topic position or somehow focalized (Mallén 1992; Zubizarreta 1998).

3.2  Nahuatl
Nahuatl is generally considered a polysynthetic language. Baker (1996) pro-
posed that polysynthesis constitutes a macroparameter the result of which is
so overarching that these languages are fundamentally different from Indo-
European languages such as Spanish. In particular, he argued that the very rich
agreement morphemes in polysynthetic languages were pronominal in nature,
thus satisfying the argument requirements of verbs. In consequence, not only
is word order variable given that the full argumental DPs can appear in a vari-
ety of non-argumental positions, but also arguments can be dropped, making
these languages subject and object drop. Note, however, that recent analyses of
Spanish (see above), at least as regards to the subject, are not incompatible with
Baker’s (1996) suggestion. Spanish and Nahuatl agreement morphemes can be
thought of as sharing the characteristic of being [+pronominal]. Nahuatl also
exhibits obligatory object agreement and there is no infinitival form of the verb.
That is, all verbal forms need to be specified for person and number. In Table 3
we see the paradigm for the present tense of an intransitive and a transitive
verb as conjugated in the region that we studied. The past tense is illustrated
in (2). Over time, a certain amount of simplification has taken place in San
Isidro Buensuceso, the village we studied, leading to the loss of distinctions that
were present in classical Nahuatl and may still be extant in other communities.3

.  EPP or Extended Projection Principle originally referred to the requirement that a Tense
Phrase project a Specifier (the subject). Currently this requirement is said to be due to unin-
terpretable EPP features on T that attract the subject to its specifier.
.  The paradigms and examples used here were collected by the authors in the village of San
Isidro Buensuceso with the help of informants.
 Verbal agreement in the L2 Spanish of speakers of Nahuatl 

Wedo not include in Table 3 the honorific forms as they are beyond the scope
of this paper.4

Table 3.  Present tense of intransitive (cry) and transitive (have) verbs in Nahuatl5
Pronouns Cry Have a house

Subject person agreement + Subject person agreement + object

stem (+number) agreement + stem (+number)

1sg (ne)5 ni -choka ni - k -pia kalli

1sg.sbj -cry 1sg.sbj - 3.obj.agr -have house
2sg (te) ti -choka ti -k -pia kalli
2sg.sbj -cry 2sg.sbj -3.obj.agr -have house
3sg (ye) choca ki -pia kalli
∅ -cry ∅ -3.obj.agr -have house
1pl (tehuan) ti -choca -h ti -k -pia -h kalli
1pl.sbj -cry -pl 1pl.sbj -3 p. obj -have -pl house
2pl (amehuan) An -choca -h An -ki -pia -h kalli
2.pl.sbj -cry -pl 1pl.sbj -3.obj.agr -have -pl house
3pl (yehuan) choca -h ki -pia -h kalli
∅ cry -pl ∅ -3.obj.agr -have -pl house

The past tense is formed by the addition of the prefix o- to the left periphery
of the verb and its obligatory agreement morphemes. In other words and simplify-
ing somewhat, we find tense > subject agreement > object agreement > verb root,
as illustrated in (2). In short, the tense morpheme precedes the person agreement
morphemes, and all are prefixes. The morpheme denoting plurality goes at the
end. There are some suffixes that have traditionally been associated with tense, for
example the suffix -s is said to signify future. We believe, however, that it is prob-
ably a marker of irrealis mood.
(2) O -ti -kim -pia ome kaltin
pst -2sg -3.obj.agr -have two houses
‘You had two houses.’

.  There is a great deal of variability regarding the spelling in Nahuatl. It is still common to
use Spanish spelling, but for some there is an effort to move away from this. The word ‘house’,
for example, can be found as calli (Spanish spelling) or kalli, as we have spelt it here.
.  Pronouns can be realized in different forms according to the degree of formality. For
example, the first person ‘ne’ (or neh) is produced as ‘nehua’ in a more formal setting or
‘nuhuatzin’ to show respect.
 Alma P. Ramírez-Trujillo & Joyce Bruhn de Garavito

In order to determine what the task facing the learner of Spanish L2, Nahuatl L1 is,
we have to try to ascertain just how different Nahuatl is from Spanish, in particu-
lar regarding subject verb agreement. As mentioned, according to Baker (1996)
and also as suggested by an anonymous reviewer, the two languages, ­Spanish
and Nahuatl, may be almost incomparable given that Nahuatl is a polysynthetic
language. However, there is no consensus. MacSwan (1998), arguing against
the Polysynthesis parameter, claimed that there was no deep difference between
Nahuatl and Spanish or English, particularly regarding the position of argument
NPs. According to MacSwan, Nahuatl is a Subject Verb Object language and, as
in Spanish, the subject is often omitted so the verb appears in first place. We do
not support MacSwan’s argument entirely, though we agree that the majority of
sentences used in production do seem to follow the pattern (Subject) Verb Object.
We also found evidence that, as in Spanish, word order is often related to discourse
pragmatic elements such as topicalization and focus. However, we have data that
seems to support the Polysynthesis Parameter based on negation, word order phe-
nomena, and noun incorporation.
Nevertheless, the important question is whether agreement is fundamentally
different in the two languages. The most salient difference is that Spanish agree-
ment markers are suffixes and in Nahuatl they are prefixes. Secondly, Nahuatl has
obligatory object agreement while Spanish does not, although it has been argued
that Spanish clitic object pronouns constitute, in fact, the realization of object
agreement (Franco 1993). It is notable that in almost all Spanish dialects in which
there is strong contact between Spanish and other languages, object clitics are used
to reduplicate direct objects (Acuña & Menegotto 1992, 1996 for Mapudungun;
Lastra 1992 for Otomí and Nahuatl; Escobar 2000 for Quechua; Sánchez 2004 for
Quechua; Díez Mendieta 2009 for Basque; Ramírez-Trujillo 2013 for Nahuatl),
evidence that these pronouns, at least for some speakers of Spanish, are indeed
object agreement markers. What seems fundamental to this paper is that both
languages are pro-drop, there is evidence that in both languages the agreement
markers are [+pronominal] and there is no evidence that somehow the process of
agreement or the mapping between the morphemes and the relevant functional
categories is different. We also find no evidence that the bundling of subject agree-
ment and tense features is different (Lardiere 2009), although, as suggested by
a reviewer, the position of the morphemic bundles is different. Finally, in both
languages, the third person is morphologically unmarked for subject agreement,
which suggests that in both languages the third person could serve as a default.
That is, as there is no morpheme indicating third person, there will be no feature
clash if a third person is used instead of a first or second person. There would be a
clash if the first person, for example, were to be used instead of the second (Bruhn
de Garavito 2003; McCarthy 2007). Naturally, there are many other differences
 Verbal agreement in the L2 Spanish of speakers of Nahuatl 

and it may well be the case that the languages differ to such an extent that the
similarities we are assuming hide a deeper significance that we have not yet found,
but this remains to be demonstrated.

4.  Second language acquisition of agreement and tense

Assuming we are on the right track, what does the Nahuatl learner of Spanish L2
have to acquire? The abstract EPP uninterpretable features will be checked in both
languages in the same way, by pronominal agreement on the verb. However, the
morphological reflex of these features will be different in Spanish and Nahuatl
(and, as noted above, their position in relation to the verb differs as well); that is,
the forms that encode the abstract features will be different, and it is these new
forms that the learners have to acquire. An anonymous reviewer suggests that the
task is therefore quite simple, a case of relexification. However, there is evidence
that even in those cases in which languages share the same abstract functional cat-
egories and features the learning of morphology is difficult. For example, although
French and Spanish share the same gender features, beginning level French L1
learners make frequent errors in gender agreement between nouns and articles
and between nouns and adjectives in L2 Spanish, including in cases of natural
gender, where one would expect that a reliance on semantics and/or French would
lead to correct production (Bruhn de Garavito & White 2002). In a similar fash-
ion, although both German and Spanish distinguish between adjectival and ver-
bal passives by the use of a different auxiliary (ser and estar in Spanish, wurden
and sein in German), German L1 speakers do not have an advantage over English
L1 speakers in recognizing the properties of passives in L2 Spanish (Bruhn de
G
­ aravito 2009).
The acquisition of agreement and tense morphology has been the subject of a
great deal of research, both within the domain of DP and within the domain of TP.
The one thing that is generally agreed upon is that second language learners have
problems producing accurate morphology and, although they exhibit fewer errors
in comprehension, comprehension is not perfect either (Lardiere 1998a, 1998b,
2000, 2003; White, Valenzuela, Kozlowska-Macgregor, & Leung 2004). The ques-
tion that arises revolves around the causes for the problems related to morphology,
or, put another way, what the lack of production and perhaps comprehension of
accurate morphology in second language learners tells us about the mental repre-
sentation of second languages.
The answer to the above question is roughly represented by two approaches
(see White 2003 for a comprehensive summary of this debate). On the one hand
it is claimed that morphology drives syntax (Rohrbacher 1999) and, therefore, a
 Alma P. Ramírez-Trujillo & Joyce Bruhn de Garavito

deficit in explicit morphology is the reflex of a deficit at the representational level.

Among others, Meisel (1997, 2011) argues that Universal Grammar (UG) is not
available in adult L2 and, as a result, the mental representation and the acquisi-
tional process are different for older learners. Other scholars argue for a local-
ized impairment. According to Hawkins and colleagues (Hawkins & Franceschina
2004; Hawkins & Hattori 2006; see also Tsimpli & Dimitrakopoulou 2007), second
language learners are not able to instantiate in their second languages features that
are not present or are valued differently in their L1. More recently, Hawkins (2009)
suggests that L2 learners, unlike L1 learners, rely on statistical information exclu-
sively, which explains, for example, their reliance on a default in their production
of features such as gender concord. Hawkins and Casillas (2008), which looked at
verbal morphology in L2, claims that, although L2 grammars are organized in the
same way as first language grammars, at the initial stages L2 vocabulary entries
differ because the overt morphological items are identified by way of the nodes
with which they co-occur, and not, as in native speaker grammars, activated by the
nodes themselves into which they are inserted. Evidence for this claim is the pro-
duction of be + bare V (I am read) in early stages of acquisition. In more advanced
speakers this approach can be restructured, but only uninterpretable features pres-
ent in the L1 may be activated. Both these positions would predict differences
between early bilinguals and adult second language learners, either because of the
acquisition process being different or because of differences in how the functional
items in the lexicon relate to the abstract categories. For Hawkins, as for Meisel,
the determining factor for the presence of a deficit at the level of functional catego-
ries is age. Therefore, the prediction is that early bilinguals should differ from late
bilinguals no matter how much input is accessed.
In contrast to the previous positions, scholars of the syntax before morphol-
ogy camp argue that syntax, including functional categories, may be intact even
when the morphology is not in place. Prévost and White (2000; see also Haznedar
2003; Lardiere 1998a, 1998b; among others) suggest that there is no fundamental
deficit but rather a problem of mapping between the respective functional cat-
egories and/or features and the appropriate morphological realization. Arguing
for this position Grüter, Lew-Williams and Fernald (2012) examined the way in
which L1 and L2 learners of Spanish used the gender of determiners as predic-
tive cues in on-line comprehension. In an eye-tracking study participants were
shown two pictures, and then heard a phrase naming one of the objects shown
with the appropriate determiner. Their eyes moved to the named picture as soon as
the determiner was pronounced. L2 learners were only able to do this with novel
words. The authors concluded that, because of the nature of the input, young chil-
dren form tight associations between determiners and nouns in the L1 lexicon (see
also Hopp 2012; Liceras, Díaz & Mongeon 2000 for similar conclusions). Older
 Verbal agreement in the L2 Spanish of speakers of Nahuatl 

learners have many more cues to identify lexical items, such as the written word
and rich context, and therefore need not pay much attention to transitional prob-
abilities and co-occurrence relations. As a result, they form weaker associations. In
other words, Grüter et al. (2012), like Hawkins (2009), highlight the importance of
the nature and frequency of the input but, in contrast to Hawkins, at the same time
they maintain there is no representational impairment. They see no contradiction
in a role for statistical learning in both L1 and L2 while at the same time assigning
a crucial role for UG. Although their study focuses on gender agreement in noun
phrases, attention to transitional probabilities may also be involved in the process
of acquisition in subject object agreement, although the mental representation
must be clearly anchored in functional categories and features provided by UG.
To summarize, researchers agree that morphology is problematic in second
language acquisition, but they do not agree on the causes. For scholars such as
Hawkins the main factor is age. Adult learners either unable to access uninterpre-
table features not found in their L1 and therefore must have recourse to differ-
ent learning processes such as a reliance on statistical information. In contrast, for
researchers such as White and Lardiere, among others, age is not a determining fac-
tor. Grüter et al. (2012) suggests that statistical information is important in both L1
and L2, as it allows associations to be formed, but type of input is important. Chil-
dren form stronger associations between features and the forms associated with
them because these features are the only cues available to them, while in adult L2
acquisition associations are weaker because other cues, including the written word
and forms of instruction, are also present in their learning context. In the present
paper we will provide evidence that the no-impairment position is the correct one.
Our research questions are the following:

1. Will Nahuatl early learners of Spanish differ from late learners in their sup-
pliance of appropriate subject verb agreement and tense morphology? If age
of acquisition is the determining factor we should find differences between
the two groups, regardless of the amount of contact with the monolingual
standard.
2. Will speakers with substantial contact with native (monolingual) Spanish
speakers differ from those with little contact in their suppliance of appropri-
ate subject verb agreement and tense morphology? If quantity and type of
input are the determining factors there should be a different between the two
groups, regardless of age of acquisition.
3. What type of errors will the Spanish L2 speakers commit? There is evidence
that L2 learners rely on the least specified forms as a default (Bruhn de
G
­ aravito 2003; McCarthy 2007). In Spanish the least specified form of the
verb is the third person and, interestingly, this is also the case for Nahuatl.
 Alma P. Ramírez-Trujillo & Joyce Bruhn de Garavito

5.  Methodology

5.1  Description of participants

The language profile of the participants was determined with an oral question-
naire. There were three groups of speakers: a group of post-puberty L2 learners
of Spanish, a small group of early bilinguals, and a small group of monolingual
Spanish speakers who spoke the standard variety of the region. The first group
was made up of 21 participants who had spoken Nahuatl since birth and acquired
Spanish as a second language. Of these we only retained 19 because 2 were unable
to carry out the translation task (see below). Some of the participants report hav-
ing learned Spanish because it was the language of the school, but they did not
complete many years of schooling. Recall also that the first years of schooling are,
at least on paper, conducted in Nahuatl. Only one participant attended high school
but did not finish. Most reported that they had really learned Spanish ‘on the street
and what I hear around places’ (en la calle y de lo que oigo por ahí). The mean age
at which the participants reported having begun to learn Spanish is 11.7, with a
range 6–20. The mean age of the participants is 38.8, with a range of 13–68.
The second group was made up of 6 participants who reported having spoken
both Nahuatl and Spanish from birth.6 The mean age of this group was younger,
19.6, ranging between 14 and 30. The third group consisted of 5 monolingual
Spanish speakers who lived in the region and were in contact with Nahuatl only.
We included this group in order to determine what the agreement rules were for
the variety spoken there.
Most of the participants own small plots of land where corn is cultivated and
farm animals graze, which again shows how different these participants are to
those found in the majority of studies into second language acquisition. However,
because working off the land does not provide for the family, most men and some
women also have other jobs in factories, markets or construction sites outside of
San Isidro. These social conditions were also noted in Hill and Hill (1986). These
facts are important to the present study as it is precisely those who work outside of
the town who are in greater contact with the standard of the region.
The description of the participants is included in Appendix 1. When organiz-
ing these data it became clear that we were dealing with two groups of people: one
who reported being equally comfortable in Nahuatl and Spanish and one who is
more comfortable in Nahuatl, regardless of their onset of acquisition. There is also
a clear association between degree of contact with Spanish and comfort with the

.  It was impossible to find more early bilinguals in the same village.
 Verbal agreement in the L2 Spanish of speakers of Nahuatl 

language. Organizing the data along these lines results in a grouping that is dif-
ferent from that related to age of onset. This information becomes crucial below.

5.2  Description of the testing procedures

Because our interviews showed the participants in this study were mostly illit-
erate in Spanish and Nahuatl, we were faced with the problem of determining
their level in Spanish. We decided to try an oral translation task from Spanish into
Nahuatl to make sure that they, at a minimum, understood Spanish relatively well.
The translations were recorded, transcribed, translated and analyzed by a Nahuatl
native speaker who was literate. We found the participants did not have the same
concept of translation that we do and they would often interpret the sentence they
heard rather than simply translate it, and try to imbed it in a Nahuatl context that
made sense to them. We do not attribute these difficulties to a lack of Spanish
comprehension but rather as evidence of how much we have to learn in this type
of cross-cultural testing. An example is the following:
(3) a. Sentence to be translated:
José llegó de Estados Unidos. ¿Lo quieres ir a ver?
‘José arrived from the United States. Do you want to go to see him?’
b. Translation given:
0-  k -ihto:s- neki 0- wa:hlaw
3sg-3.obj.agr -say - want 3sg -come
mo -te:lpo:chtsi:n we:noh tonces tla: 0- wa:law
3sg.poss -boy well then if 3sg -come
x- 0 -k- o- no:tsa,
imp-3sg-3sg-obj.agr- pst- speak,
x- 0- k- o- no:tsa wa:n nika:n
imp-3sg-3sg-obj.agr- past-speak and here
ni- k- ita- s
1sg-3.obj.agr-see- fut.sg
‘It means, this boy came, then if he comes, they will tell me and I am
going to see him.’

It is interesting to note that the introduction to the above translation, ‘it means’ is
in fact a word for word translation of the Spanish quiere decir ‘it wants to say’. After
500 years of contact with Spanish there is a great deal of influence of this language
on Nahuatl.
On the basis of the translation task we excluded two speakers who were unable
to carry it out, but we were fairly liberal in what counted as an acceptable transla-
tion. This is clearly not a fine-grained instrument.
 Alma P. Ramírez-Trujillo & Joyce Bruhn de Garavito

Besides the translation task, there were two main tasks that tested subject verb
agreement: an elicited production task and a spontaneous production task. We
only report on the elicited production task due to space limitations (see Ramírez-
Trujillo 2013 for other results).
The objectives of the production task were to elicit subject verb agreement and
also object pronouns, which we do not report on here. It consisted of 38 pairs of pic-
tures, in the first of which someone was carrying out an action with an object, and
in the second a different action with the same object. For example, the speaker saw
the following picture and was asked in Spanish to describe the actions of the person.
The expected answer was along the lines of ‘She is cutting the melon.’ Then the per-
son saw the second picture and was asked: and afterwards/before with the melon?
(¿y después, con el melón?), with the expected answer ‘She ate it.’ Because of the way
the second question was framed the response was always expected to refer to the past.

  
Figure 1.  Elicited Production Task

Of the 38 pairs of pictures, 10 were designed to elicit the first person singular
(one picture in the present, one in the past), 10 the second person singular and
10 the third person. There were 8 distracters. In the case of the first person the
participants were asked to imagine they were in the picture, and in the case of the
second person that the interlocutor was in the picture. They were told not to use
the formal usted given that this would make the second and third person indistin-
guishable regarding agreement. Distracters included the verb gustar ‘to like’. The
responses were recorded and then transcribed. They were coded for person and
for agreement, and within agreement for tense. Agreement could be matched, or
mismatched to what the answer in the standard variety would be. If there was a
mismatch the type of error was noted.

6.  Results

The first question we asked in this paper was whether there were differences in
accuracy rates in the production of subject verb agreement between early and late
 Verbal agreement in the L2 Spanish of speakers of Nahuatl 

bilinguals. The answer is negative. A repeated measures ANOVA7 failed to pro-

duce a significant difference between the three groups (control, early bilinguals,
late bilinguals) (F(2, 27) = 2.516, p = .0995), and, although there was a main effect
for sentence type (F(5, 135) = 8.887, p = .0001), there was no interaction between
group and type of sentence (F(10, 135) = 1.726, p = .0808). Furthermore, an exam-
ination of the individual results shows the early bilinguals split exactly down the
middle, with half (3 people) producing hardly any errors and half with low accu-
racy rates.
We turn then to the second question, whether input in the form of contact
with Spanish speakers in the region had an effect on accuracy. Details regard-
ing contact (see Appendix 1) show that 16 participants, including 3 of the early
bilinguals, have very little contact with Spanish, either because they hardly ever
leave the village or because they work with people who speak Nahuatl. These all
responded that they felt more comfortable speaking Nahuatl. In contrast, 9 partic-
ipants, who say they feel comfortable speaking both languages (one actually pre-
fers Spanish), travel frequently and/or work outside the village. Based on contact,
therefore, we also have three groups that we will refer to as low contact (N = 15),
high contact (N = 10), and control (N = 5).
Figure 2 represents the differences between the three groups for each of the
different person agreements, first (1), second (2) and third (3) in the two tenses,
present and past. The control group did not produce errors in agreement.

100
90
80
70
60 Control
50 More Contact
40 Less Contact
30
20
10
0
1 2 3 1 Past 2 Past 3 Past
Present Present Present

Figure 2.  Mean accuracy rates for each of the three groups according to sentence type

Results of a repeated measures ANOVA in which the independent variables

are Group (high contact, low contact, control) and Sentence Type (person/tense)

.  We realize these results are only tentative given the number of subjects in the bilingual
group.
 Alma P. Ramírez-Trujillo & Joyce Bruhn de Garavito

shows a main effect for Group (F(2, 27) = 14.698, p = .0001), a main effect for the
Sentence Type (F(5, 135) = 9.476, p = .0001) and an interaction between Group
and Sentence Type (F (10, 135) = 2.736, p = .0043). This indicates that contact with
Spanish outside the town is an important variable.
Three things are clear from Figure 2. The group of low contact is less accu-
rate than the other two groups, appears to be more accurate on the past tense
than in the present and is more accurate on the third person than in first and
second. This is confirmed in part by the statistics. A Factorial ANOVA shows
that, for both the present and the past, a significant difference is found between
the groups in the first and second persons (F (2, 27) = 13.654, p = .0001 for the
first person present; F (2, 27) = 9.75), p = .0006 for the second person present;
F (2, 27) = 8.78, p = .0012 for the first person past; F (2,27) = 7.831, p = .0021
for the second person past). In all four cases a Post Hoc Scheffe-F test showed a
significant difference between the low contact group and the high contact group
on the one hand and between the low contact group and the control group. No
difference was found between the controls and the high contact group. Regard-
ing the third person, we found no significant difference between the groups, but
there could be a tendency towards significance if the number of participants were
higher (F (2, 27) = 3.243, p = .0547 for the third person present; F (2, 27) = 3.33,
p = .051 for the third person past). These results are also confirmed by the indi-
vidual responses found in Appendix 2. The table in Appendix 2 was organized
according to accuracy on the first person present due to the fact that there is an
implicational hierarchy, high accuracy on the first and second persons predict
accuracy overall, while high accuracy on the third person, particularly the past
tense third person, is much higher for all participants and is not predictive of suc-
cess in other persons. In the next section we will turn to the third question, the
use of a default, by examining the type of errors committed.

6.1  E
 rror types
It is clear from the group and individual results that the highest accuracy rate
for the bilinguals is in the third person, particularly in the past. The high contact
group produced only two errors in the third person past (98% accuracy), and 8 in
the third person present, 6 of which were produced by the same person (L2-19)
(92% group accuracy). The low contact group produced 14 errors in total in the
third person past (90.66% accuracy) and 33 errors in the third person present
(78% accuracy). As Figure 1 shows, the accuracy rate for first and second persons
is much lower in the low contact group, below 50% in the present and below 70%
in the past. Given the hypothesis of the third person as underspecified default, we
must ask whether the most common type of error is the replacement of the second
 Verbal agreement in the L2 Spanish of speakers of Nahuatl 

and third persons by the third. As Table 4 shows, this is not the case. Both the ger-
und and the infinitive were used more frequently as a default.

Table 4.  Raw number of errors and percentages organized according to type of answer
produced
Present Past

Gerund Infinitive 3Ps Other Gerund Infinitive 3Ps Other

High contact 36 (46%)    6 (7%) 3 (4%) 3 (4%) 9 (12%) 16 (21%)   1 (1%) 4 (5%)
Low contact 130 (47%) 29 (10%) 10 (4%) 0 (0%) 40 (14%) 32 (12%) 28 (10%) 8 (3%)

Across the two groups there was a mismatch between the expected answer
and the answer produced 355 times. Of these anomalous answers, 78 were pro-
duced by the high contact group, 277 by the low contact group. As Table 4 shows,
for the former group the raw number of errors is not high but nevertheless only
4% of those errors consisted in the replacement of the expected form with the
third person. In total, 53% of errors consisted in the production of a non-finite
form realized by the gerund or an infinitive without the expected auxiliary, estar
‘to be’ for the gerund, haber ‘to have’ for the infinitive. In the past, the non-finite
form of preference was the infinitive without a main verb, but again non-finite
forms dominate. The same pattern is reflected in the low contact group: 57% of
errors consist in the use of a non-finite form in the present, and 26% in the past.
The use of the third person as a default is minimal. Regarding the category ‘other’,
it consists of unexpected errors, in which first or second person agreement mark-
ings are used incorrectly, for which we have no explanation. In (4) we show an
example of the third person produced in a context where the first is expected,
in (5) we illustrate the use of the gerund and in (6) the use of the infinitive. In (6)
the second answer should have been in the past. The responses in (5) and (6) are
important because the object pronoun is positioned correctly. We will return to
this point later.
(4) a. Use of third person for first person.
Experimenter –Imagina que tú eres el de la foto, ¿qué haces tú aquí?
‘Imagine that you are the one in the photo, what are you doing here?’
Participant: –Lo está cosiendo
  it is-3.prs sewing-ger
‘I (he?) am sewing it.’
(5) Use of the gerund for third person.
Experimenter –¿Qué hace esta señora?
‘What is this lady doing?’
 Alma P. Ramírez-Trujillo & Joyce Bruhn de Garavito

a. Participant –Comiendo el melón.

  eating-gerund the melon
‘She is eating the melon.’
Experimenter –¿Y antes con el melón?
‘And before with the melon?’
b. Participant –Cortándolo.
  cutting-gerund-it
‘She is cutting it.’
(6) Use of the infinitive for the second person.
Experimenter –Imagina que tú eres quien está en la
fotografía. ¿Qué haces aquí?
‘Imagine that you are who is in the photograph. What are you
doing here?’
a. Participant –Lucir mi sombrero. (correct in standard Spanish)
show-off-inf my hat
Experimenter –¿Y antes con el sombrero?
‘And before with the hat?’
b. Participant –Tejerlo.
  Weave-inf-it
‘She was weaving it.’
The question that these results point to is whether the use of non-finite forms in
finite contexts is clear evidence of a deficit in the area of syntax. We will briefly
turn to this question in the next section.

6.2  Morphology and syntax

The sentences in the elicitation task were quite simple. However, object clitic pro-
nouns were produced frequently (see Ramírez-Trujillo 2013), including when a
single infinitive or present participle was used. This is important because the posi-
tion of the clitic is an indication of the correct positioning of the verb. Following
Kayne (1991; see also Uriagereka 1995; among others) we assume the clitic is in
a fixed position and the conjugated verb in Spanish has moved to T, while non-
finites move to some higher position, probably C. This explains why object clitics
precede conjugated verbs (7), while they follow non-finite forms (8).8
(7) a. Lo como.
it eat-prs-1sg
‘I eat it.’

.  It is beyond the scope of this paper to examine the complete quite complex system of clitic
object pronouns in Spanish.
 Verbal agreement in the L2 Spanish of speakers of Nahuatl 

b. *Comolo.
  Eat-prs-1sg it (=(7a))
(8) a. Para comerlo.
in order to eat-inf-it
‘In order to eat it.’
b. *Para lo comer. (=(8a))
  in order to it eat-inf
c. Comiéndolo.
eating-ger-part-it
‘Eating it.’
d. *Lo comiendo. (=(8c))
  it eating-ger

In total in this task we found 36 responses that consisted of an infinitive with a

pronoun, as in (6b) and 4 in which the gerund was followed by a pronoun, as in
(5b). There were no errors in the position of the pronoun relative to the verb; that
is, the pronoun was correctly positioned following the non-finite form. We also
found no errors in the position of the pronoun in the many cases of conjugated
verbs and pronouns. It is clear that the grammar of these learners distinguishes
between the position of non-finite verbs and finites in the sentence.

7.  Discussion and conclusions

This paper set out to examine the production of agreement between the verb and
the subject in the correct production of tense in the production of Nahuatl L1
learners of Spanish. The first question we asked was whether there was evidence
here that early bilinguals differed from late learners in the production of the cor-
rect morphology. No indication of a difference between early and late bilinguals
was found. Early bilinguals, of which there were only six among the participants,
were evenly divided down the middle: half produced almost no errors and half a
very high rate of error (see Appendix 2). In the same way, the late learners include
several who made no errors, and a group that made many. Statistically, there was
no difference between the groups, but we cannot assign great validity to this given
the small number of participants in the early bilingual group.
The second question we addressed was whether the quantity and possibly the
quality of input was a determining factor. Several scholars have recently argued
for the importance of the strength of associations between forms, associations that
must depend to a certain extent on the input (Hawkins 2009; Grüter et al. 2012).
The results of this study confirm the importance of input and therefore provide
indirect evidence that the strength of associations do matter. Speakers who had
 Alma P. Ramírez-Trujillo & Joyce Bruhn de Garavito

extensive contact with the Spanish spoken in the region were significantly less
likely to make errors in agreement and/or tense, no matter what the age at which
they were exposed to Spanish. However, we argue that it is not only the quantity
of input, but also the quality, in that those who worked outside the town would
encounter standard Spanish, while those that did not leave would have heard
many instances of lack of agreement between verbs and subjects.
The importance of the quality and quantity of input is counter to what Meisel
(2011) would predict. The finding also runs counter in some measure to Hawkins,
for whom differences in the production of morphology are an indication of a syn-
tactic deficit that is related to the age of acquisition. It is, however, in agreement
with Grüter et al.’s (2012) position that strength of association is important, and
therefore the type of input matters. The strength of associations seems to us partic-
ularly relevant to the acquisition of morphology given that, within the Mininmal-
ist Program, it is part of the lexicon, although it belongs to the class of functional
elements.
The finding that the quantity and quality of input matter more than age is an
important result, but equally important is the fact that the pattern of errors was
the same for both groups. The overwhelming choice of bare non-finite forms in
a context that calls for either a conjugated main verb or a conjugated auxiliary
needs an explanation. It cannot be the result of statistical learning as the produc-
tion of bare non-finite forms in these contexts would not be frequent, and in fact
the monolingual Spanish speakers never produced them. Nor can we explain it
as a consequence of L1 transfer given that in the L1 verbal forms without agree-
ment do not exist in the language.9 The most reasonable explanation is that, for
these speakers, non-finite is a default, in fact, a default that is less specified than
the third person as it does not include agreement or tense. The main difference
between the learning context for these participants and those that have been
found to use the third person default is the fact that these are not formal learn-
ers, they are not taught in a classroom where verbal morphology is the focus of
teaching. We may suppose, therefore, that the least specified form that still gets
the meaning across is the one chosen, and this form is a non-finite form for the
native Nahuatl speakers.
The results of this paper lead to the age-old question: must we interpret them
as evidence for a deficit at the functional level? Prévost and White (2000) provided
evidence that L2 learners do not place finite forms in non-finite positions, although

.  The need to form dictionary entries has led to the inclusion of certain endings in order
to constitute citation forms. For example, verbs borrowed from Spanish are overwhelmingly
made up of the Spanish infinitive with the Nahuatl ending -oa (e.g. trabajar-oa).
 Verbal agreement in the L2 Spanish of speakers of Nahuatl 

non-finite forms may appear in positions s where a finite form is expected. ­Arguing
on the basis of this, they suggested the Missing Surface Inflection Hypothesis
(MSIH) according to which learners have appropriate representations of finiteness
and agreement in the L2 syntax but have a problem mapping the correct forms.
Recall that, in this case, the forms are what the learners have to acquire. Evidence in
our data supporting the MSIH comes from the position of clitics. We have assumed,
following Kayne (1991) that the finite verb in Spanish climbs to T where it adjoins
to the clitic, while the non-finite forms climb over the clitic to a position higher
in the derivation, perhaps in C. This accounts for the fact that clitics in Spanish
precede conjugated verbs and obligatorily follow non-finite forms when there is no
auxiliary/main verb. We found absolutely no errors in the position of clitics, either
with conjugated verbs or with non-finite forms, evidence that the speakers are cor-
rectly differentiating between the finite and non-finite forms. In other words, there
is no problem in the computational system, but only in accessing the new forms
that represent the abstract categories. The syntax is intact and problems with mor-
phology do not point to syntactic deficits.
A last point we would like to address is the possibility that the production
results for the low contact group may be reinforced by other speakers in the lin-
guistic environment who also diverge from the norm in their speech, which is
quite probable. Meisel (2011) suggests that L2 learners are the main and perhaps
the only agents of language change due to contact. Certain characteristics of the
speakers in the present sample seem to be found in many communities in contact
with Indigenous languages, for example the reduplication of the direct object by
an accusative clitic with masculine singular features, no matter what the gen-
der and number of the direct object may be (see Ramírez-Trujillo 2013) and, as
we have seen, variability in the production of morphology. These commonalities
have led to the perception that there exists a variety often referred to in Mexico
as Indian Spanish (Lastra 1992). If this is indeed a new variety, future research
should address several issues, among which are: how stable is this variety? Is
variability an inherent characteristic of contact languages, and if so, how do we
account for this variability? Will the changing sociolinguistic circumstances lead
to its e­ ventual disappearance? There is evidence that variability may be the norm
in certain individuals and communities, even in the case of monolingual speakers
(Bruhn de Garavito & Atoche 2006). In second language research we usually take
variability as an indication of some sort of problem either of mapping or at the
representation level (Bley-Vroman 2009). This cannot be the case if it is the norm
of monolingual native speakers. We believe the answer to some of these questions
is relevant to the study of languages in contact, language change, second language
acquisition, and the other well-known case of variability in production: heritage
languages.
 Alma P. Ramírez-Trujillo & Joyce Bruhn de Garavito

References

Acuña, M.L., & Menegotto, A.C. (1992). Dativo sin a y verbos pronominales sinse: Rasgos
­dialectales del español de la zona mapuche. Revista Argentina de Lingüística, 12, 13–20.
Acuña, M.L., & Menegotto, A.C. (1996). El contacto lingüístico español/mapuche en la
­Argentina. Signo y Seña, 6, 235–274.
Alexiadou, A., & Anagnostopoulou, E. (1998). Parametrizing AGR: Word order, V-movement
and EPP-checking. Natural Language and Linguistic Theory, 16, 491–539.
DOI: 10.1023/a:1006090432389
Baker, M. (1996). The Polysynthesis Parameter. Oxford: OUP. DOI: 10.1017/s0022226798227113
Bley-Vroman, R. (2009). The evolving context of the fundamental difference hypothesis. Studies
in Second Language Acquisition, 31(2), 175–198. DOI: 10.1017/s0272263109090275
Bruhn de Garavito, J. (2003). Learners’ competence may be more accurate than we Think:
­Spanish L2 and agreement morphology. In J. M. Liceras, H. Zobl, & H. Goodluck (Eds.),
Generative Approaches to Second Language Acquisition (GASLA) 6, University of Ottawa
(pp. 17–23). Somerville, MA: Cascadilla Press.
Bruhn de Garavito, J. (2009). Eventive and stative passives: The role of transfer in the acqui-
sition of ser and estar by German and English L1 speakers. In J. Collentine, M. García,
B. ­Lafford, & F. Marcos Marín (Eds.), Selected Proceedings of the 11th Hispanic Linguistics
Symposium (pp. 27–38). Somerville, MA: Cascadilla Proceedings Project.
Bruhn de Garavito, J., & Atoche, C. (2006). Variability in contact Spanish: Implications for
­second language acquisition. In C. Lefebvre, L. White, & C. Jourdan (Eds.), L2 Acquisition
and Creole Genesis. Dialogues (pp. 353–369). Amsterdam: John Benjamins.
DOI: 10.1075/lald.42.07bru
Bruhn de Garavito, J., & White, L. (2002). L2 acquisition of Spanish DPs: The status of gram-
matical features. In A. T. Pérez-Leroux & J. Liceras (Eds.), The Acquisition of Spanish
­Morphosyntax: The L1/L2 Connection (pp. 151–176). Dordrecht: Kluwer.
DOI: 10.1007/978-94-010-0291-2_6
Chomsky, N. (1981). Lectures on Government and Binding. Dordrecht: Foris.
DOI: 10.2307/2273965
Chomsky, N. (1995). The Minimalist Program. Cambridge, MA: The MIT Press.
DOI: 10.1017/s0272263197241070
Díez Mendieta, B.A. (2009). Pronombres clíticos en hablantes bilingües: Contacto lingüístico
­español – euskera. Unpublished Master thesis. The University of Western Ontario, London,
Ontario, Canada.
Escobar, A.M. (2000). Contacto social y lingüístico. El español en contacto con el quechua en el
Perú. Lima, Fondo Editorial de la Pontificia Universidad Católica del Perú.
DOI: 10.1017/s0022216x00014930
Franco, J. (1993). On object agreement in Spanish. Unpublished Ph.D. dissertation. University of
Southern California.
Grüter, T., Lew-William, C., & Fernald, A. (2012). Grammatical gender in L2: A production or a
real-time processing problem? Second Language Research, 28(2), 191–215.
DOI: 10.1177/0267658312437990
Hawkins, R. (2009). Statistical learning and innate knowledge in the development of second lan-
guage proficiency: Evidence from the acquisition of gender concord. In A. G. Benati (Ed.),
Issues in Second Language Proficiency (pp. 63–78). London: Continuum.
 Verbal agreement in the L2 Spanish of speakers of Nahuatl 

Hawkins, R., & Casillas, G. (2008). Explaining frequency of verb morphology in early L2 speech.
Lingua, 118, 595–612. DOI: 10.1016/j.lingua.2007.01.009
Hawkins, R., & Franceschina, F. (2004). Explaining the acquisition and nonacquisition of
­determiner-noun gender concord in French and Spanish. In J. Paradis & P. Prévost
(Eds.), The Acquisition of French in Different Contexts (pp. 175–205). Amsterdam: John
Benjamins. DOI: 10.1075/lald.32.10haw
Hawkins, R., & Hattori, H. (2006). Interpretation of English multiple wh-questions by Japanese
speakers: A missing uninterpretable feature account. Second Language Research, 22(3),
269–301. DOI: 10.1191/0267658306sr269oa
Haznedar, B. (2003). The status of functional categories in child second language acquisition:
Evidence from the acquisition of CP. Second Language Research, 19, 1–41.
DOI: 10.1191/0267658303sr212oa
Hill, J.H., & Hill, K.C. (1986). Speaking Mexicano: Dynamics of Syncretic Language in Central
Mexico. Tucson, AZ: The University of Arizona Press. DOI: 10.1017/s0142716400008134
Hopp, H. (2012). Grammatical gender in adult L2 acquisition: Relations between lexical and
syntactic variability. Second Language Research, 29(1), 33–56.
DOI: 10.1177/0267658312461803
Kayne, R.S. (1991). Romance clitics, verb movement, and PRO. Linguistic Inquiry, 22(4),
647–686.
Lardiere, D. (1998a). Case and tense in the ‘fossilized’ steady state. Second Language Research,
14, 1–26. DOI: 10.1191/026765898674105303
Lardiere, D. (1998b). Dissociating syntax from morphology in a divergent end-state grammar.
Second Language Research, 14, 359–375. DOI: 10.1191/026765898672500216
Lardiere, D. (2000). Mapping features to forms in second language acquisition. In J. Archibald
(Ed.), Second Language Acquisition and Linguistic Theory (pp. 102–129). Oxford: Blackwell.
DOI: 10.1017/s0022226702211925
Lardiere, D. (2003). Second language knowledge of [± past] vs. [±finite]. In J. M. Liceras, H. Zobl, &
H. Goodluck (Eds.), Proceedings of the 6th Generative Approaches to Second Language Acqui-
sition Conference (GASLA 2002) (pp. 176–189). Somerville, MA: Cascadilla Press.
Lardiere, D. (2009). Some thoughts on the contrastive analysis of features in second language
acquisition. Second Language Research, 25(2), 173–227. DOI: 10.1177/0267658308100283
Lastra, Y. (1992). Is there an indian Spanish? In J. Amastae, G. Goodall, M. Montalbetti, &
M.  Phinney (Eds.), Contemporary Research in Romance Linguistics (pp. 123–133).
­Amsterdam: John Benjamins. DOI: 10.2307/416129
Liceras, J., Díaz, L., & Mongeon, C. (2000). N-Drop and determiners in native and non-native
Spanish: More on the role of morphology in the acquisition of syntactic knowledge. In
R.  P.  Leow & C. Sanz (Eds.), Spanish Applied Linguistics at the Turn of the Millenium:
Papers from the 1999 Conference on the L1 and L2 Acquisition of Spanish and Portuguese
(pp. 67–96). Somerville, MA: Cascadilla Press.
MacSwan, J. (1998). The argument status of NPs in Southeast Puebla Nahuatl: Comments on the
Polysynthesis Parameter. Southwest Journal of Linguistics, 17(2), 101–114.
MacSwan, J. (1999). A Minimalist Approach to Intrasentential code Switching. New York, NY:
Garland. DOI: 10.1177/13670069000040030803
MacSwan, J. (2000). The architecture of the bilingual language faculty: Evidence from intrasen-
tential code switching. Bilingualism: Language and Cognition, 3(1), 37–54.
DOI: 10.1017/s1366728900000122
 Alma P. Ramírez-Trujillo & Joyce Bruhn de Garavito

Mallén, E. (1992). Subject topicalization and inflection in Spanish. Theoretical Linguistics 18:
179–208. DOI: 10.1515/thli.1992.18.2-3.179
McCarthy, C. (2006). Default morphology in second language Spanish: Missing inflection
or underspecified inflection? In J. P. Montreuil & C. Nishida (Eds.), New Perspectives on
Romance Linguistics: Selected Papers from the 35th Linguistic Symposium on Romance Lan-
guages (pp. 201–212). Amsterdam: John Benjamins. DOI: 10.1075/cilt.275.16mcc
McCarthy, C. (2007). Morphological Variability in Second Language Spanish. Unpublished Ph.D.
dissertation. McGill University, Montreal, Canada.
Meisel, J. (1997). First and Second Language Acquisition. Cambridge: CUP.
DOI: 10.1017/s0272263112000770
Meisel, J.M. (2011). Bilingual language acquisition and theories of diachronic change:
­Bilingualism as cause and effect of grammatical change. Bilingualism: Language and Cogni-
tion, 14(2), 121–145. DOI: 10.1017/s1366728910000143
Prévost, P., & White, L. (2000). Missing surface inflection or impairment in second language
acquisition? Evidence from tense and agreement. Second Language Research, 16(2),
103–133. DOI: 10.1191/026765800677556046
Ramírez-Trujillo, A.P. (2013). Producción de la concordancia en español por hablantes de náhuatl.
Unpublished Ph.D. dissertation. The University of Western Ontario, London, Ontario,
Canada.
Rice, S., Libben, G., & Derwing, B. (2002). Morphological representation in an endangered,
polysynthetic language. Brain and Language, 81, 473–486.
DOI: 10.1006/brln.2001.2540
Rohrbacher, B. (1999). Morphology Driven Syntax: A Theory of V to I Raising and Pro-drop.
Amsterdam, John Benjamins. DOI: 10.1075/la.15
Sánchez, L. (2004). Quechua-Spanish Bilingualism. Interference and Convergence in Functional
Categories. Amsterdam: John Benjamins. DOI: 10.1075/lald.35
Speas, M.J. (1994). Null arguments in a theory of economy of projection. University of Massa-
chusetts Occasional Papers in Linguistics, 17, 179–208.
Tsimpli, I.M., & Dimitrakopoulou, M. (2007). The Interpretability Hypothesis: Evidence from
wh-interrogatives in second language acquisition. Second Language Research, 23(2),
215–242. DOI: 10.1177/0267658307076546
Uriagereka, J. (1995). Aspects of the syntax of clitic placement in Western Romance. Linguistic
Inquiry, 26, 79–124.
White, L. (2003). Second Language Acquisition and Universal Grammar. Cambridge: CUP.
DOI: 10.1017/cbo9780511815065.002
White, L., Valenzuela, E., Kozlowska-Macgregor, M., & Leung, Y.K.I. (2004). Gender and num-
ber agreement in nonnative Spanish. Applied Psycholinguistics, 25, 105–133.
DOI: 10.1017/s0142716404001067
Zubizarreta, M.L. (1998). Prosody, Focus and Word Order. Cambridge, MA: The MIT Press.
 Verbal agreement in the L2 Spanish of speakers of Nahuatl 

Appendix 1

Details regarding the language profile of the participants

Participants Occupation Contact with Spanish Language they feel most

comfortable speaking

L2-1 Mason’s helper Bosses speak Spanish Both languages

L2-2 Student Secondary school is in Both languages
Spanish
L2-3 Civil judge Works in the capital of Both languages
Tlaxcala
L2-4 Mason’s helper Bosses speak Spanish Both languages
L2-6 Factory worker, lived People at work speak Both languages
in the US for two Spanish
years
L2-13 Assistant to the Speaks Spanish frequently Both languages
president of the but doesn’t leave the village
village of San Isidro
L2-19 Factory worker People at work speak Both languages
Spanish
BS-2 Housewife/ Sells shoes via catalogue Both languages
Independent work and travels frequently
BS-3 Student Secondary school is in Both languages
Spanish
BS-1 Housewife Family speaks Spanish only Spanish
L2-5 Housewife Never leaves the village Nahuatl
L2-7 Works in a large People at work speak Nahuatl
market in Puebla Nahuatl
L2-8 Housewife Never leaves the village Nahuatl
L2-9 Housewife Never leaves the village Nahuatl
L2-10 Housewife Never leaves the village Nahuatl
L2 11 Housewife Never leaves the village Nahuatl
L2-12 Farmer Never leaves the village Nahuatl
L2-14 Housewife Never leaves the village Nahuatl
L2-15 Housewife Never leaves the village Nahuatl
L2-16 Farmer Never leaves the village Nahuatl
L2-17 Housewife Never leaves the village Nahuatl
L2-18 Housewife Never leaves the village Nahuatl
BS-4 Factory People at work speak Nahuatl
Nahuatl
BS-5 Housewife Never leaves the village Nahuatl
BS-6 Housewife Never leaves the village Nahuatl
 Alma P. Ramírez-Trujillo & Joyce Bruhn de Garavito

Appendix 2

Accuracy rates for the two bilingual groups on agreement in the three persons and the two
tenses. Accuracy for the control group was 100% across the board.

Participant Contact level 1 2 3 1 2 3

Present Present Present Past Past Past

L2-3 High 100 100 100 100 100 100

L2-10 Low 100 100 100 100 100 100
BS-2 High 100 100 100 100 100 100
L2-2 High 100 100 100 100 100 90
BS-1 High 100 90 100 100 100 100
L2-4 High 100 90 100 70 100 100
BS-3 High 100 80 100 90 90 100
L2-13 High 100 90 100 100 100 100
L2-15 Low 100 100 100 90 100 100
L2-12 Low 100 50 90 100 90 100
L2-16 Low 80 80 100 40 100 100
L2-17 Low 80 90 100 20 90 90
L2-18 Low 70 20 90 100 50 100
L2-14 Low 70 30 90 100 80 80
L2-11 Low 60 20 90 100 90 100
L2-1 High 50 20 100 70 90 100
L2-9 Low 50 100 100 30 100 90
L2-5 Low 50 60 100 20 80 80
L2-6 Low 50 100 60 70 100 90
BS-5 Low 20 30 90 100 80 100
L2-7 Low 20 20 40 30 50 60
L2-19 High 20 20 40 30 20 90
BS-4 Low 20 20 30 30 20 100
L2-8 Low 10 0 10 30 0 70
BS-6 Low 0 20 80 90 80 100
 Early coda production in bilingual
Spanish and Basque

Maria-José Ezeizabarrena & Alaitz Alegria

University of the Basque Country

The study of the codas in the spontaneous speech of a Basque-Spanish bilingual

child reveals that they are produced early and continue to develop gradually,
in both languages, during the period studied (1;09–2;01). The inter-linguistic
differences attested regarding the inventory of segments, the frequency of the
target-like productions (which increase faster in Basque than in Spanish), the
differentiated effect of stress in the two languages, as well as the phonological
processes involved in the adult-dislike productions of sibilants suggest a language
specific development in the production of codas, compatible with a separate
phonological representation of the phonological systems.

1.  Introduction

There is quite an extended consensus that the presence of codas (consonant(s)

following the nucleus of the syllable (1)) in early speech is one of the developmen-
tal milestones in the acquisition of phonological complexity. However, studies do
diverge regarding the model chosen to account for the development of the syllable
structure, such as Fikkert’s (1994) parametric approach or Demuth and Fee‘s (1995)
Minimal Word view; additionally, the age and frequency of codas throughout the
first stages of language production show cross-linguistic variability. Children
acquiring English are reported to produce codas by 1;08 and Catalan children pro-
duce over 50% of the expected codas at age 1;09 (Prieto & Bosch-Baliarda 2006),
whereas Japanese children start at 1;11 (Ota 2003). Moreover, some comparative
studies (Lleó, Kuchenbrandt, Kehoe, & Trujillo 2003) report on the differences
(delay) observed in the coda production of Spanish monolingual children in com-
parison to German monolinguals as well as to Spanish-German bilingual children.
Lleó et al. (2003) conclude that the production of codas in Spanish by bilingual
children at earlier ages than monolingual children is due to the positive influence
of the language with more codas (German) which accelerates the development of
coda production in the language with fewer codas (Spanish).

doi 10.1075/ihll.3.04eze
© 2015 John Benjamins Publishing Company
 Maria-José Ezeizabarrena & Alaitz Alegria

The present research aims to describe the acquisition of codas in two geneti-
cally unrelated languages, Basque and Spanish, and to test the effect of several fac-
tors mentioned in the literature such as stress and placement in their production.
It also aims to provide one more piece of empirical evidence to the debate on the
interlinguistic influence and availability of one/two phonological systems in early
bilingualism (Fabiano-Smith & Barlow 2010) focusing on the very early spontane-
ous production of codas (1;09–2;01) of a simultaneous bilingual child. The chapter
is organized as follows: in the next subsections literature is reviewed regarding the
factors that affect early coda production in early monolingual (1.1) and bilingual
acquisition (1.2); Section 2 describes Basque and Spanish codas; Section 3 deals
with the predictions and the empirical study on codas in early Basque-Spanish
bilingualism; Section 4 includes the discussion of the results; and Section 5 con-
tains the main conclusions of the study.

1.1  Factors of early coda production

Among the set of factors proposed in the literature, frequency has been consid-
ered one of the main factors determining the early acquisition of syllabic codas.
Thus, children exposed to languages with high coda frequencies, such as E ­ nglish,
­German or Catalan, produce them at a higher rate (Bernhardt & Stemberger
1998; Borràs-Comes & Prieto 2013; Grijzenhout & Joppen 1998; Kehoe & Stoel-­
Gammon 2001; Prieto & Bosch-Baliarda 2006) than children exposed to lan-
guages with low frequencies of consonantal codas, such as Spanish or Japanese
(Lleó et al. 2003; Ota 2003). More specifically, languages with a high frequency
of (monosyllabic) CVC words containing codas will be the best candidates for
children acquiring CVC structures at earliest ages (Fikkert 1994; Levelt, Schiller
& Levelt 1999, 2000). See also Salidis & Johnson (1997) for Norwegian, Kehoe &
Stoel-Gammon (2001) for English, and Borràs-Comes & Prieto (2013) and Prieto
& Bosch-Baliarda (2006), for Catalan.
The frequency of codas has been related to the language specific phonotactic
restrictions that result in great variability in cross-linguistic rates, but Zamuner,
Gerken and Hammond (2005) explained the chronological sequencing of coda
segment production in early language acquisition in terms of the Universal Gram-
mar Hypothesis and the Language Specific Grammar Hypothesis. According to
the former and based on the coda patterns of 35 languages, these authors p ­ redict
that the earliest segments produced in coda positions will be coronals (earlier
than labials and dorsals) and sonorants (earlier than obstruents). Many languages,
including Spanish (Lleó 1997; Polo 2011), have provided data consistent with the
former hypothesis. For Basque, Saizar (2005) reports on a 1;09 year old mono-
lingual child producing different segment types in coda positions: (alveolar and
 Early coda production in bilingual Spanish and Basque 

predorsal) nasals and sibilants (100%) and laterals /l/ (40%), whilst Barreña (2003)
reported on two monolingual children producing alveolar nasal codas from 1;04,
laterals from 1;10, together with or preceding alveolar (/s̺/, /s̻/) and predorsal sibi-
lant fricatives (/∫/) as well as the predorsal affricate /t∫/, with the voiceless stops /t/,
/k/ and the apico-alveolar affricate /ts̺/ coming last. The latter hypothesis predicts
a chronological production of segments that will reflect their frequency in the
child’s input, which seems to be the case in Barreña’s study.
Stressed syllables are prosodically prominent units that tend to retain their
vocalic content, namely their syllabic information1 in diachrony, and children
seem to perceive and produce stressed syllables earlier and more frequently than
unstressed ones (Lleó 1997). Thus, the perceptual saliency of stressed syllables has
been considered to affect child coda production in different languages (­Freitas
2001; Freitas, Miguel, & Hub Faria 2001; Lleó et al. 2003; Ota 2003; Prieto  &
­Bosch-Baliarda 2006). Moreover, the earlier production of codas in word final
rather than internal positions (Goad & Brannen 2000; Rose 2000) in languages
like French (accent on the last syllable), and English (frequent final accent in
polysyllabic words) confirms the existence of an interaction between stress and
position.
However, the “privileged” nature of word final syllables for coda production
remains under discussion as it does not hold for languages like Spanish or Portu-
guese, in which penultimate word syllables are the most frequently stressed, yet
children acquiring these languages do not show any preference for final stressed
positions (Lleó 2003; Lleó et al. 2003). All these studies converge on the conclusion
that the early production of codas follows the unmarked, the most frequent stress
option in the target language, rather than a universal tendency to prefer a specific
position.
Phonological properties of codas have been considered to interact with mor-
phological features. In some cases, the delay in the production of some morpho-
logical markers, such as the -s plural suffix in Spanish has been attributed to the
low-prominence nature of the mostly unstressed fricative coda (­Ezeizabarrena
1996) whereas the morphological content of some non-prominent syllables is
interpreted to positively affect its early production in Portuguese. Freitas et  al.
(2001) showed that non-morphological fricatives in unstressed final codas were
produced less frequently than the same segments when expressing plural marking
on the verb or on the nominal phrase. In contrast, some studies have shown that

.  For instance, in the diachronic processes from Latin to Spanish, vowel shortening and
vowel omission were mostly attested only in unstressed syllables (Lloyd 1993).
 Maria-José Ezeizabarrena & Alaitz Alegria

the same segments are realized productively earlier in non-morphological codas

than in morphological ones in Spanish (Polo 2013; Prieto & Bosch-Baliarda 2006)
and others observed some preference for medial codas in Spanish, “regardless
of the fact that certain final consonants have morphological import” (Lleó 2003,
p. 279). Among the morphological codas, stressed (monosyllabic) codas like vas
[bas̺] ‘you go’ are produced correctly earlier than unstressed ones, as in [′ko.res̺]
in Spanish, initially produced as [′ko.reØ] (Polo 2013). Moreover, there are lan-
guages in which a homophonous coda corresponding to two different morphemes
may show a different rate of production depending on the type of morpheme. In
Basque, productivity rates of -k codas vary depending on whether this ending in a
word like umeak is the expression of ergative case as in ‘child-singular-Ergative’ or
of plurality as in ‘child-pl-Absolutive’2 (Elosegi 1998; Ezeizabarrena & Larrañaga
1996). Furthermore, once stress and word position are taken into account, the
lack of differences between morphological and non-morphological codas led Lleó
(2003) to conclude that it is the prosody, rather than the morphology, leading the
child in the acquisition of the syllabic complexity. Thus, no clear predictions can
be made regarding the effect of stress or the morphological nature of the coda in
its production.

1.2  Coda production in early bilingualism

Finally, bilingual language acquisition and the potential inter-linguistic influence
extend the list of factors affecting the early production of codas. Similarly to what
is observed in the lexical and grammatical domains, the debate on the shared
(­Celce-Murcia 1978; Deuchar & Clark 1996; Leopold 1949, 1971; Vogel 1975) ver-
sus the separated phonological systems (Paradis (1996) for English-German and
Paradis (2001) for English-French; Johnson & Lancaster (1998) for Norwegian-­
English; Garlant (2001) for Spanish-English a.o.) has promoted the amount of
research in early bilingual acquisition. For instance, Schnitzer and Krasinski (1994)
proposed a unitary system of consonants but a separated one for vowels in Spanish
and English, whereas Ingram (1981, 1982) argues for a separated system in both
domains by age 2. More specifically, the research on the bilingual acquisition of
codas has provided interesting conclusions related to the input. Under the assump-
tion that the input the child is exposed to has an influence on her development,
many authors (Kehoe & Stoel-Gammon 2001; Lleó 2003; Prieto & ­Bosch-Baliarda
2006) predict that bilingual children will produce more CVC syllables in the lan-
guage in which they are more frequently exposed. M ­ oreover, among the different

.  Notice that absolutive case is marked with a zero morpheme in Basque.
 Early coda production in bilingual Spanish and Basque 

forms of inter-linguistic influence in bilingual language development (e.g. feature

transfer, acceleration or delay (Paradis & Genesee 1996; Pearson, Fernandez &
Oller 1993), Lleó et al. (2003) conclude that higher rates of codas in the Spanish of
bilinguals versus that of monolinguals can be interpreted as evidence of the posi-
tive bilingual influence of the language with higher frequency of codas (German,
English) accelerating coda production in the language with less codas.

2.  Basque and Spanish codas

Basque and Spanish share many phonological properties, one of which is syllabic
structure. Both languages may have from zero to a maximum of two consonants
at the onset position. The rhyme usually contains a single (short) vowel at the
nucleus position, preceded or followed by a glide in diphthongs, and zero in open
syllables or one to two consonants at the coda position in closed syllables, and
the general pattern for closed syllables is a single consonant coda, as represented
in Figure 1. According to Jauregi (2003 and references therein), closed syllables
are much less frequent than open syllables in both languages: 27.6 % vs. 72.4% in
Basque and 31.8% vs. 59.2% in Spanish).

Syllable

Onset Rhyme

Nucleus Coda
Basque (C(C)) V (C(C))
Spanish (C(C)) V (C(C))

Figure 1.  Syllabic structure in child Basque and Spanish input

Nevertheless, some instances of two consonant codas can be found in both

languages. In Basque, complex codas are only found in a reduced list of words
such as bart [bart] ‘last night’ or beltz [belts̻] ‘black’ and a some proper nouns
like Erlantz [erlants̻]. Even the complex coda of the seemingly frequent adjective
beltz ‘black’ cannot be considered frequent in the input, since in most cases it
appears followed by the determiner -a ‘the’ or some vowel-initial demonstrative
which causes resyllabification of the word into bel-tza [bel-ts̻a] ‘the black (one)’ or
into beltz hau [bel-ts̻aw] ‘this (one) black’, so that the affricate [ts̻] abandons the
complex [lts̻] coda position and occupies the onset of the following syllable in the
new prosodic word. In Spanish, the lexical words containing complex codas, such
as transporte ‘transportation’, are virtually absent in small children’s input.
 Maria-José Ezeizabarrena & Alaitz Alegria

As far as segmental inventories go, Hualde (2003) indicates that Basque codas
present more segmental diversity in final position than in the medial one. In medial
position, the inventory is restricted to nasals that share their dento-alveolar (apical
to predorsal) place of articulation with many of the following consonants: liquids
/l/, /r/, sibilant alveolar fricatives like the apico-alveolar /s̺/, the lamino-alveolar
/s̻/ and the pre-dorsal /∫/, where inter-morphemic boundaries may turn into /ns̻/
(mendiranzko ‘bound for the mountain’) and /ls̻/ (belztu ‘make or become black’)
sequences. Stops and affricates are not allowed in medial codas. In contrast, in
addition to the continuants /n/, /l/, /r/, /s̺/, /s̻/ and /∫/ observed in medial position,
final codas may include the affricate consonants /ts̺/, /ts̻/, /t∫/ and the stops /t/, /k/.
In some cases, the sequences /nts̻/ (antz ‘resemblance’), /lts̻/ (beltz ‘black’), /rts̻/
(umezurtz ‘orphan’), /rts̻/ (erts ‘to close’), /rt/ (idort ‘to dry out’) and /nt/ (galant
‘gallant’) may also occupy coda positions.
Spanish has a more reduced inventory: /n/, /m/, /l/, /r/, /s̺/3 and /θ/ in medial
position and /n/, /l/, /r/, /s̺/, /θ/ and /δ/ in final ones, though complex codas such
as /ns̺/, /ls̺/, /rs̺/ and /bs̺/ may also appear in final position with infrequent words
such as transporte ‘transport’, solsticio ‘solstice’, perspicaz ‘acute or perceptive’ or
obstáculo ‘obstacle’.
Noticeably, some (intra)dialectal variation can be observed in nasals and sibi-
lants regarding their place of articulation and/or voice as a consequence of assim-
ilative processes, such is the case of hanka > [aŋka] ‘foot’; enborra > [embora]
‘trunk’; esne >[ezne] ‘milk’ in Basque and mismo [mizmo] ‘same’ in Spanish.
As for prominence, stress is not phonologically contrastive in Basque. The
most widespread pattern in standard as well as many non-standard varieties of
Basque is for the main stress to fall on the second syllable of the word or phrase,
usually with secondary prominence on the last syllable: alába ‘daughter’, alábaren-
tzat dà ‘it is for the daughter’, lagúnarì ‘to the friend’ though dialectal diversity
has also been attested in the accentual system (Hualde 2003). In contrast, it has
been shown recently that even if final stressed syllables are less frequent (6%)
than medial stressed ones (22.4%) in Spanish, the frequency of codas is much
higher among the former, regardless of whether they are stressed (47.27% have
codas) or not (47.51% have codas) versus medial stressed (25.75% have codas) and
medial unstressed ones (30.51% have codas). Therefore, taking into account both
­prominence-based and frequency-based hypotheses, children acquiring Spanish
are predicted to show some preference for the production of codas in final sylla-
bles, regardless of whether they are stressed or not (Borràs-Comes & Prieto 2013).

.  Usually transcribed as /s/ in Spanish regardless of its laminal (in some varieties) or the
(more) apical pronunciation such is the case of the variety of Spanish the child is exposed to.
 Early coda production in bilingual Spanish and Basque 

One more fact to be taken into account is the morphological nature of codas.
Both Spanish and Basque are richly inflected languages in which codas are very
frequently (alone or as part of) overt expressions of morphosyntactic features such
as plurality, person, case, etc. Basque is an agglutinative postpositional language
with very richly inflected nominal and verbal morphology. In Basque, many very
frequent morphological markers such as plural -k, ergative case –(e)k, locative case
–(e)n, past tense –(e)n, first person singular -t, plural verb inflection –(a)z /s̻/, are
word final codas (or rhymes). Additionally, the coda can be part of the morpheme
like /k/ in the prolative -tik ‘from’, or /n/ in the associative –(re)kin ‘with’ and so on.
In Spanish, the marking for nominal plural –(e)s, 2nd person -s /s̺/and verbal
plural -n are also phonologically realized as codas and in some cases the coda is
part of a longer morphological marking such as -mos (1st person plural), -ais (2nd
person plural). Nasals and liquid codas are also part of frequently used functional
words such as /n/ in con ‘with’, un (masculine singular indefinite article), /r/ in por
‘for’, /l/ in el (masculine singular definitite article), etc.

3.  Codas in the early production of Spanish and Basque

3.1  P
 redictions
Some predictions can be made about the early simultaneous acquisition of Basque
and Spanish based on the acquisition literature and the specific features of codas
in these languages.

a. Frequency. Children acquiring languages in which closed syllables are not very
frequent like in Basque (25%–30% of closed syllables, about 60% in final posi-
tions) and Spanish (about 30% of closed syllables) as shown by Jauregi (2003)
are expected to produce codas at a similar age, and later than in languages with
frequent codas such as English, German or Catalan (Bernhardt & Stemberger
1998; Grijzenhout & Joppen 1998; Kehoe & Stoel-Gammon 2001). However,
languages with varied kinds of codas in terms of segments (Basque) and/or
complexity are expected to show an earlier development in the acquisition of
codas than languages with less segmental (Spanish) and structural variability
and productivity (Borràs-Comes & Prieto 2013; Lleó et al. 2003; Polo 2011).
Furthermore, the early production of coronal (alveolar and predorsal) and
sonorant consonants is predicted by both the Language Specific Hypothesis
and the Universal Grammar Hypothesis, whereas the early presence of frica-
tives, affricates and stops is only predicted by the former (Lleó 2003; Polo 2011;
Prieto 2006; Prieto & Bosch-Baliarda 2006; Zamuner et al. 2005).
 Maria-José Ezeizabarrena & Alaitz Alegria

b. Prominence and position. Stressed syllables may be better candidates for the
target production of codas than unstressed ones in both languages. Follow-
ing Borràs-Comes and Prieto (2013), who take into account prominence and
­frequency-based hypotheses, the child is expected to produce (more or earlier)
codas in final syllables than in medial ones in Spanish, regardless of whether
they are stressed or not. As for Basque, no clear prediction can be made since
the main stress falls on medial syllables (mostly on the second syllable of the
word) and the secondary one falls on the last syllable of the word or phrase.
c. Grammar. Morphological codas may show higher production rates than
non-morphological ones due to their high frequency in both richly inflected
languages, in which frequent morphological marking such as plural in the
nominal inflection, person in the verbal inflection, etc., are frequently real-
ized as word final codas in both languages. However, the delayed production
of plural suffix -s in Spanish (Lleó 2003) and the ergative and plural -k ending
in Basque (Ezeizabarrena & Larrañaga 1996) predicts an initially inconsistent
production of both voiceless obstruents.
d. The exposure to two languages with similar syllabic structures and coda rates
predicts a similar development in the acquisition of codas. However, the pres-
ence of different segments in coda position in the two languages may improve
the bilingual’s sensitivity to this syllabic structure and cause that the bilingual
child acquire/produce closed syllables at earlier ages than the monolinguals
acquiring any of the languages.

3.2  Method and corpus

The current study is based on partial data available from a longer longitudinal case
study of a simultaneous early bilingual acquiring Basque and Spanish, two geneti-
cally unrelated languages, through spontaneous exposure in the family environment.
The main source of the early production analyzed in the current empirical study
comes from a set of video-recorded sessions of a bilingual male child, Mikel, inter-
acting spontaneously with Basque and Spanish interlocutors during a period of five
months between 1;09 and 2;01. Mikel and his older sister belong to a bilingual family
in which the father, a (near monolingual) native speaker of Spanish, interacts with
his children and wife in this language, whilst the mother, a Spanish-Basque native
bilingual speaker of Biscayan Basque uses this variety when speaking to her children.
In the presence of the father, the wife and children use Spanish, but Basque is the
language the children use when addressing their mother as well as when addressing
each other. Though Basque can be considered the dominant language in the chil-
dren’s proximate environment, during the period of study, Mikel is also very regu-
larly exposed to the Spanish language. The recordings took place with a frequency
of approximately two weeks, in two sessions of about 30 minutes in each language
 Early coda production in bilingual Spanish and Basque 

(9 in Spanish and 8 in Basque), in which the child interacted with (at least) one of the
parents in a natural lunch, dinner, play or storytelling setting.
The video-recordings were first transcribed and codified in the collabora-
tive HEGEHJ-BUSDE project between the University of the Basque Country in
­Gasteiz, Spain, and the University of Hamburg in Germany.4 Phonology was not
the main goal of the materials recorded in the project mentioned, and therefore, all
the recordings included in the analysis for the current study were re-transcribed
phonetically, using the IPA and Praat 4.4.17. To that purpose, for every linguistic
expression, all the words containing some CVC syllable in the target form were
codified twice in order to take into account both the child’s actual production and
the intended target (Alegria 2007).5
In order to make sure that the segments under study were real codas, all the
“apparent” (C)VC syllables, namely the ones followed by vocalic onsets in the
same prosodic word and susceptible to resyllabification (CVC-V > CV-CV) were
excluded from the study, as in Example (2).
(2) a. ez dakit ipintzen ‘I don’t know (how) to put it’
b. el autobús ‘the bus’

The production of onomatopoeic expressions, pieces of songs and proper nouns

(Mikel, Maialen, Charly, Mosku) were included, as well as cognates and Basque
loan words of Romance or Spanish origin such as asustau/asustar ‘scare’ or aber/
a ver ‘let’s see’. In contrast, unclear expressions, such as the ones produced too
quietly or the ones overlapping with other interlocutors’ productions or environ-
mental noise, were excluded from the analysis. Mixed utterances containing items
from the two languages were also excluded from the current study.

3.3  R
 esults
There is some evidence that Mikel produced some CVC syllables during his one-
word stage (han ‘there’ at 1;07, MLU = 1) even before the first recording of the
present study at age 1;09, but the precise age at which he produced his first coda is
not available to us.
The details of Mikel’s corpus from 1;09 to 2;01 are presented in Table 1. They
contain data regarding the total number of utterances in each recording, the

.  We are very thankful to the child and his family, to the promoters of both research groups,
Itziar Idiazabal and Jürgen M. Meisel, respectively, and to the collaborators: A. Barreña,
M.Almgren, A. Mahlau, M.P. Larrañaga, A. Hernández-Cembellín, J. Mercado, J. Cárdenes,
Nuria Acacio and Luis Moreno.
.  It should be noted that glides produced in contexts of target consonants have been in-
cluded in the study as (target-deviant) codas.
 Maria-José Ezeizabarrena & Alaitz Alegria

­ umber of expected CVC target syllables and the CVC produced by the child in
n
spontaneous conversations during the five months of the study.6

Table 1.  Details of the sample. Mikel 1;09–2;01

AGE Basque Spanish

Recorded Utterance (C)VC Produced Recorded Utterance (C)VC Produced

time (min.) N Context (C)VC time N Context (C)VC
N N (min.) N N

1;09  60’ 419 202 49  60’ 212 51 4

1;10   60’ 228 80 31  30’ 124 33 10
1;11   30’ 214 84 23  60’ 384 114 23
2;00   60’ 444 312 132 180’ 382 119 33
2;01   30’ 232 132 66  30’ 223 83 25
Total 240’ 1537 810 301 360’ 1325 400 95

The data in Figure 2 indicate an increase in the production of codas in both

languages during the studied period, with a consistently higher rate in Basque
than in Spanish throughout this time. With the exception of the first recordings,
in which the number of contexts was quite low in Spanish, a steady increase in the
between-language difference is observed throughout the studied period, from 9%
on 1;10 to 20% three months later.

60%

50%
% coda production

40%
Basque
30% Spanish
20%

10%

0%
1;09 1;10 1;11 2;00 2;01

Figure 2.  Mikel’s rates of codas produced out of the intended (C)VC syllables in both
languages by month (1;09–2;01)

.  Both, target-like and target-deviant CVC productions by the child are included in this
category. For instance, the target-like /pur/ and the deviant /pul/ would be counted as a “pro-
duced CVC” when the child is attempting to produce apurtu ‘break’).
 Early coda production in bilingual Spanish and Basque 

From the first recording, the bilingual child produces over 20% of the expected
codas in Basque, a rate he achieves in Spanish one month later. However, his coda
production increases from age 2 onwards in Basque, whereas in Spanish there is
an increase at age 1;10, but no increase afterwards.

Table 2.  Mikel’s number and rates of codas produced in the two languages out of the
intended (C)VC syllables
Age 1;09 1;10 1;11 2;00 2;01 Total

Basque 49/202 31/80 23/84 132/312 66/132 301/810

24% 39% 27% 42% 50% 37%
Spanish 4/51 10/33 23/114 33/119 25/83 95/400
8% 30% 20% 28% 30% 24%
Difference 16% 9% 7% 14% 20% 12.84%

As Table 2 shows, the number of closed syllable contexts is higher in the

Basque sample (810 contexts) than in the Spanish one (400 contexts). Based on
the data described, a general tendency to omit codas is attested in the two lan-
guages of the bilingual during his earliest stages of language production, though
the longitudinal data suggest that the developmental pattern observed seems to be
language specific: Spanish rates remain at around 30%, whereas Basque produc-
tion rates reach 50% at 2;1.
All the codas in the over 1200 target forms analyzed in both languages
throughout this period contain only one segment and the child only produced
simple codas in his approximately 400 closed syllables. Thus, four characteristics
were taken into account for their classification, namely: (a) the features of the seg-
ment (target and child production), (b) the position of the CVC syllable in the
word; (c) the stressed or unstressed nature of the syllable and (d) the morphologi-
cal or non-morphological nature of the coda.

3.3.1  Segment inventory

The inventory of segments in coda position produced by Mikel during the studied
period is quite wide, though it differs from Basque to Spanish. The presence of
target nasals, liquids and sibilants is attested in the corpus of both languages (3),
whereas stops and affricates are only attested in Basque, as required (3a).
(3) a. Segment inventory in Basque codas: /n/, /l/, /r/, /s̺/, /s̻/, /ts̻/, /t/, /k/
b. Segment inventory in Spanish codas: /n/, /m/, /l/, /r/, /s̺/

The data in Table 3 indicate that the production of codas varies across segments in
both languages. The fact that the sample is larger in Basque (1537 utterances) than
 Maria-José Ezeizabarrena & Alaitz Alegria

in Spanish (1325 utterances) could affect the higher total coda numbers in the for-
mer language. Nevertheless, target consonantal codas are generally more frequent
in Basque than in Spanish, though the number of contexts also varies across the
segments. Whereas in Basque the most frequent consonantal coda is the alveolar
nasal /n/ (329 contexts) followed by the sibilants, mostly the alveolar fricatives,
the apico-alveolar /s̺/ and the lamino-alveolar /s̻/, with 215 contexts, in Spanish
the most frequent target coda at these ages is the sibilant /s̺/ with 184 instances,
followed by the coronal nasal /n/ with 86 (Table 3). A look at the rates of produc-
tion out of the number of (target) contexts in Table 3 indicates that nasals are the
most frequently produced, and at a similar rate in both languages (57% in Basque
and 63% in Spanish), followed by the lateral (33% in Basque and 19% in Spanish)
and by the sibilants (32% in Basque and 10% in Spanish). The other segments –/r/,
/t/ and /k/– are very distant from the target (production between 10% and 1%).

Table 3.  Number and rates of target coda production by segment and language
Nasals Liquids Sibilants Stops Non-target
segments
n m l r t k

Basque 187/329 * 14/42 8/82 68/215 2/57 3/ 85 19

(1;09–2;01) 57% 33% 10% 32% 3% 3%
Spanish 54/ 86 3/6 9/47 1/77 19/184 * * 9
(1;09–2;01) 63% 50% 19% 1% 10%
*: unattested

But not all consonants produced in coda position are target-like. In addition
to the target-likely produced segments listed in (3), Mikel produces 19 instances
of target-deviant substitution in Basque. For instance, many of the 68 instances
of Basque sibilants in Table 3 correspond to fricatives such as /s̺/, /s̻/, /∫/, /ç/ and
affricates such as /ts̻/, /ts̺/, /t∫/, some of them alternating with the apical /s̺/ and
laminal fricatives /s̻/, or the affricates /ts̺/ and /ts̻/, respectively. We collapsed all
such production into the “sibilant” category in Table 3, where we analyzed sibilant
fricatives and affricates as allophonic productions of the corresponding voiceless
fricatives and affricates.7 Among the (really) target-deviant codas produced in
Basque (4a–m), and indicated as “non-target segments” in Table 4, some glides
/j/ (4d, e, m) are found to be produced instead of voiceless sibilants (4d) or later-
als (4e, m); voiced (4j) as well as voiceless velar fricatives (4f) replacing voiceless

.  Notice that in the dialect the child is exposed to, the apical/laminal distinction is not
phonological.
 Early coda production in bilingual Spanish and Basque 

stops (4f, j) and alveolar fricatives (4i). In some cases, the vibrant /r/ (4a, c) and the
alveolar nasal /n/ (4b) are also replaced by the flap /ɹ/, the lateral /l/ (4a, c) and the
(pre)palatal /ɲ/. In Spanish (4m-q) Mikel produces 9 substitutions. Among them
7 correspond to segments not included in (3b), such as the palatal glide /j/ (4m),
the flap /ɹ/, the voiceless velar /x/ (4p), and the dental /d/ (4q) produced instead
of other consonants, mostly liquids. The flap /ɹ/ and the (pre)palatal /ɲ/ are also
attested in place of the vibrant /r/ and the alveolar nasal /n/ respectively.
(4) Some substitutions in Basque (a–o) and Spanish (m–q)
a. [r] > [l]: [‘s̺al.tu] for sartu [s̺artu]
b. [n] > [l]: [‘a.lal] for holan [olan]
c. [r] > [l]: [a.’pul.tu] for apurtu [apurtu]
d. [s̻] > [j]: [‘ej.ta.a.’pul.tu] for ez da apurtu [es̻tapurtu]
e. [l] > [j]: [boj.te.’re.ta] for boltereta [bolteɹeta]
f. [k] > [x]: [s̻ux] for zuk [s̻uk]
g. [k] > [n]: [nin] for nik [nik]
h. [n] > [t]: [amet] for hamen [amen]
i. [s̺] > [x]: [ax.’ka.tu] for askatu [as̺katu]
j. [k] > [g]: [‘awn.ts̻eg] for ahuntzek [awnts̻ek]
k. [r] > [n]: [s̻ien da] for zer da [s̻erda]
l. [l] > [n]: [tantaw] for soltau [soltaw]
m. [l] > [j]: [‘mi.kej] for Mikel [mikel]
n. [l] > [s̺]: [mikes̺] for Mikel [mikel]
o. [r] > [l], [j] and [x]: [a.’ßel], [a.’ßej], [a.’ßex] for a ver [aßer]
p. [l] > [x]: [‘u.ka:.’lox] for un caracol [unkaɹakol]
q. [r] > [d] and [ɹ] [pin.’tad], [pin.’taɹ] for pintar [pintar]
Though it is a much less frequent phenomenon than omission in coda positions,
Examples in (3) indicate that substitution is also attested in the coda production of
this child in both languages (19 items in Basque and 9 in Spanish). It occurs in dif-
ferent contexts and affects different segments as shown in (3). Thus, since in most
cases the codas produced correspond to the target (allophonic variant) production
of the segment intended and the replacement mostly corresponds to the phono-
tactics of the target language, we take the child’s productions to indicate a highly
developed phonological representation of both systems.
Figures 3 and 4 plot the rates of production of some specific segments in the
coda contexts in which they are required. As a general observation, it can be said
that during the period 1;9–2;0 all the segments are generally omitted, with the
exception of the nasals. In Basque and Spanish only the alveolar nasal approaches
adult-like production, achieving a production rate of 60%–80% in the required
contexts by age 2. Following this, in Spanish the bilabial /m/ suddenly begins to
be produced correctly at age 2;01 where its rate fluctuated between 0% and 40%
 Maria-José Ezeizabarrena & Alaitz Alegria

in the preceding period. The rest of consonants are produced quite rarely, ranging
their target-like production from 0% to 60% in the period of study in Basque and
from 0% to 40% in Spanish. Sibilants and liquids show a steady increase from 1;11
onwards in Basque, pointing towards a consolidation of their production in the
later period.

80
70
60
n
% production

50 l
40 r
sibilant
30
t
20 k

10
0
1;09 1;10 1;11 2;00 2;01

Figure 3.  Rates of target production of each segment in Basque codas

100
90 n
80 l
70
% production

r
60 sibilant
50
m
40
30
20
10
0
1;09 1;10 1;11 2;00 2;01

Figure 4.  Rates of target production of each segment in Spanish codas

In Basque, the alveolar nasal /n/, the sibilants and the liquids /l/, /r/ appear
earlier than the stops in children’s codas; however, in the studied period the
development of the vibrant /r/ and the two stops /t/ and /k/ is quite similar
(below 15%). Nasals and laterals seem to develop the most in both languages
during the period of study, together with sibilants (only in Basque), especially
after 1;11.

3.3.2  Placement
One of the goals of the study was to look at the potential effect of syllable (final,
medial) word placement on the production of codas. Since, in the flow of oral
 Early coda production in bilingual Spanish and Basque 

speech, many word boundaries disappear, a methodological distinction was

included in the codification of CVC syllables. The placement of the syllable in the
lexical word was codified in the case of mono-lexemic utterances, but in multi-­
lexemic prosodic words, the (medial or final) placement of the syllable was codi-
fied regarding its position inside the prosodic word. Consequently, codas codified
as codas in final position of the utterance were all at word final position (see
­Examples (5a) and (5c) for Basque and (5b) and (5d) for Spanish), whilst codas
in medial position of the prosodic word included word internal codas ((6a) for
Basque, (6b) for Spanish) as well as codas which, despite their word final position,
are placed at some position inside the prosodic word ((6c) for Basque, (6d) for
Spanish).
(5) a. hamen ‘here’
b. zapatillas ‘shoes’
c. hau kendu gura dot ‘I want to put this out’
d. en el tren ‘in the train’
(6) a. konpondu ‘(to) repare’
b. este ‘this’
c. nun dau ‘where is it?’
d. se ha caído el tren ‘he fell out of the train’

The frequent presence of the negation ez ‘no’ in Basque may distort (C)VC tal-
lying. When appearing in final positions, this particle is produced accurately for
the most part (Table 4) and the rest of the monosyllabic words are also produced
in over 50% of the contexts: bat [bat]. Medial positions are more problematic,
though, as the omission of codas even in these same particles is as frequent as
their production. Similarly, nasal and vibrant codas are inconsistently produced in
medial positions, as in apurtu egin da ['pu.tu.'in.da] ‘it is broken’, where the codas
are only attested in some of the stressed positions.

Table 4.  Production rates of codas in Basque negation ez [es̻] ‘no’

1;09 1;10 1;11 2;00 2;01 Total

N Produced/target 2/3 4/4 3/3 25/26 26/29 60/65

% 67% 100% 100% 96% 90% 92%

In Spanish, the production of medial codas is quite irregular: 0% to 8% by

age 2;00 and over 70% after that age. See for instance, sí al cuarto [s̺i.al.'ta.to]
(M 1;11;02); al túnel [al.’tu.nel] (M 2;00;00); un accidente ['un.'te:n.te] (M 2;00;20).
Proclitic contractions al ‘to the’ or determiners un ‘a, one’ tend to be produced
more frequently than other non-proclitic elements (N = 6).
 Maria-José Ezeizabarrena & Alaitz Alegria

3.3.3  Stress and word length

Stress is another of the features studied. Multisyllabic words as well as prosodic
words may have more than one closed syllable. In many cases the child tends
to produce codas in the stressed syllable and to omit (at least) the coda of the
unstressed one. The examples in (7) illustrate this preference in Basque words and
the same is illustrated in (8) for Spanish.
(7) CVC contexts in polysyllabic (prosodic) words:
a. hemendik doie ['men.ji.'o.je], ['men.di.'o.je] (M 1;11,06)
‘it goes from here’
b. hemendik [e.'men.di] (M 2;00,00)
‘from here’
c. hemendik ['men.tʃi] ['man.tʃi] (M 2;00,00)
‘from here’
d. konpondu [a.'pun.du] (M 2;00,00)
‘repair’
e. bazkaldu ['kal.du] (M 2;00,22)
‘to have lunch’
f. kontuz ['kon.tu] (M 2;01,06)
‘be careful’
g. bazkaldu [ma.'kal.du], [bo.'kal.du] (M 1;11,02)
‘to have lunch’
(8) a. sí durmiendo ['s̺i.nu.'nen.do] (M 2;00,00)
‘yes, sleeping’
b. así/ahí se engancha ['aj.ta.'kan.t∫a] (M 2;00,00)
‘so, there can be tied’
c. un coscorrón ['un.to.ko.'ko:n] (M 2;00,22)
‘a bump on the head’

Anecdotally, a few instances (N = 4) were found in Basque in which codas were

omitted in the stressed (primary) syllable, but produced in the stressed (­secondary)
syllable of the same word (9). No similar examples are found in Spanish.
(9) a. hau ipintzen da ['a.'pi.cen.'ða] (M 1;09)
‘this is put in this way’
b. bestien ['be.cen] (M 1;09)
‘in the other one’
c. bazkaltzen [ke.'ci.cen] (M 1;09,22)
‘having lunch’
d. ez dot ikusten [ta.'ßu:.cen] (M 2;01;06)
‘I don’t see’
 Early coda production in bilingual Spanish and Basque 

In general, stress seems to facilitate the general production of codas in Basque, as

indicated by their rate of over 40% in stressed syllables as compared to a rate of
13% in unstressed ones. In contrast, no such difference is observed in the Spanish
productions, in which coda production rates oscillate around 20% regardless of
the stressed or unstressed nature of the syllable (See Table 5).

Table 5.  Coda production in stressed and unstressed syllables in Basque and
Spanish. Mikel 1;09–2;01
Basque Spanish

Syllable Stressed Unstressed Stressed Unstressed

Prod/Target (n) 103/244 38/287 48/227 15/73

Rate 42% 13% 21% 20%

In Basque, differences between stressed and unstressed syllables are found

throughout the whole study from 1;09 onwards, (in both final, and medial posi-
tions) with the frequency of codas in both contexts differing in a range of 10% to
45% across age periods (Figure 5). In Spanish, no differences are observed in the
whole mean average. Coda rates oscillate in both stressed (range 13% to 32%) and
unstressed syllables (range 0%–30%) throughout the period of study, which com-
plicates the drawing of any conclusion regarding the effect of stress (alone) in this
language (Figure 5).

70%

60%

50%

Basque Stressed
40%
Basque Unstressed
Spanish Stressed
30%
Spanish Unstressed

20%

10%

0%
1;09 1;10 1;11 2;00 2;01 TOTAL

Figure 5.  CVC production rates in (un)stressed syllables, by month and language. Mikel
1;09–2;01
 Maria-José Ezeizabarrena & Alaitz Alegria

A look to the length of words in syllables indicates that codas are produced
more frequently in monosyllabic than in polysyllabic words in both languages
(Table 6).

Table 6.  Coda production in monosyllabic and polysyllabic words in Basque and
Spanish. Mikel 1;09–2;01
Basque Spanish

Monosyllabic Polysyllabic Monosyllabic Polysyllabic

Prod/Target 160/279 141/531 32/100 63/300

Rate 57% 26% 32% 21%

The difference is higher in Basque (31%) than in Spanish (11%), though

monosyllabic words show a similar pattern from 1;10 onwards in Basque (range
34%–65%) and in Spanish (range 20%–53%), as do polysyllabic words in Basque
(range 24%–42%) and in Spanish (range 20%–26%). See Figure 6.

70%

60%

50%
Basque Monosyllabic
40%
Basque Polysyllabic
30% Spanish Monosyllabic

20% Spanish Polysyllabic

10%

0%
1;09 1;10 1;11 2;00 2;01 TOTAL

Figure 6.  Coda production rates in monosyllabic and polysyllabic words by month and
language. Mikel 1;09–2;01

Data in Figure 7 show that word length, position and stress, all interact with
each other in Basque, as codas are produced the most frequently in monosyllabic
final words (mean 77.2% and rates over 70% in all recordings), followed by mono-
syllabic words in medial positions (mean 41.6%, range 15%–59% across record-
ings), as well as stressed syllables in medial (mean 43%, range 7%–55%) and
stressed syllables in final positions inside polysyllabic words (mean 43%, range
0% to 50%). The lowest ones correspond to unstressed syllables inside polysyllabic
 Early coda production in bilingual Spanish and Basque 

words regardless of their final (mean 13%, range 0% to 26%) or medial position
(mean 14% range 0% to 25%).

100
90
Final stressed
80 syllables
70 Final unstressed
syllables
60 Medial stressed
syllables
50
Medial unstressed
40 syllables
Monosyllabic final
30
words
20 Monosyllabic medial
words
10
0
1;09 1;10 1;11 2;00 2;01

Figure 7.  Rates of coda production in Basque by syllable type (stress and placement) and age

In Spanish, the production of codas seems to be slightly favored in monosyl-

labic words (mean over 30% regardless of position) as compared to polysyllabic
words, (around 20%), though rates of coda production show more variation in
monosyllabic final (mean 32%, range 0% to 100%) than in medial monosyllabic
words (mean 30%, range 0%–43%) as shown in Figure 8. In polysyllabic words
coda production rates are lower, of around 20%, regardless of stress and position
(21% in final stressed, 22% unstressed final syllables and 21% and 19% in medial
stressed and unstressed syllables respectively). Figures 7 and 8 do not suggest any
development by age throughout the period of study. Codas are almost target-like
in final monosyllabic words in some recordings at age 1;10 and after 2;00, but
their low production in the same contexts at age 1;11 (1 coda out of 12 contexts),
as well as the high production of codas (80%) in unstressed final syllables at age
2;00 weaken any conclusion related to development by age. In most recordings and
conditions, rates of coda production remain below 50% during the five months
period.
Figure 8 suggests that stress has either no effect or a possible negative effect
on the production of codas, since, at least at age 2;00, codas tend to be more fre-
quently produced in unstressed syllables. Interestingly, however, stress seems to
have some effect on medial syllables from 1;11 onwards, when closed syllables
are much more frequent. It is noteworthy that coda contexts are more frequent in
medial (N = 213) than in final position (N = 83). Therefore, no clear conclusions
can be drawn regarding the effect of stress or final versus medial position of the
 Maria-José Ezeizabarrena & Alaitz Alegria

100
90
Final stressed syllables
80 Final unstressed
70 syllables
Medial stressed
60 syllables
50 Medial unstressed
syllables
40
Monosyllabic final
30 words
20 Monosyllabic medial
words
10
0
1;09 1;10 1;11 2;00 2;01

Figure 8.  Rates of coda production in Spanish by syllable type (stress and placement) and age

syllable in the production of codas in Spanish by this bilingual child, at least dur-
ing the studied period.

3.3.4  Morphological codas

A final classification was conducted distinguishing morphological and non-­
morphological closed syllables in order to test the potential effect of the gram-
matical nature (and consequently, the effect of frequency) in coda production.
In the Basque sample, the attested (C)VC syllable targets in which the coda
involved in morphological marking in the nominal domain are the following: erga-
tive -k, absolutive plural -k, inesive -n, ablative -tik and instrumental -z. In the
verbal domain the attested markings were verbal imperfective aspect -t(z)en, first
singular person -t as in dakit ‘I know’, dekot ‘I have’ and plural -z like in dauz ‘they
are’. The few instances in which the child produced a morphological item different
from the one expected were excluded from the phonological analysis. One such
case was the production of the perfective infinitive or participle ipini ‘put’ (M 1;10
and M 2;00) instead of the imperfective ipintzen ‘putting’. The discontinuous curves
in Figure 9 indicate that codas are more frequently produced in non-grammatical
contexts (range 20% to 70%), than in grammatical ones (range 0% to 33%).
In the Spanish sample, the attested grammatical codas were the plural marker
-s in the nominal domain, and in the verbal domain, the -ar/er/ir infinitival suf-
fixes, the 3rd person plural -n (e.g. están ‘they are’) and the 1st plural -s in vamos
‘we go, let’s go’. Moreover, some functional monosyllabic elements such as the
masculine determiners un ‘one, a’ and el ‘the’ as well as some instances of the
preposition-article contraction al ‘to the’ or the preposition con ‘with’ are pseudo-­
cliticized to the following nominal element. Except for the prepositional function
words, the other grammatical codas correspond to the plural suffix /-s/ in most
cases, f­ollowed by the infinitive markers -ar/er/ir and person marking. No clear
 Early coda production in bilingual Spanish and Basque 

80%
70%
60%
50% Final -morf
Final +morf
40%
Medial -morf
30% Medial +morf

20%
10%
0%
1;09 1;10 1;11 2;00 2;01

Figure 9.  Target production of (non-)morphological codas in medial and final syllables in
Basque by Mikel (% out of contexts)

­ ifferences were observed in the production of grammatical codas in stressed vs.

d
unstressed syllables in Spanish.
In Spanish, a steady increase is observed in the low rates of coda production
in medial positions, where the rates range between 0% and 35% (See Figure 10).
In final position, rates of coda production oscillate between the two extremes (0%
to 100%) but it must be recognized that the highest rate of coda production (60%)
in final morphological contexts at 1;10 corresponds to three produced codas out
of five target ones; furthermore, even in the last recording of apparently target
produced non-morphological codas, the number of items remains very scarce
(N = 3). Therefore, the grammatical nature of the coda does not seem to affect its
production in either language, at least not positively.

100%
90%
80%
70%
Final -morf
60%
Final +morf
50%
Medial -morf
40%
Medial +morf
30%
20%
10%
0%
1;09 1;10 1;11 2;00 2;01

Figure 10.  Target production of (non-)morphological codas in medial and final syllables in
Spanish by Mikel (% out of contexts)
 Maria-José Ezeizabarrena & Alaitz Alegria

4.  Discussion

The simultaneous Basque-Spanish bilingual child studied in the period from

1;09 to 2;01 frequently omits codas in monosyllabic CVC words in his spon-
taneous production, for instance /'xa/ for jan ‘eat’ /xan/, or /'ba/ for bat /bat/
‘one’ (N = 3), /'a/ for han /an/ ‘there’ (N = 22) in Basque or /'ma/ for más ‘more’
(N = 12) in Spanish. But, at the same time, the codas produced, among others,
in the 36 instances of the deictic adverbial /an/ ‘there’ along the study (preceded
by a very early instance of this monosyllabic CVC word in an earlier recording
at 1;07) are evidence for the early production of codas, at the one-word stage, at
ages at which some CVCV words are also found. Yet, methodological limitations
do not allow us to confirm or refute the existence of a CV stage in the acquisi-
tion of Spanish or Basque, preceding the CVC stage found in other languages
(Demuth & Fee 1995; Fikkert 1994).
Similarly to observed for Catalan (Prieto 2006) coda production is attested
quite extensively from the first recordings of the current study (1;09) in both
languages and their production shows some increase with age throughout the
recordings analyzed between 1;09 and 2;01. In Basque the largest increase takes
place from 1;11 to 2;01 (27% to 50%), whilst the largest overall increase (from
8% to 33%) is found in Spanish during 1;09 and 1;10, and the second one of
about 10% between 1;11 and 2;01 corresponds with other studies (Lleó et al.
2003; Polo 2011). Thus, Mikel’s early production of codas in the two languages
is marginally higher (range 2% to 16%) but consistent with two out of three
monolingual Spanish longitudinal corpora studied by Lleó et al. (2003), in
which a low initial production of codas (range 4% to 33%) around age 2 is
followed by periods of slight increase in their production in the succeeding
months.
So far, no data are available regarding the frequency of CV and CVC sylla-
bles in this child’s Basque and Spanish input; however, the rates of closed syllables
found in several Basque and Spanish corpora, ranging from 25% to 30% (Jauregi
2003), suggest that children acquiring Basque and Spanish may have a regular
exposure to codas, though at a lower rate than in languages like Catalan, German
and English (Prieto & Bosch-Baliarda 2006; Saceda 2005).
The corpus analyzed provides, however, data regarding the frequency of
codas (intended and produced) in the child’s output in each of the languages.
In general, the number of target contexts in the child’s productions is higher in
the Basque sample (N = 810) than in the Spanish one (N = 400), and the number
of coda instantiations is also larger in the Basque corpus (36.8% versus 24% in
 Early coda production in bilingual Spanish and Basque 

Spanish).8 In fact, the data plotted in Table 7 (reported by Lleó et al. 2003 and
Alegria 2007, respectively) reveal a consistently higher production of codas in
Basque than in Spanish despite a slight difference that ranges between 7% and
20% in the months of the longitudinal study, which results in a higher total rate
of 12% in Basque over Spanish. Thus, the prediction that since children acquir-
ing languages with frequent and/or varied kinds of codas (Basque) will produce
codas (more or) earlier than children acquiring languages with less segmental
and structural variability and productivity (Spanish) as mentioned by Lleó et al.
(2003), Polo (2011) and Borràs-Comes and Prieto (2013) seems to hold for the
general data obtained in the longitudinal study.

Table 7.  Longitudinal data of coda production in child language

CHILD Profile language 1;09 1;10 1;11 2;00 2;01 2;02 2;03

José Mono Spanish 12% 17% 8% 26% * 41% 37%

María Mono Spanish * 4% * 6% 30% 11% 24%
Miguel Mono Spanish * 58% * 61% * 69% 70%
Simon Bi Span(-Ger) 0% 32% 52% 70% * 73% 60%
Stefan Bi Span(-Ger) 0% 67% * 50% 71% 63% 56%
Robert Bi Span(-Ger) * * 50% 76% 81% 86% 44%
Mikel Bi Span(-Ger) 8% 30% 20% 28% 30% * *
Mikel Bi Basque(-Span) 24% 39% 27% 42% 50% * *

*: data not available.

The nature of the segments, their position and their prominence seems to have
an effect on the earlier/later consistent coda production in Basque and S­ panish. As
attested in many other linguistic levels such as syntax, morphology and vocabu-
lary, the first production of any phonological segment in the spontaneous linguistic
production of a child implies neither its adult-like knowledge nor its use at any
phono-syntactic position (Barreña 2003; Lleó 1997; Prieto & Baliarda 2006). For

.  Notice that along this study not only lexical but also functional words are included in the
counting of target and produced codas. For this reason coda rates in the current study may not
be directly comparable to other studies based on lexical items (Polo 2011; Prieto 2006; etc.) or
(pseudo)words (Borràs-Comes & Prieto 2013).
 Maria-José Ezeizabarrena & Alaitz Alegria

Basque, Barreña (2003) observes that the chronological order in the production of
phonological items varies from onset (nasal > stops > liquids and sibilants) to coda
positions (nasal /n/ > liquids /l/, /r/ > voiceless sibilants > voiceless plosives) in the
production of a monolingual child. The frequency of coda consonants analyzed in
the production of the bilingual child between 1;09 and 2;01 in the current study
confirms Barreña’s segment sequencing in coda position in monolingual child
Basque: nasal /n/ (57%) > lateral /l/ (33%) and voiceless sibilants (32%), followed
by the vibrant /r/ (10%) and the scarcely produced voiceless plosives /t/ (3%) and
/k/ (3%). In Spanish, Mikel’s frequency order of target production of codas was:
nasals /n/ (63%), /m/ (50%), and lateral /l/ (19%) followed by /s̺/ (10%) and one
isolated instance of the vibrant /r/ (see Table 3). This pattern follows the chrono-
logical sequencing in the production of the bilingual studied by Lleó et al. (2003)
quite closely: /n/ > /θ/ > /l/ > /k/ > /s̺/ and /x/. Similarly, Polo (2011) observed
in her word elicitation task that the nasal /n/, followed by /l/ and /s̺/ are the first
candidates in the early production of Spanish codas. Thus, a similar frequency pat-
tern seems to appear in Mikel’s production in both languages regarding nasals /n/
(and also /m/ in Spanish), which are quite consistently produced by age 2;01, fol-
lowed by the lateral /l/. Also the sibilants in Basque turn out to be uttered more
and more productively, whereas the rest of segments cannot be considered con-
sistent at this age. Interestingly, the observed frequency of target-like production
seems to correlate with the frequency of the target segments in coda positions in
both languages, namely: /n/ (n(umber) of contexts = 329 in Basque, 86 in Spanish),
sibilants (N = 215 in Basque, 184 in Spanish), lateral /l/ (N = 42 in Basque, 47 in
Spanish) and stop /k/ (N = 85 in Basque), multiple vibrant /r/ (N = 82 in Basque,
N = 77 in Spanish), stop /t/ (N = 57 in Basque). This parallelism is quite consis-
tent in Mikel’s Basque, confirming the results of other monolingual and bilingual
studies in this regard (Borràs-Comes & Prieto 2013; Lleó et al. 2003; Polo 2011).
However, his Spanish data do not seem to follow the same pattern. On the one
hand, even if nasals are not the most frequent coda segment in the Spanish targets,
they are the first ones to be systematically produced, which could be explained in
terms of the double exposure and use of such units in the two languages of the
bilingual (additive effect, fourth prediction). Conversely, the high frequency of
target sibilant /s̺/ (higher than nasals, and laterals) contrasts with its low produc-
tion rate (10%) in Spanish. A possible explanation for that fact could be that the
different distribution of sibilants in the two languages (more sibilant phonemes
and also more allophonic variation in Basque than in Spanish) may cause a delay
in the acquisition of the phonological representations in the two languages, espe-
cially the ones regarding the sibilant subset. In fact, substitutions are attested in
the production of other segments such as the lateral instead of the vibrant in both
languages. However, contrary to what has been claimed for other phonological
 Early coda production in bilingual Spanish and Basque 

categories for which a shared ­representation seems to facilitate the inter-linguistic

influence and consequently the similar production pattern in the two languages,
the bilingual child seems to make use of language specific strategies in the pro-
duction of sibilants: mostly substitutions in Basque (allophonic or non-allophonic
variants which share some of their features with the segment intended) such as
[es̺], [es̻], e[θ] or [e∫] instead of ez [es̻]) and mostly omissions in Spanish ([ete]
instead of [es̺te] ‘this’).
The prosodic prominence hypothesis predicts that stressed syllables will be
better candidates for the target production of codas than unstressed ones. The
bilingual child shows a general preference for the production of codas in monosyl-
labic words and in stressed syllables in Basque, regardless of their final (33%) or
medial (42%) placement, as opposed to their low frequency in unstressed contexts
(below 15%), which is also consistent with the input frequency hypothesis, since
most words have their primary stress on the second syllable and the secondary one
on the final syllable. As for Spanish, no clear preference is observed for stressed
syllables – only in medial syllables was there attested, some, very slightly (4%)
higher coda production in stressed syllables versus unstressed ones – whereas a
similar rate of codas is attested in stressed (21%) and unstressed syllables (22%)
in final position. These data contrast with the Catalan data of 2-year-old Spanish-
Catalan bilinguals studied by Borràs-Comes & Prieto (2013), who in a repetition
and in an elicitation task of lexical words, produce more codas in final position
(over 70%) than in the medial one (around 40%), and show higher rates of pro-
duction in stressed final position (over 75%) than in unstressed final ones (over
50%). Thus, Mikel’s data are compatible with a positive effect of stress in coda
production in Basque, but not in Spanish.9
The third prediction regarding the potential effect of the morphemic nature
in the production of codas is not confirmed in either of the languages under study.
Most of the morphological codas are placed in word-final position in both lan-
guages, with the exception of some proclitic codas inside function words such as
al ‘to the’, or un ‘one’ in Spanish. Mikel’s delayed productions of morphological
codas are in line with other studies on early Spanish reporting a generalized fre-
quent omission of unstressed determiners and functional elements (Elosegi 1998;
López-Ornat 1999), as well as plural or person marking (Ezeizabarrena 1996;
Lleó 2003; Prieto & Bosch-Baliarda 2006). The morphological nature seems to
negatively affect the early production of codas, such is the case of [k] an [t] in

.  Saceda (2005) reports on differences between the coda rates produced in stressed over
unstressed syllables in the earliest production of codas of one child up to age 1;9. No clear dif-
ferences can be observed between stressed and unstressed syllables after this age.
 Maria-José Ezeizabarrena & Alaitz Alegria

Basque (always morphemic suffixes) or of [s̺], which is more frequently omitted

in Spanish (morphological and non-morphological) than in Basque (always non-
morphological). In Spanish codas, the distribution of their production/omission
does not seem to vary between morphemic (suffixes) and non-morphemic ones.
The data analyzed confirm a gradual acquisition of the coda consonants in
both languages, which also share an earlier mastering of nasals (the crosslinguisti-
cally unmarked option) as compared to the remaining consonants. However the
earliest coda samples attested in both languages are not in line with the universal
sequencing hypothesis as they include voiceless obstruents. Moreover, the fact that
the rates of targetlike production of sibilants, laterals and plosives diverge from
one language to the other is more in line with a language specific developmental
pattern that may reflect the segmental frequency in the input (Borràs-Comes &
Prieto 2013; Prieto 2006).
Two different phonological processes have been observed as operational strat-
egies during the intermediate stage preceding the target production of some coda
segments. Omission is the most frequent process for all segments. In the case of
sibilants, the bilingual resorts to the omission phonological process in Spanish,
whereas in Basque, both substitution and omission are frequently attested strate-
gies. Data such as these reveal the child’s high sensitivity to inter-linguistic dis-
tinctions, as a first step towards a separate language specific representation of the
phonological systems, as it has been also proposed for adult Basque-Spanish bilin-
gual aphasia (Munarriz & Ezeizabarrena 2009).
Lastly, the slightly higher rate of codas produced in Spanish by this bilingual
vis-à-vis some monolingual corpora is also compatible with the inter-linguistic
influence hypothesis, which predicts an increasing influence of the language with
more codas (Basque) on Spanish, the language with less productive use of them
(in line with Lleó et al. 2003; Polo 2011). There are, however, two facts that lead us
to follow that reasoning with caution in the case of the Basque-Spanish bilingual.
First, there is the case of one Spanish monolingual child studied by Lleó et  al.
(2003), Miguel, who produces relatively high coda rates (ranging 58% to 70%)
between age 1;10 and 2;02, which are comparable with, or even higher than, some
bilinguals’ initial rates (32%, 67% at age 1;10, See Table 7), and considerably higher
than our Basque-Spanish bilingual’s (below 35% between 1;10 and 2;01). This sug-
gests that the linguistic inter-individual variation observed among monolinguals
needs to be described/explained in order to avoid attributing monolingual versus
bilingual differences to inter-individual ones. Secondly, even if codas may be more
frequent in the Basque than in the Spanish child input, the absence of strong pro-
duction of codas in Basque (below 50% even in final recordings) weakens, at least
in this specific case, the plausibility of the higher production of codas in bilin-
guals as a consequence of the inter-linguistic influence causing acceleration in
the ­acquisition of codas in the other language. We are nevertheless aware that the
 Early coda production in bilingual Spanish and Basque 

scarcity of studies on child coda production in Basque and the limitations of the
current sample present an obstacle to the confirmation/refutation of this hypoth-
esis and the statement of any conclusive remark in this regard. Further research is
required in order to test this hypothesis.

5.  C
 onclusions

The longitudinal study of the coda productions in Basque and Spanish by a bilin-
gual child during the period 1;9–2;1 reveals that:

1. Despite an initial frequent omission of codas, their production rate increases

gradually in both languages from 1;09 onwards, though faster in Basque than
in Spanish, and earlier for some segments (nasals) than for others (sibilants
and plosives) in both languages.
2. The language specific development in the inventory and the production rates
of the most productive segments during the earliest production of codas,
which only partially reflect their frequency in the corresponding target lan-
guage, as well as the effect of factors such as stress, which affects their pro-
duction in Basque more than in Spanish, and the phonological processes
observed in the target-deviant production of sibilants (omission and substi-
tution in Basque, vs. mostly omission in Spanish) point towards a language
specific representation of the (morpho-)phonological features of the phone-
mic segments, compatible with a separate phonological representation of the
phonological systems.

Acknowledgments

The research reported in this paper was partially supported by the Ministerio de
Ciencia e Innovación (FFI2012-32212) and the Basque Government (IT-676-13).
The authors are very grateful to Nuria Polo, to the two anonymous reviewers and
to the editors of the volume for their valuable comments.

References

Alegria, A. (2007). Haur elebidun baten silaba amaierako kontsonanteen jabekuntza. Unpub-
lished Master thesis. University of the Basque Country.
Barreña, A. (2003). Euskararen sistema fonologikoaren jabekuntzari buruzko oharrak. In
X. Makazaga & B. Oyharçabal (Eds.), Euskal gramatikari eta literaturari buruzko ikerketak
XXI mendearen atarian (pp. 161–178). Bilbao: Euskaltzaindia.
 Maria-José Ezeizabarrena & Alaitz Alegria

Bernhardt, B.H., & Stemberger, J.P. (1998). Handbook of Phonological Development. From a Non-
linear Constraints-based Perspective. San Diego, CA: Academic Press.
DOI: 10.1017/s0952675799003723
Borràs-Comes, J., & Prieto, P. (2013). The acquisition of coda consonants by Catalan and S­ panish
children: Effects of prominence and frequency of exposure. Probus, 25(1), 1–24.
DOI: 10.1515/probus-2013-0001
Celce-Murcia, M. (1978). The simultaneous acquisition of English and French in a two-year-
old child. In E. Hatch (Ed.), Second Language Acquisition: A Book of Readings (pp. 38–53).
Rowley MA: Newbury House.
Demuth, K., & Fee, E.J. (1995). Minimal Words in Early Phonological Development. MS, Brown
University, Providence, RI & Dalhousie University, Halifax, CA.
Deuchar, M., & Clark, A. (1996). Early bilingual acquisition of the voicing contrast in English
and Espanish. Journal of Phonetics, 24, 275–305. DOI: 10.1006/jpho.1996.0019
Elosegi, K. (1998). Kasu eta preposizioen jabekuntza-garapena haur elebidun batengan. Bilbao:
Universidad del País Vasco.
Ezeizabarrena, M.J. (1996). Adquisición de la morfología verbal en euskera y castellano por niños
bilingües. Bilbao: Universidad del País Vasco.
Ezeizabarrena, M.J., & Larrañaga, M.P. (1996). Ergativity in Basque: A problem for language
acquisition? Linguistics, 34, 955–991. DOI: 10.1515/ling.1996.34.5.955
Fabiano-Smith, L., & Barlow, J.A. (2010). Interaction in bilingual phonological acquisition:
­Evidence from phonetic inventories. International Journal of Bilingual Education and Bilin-
gualism, 13(1), 81–97. DOI: 10.1080/13670050902783528
Fikkert, P. (1994). On the Acquisition of the Prosodic Structure. Dordrecht: ICG Printing.
Freitas, M.J. (2001). Syllabic constituency and segmental emergence: Evidence from the acquisi-
tion of European Portuguese. In M. Almgren, A. Barreña, M.J. Ezeizabarrena, I. ­Idiazabal, &
B. MacWhinney (Eds.), Research on Child Language Acquisition, 2 (pp. 45–57). Somerville:
Cascadilla Press. DOI: 10.1075/lald.24.04fre
Freitas, M.J., Miguel, M., & Hub Faria, I. (2001). Interaction between prosody and morphosyn-
tax: Plurals within codas in the acquisition of European Portuguese. In J. ­Weissenborn &
B.  Höhle (Eds.), Approaches to Bootstrapping: Phonological, Lexical, Syntactic and Neu-
rophysiological Aspects of Early Language Acquisition, II (pp. 45–57). Amsterdam: John
Benjamins.
Garlant, M. (2001). Spanish Phonological Patterns of Young Spanish-English Bilinguals. Unpub-
lished Master thesis. University of Arizona, Tucson.
Goad, H., & Brannen, K. (2000). Syllabification at the right edge of words: Parallels between
child and adult grammars. McGill Working Papers in Linguistics, 15, 1–16.
Grijzenhout, J., & Joppen, S. (1998). First steps in the acquisition of German phonology: A case
study. SFB 282 Working Paper No.110.
Hualde, J.I. (2003). Phonology. In J. I. Hualde & J. Ortiz de Urbina (Eds.), A Grammar of Basque
(pp. 15–71). Berlin: Mouton de Gruyter. DOI: 10.1515/9783110895285
Ingram, D. (1981/1982). The emerging phonological system of an Italian-English bilingual
child. Journal of Italian Linguistics, 95–113.
Jauregi, O. (2003). Euskal testuetako silaba egituren maiztasuna diakronikoki. ASJU, 37(1),
393–410.
Johnson, C.E., & Lancaster, P. (1998). The development of more than one phonology: A case
study of a Norwegian-English bilingual child. The International Journal of Bilingualism, 2,
265–300.
 Early coda production in bilingual Spanish and Basque 

Kehoe, M.M., & Stoel-Gammon, C. (2001). Development of syllable structure in English-­

speaking children with particular reference to rhymes. Journal of Child Language, 28,
393–432. DOI: 10.1017/s030500090100469x
Leopold, W. (1949/71). Speech Development of a Bilingual Child, a Linguists’ Record, (Vols. 1–4).
New York: AMS Press.
Levelt, C.C., Schiller, N.O., & Levelt, W.J. (1999/2000). The acquisition of syllable types. Lan-
guage Acquisition, 8, 237–264. DOI: 10.1207/s15327817la0803_2
Lleó, C. (1997). La adquisición de la fonología de la primera lengua y lenguas extranjeras. Madrid:
Visor. DOI: 10.5209/rev_rfrm.2013.v30.n1.42607
Lleó, C. (2003). Prosodic licensing of coda in the acquisition of Spanish. Probus 15, 257–281.
DOI: 10.1515/prbs.2003.010
Lleó, C., Kuchenbrandt, I., Kehoe, M., & Trujillo, C. (2003). Syllable final consonants in ­Spanish
and German monolingual and bilingual acquisition. In N. Müller (Ed.), (In)vulnerable
Domains in Multilingualism (pp. 191–220). Amsterdam: John Benjamins.
DOI: 10.1075/hsm.1.08lle
Lloyd, P.M. (1993). Del latín al español. Madrid: Gredos.
López-Ornat, S. (1999). La adquisición del lenguaje. In M. Vega & F. Cuetos (Eds.), Psico-
lingüística del español (pp. 469–534). Madrid: Trotta.
Munarriz, A., & Ezeizabarrena, M.J. (2009). Parafasias fonológicas en el habla de un bilingüe
adulto. In C. M. Bretones Callejas (Ed.), Applied Linguistics Now: Understanding Language
and Mind (pp. 877–888). Almería: Universidad de Almería.
Ota, M. (2003). The Development of Prosodic Structure in Early Words. Amsterdam:
John ­Benjamins. DOI: 10.1017/s0022226704293239
Paradis, J. (1996). Phonological differentiation in a bilingual child: Hildegrad revisited. In
A. Stringfellow, D. Chana-Amitay, E. Hughes, & A. Zukowski (Eds.), Proceedings of BUCLD
20 (pp. 528–539). Somerville, MA: Cascadilla Press.
Paradis, J. (2001). Do bilingual two-year-olds have separate phonolgy systems? International
Journal of Bilingualism, 5, 19–38. DOI: 10.1177/13670069010050010201
Paradis, J., & Genesee, F. (1996). Syntactic acquisition in bilingual children: Autonomous or
interdependent? Studies in Second Language Acquisition, 18, 1–25.
DOI: 10.1017/s0272263100014662
Pearson, B.Z., Fernandez, S.C., & Oller, D.K. (1993). Lexical development in bilingual infants
and toddlers: Comparison to monolingual norms. Language Learning, 43(1), 93–120.
DOI: 10.1111/j.1467-1770.1993.tb00174.x
Polo, N. (2011). Notas sobre la adquisición fonológica bilingüe del español: El desarrollo de las
codas en español como lengua materna en hablantes bilingües del País Vasco. Oihenart,
26, 399–421.
Polo, N. (2013). Adquisición del componente fonológico y morfológico del español como len-
gua materna. Talk at the II Jornadas de Lingüística Hispánica. Lisbon, Universidade de
­Lisboa, 15th April 2013.
Prieto, P. (2006). The relevance of metrical information in early prosodic word acquisition:
A comparison of Catalan and Spanish. Language and Speech, 49(2), 231–259.
DOI: 10.1177/00238309060490020501
Prieto, P., & Bosch-Baliarda, M. (2006). The development of codas in Catalan. Catalan Working
Papers in Linguistics, 5, 237–272.
Rose, Y. (2000). Headedness and Prosodic Licensing in the L1 Acquisition of Phonology. Unpub-
lished Ph.D. dissertation. McGill University, Montreal.
 Maria-José Ezeizabarrena & Alaitz Alegria

Saceda, M. (2005). Adquisición prosódica en español peninsular septentrional: la sílaba y la

palabra prosódica. Unpublished Master thesis. Universitat Autònoma de Barcelona,
Barcelona.
Saizar, M. (2005). Euskararen jabekuntza: silaba itxiak edo kontsonantez amaitzen diren koda-
egiturak. Unpublished Master thesis. University of the Basque Country, Vitoria-Gasteiz.
Salidis, J., & Johnson J. S. (1997). The production of minimal words: A longitudinal case study
of phonological development. Language Acquisition, 6(1), 1–36.
DOI: 10.1207/s15327817la0601_1
Schnitzer, M.L., & Krasinski, E. (1994). The development of segmental phonological production
in a bilingual child. Journal of Child Language, 21, 585–622.
DOI: 10.1017/s0305000900009478
Vogel, A. (1975). One system or two: An analysis of a two-year-old Romanian-English bilin-
gual’s phonology. Papers and Reports on Child Language Development, 9, 43–62.
Zamuner, T.S., Gerken, L.A., & Hammond, M. (2005). The acquisition of phonology based on
input: A closer look at the relation of cross-linguistic and child language data. Lingua, 115,
1403–1426. DOI: 10.1016/j.lingua.2004.06.005
 The locative paradigm in the L2 Spanish
of Catalan native speakers

Silvia Perpiñán
The University of Western Ontario

This chapter explores the expression of locative and existential constructions

in the Spanish of native speakers of Catalan (N = 20). Three (micro)parametric
differences are investigated: the definiteness effect, the eventive effect, and the use
of clitics. An oral production task (OPT) and an acceptability judgment task
(AJT) were analyzed; a control group of native Spanish speakers was included
(N = 20). Results of the OPT indicated underuse of estar, and an increase in
clitics in the grammar of Catalan-Spanish bilinguals; the AJT showed that
bilinguals accepted ungrammatical sentences with ser to locate in Spanish, and
ungrammatical definite DPs with haber, displaying significant differences in their
implicit grammars and a certain degree of functional convergence and language
interference.

1.  Introduction

This study investigates the use of Spanish as a second language by native speakers
of Catalan. Spanish and Catalan are the two co-official languages of Catalonia,
and their amount of use mainly varies according to the geographical area where
it is spoken as well as issues such as language choice and national identity. Given
the sociolinguistic reality of Catalonia, all Catalan native speakers are, to differ-
ent degrees, bilingual in both Catalan and Spanish. This study investigates the
Spanish of Catalan native speakers who declare themselves dominant in Catalan,
use Spanish only when necessary and have a clear ethnolinguistic identification
with Catalan. The purpose of this study is twofold. On the one hand, it proposes
that the Spanish of these Catalan speakers shares some similarities with that of
other second language learners, displaying similar traits to those of interlanguage
grammars. On the other hand, it provides an analysis of the language of these
Catalan-dominant speakers under the framework of experimental generative SLA,
something that is not common in the Spanish-Catalan bilingualism literature.

doi 10.1075/ihll.3.05per
© 2015 John Benjamins Publishing Company
 Silvia Perpiñán

Recently, there has been an increase in the much-needed description of the

Spanish language spoken by Catalan bilinguals (see Blas Arroyo 2011; Sinner 2004;
Sinner & Wesch 2008 for overviews). Practically all of these studies are enclosed in
the field of contact linguistics and/or a variationist framework, with a sociolinguist
focus. The purpose of this chapter is to look at the same overall phenomenon from
a different perspective: contact linguistics as the byproduct of second language
acquisition. Whereas it does not aim to deny the relevance of sociolinguistic vari-
ables, or the field of language contact, here we are interested in the outcomes of
bilingualism as instances of language development. Therefore, we believe that the
methodology and theoretical framework provided by generative SLA is an appro-
priate model to describe the language of these bilinguals, and can provide us with
a better understanding of their implicit grammars. Furthermore, we believe that
language contact starts at the individual level, in the bilingual mind. Thus, it is cru-
cial to understand this bilingual mind and the process of development of a second
language in order to understand a bilingual society.
In particular, this study investigates the expression of location and existen-
tial constructions in Spanish by Catalan-dominant bilinguals, a topic generally
understudied in the literature of the acquisition of Spanish. Some studies have
been conducted on the acquisition of the copular contrast ser/estar ‘to be’ in L2
Spanish, but mostly in attributive contexts when combined with adjectives and
not with PPs. In this investigation, we also explore existential sentences, which
have crucial differences with those of Catalan, as explained in Section 3. The
chapter is organized as follows: Section 2 describes the variety known as C ­ atalan
­Spanish. Section 3 describes the linguistic phenomenon and its properties in the
two languages. ­Section 4 reviews the main studies on SLA on this linguistic topic.
­Section 5 explains the methodology of this study and Section 6 presents the results.
­Section 7 discusses the main findings and conclusions.

2.  The Spanish of Catalan speakers

According to the last survey on language uses in Catalonia,1 approximately

3.4 million people identify Spanish as their first language while slightly less than
2 million identify Catalan as their first language. Only 236,000 people acknowl-
edge having both languages as their first languages. Therefore, even though bilin-
gualism is an essential feature of the identity of Catalonia and Catalonians, and
nearly all Catalonians can understand and speak both languages, language choice

.  Idescat i Secretaria de Política Lingüística del Departament de Vicepresidència 2008.

 The locative paradigm in the L2 Spanish of Catalan native speakers 

and language identity is a key element in the linguistic reality of these bilinguals
(Boix-Fuster & Sanz 2008). Moreover, the family origin and the neighborhood
in which the bilingual is born and raised are crucial components that ultimately
condition their level of proficiency in the two languages. For instance, according to
the same survey, the average percentage of the population that has Catalan as their
first language is 40.4%. However, this figure multiplies 1.5 to 2 times in central
Catalonia (comarques centrals), whereas the percentage is significantly lower in
the metropolitan area of Barcelona. Given these numbers, it is reasonable to think
that Spanish or Catalan in certain speakers develops as an early second language
depending on the geographical area and/or the family origin.
Nonetheless, most of the studies on the Spanish variety spoken by Catalan-
speakers, Catalan Spanish (CS), are general descriptions of the characteristic fea-
tures that make this variety different from General Spanish (GS), and most of them
advocate for a variationist approach as the most appropriate way to fully describe
it (Blas Arroyo 2004; Vann 2002; Wesch 1997). Blas Arroyo (2011) describes the
Spanish in contact with Catalan as a variety that results from linguistic and extra-
linguistic factors, the result of multiple causation (Thomason & Kaufman 1988).
The position of Blas Arroyo is that most of those factors are directly related to
the contact situation of the two languages. For example, the increased use of the
analytic future in CS compared to the form ir a + infinitive (go to + inf.), consider-
ably more common in other variants of Spanish (Wesch 1997), is believed to be
caused by the presence of Catalan since Standard Catalan (SC) does not employ
the periphrasis with future meaning.
Other forms that characterize the Spanish in Catalan-speaking territories are
innovations or forms that are not present in GS but that could also be explained
internally. For instance, the common use of ves (‘go’) as the second person impera-
tive form of the verb ir (‘ve’ in GS) can be explained by internal and external rea-
sons: Catalan has the form vés, so it could be just an instance of transfer, but there
is also a strong tendency in GS to mark all of the 2nd person singular forms with
the morpheme -s (trajistes instead of trajiste), so both explanations are possible,
and the most probable explanation is that both factors act together.
Besides the increased frequency of forms due to convergence and the innova-
tions within the possibilities of the receptive language, Blas Arroyo (2011) men-
tions cases of loanwords, calques, and language interference, such as the sibilant
voicing across word boundaries or the devoicing of final stops. In morphosyntac-
tic terms, he identifies reinterpretation as a distinctive attribute. This is when two
linguistic features no longer keep the original opposition of the receptive language,
in this case Spanish, but adopt the opposition they have in the other language,
Catalan. The use of ser/estar in locative constructions, the focus of our investiga-
tion, is given as an example of reinterpretation. These are cases in which these
 Silvia Perpiñán

speakers use the verb ser to express the location of objects or people, as it is used
in SC. However, GS requires estar in these contexts. The following examples are
from Serrano (1996):
(1) a. Mamá es a casa de la abuelita.
Mom be-Ser at home of the grandma
‘Mom is at grandma’s house.’
b. El pan es a la mesa.
The bread be-Ser at the table.
‘The bread is on the table.’

As we will see, very few of these ‘reinterpretations’ of the copulas were found in
our production data, but more subtle differences of locative and existential con-
structions were discovered. We believe that only with an experimental method-
ology will some of those distinctive features be discovered. This is precisely the
purpose of this chapter: to provide experimental data on one particular linguistic
variable to contribute to the description of the Spanish used by Catalan speakers.
It has been proven that generative SLA methodology is a useful and enlightening
approach to investigating other language contact situations. For instance, Sánchez
(2003, this volume) investigated functional interference and functional conver-
gence processes in Quechua-Spanish bilinguals. Both processes are considered to
be the byproduct of feature reassembly in the L2 (Lardiere 1998, 2008, 2009), that
is, the mapping between grammatical features and overt morphology.
According to Sánchez (2003, p. 13), “functional interference consists of the
activation of functional features in one language triggered by input in the other
language.” The result of this process is frequently a syntactic change, since a gram-
maticalized feature of one language gets mapped into different morphology in the
other language. We find an example of this phenomenon in the Spanish spoken
by Southern Quechua-Spanish bilingual children, who had problems interpret-
ing sentences with the dative clitic le (Kalt 2003). These bilinguals were assigning
other meanings to the clitic le, namely meanings that Quechua grammaticalizes
with overt morphology such as oblique or locative functions, but that Spanish
does not. Therefore, these bilinguals were interpreting the Spanish dative clitic
not only for dative functions, but also for oblique or locative meanings, producing
functional interference in the clitic system: features from the L1 such as oblique or
locative functions are mapped onto L2 morphology that originally did not convey
this meaning. Conversely, functional convergence produces a fusion of functional
features and creates the same new mappings between features and morphology in
the two languages. This occurs when the input from one language frequently acti-
vates a feature not previously activated in the other language (Sánchez 2004). For
instance, when Quechua-Spanish bilinguals use Spanish imperfect and pluperfect
 The locative paradigm in the L2 Spanish of Catalan native speakers 

forms not so much to express aspectual differences, but to express evidentiality,

a feature that is present in Quechua morphology but not in Spanish morphology.
With this theoretical framework in mind, we ask whether Catalan-Spanish bilin-
guals also display functional interference and/or functional convergence in their
grammars.

3.  Th
 e linguistic phenomenon: The expression of existentials
and locatives

Assuming a unified account for locative and existential constructions (Freeze

1992; Lyons 1967; Rigau 1997), there is a universal locative paradigm with three
possible surface structures related syntactically: (a) the predicate locative, with
examples in (2); (b) the existential structure, as in (3); and (c) the possessive con-
struction with have, exemplified in (4). GS employs the verb estar for the predicate
locative (2a) except for eventive themes, in which case ser is used (2c). SC, on the
other hand, employs ser with object and event subjects (2b, 2d), although estar is
constantly gaining ground to locate objects, especially in some areas of Barcelona,
and in the southern varieties (Sanz & González 1995; Solà 1994).
(2) Predicate Locative
THEME COPULA
LOCATIVE
a. Las llaves están en el cajón. GS: be-estar
b. Les claus són al calaix. SC: be-ser
‘The keys are in the drawer.’
c. La fiesta es en casa de Joan. GS: be-ser
d. La festa és a la casa d’en Joan. SC: be-ser
‘The party is at Joan’s place.’
(3) Existential
CL COPULA THEME LOCATIVE
a. [e] Hay unas llaves en el cajón. GS: have
b. *(Hi) ha unes claus al calaix. SC: have
‘There are some keys in the drawer.’
(4) Possessive
a. Tengo las llaves. GS: tener
b. Tinc les claus. SC: tenir
‘I have the keys.’

According to Brucart (2012), the main parametric difference between Catalan

and Spanish in the locative system lies in the preposition system. Notice the
 Silvia Perpiñán

­ ifference in use between the preposition en and a in Catalan and Spanish (2a,
d
2c vs. 2b, 2d). Brucart argues that GS Estar + en = SC Ser + a, the main differ-
ence being that Catalan preposition a can express the notion of path, which in
turn values the uninterpretable feature in the functional projection. In Spanish
locatives, except with eventive (2c) or path nominals, only estar can value this
uninterpretable feature. In Standard Catalan, since the preposition can do it, the
default copula ser appears (2b, 2d). Catalan and Spanish further differ in the pres-
ence of the locative proform hi. The clitic is needed in Catalan haver construc-
tions, even if the locative PP or AdvP appears in the existential sentence (3b), in
which case this PP locative would occupy a peripheral position in the sentence
(Rigau 1994). Spanish has the lexicalized locative proform y attached to the verb
haber in the present tense, but not in other tenses, so it is not a productive strat-
egy and it is not perceived as a locative proform (3a). Longa, Lorenzo and Rigau
(1998) proposed that Spanish (also Galician and Asturian) has a silent clitic [e]
in these cases (3a). On the other hand, the Catalan clitic is not required with the
locative ser construction (2b, 2d), although it needs to appear if the locative PP
is not present as (5a) below indicates. It is argued that Spanish would resort to a
silent clitic in these cases (5b).
(5) a. Ell *(hi) és. SC. be-ser + CL
He CL BE
b. Él [e] está. GS. be-estar
‘He is (there).’

Longa et al. (1998) also proposed that due to the poor Spanish clitic paradigm, this
language ‘recycles’ accusative clitics for partitive, locative and even subject uses, as
shown in (6b, 7b). SC, on the other hand, cannot have an accusative clitic with the
locative 〈have〉 (7a), according to these authors, because of the incompatibility
of the accusative clitic with hi.2 The optionality of the Spanish accusative clitic,
marked with the parenthesis, indicates that these sentences are also grammatical
with object drop or a null object.
(6) a. No hi ha pa. → No *(n’) hi ha. SC
Not CL has bread not   CLpart CL has
b. No hay pan.     → No (lo) hay. GS
Not has bread   No (CLacc) has
‘There is no bread.’

.  The technical details have been omitted since they are not relevant to the focus of this
study. Longa et al. (1998) assumed an AgrP to which hi adjoins. In AgrP, the combination of
the person features produces the derivation to crash.
 The locative paradigm in the L2 Spanish of Catalan native speakers 

(7) a. No hi havia el gos. → *No l’ hi havia. SC

not CL have the dog not CLacc CLloc have
b. No había un perro   → No (lo) había. GS
not had a dog   not CLacc had
‘There wasn’t the/a dog.’

Furthermore, as (7) shows, Catalan, unlike Spanish or English, does not present
the definiteness effect (DE). This restriction is mainly based on the semantics of
the theme. Milsark (1977) distinguished between weak and strong expressions in
English and argued that strong expressions are not allowed in the existential there
construction, but weak expressions are. Weak expressions in English are indefinite
markers such as a, some, cardinal numbers or bare plurals. Strong expressions in
English are definite markers such as the, all, every, each, possessives or personal
pronouns. Thus, definite noun phrases are usually disallowed with an existential
construction, being in a sort of complementary distribution with the locative par-
adigm, which usually requires a definite noun phrase (cfr. 2a vs. 3a). However, the
version of (3b) with a definite noun phrase (8a) is perfectly possible in Catalan, but
not in Spanish (8b), illustrating an important parametric difference between these
two Romance languages.
DEFINITESS EFFECT
(8) a. Hi ha les claus al calaix. SC
CLloc has the keys in-the drawer
b. #Hay las llaves en el cajón. GS
 Has the keys in the drawer
‘There are the keys in the drawer.’

The contrast is even clearer with proper nouns:

(9) a. Hi ha el Joan a casa. SC
CLloc has the Joan at home
b. *Hay Joan en casa. GS
 Has Joan at home
‘There is Joan at home.’

Lyons (1999) characterizes definite nouns as more familiar, more individualized,

and more identifiable than indefinites. Their reference points towards a unique
element, which the hearer and the speaker share. Freeze (1992) and White,
Belikova, Hagstrom, Kupish and Özçelik (2012), among others, suggested that
the definiteness effect is a universal restriction observed in most natural lan-
guages, particularly in affirmative sentences. This restriction, not applicable to
list-readings, has been explained in semantic terms (Milsark 1977), in syntactic
terms (Safir 1985), and in pragmatic terms (Lumsden 1988). The basic pragmatic
 Silvia Perpiñán

idea, defended by Holmback (1984) is that existential constructions introduce a

new entity in the discourse; therefore the new element needs to be unrelated to
the preceding discourse, that is, not previously defined (i.e. indefinite). In any
case, Lyons (1999) considers it a semantics-pragmatic constraint, not a syntactic
one, with a broad crosslinguistic representation, more present in some languages
than in others. Catalan, as well as Italian, are examples of two languages that do
not seem to observe this constraint.
In theoretical terms, the recent analyses for the Spanish copulas (Brucart
2010, 2012; Gallego & Uriagereka 2011; Zagona 2012) agree on considering ser a
more basic verb, the default copula, and estar a sort of derivation of ser with the
merging of a preposition of terminal coincidence: estar = ser + P. According to
Gallego and Uriagereka (2011), this general idea that estar implies something else
compared to ser, that it needs contextual specifications, is corroborated by the fact
that the Catalan locative/existential construction with ser requires a locative clitic
to be equivalent to the Spanish estar, as the contrast in (4) shows.
To recapitulate, the main lexical, morphosyntactic and semantic differences in
the locative paradigm between Catalan and Spanish are the following:

a. SC uses ser instead of estar in locative constructions, although it has been

argued that this difference is progressively disappearing, displaying large con-
vergence in some dialectal areas.
b. SC needs the overt clitic hi in the existential construction, either with haver
or with ser. Moreover, GS has an impoverished clitic system and expresses
partitive meanings with accusative clitics or with an empty element. SC, on
the other hand, has a rich clitic system that needs to be used maximally.
c. SC does not present the definiteness effect, whereas GS does.

Given these differences, we question whether L1 Catalan – L2 Spanish speakers are

able to fully acquire the distribution of estar in locative predicates and observe the
restriction on definite DPs in Spanish existential constructions. Furthermore, we
wonder how the bilingual mind will restructure her clitic system into a smaller one.
In particular, how a speaker who is used to maximally expressing all the arguments
of the sentence through clitics will manage with a reduced morphological paradigm.

4.  The acquisition of locatives and existentials in L2 Spanish

The acquisition of the Spanish ser/estar contrast is one of the characteristic diffi-
culties for speakers whose native language has a unique to-be system. Some stud-
ies have been devoted to the acquisition of this contrast in Spanish L2 speakers,
 The locative paradigm in the L2 Spanish of Catalan native speakers 

particularly in English-speaking learners (Ryan & Lafford 1992; VanPatten 1985,

1987), but practically no other language combination has been investigated. Most
of the studies have explored the acquisition of the copulas when combined with
adjectives (Bruhn de Garavito & Valenzuela 2006; Geeslin 2002; Schmitt 1992;
Schmitt & Miller 2007), but not so much in other contexts, with the exception
of the studies on child psychology carried out by Sera and colleagues (Sera 1992;
Sera, Bales & del Castillo 1997). The studies with English-speaking learners have
shown that estar for locative contexts appears late in the development of the copu-
las. VanPatten (1985, 1987) proposed five developmental stages for the acquisition
of the ser/estar contrast. First, there is a short period of omission of the copula, fol-
lowed by a period of overproduction of ser for all contexts (second developmental
stage). Estar appears in the third stage, but only as an auxiliary for progressives. It
is not until the fourth developmental stage that estar is used for locative contexts
(later according to Ryan & Lafford 1992). The final developmental stage is when
both ser and estar are used with adjectives. Ryan and Lafford (1992) tested English-
speaking students in 3 different moments, within a four and a half month study
abroad program. Overall, they found that the use of ser was much more accurate
(90%) than that of estar (40%–70% accuracy). The patterns of errors were also dif-
ferent. With ser the most common mistake was to omit the copula, while with estar
the most common mistake was to substitute it for ser, particularly in the locative
contexts.
Despite the late appearance of estar in the language of L2 learners, several
variationist studies have documented a progressive historical replacement of ser
with estar, especially when the copulas are combined with adjectives. This phe-
nomenon has been attested in monolingual contexts (Gutiérrez 1992), bilingual
acquisition (Gutiérrez 2003; Ortiz-López 2000; Silva-Corvalán 1986, 1994) and L2
acquisition (Geeslin 2001, 2002). In the case of Catalan, this gradual substitution
of ser by estar has also been attested and explained as an internal change acceler-
ated by contact with Spanish (Sanz & González 1995; Solà 1987). For instance,
Solà (1987) summarizes Silva-Corvalán’s (1986) study about the Spanish in Los
Angeles and suggests that a similar process of an accelerated expansion of estar,
an already ongoing internal change, is happening in present Catalan as the result
of language contact.
Two sociolinguistic studies have investigated the copula selection in Catalan-­
Spanish bilingualism. Sanz and González (1995) investigated ser and estar in
Tortosí Catalan, a southern variety of Catalan. They interviewed nine bilingual
speakers from 5 to 61 years of age in a predominantly Catalan-speaking village.
The authors analyzed the uses of ser/estar with adjectives (in obligatory and in
free variation contexts) and in locative structures that the participants produced
in spontaneous, controlled production, and grammaticality judgments. Overall,
 Silvia Perpiñán

Sanz and González (1995) found extension of estar over ser in both contexts, with
adjectives and locatives, despite the prescriptive rules proposed by Catalan gram-
marians; the overextension is larger in younger speakers. In locative contexts, they
did not find a single instance of ser with animate or inanimate subjects, displaying
a complete replacement of ser with estar in this context. They concluded that this
language change is internally motivated, since it is also attested in Spanish and
French. The language contact situation, as Silva-Corvalán (1986) argued, if any-
thing, may accelerate the process.
On the other hand, Geeslin and Guijarro-Fuentes (2008), in their variationist
study that investigated the distribution of ser/estar with adjectives in Spanish in
contact with Catalan, Basque, Galician, and Valencian3 found that Catalan and
Valencian speakers used estar less often than the monolingual Spanish variety.
They tested 141 speakers in Catalonia, and 51 in Valencia, from ages 15 to 79, with
a sentence selection task. Participants were presented with a paragraph-length
context followed by two sentences, one with ser and one with estar. They had to
choose which sentence they preferred according to the context, or mark ‘both’.
Table 1 summarizes the results.

Table 1.  Rates of copula selection, data from Geeslin and Guijarro-Fuentes (2008)
Both Estar Ser Total

# % # % # %

Catalans 149 3.8 1728 43.8 2071 52.5 3948

N = 141
Valencians  51 2.8  822 44.5  975 52.8 1848
N = 51
Monolinguals 117 5.1 1031 44.9 1148 50.0 2296
N = 83

Despite the small differences among the percentages (44.9 vs. 44.5 or 43.8),
the authors found strong significant differences in the use of estar between the

.  These authors seem to treat Valencian as a different ‘regional language’: “we have no
­evidence that Catalan and Valencian differ in their rules of copula choice, but […] we have
chosen to maintain the distinction in light of government efforts to distinguish the two and in
the absence of sociolinguistic research on copula contrast in Valencian.” (Geeslin & Guijarro-
Fuentes 2008, p. 366). While I acknowledge the research advantages of including another lan-
guage variety in their study, particularly Valencian since it presents significant differences as
compared to SC, I disagree with their language division since it is linguistically unmotivated
and politically biased.
 The locative paradigm in the L2 Spanish of Catalan native speakers 

bilinguals and the monolinguals, and concluded that Catalans and Valencians
used estar significantly less frequently. Thus, the authors argued that bilingualism
does not necessarily increase the frequency of estar, and that the language contact
situation does not automatically imply simplification of the copulas, as previously
argued. In the case at hand, it seems that the contact with Catalan and Valencian
has the opposite effect: it slows the extension of estar in the Spanish of these areas.
We can only wonder whether this effect will also be found in locative contexts,
not considered in the study. Unfortunately, Geeslin and Guijarro-Fuentes (2008)
did not discuss the type of bilinguals they tested or their language dominance or
identity, a crucial element in Catalan-Spanish bilingualism. Thus, we do not know
whether these results come from Catalan-dominant, Spanish-dominant bilinguals
or both. Geeslin and Guijarro-Fuentes (2008) did analyze some extralinguistic
variables such as the participants’ L1, but it was only significant in the Galician
group. In any case, we do not have any information about the composition of the
groups or where these data were collected.
To my knowledge, the only available published study on the acquisition of L2
Spanish and L2 Catalan attributive, locative, existential and possessive construc-
tions is Gràcia, Crous and Garganta (2008). In this study, they investigated the
oral speech of nine adult L2 learners of Spanish and nine L2 learners of Catalan
by Moroccan Arabic speakers, Romanian speakers and Chinese speakers, three
participants per group. Table 2 displays the percentages of errors in the total pro-
duction, as well as the most representative errors in each group.

Table 2.  Locative, existential and possessive constructions in L2 Spanish,

data from Gràcia, Crous and Garganta (2008)
Group % of errors in Types of errors
total production
Omission (%) Substitution (#)

Moroccan 8.48% 70% 3/34

Arabic (Estar instead of haber)
Chinese 25.23% 76% 10/84
(Tener instead of haber)
Romanian 11.73% 32% 14/54
(Estar instead of ser)

The authors conducted an analysis of errors and found that the existential
construction was the one that posed more difficulties to the learners, and the most
common error was omission of the copula (or haber), particularly in Chinese and
Arabic speakers. Per L1 group, in the Arabic group, 70% of their errors consti-
tuted a case of omission, especially in present tense. Also, they replaced the verb
 Silvia Perpiñán

estar with haber (3/34), not obeying the definiteness effect such as in Sí, hay el mar
cerca de mí ‘Yes, there is the sea close to me’. As for the Chinese speakers, 76% of
their errors were verb omissions. The remaining errors were substitutions of the
copula with the verb tener (10/84), particularly in existential constructions, such
as in es que tengo una chica de baile muy bien, (literally, ‘is that I have a girl of dance
very well), meaning ‘There is a girl that dances very well’. There are some instances
of replacement of estar with ser (6/84) in locative constructions: cuando yo era en
China ‘when I was in China’. Interestingly, the Chinese group, due to the analytical
morphology nature of their L1, committed an error that no other group did: they
sometimes omitted the verbal part of the Catalan existential form (haver-hi), but
kept the clitic part hi: perquè a Lao n’hi molta comunitat xinesos ‘because in Lao,
there a big Chinese community’. Finally, in the Romanian group, there were 32%
errors of omission and the remaining errors were substitutions, with some over-
production of estar in attributive constructions (14/54). At the same time there
was overproduction of ser in locative contexts (5/54) and in existential contexts
(7/54), especially in L2 Catalan. The authors concluded that almost all errors could
be explained either by the L1 influence or by certain degree of grammatical simpli-
fication in the acquisition process, and thus the incorrect omissions.
Research on the acquisition of semantic aspects that are also present in
the existential constructions has been mainly focused on the definiteness fea-
ture on determiners, particularly in articles (Ionin 2006; Ionin, Ko & Wexler
2004; L­ ardiere 2004; White 2003, 2008). This feature has been less investigated in
existential constructions, with the recent exception of White (2008) and White
et al. (2009, 2012). The main argument in White and colleagues’ studies is that
even though the acquisition of articles is far from being errorless, L2 learners
remarkably produce very few DE violations, a related semantic phenomenon.
For instance, in their two case studies (a Chinese fossilized L2 speaker of English
and a Turkish L2 English speaker) Lardiere (2004) and White (2003) reported
article omission but no problems with DE. These problems with article omission
but not with the DE were interpreted as a mapping problem when surfacing
abstract categories into overt morphosyntax, or an instance of missing surface
inflection (Haznedar & Schwartz 1997; Lardiere 1998; Prévost & White 2000).
White et al. (2012), in an acceptability judgment task targeting semantic inter-
pretations from English L2 learners (L1 Turkish and L1 Russian), found that
the L2 learners were overall very accurate in distinguishing grammatical and
ungrammatical cases of there-insertion. Furthermore, the L2 learners did not
transfer the lack of definiteness effect in negative existentials from their L1s.
These results were interpreted as an indication that there must be some sort of
universal principle that regulates this semantic phenomenon and its (relatively
unproblematic) acquisition.
 The locative paradigm in the L2 Spanish of Catalan native speakers 

Therefore, given the differences between Catalan and Spanish in locative and
existential constructions, and what previous studies have investigated regarding
the acquisition of the copular system, we hypothesize that if Catalan-dominant
bilinguals behave like true L2 learners, they will show a delay and/or a less fre-
quent use of estar in their grammars, probably also due to L1 transfer. On the other
hand, if Catalan-dominant bilinguals speak the more innovative Catalan Spanish
variety, the one in which estar has replaced Catalan ser in locative contexts, then
we will expect their Spanish grammars to converge with those of Spanish mono-
linguals, producing similar rates of estar in locative contexts. Regarding the differ-
ences in the clitic system, we can hypothesize that Catalan-dominant speakers will
need to somehow express those meanings grammatically, as the Quechua-Spanish
bilinguals did, probably with new mappings of syntactic or morphological ele-
ments equivalent to those clitics, or with a higher frequency of accusative clitics to
express those other meanings.

5.  The study

5.1  P
 articipants
The speakers investigated in this study are sequential Catalan-Spanish bilinguals
(N = 20), with a mean age at the time of testing of 23.6 years old (SD = 4.1). They
reported starting their exposure to Spanish once they entered into the school
system or later, between 5 and 9 years of age,4 M = 7.5 (SD = 4.4, Range 5–20).
Until then, Catalan was their exclusive language of communication, and is still
their language of preference. It would be risky and most likely inaccurate to claim
that they had never heard Spanish before that age, but for the most part, these
speakers were raised monolingually, and still develop their lives in a monolingual
manner. They all live in Central Catalonia (Manresa and surrounding villages).
Two control groups of native Spanish speakers were used, one for the oral pro-
duction task, and another one for the acceptability judgment task (AJT) since
the data were collected in different periods. The control group of native speak-
ers for the oral task is slightly more heterogenic than the one used for the AJT.
The first control group of Spanish native speakers (N = 20) was tested in the
USA and in Spain: 15 Spaniards, one Venezuelan, one Colombian, one Mexican,
one A­ rgentinean and one Peruvian. 11 of these Spaniards were born and raised
in monolingual areas of Spain, but 4 of these Spaniards were born and raised

.  In Catalonia, the school system is obligatorily in Catalan; Spanish language is not intro-
duced until 3rd grade.
 Silvia Perpiñán

in Catalonia, although none of them lived in Catalonia at the time of testing.

These 4 Catalan Spaniards come from very Spanish-speaking neighborhoods of
the metropolitan area of B
­ arcelona. The second group of Spanish native speakers
consisted of monolingual speakers (N =21), 16 from the central area of Spain and
5 from El Salvador. They have never lived abroad or outside of their hometown.
Their mean age at testing was 27.6 (SD = 5.3).

5.2  M
 ethodology
All participants completed a linguistic background questionnaire, asking ques-
tions about language use with family and friends, and language preferences; these
responses were used to classify the participants into groups. They also completed a
proficiency test, which consisted of parts of the grammar and vocabulary sections
of the superior level of DELE (Diploma de Español como Lengua ­Extranjera),
k = 23. Spanish speakers and Catalan speakers had exactly the same mean accu-
racy in their proficiency results: 94% of correct responses.
There were two main experimental tests, an Oral Production Task (OPT) and
a web-based Acceptability Judgment Task (AJT). The OPT consisted of a ‘Spot
the Difference Task’, with 5 pairs of very similar pictures that participants had to
describe. In particular, they had to verbally identify the differences between the
two pictures. The bilingual group produced a total of 17,221 words, with an aver-
age of 861 words per person, whereas the monolingual group produced noticeably
less, a total of 11,241 words, with an average of 562 words per speaker. The utter-
ances localized differences about the (in)existence of certain animals or objects, or
depicted differences in position, color, and size of the main characters; for instance,
they could explain that in picture A there were five birds but in picture B there were
only four birds in the upper left side (see Figure 1). The task elicited both existential
and locative constructions. Monolinguals were asked to compare picture A with
picture B, whereas bilinguals received slightly different instructions; specifically,
they were asked to compare the pictures as if they belonged to different periods of
time (present vs. past). The purpose of these instructions was to elicit past verbal
forms, since instances of agreement with the theme in Spanish cannot occur in
Spanish (había(n) libros vs. hay libros). Nonetheless, we believe this difference does
not affect the overall results. An example of a scenario is given in Figure 1.
The web-based AJT targeted locative structures with ser and estar, with objects
and events as subjects (10). It also tested the definiteness effect with haber and
estar, in simple sentences (11) and in relative clauses (12). This makes a total of
nine conditions, five tokens each, 45 target items, in an AJT of 110 items. Partici-
pants had to judge these sentences on a scale from 1 to 100 in a slider bar, where
1 meant ‘sounds bad, it is an impossible sentence in the language’, and 100 meant
‘sounds natural, it is a possible sentence of the language’.
 The locative paradigm in the L2 Spanish of Catalan native speakers 

Figure 1.  Sample of a ‘Spot the Difference Task’

(10) *Ser/Estar with Objects

a. El libro *es/está encima de la mesa.
The book *ser/estar on-top of the table
‘The book is on the table.’
Ser/ *Estar with Events

b. La reunión es/*está en el hotel Majestic.
The meeting be/estar in the hotel Majestic
‘The meeting is in the hotel Majestic.’
(11) Haber with *definites and indefinites.
a. Hay *las/unas llaves encima de la mesa.
have  the/some keys on-top of the table
‘There are the/some keys on the table.’
Estar with definites and ??indefinites

b. El/??un libro está encima de la mesa.
The/??a book estar on-top of the table
‘The/ a book is on the table.’
(12) Relative Clauses: ?Haber/Estar with definite antecedents
a. Las llaves que ?hay/están en la puerta son mías.
The keys that    have/estar on the door are mine.
‘The keys that are at the door are mine.’

6.  Results

6.1  Results of the Oral Production Task

All utterances produced in the OPT were transcribed and checked by two native
speakers. Later, these were coded with the CLAN program from the CHILDES
system 〈http://childes.psy.cmu.edu/clan〉 by two Spanish-speaking researchers. All
 Silvia Perpiñán

predicates that expressed location or an existential meaning were coded. That is,
all occurrences of haber with an existential meaning, as well as the occurrences of
estar and ser accompanied by a locative adverb or a prepositional phrase express-
ing location, and llevar ‘to carry, to wear’, and tener ‘to have’ as possessives. Verbs
such as existir ‘to exist’, aparecer ‘to appear’, their negative counterparts such as fal-
tar or hacer falta ‘to lack, to miss’, if followed by a locative expression were included
under the category ‘other’. Also verbs such as ver ‘to see’ with a presentational/­
existential meaning in sentences such as veo una ardilla en la foto A (‘I see a squir-
rel in photo A’) were also included under the ‘other’ class. In addition, verbs were
coded as first or second mention (most of the utterances are comparisons between
the two pictures), and the theme was classified according to its semantics and form:
as definite or indefinite, quantificational, negative, pronominalized or elided.
A total of 2309 tokens were analyzed, 1440 produced by the Catalan speakers,
and 869 by the Spanish speakers. Overall, both groups had the same general pat-
terns in verb selection to express location and existential constructions: the most
common verb was haber, followed by the verb estar. In the case of the S­ panish
speakers, the next most frequent verb was tener, the possessive one, but in the
case of the Catalan speakers, llevar, a verb rarely used by Spanish speakers, was
more common than tener. A Chi-Square test with the counts per group and verb
(haber, estar, tener, llevar) displayed a significant association between the use of
these verbs and the group χ2 (3) = 100.360, p < .001, Cramer’s V= .226. Percentages
of use were also calculated by participant and verb, and are displayed in Figure 2.

45

40

35

30 Haber
Estar
25 Tener
Llevar
20 Other
Ellipsis
15 Ser

10

0
L1 Catalan L1 Spanish

Figure 2.  Frequency of lexical verb in locative and existential contexts

by type of speaker (in %)
 The locative paradigm in the L2 Spanish of Catalan native speakers 

A multivariate ANOVA was conducted with the percentages of verb use as the
dependent variables, and group as the independent variable. The results indicated
significant differences in the use of estar F(1,38) = 8.1, p = .007; haber F(1,38) =
13.43, p = .001; llevar F(1,38) = 36.57, p < .001; tener F(1,38) = 4.229, p = .047 and
ellipsis F(1,38) = 5.074, p = .030, but not in the use of ser or other verbs (p > .01).
Thus, despite the overall similar tendency in the use of the locative verbs, we can
see significant differences in the use of haber, which Catalan speakers used more
frequently (42% vs. 29 %), and in the use of estar, which they used significantly
less frequently (17% vs. 24%). The use of ser, a previously reported characteristic of
Catalan Spanish for locative constructions was not found significant (.2% vs. .8%);
only 3 tokens were found in the Catalan speakers as in (13a), and 8 in the Spanish
speakers (13b).

(13) Uses of SER in locatives

a. Bajando por el vestido, otra diferencia es en el color
Descending by the dress, other difference ser in the color
de la gema.
of the gemstone
‘Going down by the dress, there is another difference in the gemstone.’
 Catalan speaker # 129
b. De las tres manzanas que hay en la foto A, en la
Of the three apples that have in the photo A, in the
foto B son cuatro en la parte arriba derecha.
photo B ser four in the part upper right
‘Of the three apples that are in the picture A, in picture B there are four
in the upper right part.’ Spanish speaker # 58

With respect to the differences in the clitic system, we coded not only the use of
accusative clitics, but also the reference to the theme in the second part of the
comparison (with second mention verbs), and whether the DPs were overtly pro-
nounced, pronominalized, or partially elided (Numeral + elided NP). As explained
before, Catalan maximizes the use of clitics (accusative, partitive, locative), and
these are obligatory in argumental positions. Spanish has only accusative clitics,
and employs a null clitic more often than Catalan does. The clitic then can only
appear with verbs that assign accusative case such as llevar, tener, and haber. For
this calculation, we first classified the locative and existential verbs as first or sec-
ond mention, and then we counted the appearance of repeated DPs and clitics in
the second part of the comparison, where they are legitimated. An example of this
coding is in (14).
 Silvia Perpiñán

(14) a. El rey antes tenía cuatro símbolos en la barriga, ay,

the king before had four symbols on the belly, oops,
antes no tenía tres símbolos en la barriga y ahora tiene
before not has three symbols on the belly and now has
tres símbolos en la barriga.
three symbols on the belly

‘Before, the king had four symbols on the belly, oh, no, before he did
not have three symbols on the belly and now he has three symbols on
the belly.’ Catalan Speaker # 132
b. Antes había una casa sin palos en el tejado y
before had a house without sticks on the roof and
ahora sí los tiene.
now yes cl has
‘Before, there was a house with no sticks on the roof, and now it does
have them.’ Catalan Speaker # 123

In Example (14a), the last tres símbolos en la barriga (‘the last three symbols on the
belly’) was considered a repeated NP, and coded as such, since the speaker could
have chosen to refer to it with a negative polarity item such as nada (‘nothing’), with
an accusative clitic lo (‘it’) as in (14b), or with an empty element. With respect to the
verbs, había in (14b) was considered first mention, but tiene was considered second
mention. Similarly, the first two tenía from (14a) are considered first mention verbs,
but tiene was coded as second mention. Table 3 includes the tokens and percentages
of locative and existential verbs used in first and second mention as well as the use of
accusative clitics and repeated DPs. The percentages were calculated over the total of
second mention verbs, the context in which a clitic would be legitimized.

Table 3.  Frequency and distribution of locative verbs, clitics, and repeated DPs
1st mention verbs 2nd mention verbs Accusative CL Repeated NP

L1 Catalan 916 524 90 73

Tokens and % 63.6% 36.4% 17.17% 13.93%
L1 Spanish 578 291 31 27
Tokens and % 66.5% 33.5% 10.65% 9.28%

The percentages of clitic and repeated DPs use per participant were submitted
to an independent sample t-test and results indicated that these variables are not
statistically significant at the.05 α level, but both were marginally significant (Per-
centage of Repeated DPs use, t (38) = 1.82, p (2-tailed) = .076; Percentage of Clitic
use, t (38) = 1.891 p (2-tailed) = .064).
The definiteness of the theme was also analyzed in the production data. Per-
centages of appearance of haber with a definite theme were very similar between
 The locative paradigm in the L2 Spanish of Catalan native speakers 

the two groups (K = 28, 4.6% in the Catalan speakers vs. K = 11, 4.4% in the
S­ panish speakers, p > .1). The percentages were calculated from the overall use
of haber. However, there are some differences between the two groups: on the
one hand, in the L1 Catalan group, 9 speakers produced haber together with a
definite NP, whereas in the L1 Spanish group, only 4 speakers did, precisely the
participants who were born in Barcelona. On the other hand, whereas the Catalan
speakers used a definite NP in both, simple (k = 9, as in 15a) and relative clauses
(k = 18), a definite NP was only found as the antecedent of an existential relative
clause in the Spanish speakers production as in (15b). In a relative clause structure,
assuming that the nominal antecedent raises from the subordinate clause (Kayne
1994), the definiteness constraint is not violated because the determiner of the
relative clause antecedent is external to the subordinate clause.5
(15) a. En el antes no está el pájaro pero hay como un
in the before not estar the bird but have like a
cuadro o una ventana y en la imagen de ahora sólo
painting or a window and in the image of now only
hay el pájaro.
have the bird
‘The bird is not there in the before (picture) but there is like a painting
or a window and in the image of now, there is only the bird.’
 Catalan Speaker # 20
b. En la foto A el ratón que hay al lado del
In the photo A the mouse that have to-the side of-the
príncipe está más cerca.
prince estar more close
‘In picture A, the mouse that is next to the prince is closer.’
 Spanish Speaker # 58

Interestingly, two linguistic features were found in the speech of the Catalan speak-
ers that were not produced by the Spanish speakers: the occasional presence of the
Catalan locative clitic hi, and agreement errors with the verb haber. Four Catalan
speakers produced 9 instances of the clitic hi. This clitic was produced with the
past form of haber, había (16a); only one instance of this clitic was found with
estar (16b). The fact that the clitic was hardly produced with a different verb, and
the almost homophonic nature of había and están with their Catalan counterparts
(havia, estan) could indicate that this is a case of language transfer or even code-
switching rather than a linguistic innovation into their L2 Spanish.

.  I owe this observation to an anonymous reviewer.

 Silvia Perpiñán

(16) a. Antes hi había cuatro pájaros encima de la foto a

before CLloc had four birds on-top of the photo to
la izquierda, ahora hay sólo tres pájaros.
the left now have only three birds
‘Before there were four birds on top of the picture to the left, now there
are only three birds.’
b. Antes no habían estos tres palos y actualmente sí que
before not had these three sticks and nowadays yes that
hi están estos tres palos.
CLloc estar these three sticks.
‘Before there were not these three sticks and now they are there, these
three sticks.’ Catalan speaker # 132

6.2  Results of the Acceptability Judgment Task

In the AJT, participants had to rate the sentences with a slider between 1 and 100,
so the average of their ratings is out of 100. With respect to the use of ser and estar,
mean results of the AJT variables described in (10) were submitted to a repeated
measures ANOVA with type of verb (ser, estar) and eventiveness of the subject
(event, object) as the within-subjects variables, and L1 group as the independent
between-subject factor. Results indicated a main effect of verb F(1,39)  =  15.29,
p  <  .001, a main effect of eventiveness F(1,39) = 8.932, p = .005, and signifi-
cant interactions of verb*group F(1,39) = 22.85, p < .001, eventiveness*group
F(1,39) = 36.15, p < .001 and verb*eventiveness F(1,39) = 491.51, p < .001. When
we look at the mean responses, plotted in Figure 3, we can interpret these results:
all participants recognized that ser is used with events, and estar is not appropriate
in this context, but while the monolingual group categorically rejected the use of
ser with objects, the Catalan speakers did not, showing certain acceptance of this
verb to locate objects.
Regarding the definiteness of the theme and its interaction with the selec-
tion of the existential verb, as described in (11), a repeated measures ANOVA
with the mean judgments on those sentences displayed a main effect of verb
F(1,39) = 40.04, p < .001, but not of definiteness F (1,39) = 3.199, p = .081; how-
ever, definiteness*group interacted significantly F (1,39) = 10.10, p = .003, so did
verb*group (F (1,39) = 7.23, p = .010), and verb*definiteness (F (1,39) = 272.64,
p  <  .001). When we look at the mean percentages, shown in Figure 4, we can
interpret the lack of a main effect of definiteness as the result of Catalan speakers
accepting definite themes with the verb haber, something categorically rejected by
the monolingual speakers. Estar with indefinite themes, on the other hand, seems
to be more acceptable for both groups.
 The locative paradigm in the L2 Spanish of Catalan native speakers 

100
90
80
70
60
L1 Catalan
50
L1 Spanish
40
30
20
10
0
*Ser Estar Ser *Estar

Object Event

Figure 3.  Mean acceptability rates of ser and estar uses and the eventiveness effect

100

90

80

70

60
L1 Catalan
50
L1 Spanish
40

30

20

10

0
#Haber Estar Haber Estar

Definite Indefinite

Figure 4.  Mean acceptability rates of haber and estar uses and the definiteness effect

Finally, the definiteness effect on relative clauses (12), as the ones produced
by Spanish speakers (15b) was also considered. The repeated measures ANOVA
indicated a main effect of verb (haber vs. estar), F(1,39) = 7.5, p = .009, but no
significant interaction between verb*group (p > .1), indicating that both groups
answered similarly to this effect. Figure 5 displays the mean acceptability rates of
these relative clauses by group.
 Silvia Perpiñán

100
90
80
70
60
L1 Catalan
50
L1 Spanish
40
30
20
10
0
Haber Def. RC Estar Def. RC

Figure 5.  Mean acceptability rates on relative clauses with a definite antecedent by verb

7.  Discussion

The oral production data showed a general underuse of estar by the Catalan speak-
ers compared to the use of Spanish speakers. This finding is consistent with the data
found in typical adult L2 learners which showed a delay/underuse in the acquisi-
tion of estar (Perpiñán, 2014; Ryan & Lafford 1992; VanPatten 1987). This find-
ing is also compatible with an explanation in terms of L1 influence, since estar is
not used for locative purposes in Standard Catalan. Similarly to what G ­ eeslin and
Guijarro-Fuentes (2008) found, we can also conclude that the presence of Catalan
produces a reduced use of estar in Spanish in locative constructions, despite the
documented overextension of this verb in Catalan (Solà 1994). We can only specu-
late these speakers have not adopted estar as their regular locative verb in C­ atalan,
at least not to the extent that other dialectal varieties have (Sanz & González 1995).
The underuse of estar in the locative paradigm of these bilinguals seems to be
compensated with a higher use of haber and llevar (‘to carry, to wear’), and not
with an overproduction of ser for locative purposes, unlike in other L2 studies
(Ryan & Lafford 1992; VanPatten 1987). Besides the fact that the native language
of these bilinguals does present estar, we believe this could be due to the nature
of these bilinguals: these Catalan speakers, as their proficiency test showed, are
highly proficient in Spanish, and as such they are not at the first or second devel-
opmental stage in the acquisition of ser/estar. It is an empirical question, though,
whether Catalan children would overproduce ser in the early stages of exposure to
Spanish. Nevertheless, when we examined the intuitional data collected through
 The locative paradigm in the L2 Spanish of Catalan native speakers 

the AJT we can see that there is, indeed, a significant difference in the implicit
grammar of these speakers. Catalan speakers did not strongly reject sentences
with the verb ser to locate inanimate individuals, showing that in their bilingual
mind, ser can still function as a locative verb, independently of the preposition.
Following B ­ rucart (2012), Catalan ser can appear in locative contexts when com-
bined with the preposition a (Spanish Estar + en = Catalan Ser + a). This variable
(ser + object) in the AJT had 5 tokens, each with a different preposition or locative
adverb, in particular: encima de (‘on top of ’), cerca de (‘close to’), entre (‘between’),
en (‘in’, ‘on’) and a (‘at’). All sentences had a mean range of acceptability between
46–59 (out of 100) in the Catalan group, except for the sentence with encima6
which had a mean rate of 17. Thus, we cannot infer that it is the presence of a
in Spanish that legitimizes the acceptance of ser as a locative. To a certain extent
these bilinguals have a converged grammar in which the two copulas can serve for
locative purposes, independently from the preposition and/or the semantics of
the subject. This could be the result of the frequent use of locative ser in Catalan,
which would activate the uninterpretable feature needed in Spanish to yield the
locative interpretation for ser in the bilingual grammar. This activation, though,
is not strong enough to render actual production data, but it certainly affects the
implicit grammar of these bilinguals.
With respect to the clitic system, Catalan speakers employed more accusa-
tive clitics than the Spanish speakers did, but the difference was only marginally
significant between the two groups. Although production data seem to indicate
that Catalan speakers need to fill the argumental positions either with overt clitics
or with full (repeated) DPs more frequently than Spanish speakers do, we cannot
firmly conclude that this is a case of functional interference similar to the one we
described for Quechua-Spanish bilinguals in which the bilinguals assigned dif-
ferent interpretations to Spanish clitics. To investigate that, better-controlled data
would be needed, and unfortunately no clitics with existential constructions were
included in the AJT. This area remains open for future research.
Finally, regarding the semantic properties of the theme in existential con-
structions, whereas the AJT results of the relative clauses tell us that definite ante-
cedents are allowed in existential relative clauses for both Spanish and Catalan
speakers, as expected, the results on simple sentences clearly display the inabil-
ity for Spanish speakers to form existential sentences with haber + definite NP
(M = 4.5), strongly observing the DE. However, Catalan speakers produced some

.  These are the sentences included in this variable: El libro es encima de la mesa; el estadio
de fútbol es cerca del río; la biblioteca es entre el hotel y la universidad; la puerta es a la izqui-
erda de la casa; la televisión es en el salón.
 Silvia Perpiñán

existential sentences with a definite NP, and moderately accepted them in the AJT
(M = 44), showing that their grammars worked differently in that respect from the
Spanish monolingual grammars. Unlike White et al. (2012), which did not find
transfer of the definiteness restriction, or the lack of, in the L2 English of Russian
and Turkish speakers, we did find some type of transfer or interference from L1
Catalan into L2 Spanish. Exactly what is transferred from the L1 and what the
specific nature of this parametric difference is, however, we can only speculate. In
fact, we do not know how these speakers interpret the sentences. We know that
there is a strong cross-linguistic tendency to observe the DE, to the point that sev-
eral authors have considered it to be universal (Milsark 1977; White et al. 2012).
If this effect is a pragmatic effect, as several authors have argued (Lumsden 1988),
it could be possible that these definite DPs are interpreted not so much as definite
but as specific (and maybe indefinite) for these Catalan speakers, and that definite
Catalan DPs do not have the same reference in existential sentences that Spanish
definite DPs would. It is apparent that Catalan definite articles do not imply a pre-
supposition failure, which would disallow them from existential contexts. Accord-
ing to Lyons (1999), definiteness refers to the identification of a unique referent
shared by the speaker and the hearer, and specificity refers to a particular element,
but not necessarily identifiable to the hearer. A test on sentence comprehension
with the appropriate context would be necessary in order for us to pinpoint which
pragmatic or semantic property these speakers are transferring, but it is reasonable
to think that in a sentence such as En casa hay la televisión en el salón ‘At home,
there is the TV in the living room’, the one that scored the highest ratings in the
AJT (M = 65), the Catalan speaker refers to a generic TV, which is not exactly
identifiable to the hearer. If that is correct, then we could have a case of fluctuation
between parameter settings deciding whether DE in Catalan depends on definite-
ness or specificity (Ionin, Ko & Wexler 2004); however, as an anonymous reviewer
points out, the fact that Catalan existentials accept proper nouns clearly indicates
that Catalan allows definite and specific DPs in this position.
Another option would be to consider this difference in syntactic terms, as
argued by White et al. (2012) for negative definite existentials in Turkish and
­Russian, and assume that, somehow, the theme in existential constructions in
­Catalan can escape the domain of existential closure, and then it can be inter-
preted as indefinite. This could be due either because Catalan subjects can move
more freely in Catalan than in Spanish, or because the existential closure domain
is smaller for languages such as Catalan or Italian which do not to obey the DE. If
this were the case, then it would be the syntactic property that is transferred and
not the semantics or the pragmatic interpretation. Thus, we could conclude that
the crosslinguistic variation between Catalan and ­Spanish in this respect lies on a
different mapping between semantic and syntactic properties.
 The locative paradigm in the L2 Spanish of Catalan native speakers 

8.  C
 onclusions

To recapitulate, this study has shown some divergent acquisition of L2 Spanish by

L1 Catalan speakers. These differences are more evident when we employ experi-
mental methodology to explore them. The semi-controlled production data from
the Oral Production Task displayed general differences in the expression of loca-
tive and existentials such as the underuse of estar in comparison to the Spanish
native speaker. However, it is only in the Acceptability Judgment Task that signifi-
cant differences obtained in aspects such as the definiteness effect or the eventive-
ness effect, indicating that the implicit grammar of these bilinguals presents some
type of functional convergence in the consideration of ser as a possible locative
verb with inanimate individuals, and the use of definite DPs in existential contexts.

Acknowledgements

I would like to thank The University of Western Ontario for a SSHRC Internal Research Award
to Silvia Perpiñán, and a New Research and Scholarly Initiative Award (ADF) to Joyce Bruhn de
Garavito and Silvia Perpiñán.

References

Blas Arroyo, J.L. (2004). El español actual en las comunidades del ámbito lingüístico catalán.
In R. Cano Aguilar (Ed.), Historia de la lengua Española (pp. 1065–1086). Mexico, DF:
Editorial Ariel.
Blas Arroyo, J.L. (2011). Spanish in contact with Catalan. In M. Díaz-Campos (Ed.), The Hand-
book of Hispanic Sociolinguistics (pp. 374–394). Malden, MA: Wiley-Blackwell.
DOI: 10.1002/9781444393446.ch18
Boix-Fuster, E., & Sanz, C. (2008). Language and identity in Catalonia. In M. Niño Murcia &
J. Rothman (Eds.), Bilingualism and Identity: Spanish at the Crossroads with other ­Languages
(pp. 87–106). Amsterdam: John Benjamins. DOI: 10.1075/sibil.37.07boi
Brucart, J.M. (2010). La alternancia ser/estar y las construcciones atributivas de localización. In
A. Avellana (Ed.), Actas del V encuentro de gramática generativa (pp. 115–152). Neuquén:
Editorial Universitaria del Comahue.
Brucart, J.M. (2012). Copular alternation in Spanish and Catalan attributive sentences. Revista
de Estudos Linguísticos da Universidade do Porto, 7, 9–43.
Bruhn de Garavito, J. & Valenzuela, E. (2006). The status of ser and estar in late and early bilin-
gual L2 Spanish. In C. A. Klee & T. L. Face (Eds.), Selected Proceedings of the 7th Conference
on the Acquisitino of Spanish and Portuguese as First and Second Languages (pp. 100–109).
Sommerville, MA: Cascadilla Proceedings Project.
Freeze, R. (1992). Existentials and other locatives. Language, 68(3), 553–595.
DOI: 10.2307/415794
 Silvia Perpiñán

Gallego, A.J., & Uriagereka, J. (2011). From ser to estar: A study of the Spanish copulas. Ms. Uni-
versitat Autònoma de Barcelona & University of Maryland.
Geeslin, K.L. (2001). Linguistic simplification: Past, present and future links to second language
acquisition. In X. Bonch-Bruevich, W.J. Crawford, J. Hellermann, C. Higgins, & H. Nguyen
(Eds.), Selected Proceedings of the 2000 Second Language Research Forum (pp. 271–291).
Somerville, MA: Cascadilla Press.
Geeslin, K.L. (2002). The acquisition of Spanish copula choice and its relationship to language
change. Studies in Second Language Acquisition, 24(3), 419–450.
DOI: 10.1017/s0272263102003030
Geeslin, K.L., & Guijarro-Fuentes, P. (2008). Variation in contemporary Spanish: Linguistic pre-
dictors of estar in four cases of language contact. Bilingualism: Language and Cognition,
11(3), 365–380. DOI: 10.1017/s1366728908003593
Gràcia, L., Crous, B., & Garganta, L. (2008). Typological differences and second language acqui-
sition 2: Attributive, locative, existential, and possessive constructions. Revista Española de
Lingüística, 38(2), 47–74.
Gutiérrez, M.J. (1992). The extension of estar: A linguistic change in progress in the Spanish of
Morelia, Mexico. Hispanic Linguistics, 5(1–2), 109–141.
Gutiérrez, M.J. (2003). Simplification and innovation in US Spanish. Multilingua, 22, 169–184.
DOI: 10.1515/mult.2003.009
Haznedar, B., & Schwartz, B.D. (1997). Are there optional infinitives in child L2 acquisition?
Proceedings of the Annual Boston University Conference on Language Development, 21(1),
257–268.
Holmback, H. (1984). An interpretive solution to the definiteness effect problem. Linguistic
Analysis, 13(3), 195–215.
Ionin, T. (2006). This is definitely specific: Specificity and definiteness in article systems. Natural
Language Semantics, 14(2), 175–234. DOI: 10.1007/s11050-005-5255-9
Ionin, T., Ko, H., & Wexler, K. (2004). Article semantics in L2 acquisition: The role of specificity.
Language Acquisition, 12(1), 3–69. DOI: 10.1207/s15327817la1201_2
Kalt, S.E. (2003). Second Language Acquisition of Spanish Morpho-Syntax by Quechua-­
Speaking Children. Unpublished Ph.D. dissertation, University of Southern California,
Los Angeles, CA.
Kayne, R. S. (1994). The Antisymmetry of Syntax. Cambridge, Mass: MIT Press.
Lardiere, D. (1998). Dissociating syntax from morphology in a divergent L2 end-state grammar.
Second Language Research, 14(4), 359–375. DOI: 10.1191/026765898672500216
Lardiere, D. (2004). Knowledge of definiteness despite variable article omission in second lan-
guage acquisition. Proceedings of the Annual Boston University Conference on Language
Development, 28(1), 328–339.
Lardiere, D. (2008). Feature assembly in second language acquisition. In J. M. Liceras,
H. ­Goodluck, & H. Zobl (Eds.), The Role of Formal Features in Second Language Acquisition
(pp. 106–140). Mahwah, NJ: Lawrence Erlbaum Associates.
Lardiere, D. (2009). Some thoughts on the contrastive analysis of features in second language
acquisition. Second Language Research, 25(2), 173–227.
DOI: 10.1177/0267658308100283
Longa, V.M., Lorenzo, G., & Rigau, G. (1998). Subject clitics and clitic recycling: Locative sen-
tences in some Iberian Romance languages. Journal of Linguistics, 34(1), 125–164.
DOI: 10.1017/s0022226797006853
Lumsden, M. (1988). Existential Sentences: Their Structure and Meaning. London: Croom Helm.
DOI: 10.1017/s0022226700012329
 The locative paradigm in the L2 Spanish of Catalan native speakers 

Lyons, J. (1967). A note on possessive, existential, and locative sentences. Foundations of Lan-
guage, 3, 390–396.
Lyons, C. (1999). Definiteness. Cambridge: CUP. DOI: 10.1017/cbo9780511605789
Milsark, G.L. (1977). Toward an explanation of certain peculiarities of the existential construc-
tion in English. Linguistic Analysis, 3(1), 1–29.
Ortiz-López, L. (2000). La extensión de estar en contextos de ser en el español de Puerto Rico:
¿Evolución interna o contacto de lenguas? Boletín de la Academia Puertorriqueña de la
Lengua Española, 98–118.
Perpiñán, S. (2014). Locative and existentials in L2 Spanish: The acquisition of semantic con-
trasts among ser, estar and haber. Second Language Research, 30(4), 485–513.
DOI: 10.1177/0267658313518215
Prévost, P., & White, L. (2000). Missing surface inflection or impairment in second language
acquisition? Evidence from tense and agreement. Second Language Research, 16(2),
103–133. DOI: 10.1191/026765800677556046
Rigau, G. (1994). Catalan presentational sentences and the properties of Agr nodes. In G. Cinque,
J. Koster, J.-Y. Pollock, L. Rizzi, & R. Zanuttini (Eds.), Paths towards Universal Gram-
mar: Studies in Honor of Richard S. Kayne (pp. 343–359). Washington, DC: G ­ eorgetown
­University Press.
Rigau, G. (1997). Locative sentences and related constructions in Catalan: ésser/haver alter-
nation. In A. Mendikoetxea & M. Uribe-Etxebarria (Eds.), Theoretical Issues at the
Morphology-­syntax Interface (pp. 395–421). Bilbao & San Sebastián: Universidad del País
Vasco & Diputación Foral de Gipuzkoa.
Ryan, J., & Lafford, B. (1992). Acquisition of lexical meaning in a study abroad environment:
Ser and estar and the Granada experience. Hispania, 75(3), 714–722. DOI: 10.2307/344152
Safir, K.J. (1985). Syntactic chains. Cambridge: CUP.
DOI: 10.1017/s0022226700011932
Sánchez, L. (2003). Quechua-Spanish Bilingualism: Interference and Convergence in Functional
Categories. Amsterdam: John Benjamins. DOI: 10.1075/lald.35
Sánchez, L. (2004). Functional convergence in the tense, evidentiality and aspectual systems of
Quechua Spanish bilinguals. Bilingualism: Language and Cognition, 7(2), 147–162.
DOI: 10.1017/s136672890400149x
Sánchez, L. (2015). Crosslinguistic influences in the mapping of functional features in
­Quechua-Spanish Bilingualism. In T. Judy & S. Perpiñán (Eds.), The Acquisition of Spanish
in Understudied Language Pairings (pp. 21–47). Amsterdam: John Benjamins.
Sanz, C., & González, M.J. (1995). Ser and estar in Tortosi Catalan: Language contact, language
variation, and language change. Sintagma, 7, 5–25.
Schmitt, C. (1992). Ser and estar: A matter of aspect. In K. Broderick (Ed.), Proceedings of NELS
22 (pp. 411–426). Amherst, MA: GLSA.
Schmitt, C., & Miller, K. (2007). Making discourse-dependent decisions: The case of the copulas
“ser” and “estar” in Spanish. Lingua, 117(11), 1907–1929.
DOI: 10.1016/j.lingua.2006.11.007
Sera, M. (1992). To be or to be: Use and acquisition of the Spanish Copulas. Journal of Memory
and Language, 31(3), 408–427. DOI: 10.1016/0749-596x%2892%2990021-o
Sera, M.D., Bales, D.W., & del Castillo, J. (1997). Ser helps Spanish speakers identify “real” prop-
erties. Child Development, 68(5), 820–831. DOI: 10.2307/1132035
Serrano Vázquez, M. del C. (1996). Interferencias léxicas y semánticas en una situación de con-
tacto entre dos lenguas, catalán y castellano. Diálogos Hispánicos, 18, 375–394.
 Silvia Perpiñán

Silva-Corvalán, C. (1986). Bilingualism and language change: The extension of estar in Los
Angeles Spanish. Language, 62(3), 587–608. DOI: 10.1353/lan.1986.0023
Silva-Corvalán, C. (1994). The gradual loss of mood distinctions in Los Angeles Spanish. Lan-
guage Variation and Change, 6(3), 255–272. DOI: 10.1017/s095439450000168x
Sinner, C. (2004). El castellano de Cataluña: Estudio empírico de aspectos léxicos, morfosintácti-
cos, pragmáticos y metalingüísticos. Tübingen: Max Niemeyer.
DOI: 10.1515/9783110933871
Sinner, C., & Wesch, A. (Eds.). (2008). El castellano en las tierras de habla catalana. Madrid &
Frankfurt: Iberoamericana &Vervuert. DOI: 10.1558/sols.v5i1.177
Solà, J. (1987). Qüestions controvertides de sintaxi catalana (1st ed.). Barcelona: Edicions 62.
Solà, J. (1994). Sintaxi normativa: Estat de la qüestió (1st. ed.). Barcelona: Editorial Empúries.
Thomason, S.G., & Kaufman, T. (1988). Language Contact, Creolization, and Genetic Linguistics.
Berkeley CA: University of California Press. DOI: 10.1017/s0022226700014791
Vann, R. (2002). El castellà catalanitzat a Barcelona: Perspectives lingüístiques i culturals. Cata-
lan Review, 15(1), 117–131.
VanPatten, B. (1985). The acquisition of ser and estar by adult learners of Spanish: a preliminary
investigation of transitional stages of competence. Hispania, 68, 399–406.
DOI: 10.2307/342218
VanPatten, B. (1987). Classroom learners’ acquisition of ser and estar: accounting for develop-
mental patterns. In B. VanPatten, T.R. Dvorak, & J. Lee F. (Eds.), Foreign Language Teach-
ing: A Research Perspective (pp. 61–75). Cambridge MA: Newbury House.
Wesch, A. (1997). The Spanish spoken in Barcelona and Catalan influence. An outline of a
research program. Verba, 24, 287–312.
White, L. (2003). Fossilization in steady state L2 grammars: Persistent problems with inflec-
tional morphology. Bilingualism: Language and Cognition, 6(2), 129–141.
DOI: 10.1017/s1366728903001081
White, L. (2008). Definiteness effects in the L2 English of Mandarin and Turkish speakers.
Proceedings of the Annual Boston University Conference on Language Development, 32(2),
550–461.
White, L., Belikova, A., Hagstrom, P., Kupisch, T., & Özçelik, Ö. (2009). Restrictions on Definite-
ness in L2 English. Proceedings of the Annual Boston University Conference on Language
Development, 33(2), 622–633.
White, L., Belikova, A., Hagstrom, P., Kupisch, T., & Özçelik, Ö. (2012). Restrictions on definite-
ness in second language acquisition: Affirmative and negative existentials in the L2 English
of Turkish and Russian speakers. Linguistic Approaches to Bilingualism, 2(1), 54–89.
DOI: 10.1075/lab.2.1.03whi
Zagona, K. (2012). Ser and estar: Phrase structure and aspect. In C. Nishida & C. Russi (Eds.),
Building a Bridge between Linguistic Communities of the Old and the New World. ­Current
Research in Tense, Aspect, Mood and Modality. (Vol. XIV, pp. 303–327). Amsterdam:
Rodopi.
part ii

Spanish as an L2 in a non-bilingual society

 The acquisition of Spanish in a bilingual
and a trilingual L1 setting
Combining Spanish with German,
French and Catalan

Laia Arnaus Gil & Natascha Müller

Bergische Universität Wuppertal

This study focuses on the early acquisition of Spanish predicative adjectives

accompanied with copula selection and the position of attributive adjectives
in longitudinal data from simultaneous Spanish-German and Spanish-French
bilinguals and from Spanish-Catalan-German trilingual children. As for copula
selection, our results show cross-linguistic influence in the Spanish-German/
French bilinguals, resulting in a delay of the acquisition of Spanish. As to
adjective order, influence cannot be conclusively studied. For both linguistic
properties, bilingual children overgeneralize the less complex analysis, mainly
if it competes with the more complex analysis in one language. A subtle
positive effect in the Spanish of the trilinguals for both grammatical domains
investigated suggests the high impact of syntactic properties and the less
significance of the number of languages.

1.  Introduction

The theoretical research on bilingual first language (2L1) acquisition has clearly
shown that bilingual children are able to separate their languages early on (see
Müller, Kupisch, Schmitz & Cantone 2011 for an overview). However, in recent
decades, it has been argued that the two languages influence one another. More
importantly, this influence is claimed to be independent of language domi-
nance, yet dependent on linguistic aspects of the grammatical phenomenon
in q ­ uestion (­Schmeißer, Hager, Arnaus Gil, Jansen, Geveler, Eichler, Patuto &
­Müller, to appear)). If this is true, then the language combination should ­matter
if c­ ross-linguistic influence is to be studied (Müller & Patuto 2009; Patuto 2013).
In this chapter, the empirical examination of Spanish as one of two/three first

doi 10.1075/ihll.3.06gil
© 2015 John Benjamins Publishing Company
 Laia Arnaus Gil & Natascha Müller

l­anguages contributes to the theory of bilingual L1 acquisition by providing

important information regarding the selection of copula verbs and the position
of attributive adjectives in early multilingual children. The multilingual children
acquire Spanish in different language combinations so they are presented with
syntactic structures that differ according to the complexity required in the pro-
cess of syntactic derivation. If confronted with derivations with differences in
complexity, the young monolingual or bilingual child will prefer the less complex
analysis (­Müller & Hulk 2001). It is in this respect that language acquisitionists
could contribute to linguistic theory for a better understanding of the involved
grammatical phenomena in the different languages.
Studies on early bilingualism and trilingualism with Spanish being one of
the first languages have focused mainly on the overall acquisition of these L1s
or on child input (Castro & Gavruseva 2003; Silva-Corvalán 2003 for English-
Spanish bilinguals; Austin 2001; Barreña 1997; Ezeizabarrena 1996, 1997a, 1997b,
2002; Idiazábal 1996; Larrañaga 2000 for Basque-Spanish; Lleó 2001a, 2001b for
G­ erman-Spanish; Vila & Cortés 1991 for Catalan-Spanish; Barnes 2011 for Basque-
Spanish-English trilinguals; Montanari 2009 for Spanish-English-­Tagalog). Little
attention has been paid to some linguistic phenomena, such as the ones under
study in this paper. The focus of the present paper lies on two linguistic phenom-
ena which relate intimately the clause and the nominal level: the use of adjectives
in verbal and nominal expressions. Spanish adjectives within nominal expressions
can generally appear before or after the noun, or both. At the clause level, S­ panish
adjectives allow either only one of two copula verbs or both. Catalan shares most
of these properties with Spanish (for different uses, see ­Perpiñán, this volume).
German and French, on the other hand, differ from Spanish by providing only one
phonologically realized copula form and by restricting the positional possibilities
for adjectives within the noun phrase,1 with German only exhibiting prenominal
adjectives. We will see examples of these structures in the following section. In
order to study copula selection and adjective placement, we focus on some multi-
lingual language combinations, all sharing Spanish as one of the L1s. Specifically,
we examine early bilinguals of Spanish-German and Spanish-French as well as
early trilinguals of Spanish-Catalan-German.

.  French actually has both pre- and postnominal adjectives in the DP. Yet, French adjectives
tend to make use exclusively of one position. For example, color adjectives are always post-
nominal and short adjectives such as petit only occur prenominally (Bouchard 2002).
 The acquisition of Spanish in a bilingual and a trilingual L1 setting 

2.  Theoretical framework

2.1  The copula verb(s)

2.1.1  The dual Spanish and Catalan system
In this section, we briefly describe the distribution of the copula verbs in Spanish
(for a detailed description, see Arche 2006; Arnaus Gil 2013; Camacho 2012; Lema
1995; Zagona 2009; among many others). Spanish has two copula verbs, SER and
ESTAR, which sometimes exclusively select one predicate type, whereas in  other
cases, both compete for the same predicative context. Examples (1) and (2) illus-
trate this distinction. In this case, it has been proposed in the literature that by
selecting the copula verb SER, a permanent, inherent property is associated with
the predicate (2a), whereas by using ESTAR, a temporary, accidental property is
assigned to the complement (2b).
(1) a. Aquella variable matemática es /*está constante.
That variable mathematic is-ser /*is-estar constant
b. Cuando juega el Barça, la ciudad *es/está desierta.
When plays the Barça, the city *is-ser/is-estar deserted
(2) a. Silvia es alegre, da igual que
Silvia is-ser happy, it doesn’t matter whether
tenga un buen o mal día.
(she) has a good or bad day
b. A Tomás le ha llegado el regalo
To Tomás Cl-Dat has arrived the present,
por eso está alegre.
that is why (he) is-estar happy

As can be seen in Example (1a), SER is the only copula allowed in combination
with adjectives that can only express a permanent property. By contrast, ESTAR
has to be selected when the predicate expresses a temporal property (1b). The
examples in (2) illustrate the different meaning of the adjective when accompa-
nied by one of the copula verbs: when SER is selected, an inherent property is
assigned to the adjective (2a); when ESTAR is chosen, a temporal reading of the
same adjective is obtained (2b). The difference between permanent and temporal
properties was first captured by Carlson (1977) with the dichotomy Individual-
Level and Stage-Level Predicates (ILP and SLP, respectively): whereas the former
expresses a property of the individual corresponding to a general norm, i.e. a per-
manent property, the latter restricts the quality in a specific temporal limit (or
stage) of the same individual.
 Laia Arnaus Gil & Natascha Müller

The examples in (1) and (2) reveal that there are two groups of adjectives in
Spanish. In lexical adjectives, the expression of a permanent or a temporal property
is encoded in the lexical entry (cf. the examples in 1). The syntactic derivation will
be chosen correspondingly. A second group of adjectives does not have any lexi-
cal specification of a permanent/temporal property (cf. the examples in (2) above
and (3) below). Both the temporal and the permanent property will be expressed
via a particular syntactic derivation. In order to derive the different constructions,
Arnaus Gil (2013) proposes an analysis for Spanish SER in which T selects a vP as
in structure (3a).
(3) a. Juan es guapoILP
Juan is-ser nice
b. La profesora está guapaSLP
The teacher looks-estar nice

TP TP

DP T’

Juank DP T’
T vP
La profesora
tk v’
T Temp. AP
v Perm. AP está guapa
es guapo

Evidence for this assumption comes from Spanish passive constructions, where
SER is used (La canción ha sido escuchada por la banda ‘The song has been-SER
heard by the band’), i.e. the form of the copula which activates a vP according
to Arnaus Gil (2013). Further evidence for the existence of a vP layer for SER
comes from causatives and gerundive constructions. Spanish SER can be used in
causatives as in El profesor hace ser más valiente al alumno (‘the teacher makes be-
SER more courageous to the student’), thus activating a vP layer. SER in v must
move further to T in order to check tense and agreement features within the TP.
In gerundive constructions, SER can also be used in its gerundive form, accom-
panied by the auxiliary ESTAR as in El vendedor está siendo muy pesado (‘The
salesperson is-ESTAR being-SER very boring’). In contrast to SER, ESTAR can
neither be passivizer (*La canción ha estado escuchada por la banda ‘*The song
 The acquisition of Spanish in a bilingual and a trilingual L1 setting 

has been-ESTAR heard by the band’)2 nor does it play a role in causativization
(*Su madre hace estar descalza a María ‘*Her mother makes be-ESTAR barefoot
to Maria’)3 nor can it be used in its gerundive form (*Juan está estando afónico
‘*Juan is-ESTAR being voiceless’). Arnaus Gil (2013) deduces from these obser-
vations that ESTAR is directly inserted in T, thus a vP layer can remain inactive
(Example (3b)). Whether T selects an AP or a small clause cannot be discussed
here (cf. Arnaus Gil 2013).
In contrast to this theoretical proposal, there are other syntactic approaches
for the analysis of Spanish copula verbs SER and ESTAR that assume a more elabo-
rated syntactic derivation for ESTAR rather than for SER (Brucart 2012; Camacho
2012; Gallego & Uriagereka 2011; Zagona 2009). Following Schmitt (1992) and
Fernández Leborans (1999), it is postulated that ESTAR has aspectual properties
which are captured by means of a temporal internal structure, something that SER
lacks. This implies, for Camacho (2012), that the aspectuality of ESTAR takes the
form of an inchoative feature, since ESTAR expresses the beginning of the time in
which the perfective property is interpreted. For Zagona (2009), ESTAR is gener-
ated in the vP layer, takes a prepositional category P with an uninterpretable feature
[uP] which, in turn, has to be checked with its predicate, also containing a feature
that is evaluated with [uP]. In this respect, our proposal goes against the existence
of the vP layer for adjectival participles and its absence for non-stative events, i.e.
passive constructions, as Zagona (2009, p. 2) shows: los exámenes están corregidos
mañana vs. los exámenes serán corregidos mañana, respectively. She postulates that
the P-category with the uninterpretable [uP] feature is needed for ESTAR to be
located in the vP layer and in order to be checked with the adjectival participle.
By contrast, passive constructions are considered non-stative events and therefore

.  There are adjectival passives which are indeed built with ESTAR as in El pollo estaba
bien cocinado ‘the chicken was-ESTAR good cooked’ (Bruhn de Garavito 2009, p. 1) and are
claimed to be built in the lexicon (Wasow 1977, as cited in Bruhn de Garavito 2009). However,
De Miguel (1999, 2004) and Marín (2004) suggest that there is actually a case of a passive con-
struction which is built with ESTAR and needs the preposition por to incorporate the activity
of the agent as in Las alergias están provocadas por muchas causas ‘(the) allergies are-ESTAR
caused by several causes’ (De Miguel 2004, p. 24).
.  Sentences such as the one above and El maquillaje hace estar guapa a María ‘The make-up
makes be-ESTAR pretty to Maria’ as well as La madre hace estar quietos a los niños ‘The mother
makes be-ESTAR quiet to the children’ seem to elicit variability in the responses of native
speakers’ grammaticality judgments. Therefore, we could speak here of dialectal differences in
the use or avoidance of causative structures with ESTAR.
 Laia Arnaus Gil & Natascha Müller

they convey a path. This implies that neither the copula nor the verbal passive have
any uninterpretable feature to be checked. To sum up, the need for a vP layer is
due to purely grammatical reasons, forced by the selection of a P-category with an
uninterpretable feature [uP] for ESTAR. According to Arnaus Gil (2013), tempo-
ral anchoring as the major distinguishing exponent between both copulas could
well be captured by means of the (in)activity of the vP layer, as the examples in (3)
show. In this vein, we propose that the activation of the vP layer is necessary for
SER for passives and causatives as well as its gerundive form siendo ‘being-SER’
for progressive constructions. ESTAR leaves the functional vP layer inactive and
by this syntactic property does not permit the three syntactic constructions char-
acteristic for SER.4 In other words, the syntactic analysis of SER is more complex
than the one for ESTAR since it involves more movement steps.
Let’s turn now briefly to Catalan. The next examples are the Catalan counter-
parts of (1) and (2).
(4) a. Aquella variable matemàtica és / *està constant.
That variable mathematic is-ésser / *is- estar constant
b. Quan juga el Barça, la ciutat
When plays the Barça, the city
és / està deserta.
is-ésser / is- estar deserted
(5) a. La Sílvia és alegre, és igual
Silvia is-ésser happy, (it) doesn’t matter
si té un bon o mal dia.
whether (she) has a good or bad day
b. A en Tomàs li ha arribat el regal,
To Tomás Cl-Dat has arrived the present,
per això està alegre.
that is why (he) is-estar happy

Generally speaking, the copulas ÉSSER and ESTAR are allowed in the same con-
texts as in Spanish. However, there are also some differences: The Catalan adjec-
tive deserta ‘deserted’ allows both ÉSSER and ESTAR (4b), whereas Spanish
only selects ESTAR (1b). Several Catalan adjectives seem to behave like deserta
‘deserted’ and therefore it can be claimed that property assignment in Catalan

.  One reviewer poses the question as to whether assigning an IL/SL quality after insertion of
the copula in v (SER) or T (ESTAR) matters. In fact, property assignment takes place in a head-
complement relation, both for SER and for ESTAR, as transitive verbs assign accusative case
to their complements; therefore, at first sight, no difference should be observed in this respect.
 The acquisition of Spanish in a bilingual and a trilingual L1 setting 

takes place ­syntactically far more frequently than in Spanish. Along the same lines,
Catalan locative expressions can also pattern with adjectives like deserta, as Ramos
Alfarín (2008) observes:
(6) a. El gos és a la seva caseta.
The dog is-ésser at the his house
b. El gos està a la seva caseta.
The dog is-estar at the his house

The Catalan locative examples in (6) show that both copulas are allowed in certain
contexts and/or dialects, whereas in Spanish only ESTAR is permitted in this con-
text (for more on Catalan locatives, see Perpiñán, this volume). As in (5) and (6),
the use of ÉSSER or ESTAR determines the permanent or temporary property
of the predicate. Arnaus Gil (2013) concludes that Catalan predicates, which are
allowed with both copulas verbs, necessitate a syntactic derivation which makes use
of the vP layer for both temporal and permanent properties. She observes for the
Catalan system that predicates that can be used with ÉSSER can also be used with
ESTAR but not vice versa (hence, the higher frequency of ÉSSER in comparison
with ESTAR, cf. Ramos Alfarín 2008). Moreover, Arnaus Gil (2013, p. 47) observes
that Catalan participles can be accompanied by both ÉSSER and ESTAR and inter-
preted as passive structures as in Aquesta figura és/està feta a mà per en Joan ‘This
figure is-ÉSSER/ESTAR done handmade by the Joan’, thus activating the vP layer
also for the latter. Therefore, she suggests a syntactic analysis in which ESTAR is
generated lower in the tree than ÉSSER, as the following examples in (7) show:
(7) a. En Joan està guapoSLP.
The Joan looks-estar nice
b. En Joan és guapoILP.
The Joan is-ésser handsome

TP
TP

DP T’
DP T’

En Joank T vP En Joank T vP
és
tk v’ tk v’

v SLP AP ILP v AP
guapo
està guapo

 Laia Arnaus Gil & Natascha Müller

It is thus the syntactic derivation of ESTAR which is more complex than ÉSSER in
Catalan, since the former involves more movement steps.
In what follows, the copula systems of German and French, SEIN and ÊTRE
respectively, will be presented and compared to the Spanish and Catalan copula
verbs SER/ÉSSER and ESTAR.

2.1.2  The unitary Germanic/French system

German (8) and French (9) only have one copula verb, SEIN and ÊTRE respec-
tively. The Spanish examples presented in (1) and (2) have translations that use
only one copula verb:
(8) a. Diese mathematische Variable ist konstant.
That mathematic variable is-sein constant
b. Wenn Barça spielt, ist die Stadt leer.
When Barça plays, is-sein the city deserted
c. Silvia ist glücklich, es ist egal
Silvia is-sein happy, it doesn’t matter
ob sie einen guten oder schlechten Tag hat.
whether she a good or bad day has
d. Tomás hat das Geschenk erhalten,
Tomás has the present received,
deswegen ist er so glücklich.
that is why is-sein he so happy
(9) a. Cette variable mathématique est constante.
That variable mathematic is-être constant
b. Quand Barça joue, la ville est déserte.
When Barça plays, the city is-être deserted
c. Silvia est contente, n’importe
Silvia is-être happy, it doesn’t matter
qu’ elle a passé une bonne ou mauvaise journée.
(whether) she has had a good or bad day
d. Le cadeau pour Tomás est arrivé, c’ est la
The present for Tomás has arrived, that is the
raison pour laquelle il est content.
reason why he is-être happy

As the German and French examples show, both SEIN and ÊTRE can be employed
in all predicative contexts in which two copula verbs compete in Spanish (2) and
Catalan ((4b)–(5)). The permanent or temporal reading of the German and French
adjective glücklich/content(e), respectively, is conveyed, although there is only one
phonetically realized copula in these languages. The approach we follow is that
 The acquisition of Spanish in a bilingual and a trilingual L1 setting 

unitary copula systems such as German or French will assign the corresponding
permanent/temporal property to predicates by means of a different syntactic posi-
tion in the derivation, as the following structures in (10) show, taken from Arnaus
Gil (2013, p. 108).
(10) a. Das Auto ist kaputtSLP.
The car is-sein broken
b. Hans ist intelligentILP.
Hans is-sein intelligent
TP TP

DP T’ DP T’

Das Autok Hans

T vP T Perm. AP
ist intelligent
tk v’

AP Temp. v
kaputt ist

c. La voiture est casséeSLP.

The car is-être broken
d. Hans est intelligentILP.
Hans is-être intelligent
TP TP

DP T’ DP T’

La voiturek Hansk
T vP T vP
est

tk v’ tk v’

v Temp. AP v AP
Perm. intelligent
est cassée

In German and French, SEIN and ÊTRE figure in passive constructions: Die Haare
sind gefärbt/les cheveux sont coloriés ‘the hair is coloured’. These constructions
are interpreted as stative passives, i.e. they express a change of state – a temporal
 Laia Arnaus Gil & Natascha Müller

S­ tage-Level-property. Arnaus Gil (2013) takes this observation as evidence in favor

of the assumption that a vP layer is necessary in order to derive the copulas SEIN
and ÊTRE for temporal properties. In this respect, French and German resemble
the dual copula system of Catalan in that they provide a syntactic derivation which
contains a vP layer for predicates with a temporal reading. Yet, German, in con-
trast to Catalan, generally forms the passive with a particular verb werden which
differs from the copula SEIN: der Apfel wird gegessen ‘the apple is-WERDEN eaten’
and not *der Apfel ist gegessen ‘*the apple is-SEIN eaten’. If the copula SEIN always
instantiated a vP, we would predict that passives are generally formed with SEIN,
but they are not. We can deduce from this observation that the copula SEIN does
not always instantiate a vP in German and therefore it is sometimes inserted in T
(as Spanish ESTAR).
In sum, we conclude that both German and French have copula systems
which are characterized by having one phonological realization of the copula verb
(i.e. SEIN/ÊTRE), with both the permanent and temporary reading being con-
veyed for the same adjectival predicate. As such, they will henceforth be treated
together. Differently, Spanish and Catalan share the characteristic of having two
copula verbs, which, when competing, assign the corresponding inherent or tem-
poral quality to the predicate. However, while Catalan and French5 are the only
languages under study that always make use of derivations that contain a vP layer,
Spanish and German need structures without this layer, i.e. the copula is directly
inserted in T. Despite this first resemblance, Spanish and German also exhibit
some differences as to the type of predicate’s reading and its correspondence to
syntactic structures: While a vP layer is necessary with IL-properties in Spanish,
German makes use of this functional layer with SL-properties.

2.1.3  Previous studies on the early acquisition of the Spanish SER and ESTAR
The first studies on Spanish copula acquisition date from the 1970s and 1980s.
In her early studies, Hernández Pina (1979, 1984) examines the acquisition of
Spanish by a monolingual child up to 3;0. The first copula productions found in
the data are with ESTAR to express a state of the subject referent as in nena tá
cota da (la nena está acostada ‘the girl is sleeping’). During this very first period,
the studies from Hernández Pina (1979, 1984) and Corpas (2007) also observe
that copula omissions are frequent, although it seems that SER is more prone to
omission. Since adjectives are those predicative contexts which tend to capture
most of the omissions, Sera, Bales and del Castillo Pintado (1997) carried out an

.  Remember that French uses ÊTRE for both passive constructions and state passives and
that is why, under this proposal, a vP layer is always needed.
 The acquisition of Spanish in a bilingual and a trilingual L1 setting 

experiment with L1 Spanish and L1 English children that found evidence for pos-
tulating that the dual copula system benefited the children by helping them dif-
ferentiate between temporal and permanent properties. Comparing the early L1
monolinguals with their adult peers, the former performed less target-like than
the latter, specifically by using ESTAR with eventive locatives (la fiesta es/*está en
el jardín ‘the party is-SER in the garden’) and by producing the wrong copula with
adjectival predicates, concretely with those that can select both SER and ESTAR,
with ESTAR being the copula that tends to be more frequently overgeneralized
(cf. Crain & Thornton 1998; Holtheuer 2009; Schmitt & Miller 2007; Schmitt,
­Holtheuer & Miller 2004; Sera 1992 for target-deviant uses of Spanish copula
verbs with adjectival predicates). These adjectives can be claimed to be more dif-
ficult to acquire since doing so requires knowing the implications of selecting the
appropriate copula verb.
In summary the aforementioned studies have shown that the early acquisition
of Spanish copulas is, in general, far from problematic. L1 Spanish children are
capable of assigning the corresponding permanent or temporal property to the
subject DP by means of selecting SER or ESTAR respectively, and perform more
accurately than their L1 English peers in this respect. Nevertheless, adjectival
predicates that allow both copulas are difficult and represent most of the children’s
target-deviant productions during their first three years, yet at a very low rate.
Much has also been said about Spanish copula acquisition in early bilingual-
ism, although mainly for the language pair Spanish-English (Fernández Fuertes &
Liceras 2010; Krasinski 2005; Liceras, Fernández Fuertes, de la Fuente & T ­ ercedor
Sánchez 2010; Sera et al. 1997; Silva-Corvalán & Montanari 2008; among oth-
ers). Within this body of work, some studies indicate that the copula asymme-
try influences the acquisition of SER and ESTAR negatively (Krasinski 2005;
­Silva-Corvalán & Montanari 2008), whereas other investigations suggest that it
is the lexical transparency of SER for permanent and ESTAR for temporary prop-
erties that favors copula acquisition in English, measured here by means of BE
omissions (Fernández Fuertes & Liceras 2008; Fernández Fuertes & Liceras 2010;
Liceras et al. 2010).
For example, Silva-Corvalán and Montanari (2008) report late acquisition of
ESTAR and the use of ESTAR for existential constructions and in SER-contexts,
such as in (11), which indicate there is cross-linguistic influence from English into
Spanish:
(11) *El gato estaba malo (= tiene mal carácter) Nicolás 2;7,16
*The cat is-estar bad (= has a bad character)

The study by Arnaus Gil (2013) investigates early bilinguals who acquire Spanish in
combination with either a Germanic or a Romance language, i.e. German-Spanish
 Laia Arnaus Gil & Natascha Müller

and French-Spanish and observes similar results as with the early Spanish-English
bilinguals. Our present study on the early acquisition of Spanish by Spanish-­
German bilinguals will include the data analyzed in Arnaus Gil (2013). For this
reason, we summarize the results of Arnaus Gil (2013) together with the results of
the present study.
In contrast to the results illustrated above, the longitudinal studies of Liceras
and her colleagues (Fernández Fuertes et al. 2008; Fernández Fuertes & Liceras
2010; Liceras et al. 2010) suggest a positive influence from Spanish to English
since they observe that the English copula BE is produced much more frequently
than what has been observed in the literature for early L1 English children
(Becker 2000). In the same vein, Sera et al. (1997) tested older Spanish-English
bilinguals between 3;8 and 7;9 on property assignment and compared them with
monolinguals. The authors observed that bilinguals and monolinguals assigned
temporal properties similarly well, yet permanent properties were much more
difficult to specify in English than in Spanish. This finding could thus suggest that
Spanish has a positive influence on English when it comes to the assignment of
IL-/SL-properties.
In a nutshell, whereas some studies on Spanish-English bilingual children
seem to indicate a delay in the acquisition of SER and ESTAR, other researchers
have found that it is the lexical transparency of SER for permanent and ESTAR
for temporary properties that favors copula acquisition in English, measured here
by means of BE omissions and therefore no negative cross-linguistic influence
towards Spanish has been found. The next section presents the context in which
adjectives can occur, namely in an attributive constellation, i.e. accompanying a
noun in a DP.

2.2  Attributive adjectives

2.2.1  The Spanish adjective system
Scalar or qualitative adjectives such as famosa, rubia (Bouchard 1998; Gumiel-
Molina & Pérez-Jiménez 2012) are the most frequent adjective type in the speech
of young children. They assign a certain property to the objects denoted by the
noun. From a syntactic perspective, they are allowed in both pre- and postnomi-
nal position in Spanish and Catalan (Picallo 2002, p. 1655), although the post-
nominal position is more frequent. Adjectives that allow both positions are called
non-­categorical adjectives. The different adjectival positions found in Spanish and
Catalan produce meaning differences. Bouchard (1998, p. 146) observes that post-
nominal adjectives “assign properties to the set of individuals determined by the
components of an N”, whereas prenominal adjectives “express less concrete prop-
erties that belong naturally to the N”. Demonte (2008) further claims that whereas
 The acquisition of Spanish in a bilingual and a trilingual L1 setting 

the first map properties onto individuals and thereby produce a new extension of
the noun with which they occur, the latter do not modify the extension of N but
predicate the whole group of objects denoted by the N. Crucially, the change of
meaning is restrictive in the case of postnominal adjectives and non-restrictive in
the case of prenominal adjectives.
Still, not all Spanish adjectives allow both positions. As Demonte (2008) and
other linguists have observed, one group of Spanish adjectives is only allowed
either exclusively before or after the noun, such as the classifying or relational
adjectives eléctrico ‘electric’ or nuclear ‘nuclear’ and modal or epistemic adjectives
such as presunto ‘alleged’, as the examples in (12) and (13) show:
(12) a. el cable eléctrico/*el eléctrico cable
the cable electric/*the electric cable
b. la energía nuclear/*la nuclear energía
the energy nuclear/*the nuclear energy
(13) El presunto cómplice/*el cómplice presunto
The alleged accomplice/*the accomplice alleged

Adjectives which occur in only one position are generally called categorical adjec-
tives. As for the syntactic derivation of adjectives, some linguists have proposed
an adjunction analysis to the left or to the right within NP (Abney 1987). Later
analyses have postulated a common base position for all adjectives. The different
surface positions are derived by means of N and/or Adj movement (Alexiadou
2001; Cinque 1995, 2010; Demonte 2008; Kayne 1994). In more concrete terms,
adjectives are merged in a position within the DP which indicates their predica-
tive source, as in (14) (Bernstein 2001; Cinque 2010; Kayne 1994). The derivations
in (14), taken from Arnaus Gil, Eichler, Jansen, Patuto and Müller (2012, p. 245),
illustrate a pre- and a postnominal position of a Spanish scalar adjective, following
Kayne’s (1994) Universal Base Hypothesis:
(14)
[DP [D la] [CP [AP pálida] [C] [IP [NP luz] [I [AP pálida]]]]] ‘the pale light’

 [DP [D la] [FP [F luz] [CP [AP pálida] [C [N luz]] [IP [NP luz] [I [AP pálida] ]]]]] ‘the light pale’

The first movement step of the scalar adjective is similar for a pre- and a post-
nominal position, namely to Spec CP. For adjectives that appear after the noun,
other derivational steps are needed. The next derivational step involves movement
of the NP-head in Spec IP to C and further raising to F, located between DP and
 Laia Arnaus Gil & Natascha Müller

CP. F stands for Functional category and might equal Number. In other words,
prenominal adjectives require AP-movement only, whereas postnominal ones are
the result of more derivational steps (AP- and NP-movement). Thus, the syntactic
derivation of prenominal adjectives is more economical than the one of postnomi-
nal adjectives.
Spanish and Catalan scalar adjectives are mainly postnominal in French
and exclusively prenominal in German, therefore, the following question arises:
how do French and Germanic languages convey a restrictive and non-restrictive
meaning when the language only allows one position? One proposal is provided
by Demonte (2008) who postulates N-movement after PF to convey the restrictive
interpretation.

2.2.2  Previous studies on the acquisition of Spanish adjectives

Adjective placement is not problematic in monolingual Spanish children: post-
nominal and prenominal placement is mastered by the children from early on
(cf. Rizzi, Arnaus Gil, Repetto, Geveler & Müller 2013 and the literature therein).
To our knowledge, the only study on adjective placement in bilingual children
with Spanish is that of Rizzi et al. (2013). However, the study only examines two
German-Spanish bilingual children and therefore it can only be considered as a
rough idea about the acquisition path. Since the present study includes these two
children, we will summarize the results of Rizzi et al.’s (2013) study of the Spanish-
German children together with the results of the present study.

3.  P
 redictions for the acquisition of copula selection and
adjective placement

Research in the field of early child bilingualism has shown that the child’s two lan-
guages may influence each other (cross-linguistic influence). According to Müller
and Hulk (2001), Hulk and Müller (2000) and Müller and Patuto (2009), cross-
linguistic influence is a child-internal process. Hulk and Müller (2000) and Müller
and Hulk (2001) have defined two conditions under which cross-linguistic influ-
ence is likely to occur, one of which is shared by many researchers (cf. Serratrice
& Sorace 2003; Serratrice, Sorace & Paoli 2004; among others). The first condition
claims that “the vulnerable grammatical phenomenon is an interface property, e.g.
a grammatical property located at the interface between syntax and pragmatics.”
(Hulk & Müller 2000, p. 228).
In the previous section, the description of the grammatical phenomena under
study demonstrated that copula verbs and adjective placement are phenomena
located at the syntax-semantics interface. As such, we are faced with a ­grammatical
phenomenon located at an internal interface. According to Sorace and Serratrice
 The acquisition of Spanish in a bilingual and a trilingual L1 setting 

(2009), cross-linguistic influence is not expected here, but, as we will see, this does
not seem to be the case.
The second condition for cross-linguistic influence is that the vulnerable
grammatical phenomenon must be expressed in the two languages in such a way
as to show overlap of the two constructions at the surface. Müller and Patuto
(2009, p. 23) have refined this condition in the following way: “The surface strings
of the two languages A and B are derivable in terms of the syntactic derivation of
one language (which is less complex).”
The direction of the cross-linguistic influence is related to computational
complexity (i.e. a syntactic analysis in language A is more complex if it requires
more movement steps than the derivation in language B; cf. Gavarró 2003; Hulk &
Zuckerman 2000; Zuckerman 2001). Therefore, we hypothesize that, when con-
fronted with a grammatical phenomenon which has a complex and a less complex
derivation available, the bilingual child will use the computationally less complex
derivation available for both languages.
After all, the two linguistic phenomena under investigation, copula selec-
tion and adjective placement, show diverging syntactic structures among the
languages, which makes it possible to observe cross-linguistic influence. In addi-
tion, the grammatical structures in the studied languages exhibit differences in the
computational complexity of their syntactic derivations. This enables us to predict
whether the language that is in fact affected by cross-linguistic influence is also the
language with the more complex syntactic analysis.
With these theoretical preliminaries in mind, our predictions for the two
grammatical domains analysed herein with a focus on Spanish are:

1. For the acquisition of Spanish copula verbs, the child will frequently make use
of the less complex analysis. Regarding production, we expect an overgener-
alization of the Spanish copula verb ESTAR to SER contexts, since ESTAR is
syntactically less complex, following our proposal (cf. (3)). Since the analysis
of the Catalan data enables us to underline the plausibility of the syntactic
analysis of the copula verbs in Spanish, we formulate the inverse predication
for the Catalan data, namely that in Catalan, ÉSSER will take over ESTAR
contexts for exhibiting the less complex derivation (cf. (13)).
2. For the acquisition of adjective placement in Romance, the bilingual child has
a less (A+N) and a more complex syntactic operation (N+A) available. We
predict that s/he will make use of the less complex analysis for those contexts
in which the more complex, postnominal analysis is required.

In what follows, the longitudinal production data analyzed in this study is pre-
sented with the main results on acquisition of copula verbs and adjectives in the
Romance languages in early bilingual and trilingual children.
 Laia Arnaus Gil & Natascha Müller

4.  Methodology

4.1  P
 articipants
In Table 1, the relevant information concerning the multilingual children is pre-
sented. In addition to the data from ten multilingual children,6 two monolingual
Spanish children from the CHILDES database (MacWhinney 2000) were also
included as a control group. Both the bilingual and trilingual children were raised
simultaneously with two/three languages and have received input of at least two
of their L1s at home. The languages we focus on are Spanish (Sp), German (Ger),
French (Fr) and Catalan (Cat). The subjects were recorded in regular intervals
(mostly every two weeks) for half an hour in each language, generally starting
before the age of 2 until the 5th birthday. This allowed us to examine the linguistic
data of approximately 50 recordings per language (cf. Table 1). The children who
lived in Germany were visited at their homes by a research team formed by two
monolingual native speakers of the languages in which the recordings took place.
In the Romance countries, the parents carried out these recordings. The interac-
tions with the children took place in monolingual situations: while one person was
playing with the child, the other was video-recording.

Table 1.  Monolingual and multilingual child data

Language Number of Total number
Child Age
combination recordings of utterances

Arturo Sp- Ger 2;3–5;3 49/49 4,960/6,680

Teresa Sp- Ger 1;5–5;0 56/55 6,857/8,233
Lucas Sp- Ger 1;7–3;6 35/30 5,922/5,360
Erik Sp- Ger 1;6–3;4 43/26 5,945/667
Nora Sp- Ger 1;6–3;8 27/32 1,295/1,874

(Continued)

.  The research projects were funded by the German Science Foundation (DFG) by grants
given to Natascha Müller from 2005–2013. 1. Die Architektur der frühkindlichen bilingualen
Sprachfähigkeit: Italienisch-Deutsch und Französisch-Deutsch in Italien, Deutschland und
Frankreich im Vergleich (Wuppertal 2005–2008); 2. Code-Switching bei bilingual aufwach-
senden Kindern in Deutschland, Italien, Frankreich und Spanien: Italienisch-Deutsch,
Französisch-Deutsch, Spanisch-Deutsch, Italienisch-Französisch, Italienisch-Spanisch,
­Französisch-Spanisch (Wuppertal 2009–2013). For a detailed description of the data, cf.
Cantone, Kupisch, Müller and Schmitz (2008); Müller, Kupisch, Schmitz and Cantone (2011);
Hauser-Grüdl, Arencibia-Guerra, Witzmann, Leray and Müller (2010) and Müller (in prep.).
 The acquisition of Spanish in a bilingual and a trilingual L1 setting 

Table 1.  (Continued)

Language Number of Total number
Child Age
combination recordings of utterances
Syca-Inès Sp-Fr 2;2–3;8 35/8 6,627/2,207
Frank Sp-Ger-Cat 1;11–3;4 36/0/22 6,147/438
Milena Sp-Ger-Cat 1;6–2;3 13/0/14 1,651/1,190
Eric Sp-Ger-Cat 1;9–3;2 12/15/17 213/1,587/2,491
Kilian Sp-Ger-Cat 2;3–3;7 0/25/25 0/343/3,954
Emilio Sp 1;6–4;8 29 4,212
Irene Sp 1;6–3;2 40 10.419

As can be seen from Table 1, the multilingual data includes early bilingual
Spanish-German/French and early trilingual children, the latter being raised with
Spanish, Catalan and German in Barcelona. This valuable data is unique.

4.2  The data on the acquisition of Spanish copula verbs

Figure (1) shows the overall use of the two Spanish copula verbs SER and ESTAR
in all possible predicative contexts (i.e. NP/DP, PP, AP) for monolingual, bilin-
gual Spanish-German as well as Spanish-French and trilingual children separately.
Besides, a distinction between grammatical (target-like) and ungrammatical
(target-deviant) productions is made. A child production that deviates from the
adult’s use of the copulas is represented by the insertion of the copula 1 in a context
in which an adult would expect the use of copula 2 as in (15) and is represented
as a target-deviant commission. Target-deviant omissions7 as in (16) were also
considered as deviating from the adult language.8

.  According to Liceras, Fernández Fuertes, de la Fuente & Tercedor Sánchez (2010), this
phenomenon could be a reflex of semantic/syntactic differences between permanent and tem-
poral properties. Following this idea and in terms of syntactic complexity, copula omissions
could be representatives of having derived the complex syntactic analysis but having failed
to insert the corresponding copula verb in v. In this respect, predicative contexts that do not
express a temporal property should be more prone to copula omissions.
.  Copula omissions that are also found in adult speech (e.g. answers to questions using the
copula as in (i) and enumerations as in (ii)) were not included as target-deviant omissions in
child’s speech.
(i) ¿Dónde está el libro? El libro, aquí. ‘Where is the book? The book, here.’
(ii) El osito aquí y el panda aquí. ‘The little bear here and the panda here.’
 Laia Arnaus Gil & Natascha Müller

(15) a. *María es cansada.

b. *Pedro está un doctor.
(16) a. El coche Ø vacío.
b. Juan Ø inteligente.

100% 6 41 6 2 4 79 21 1
90% 38
295 40 12 4 61
295 60
80% 129
70%
60%
50% 1461 690 466 1604 311 483
40% 1427
30%
20% 79
10%
0%
Sp

er

er

Sp

er

er
-F

-F
-G

-G

-G

-G
Sp

Sp
Sp

at

Sp

at
-C

-C
Sp

Sp
SER ESTAR

targetlike target-deviant omissions target-deviant comissions

Figure 1.  Target-like and target-deviant productions of Spanish copula verbs SER and ESTAR
by monolingual and multilingual children

In order to examine the overall frequency of SER and ESTAR that has been
produced by the monolingual and multilingual children, Figure 1 shows the
total amount of copula use in all possible predicative contexts in which SER and
ESTAR can be used (e.g. with locatives, adjectives, nouns, etc.) taking the number
of languages and the different language combinations into account and the overall
age span of the children under study. As can be seen, multilingual (i.e. bilingual
and trilingual) children show a different pattern of grammatical and ungram-
matical productions of SER and ESTAR as compared with Spanish monolinguals:
whereas the latter master the use of Spanish copulas in a nearly adult-like fashion
(with an error rate of less than 5%), Spanish-German bilinguals show difficulties
in SER-contexts, i.e. they tend to produce more target-deviant structures when
they want to assign a permanent property. In approximately only 15% of ESTAR-
contexts, Spanish-German bilinguals overuse SER or omit ESTAR. As for the
Spanish-Catalan-German trilinguals, they seem to present similar rates of target-
deviant uses of ESTAR as their Spanish-German and Spanish-French bilingual
peers, yet we observe that trilinguals tend to overgeneralize ESTAR, whereas both
bilingual groups’ target-deviant productions tend to be omissions, i.e. they omit
ESTAR more frequently. With respect to SER, we observe that trilinguals are the
 The acquisition of Spanish in a bilingual and a trilingual L1 setting 

ones who concentrate the highest rate of copula omissions, reaching 60% of the
total amount of their SER productions. Concerning the overall grammatical use
of both SER and ESTAR on the one hand, and the overall ungrammatical use of
the copulas on the other hand, the difference between monolinguals and bilin-
guals (Χ2 = 44.95 (1), p < .05), as well as that between monolinguals and trilin-
guals (Χ2 = 138.22 (1), p < .05) is statistically significant.
In what follows, we focus on predicative adjectives, since they show a mixed
pattern between licensing one copula verb exclusively on the one hand, as
­Examples (1a) and (1b) of the (lexical) adjectives constante and desierto, and a much
more free variation as the cases in (2) of the (lexical-syntactic) adjectives alegre
on the other hand. In this regard, adjectives are a good testing ground to examine
copula acquisition, since they allow us to observe the semantic and syntactic selec-
tion mechanisms that children might employ in order to assign the corresponding
property to the predicate through the selection of SER or ESTAR. Table 2 focuses on
the multilingual children and their use of the appropriate copula verb with a certain
adjective type, particularly that of the kind alegre that allows both SER and ESTAR.

Table 2.  Distribution of (un)grammatical productions with Spanish lexical-syntactic

adjectives (adapted from Arnaus Gil 2013, p. 287)
Lexical-syntactic IL Lexical-syntactic SL

  Target-like Target-deviant Target-like Target-deviant

Monolinguals 100 (98%) 2 (2%) 24 (100%) 0

Bilinguals 230 (67.25%) 112 (32,75%) 123 (95.35%) 6 (4.65%)
Trilinguals 33 (48.53%) 35 (51,47%) 11 (100%) 0

Table 2 shows that multilingual children seem to behave similarly with respect
to the (un)grammatical productions of SER and ESTAR with lexical-syntactic
adjectives: their target-deviant productions are found mainly with IL-adjectives
(32.75% and 51.47%, respectively). That is, both bilinguals and trilinguals seem
to demonstrate more problems in assigning an intrinsic property to the referent;
e.g. when the guapa-adjective type needs to obtain an IL-meaning, multilingual
children frequently insert ESTAR (i.e. overgeneralizations of ESTAR) or omit SER,
both productions being inappropriate in the context which has been produced.
Examples (17) and (18) show some examples:
(17) a. To (=este) *tá mu (=muy) bueno. Teresa (3;1,0)
This *is-estar very good
[R: the child shows the adult a crocodile and the adult asks her to put the
animal away because she is scared. The child then produces the sentence
and means that the crocodile has a good heart.]
 Laia Arnaus Gil & Natascha Müller

b. Esto *tá marrón. Lucas (3;0,17)

This *is-estar brown
[R: the child and his father are drawing several animals in a book. The child
realizes that the lion is already drawn and has the color brown. Then, he
produces the sentence.]
(18) a. Esto también rojo. Erik (2;1,4)
This (is-ser) also red
[R: the child is playing with Legobricks that are different colors and is
­collecting the red ones with the aid of his father.]
b. Sol grande. Frank (2;10,9)
(The) sun (is-ser) big
[R: the child and his mother are reading a book together. He points at
­different objects in the picture and tells his mother what he sees.]

Table 3 presents the (un)grammatical productions of SER and ESTAR accompa-

nying adjectives like the ones of the type constante or desierto. Recall that these
adjectives exclusively select one copula verb.

Table 3.  Distribution of (un)grammatical productions with Spanish lexical adjectives

(adapted from Arnaus Gil 2013, p. 287)9
Lexical IL Lexical SL

Target-like Target-deviant Target-like Target-deviant

Monolinguals 55 (98.21%) 1 (1.79%) 104 (92.86%) 8 (7.14%)

Bilinguals 94 (63.95%) 53 (36.05%) 327 (89.59%) 38 (10.41%)
Trilinguals 4 (50%) 4 (50%)9   38 (70.37%) 16 (29.63%)

Interestingly, these data show that bilingual and trilingual children consider
lexical adjectives and the quality they are assigned to quite differently, at least
when temporal properties are concerned: While bilinguals seem to have captured
the temporal property of SLP by using ESTAR in a target-like manner, trilinguals
have more issues in assigning the temporal reading to the adjective (29.63%).
Some examples are given in (19) for the bilingual group and in (20) for the trilin-
guals (Arnaus Gil (to appear)):

.  Since the total amount of lexical IL adjectives for the trilingual group is very low (8 cases)
and the target-like and target-deviant productions are equally low (4 cases each), we will not
take these productions into consideration.
 The acquisition of Spanish in a bilingual and a trilingual L1 setting 

(19) a. *Están muy tan frío. Arturo (3;2,10)

(They) *are-estar very so cold
b. *Ése muy rápido. Lucas (2;1,14)
  This (is-ser) very fast
(20) a. *Ahí ya es sucio. Frank (3;1,4)
 Here already *is-ser dirty
b. *Ése roto. Frank (2;5,2)
  This (is-estar) broken10

Keeping Table 3 and the Examples (17)–(20) above in mind, we have observed
that trilinguals seem to evidence some difficulties in assigning the temporal prop-
erty to its lexical adjective, whereas bilinguals’ target-deviant uses mostly involve
adjectives assigning a permanent property. This might indicate that Spanish lexi-
cal adjectives are treated differently by bilinguals and trilinguals, but what distin-
guishes actually both multilingual groups? Trilinguals learn Catalan as another
L1, apart from Spanish and German, and so we suggest that the linguistic repre-
sentation of property assignment in Catalan (and German) works differently from
Spanish: while Catalan and German assign temporal properties in v and perma-
nent readings in T, Spanish syntactically mirrors this proposal, i.e. Spanish tem-
poral properties are codified in a head-complement relation in T and permanent
readings under v (cf. Section 2.1 for a detailed description).
Figure 2 below shows the target-like and target-deviant productions of the
Catalan copula verbs ÉSSER and ESTAR, i.e. the (un)grammatical assignment
of an IL/SL property, respectively. As mentioned in Section 2.1.1, Catalan adjec-
tives vary more freely than in Spanish, i.e. they can frequently be accompanied
by both copulas. As a consequence, and as opposed to Spanish, we do not divide
Catalan adjectives into lexical and lexical-syntactic adjectives, since they can gen-
erally be preceded by both ÉSSER and ESTAR, but we treat them as one single
group, i.e. they receive the corresponding aspectual interpretation syntactically.
However, in order to be able to compare the Spanish and Catalan adjectives,

.  Omissions and commissions have been considered ungrammatical productions, since
they deviate from adult speech. Yet, we do find omissions in the longitudinal production data
that have been considered grammatical because they represent copula ellipses and therefore
they do not belong to the group of target-deviant omissions. For the grammatical cases, we
show in (i) an omission of Arturo (3;10,13):
(i) Adult: y se enfadó mucho con el gato/hm/¿estaba contento el gato? /
Child: no /
Adult: ¿no?/¿qué estaba? /
Child: triste
 Laia Arnaus Gil & Natascha Müller

those ­Spanish adjectives considered to belong to the lexical adjectival group (e.g.
inteligente, europeo, cansado, cerrado, etc.) are classified here as lexical in Catalan.
­Examples (21) are extracted from the trilingual data.

100%

90%

80% 13

70%

60%

Target-deviant
50% 10 Target-like

40%

30% 18

20%

10%

0%
ÉSSER ESTAR

Figure 2.  (Un)grammatical productions of Catalan copulas with lexical adjectives

by trilinguals

(21) a.   globo tallat? Eric (1;11,4)

(is-estar) (the) balloon cut?
b. La manguera enrollada. Kilian (3;2,5)
The hose (is-estar) coiled up.
c.   bo. Eric (1;10,11)
(is-estar) good.
[R: the child takes an apple and starts eating it. He wants to give his
father a bite and offers him the apple. Later on, he continues eating it.]

As Figure 2 shows, trilingual children have, in fact, some difficulties when using
ESTAR in Catalan, that is to say, when assigning a temporal property to the adjec-
tival predicate. These target-deviant productions are exclusively ESTAR omissions,
as the examples in (21) show. Examples (21a) and (21b) show deverbal adjectives,
also known as adjectival participles and constitute 12 of 13 ESTAR omissions.
 The acquisition of Spanish in a bilingual and a trilingual L1 setting 

The question that arises here is whether these productions are instances of copula
or auxiliary omission. If it was auxiliary omission, these adjectival forms would
be referred to as root non-finite constructions, a phenomenon which has been
widely studied in the literature by Hyams (2001), Wexler (1994, 1998) and Rizzi
(1993/1994), among many others. For the multilingual children analyzed in this
paper, no Root Infinitive Stage has been identified (cf. Arnaus Gil 2013). Thus,
the examples in (21) should be considered adjectives with a verbal source, since
they express and encode a temporal property, a state of the subject referent as the
Spanish lexical adjectives descalzo ‘barefoot’ and cansado ‘tired’. With regard to the
copula use for permanent properties, the trilinguals master the use of ÉSSER with
ease, exhibiting an error-free acquisition.
In sum, we conclude that the bilingual Spanish-German children show cross-
linguistic influence and this influence results in a tendency towards target-deviant
uses of Spanish copula verbs as other authors for the Spanish-English bilingual
acquisition have also observed (Silva-Corvalán & Montanari 2008) and in con-
trast to the findings of Liceras’ research group. The analysis of the trilingual chil-
dren further suggests that they are able to capture the different syntactic property
assignment for Catalan (i.e. permanent properties are assigned in T and temporal
properties in v) and Spanish (although they show some difficulties by the latter)
rather than treating Spanish and Catalan copulas similarly because of their two
phonologically copula forms SERSp/ÉSSERCat and ESTARSp/Cat. In Section 3, we
hypothesized that we expect an overgeneralization of the Spanish copula verb
ESTAR to SER contexts, since ESTAR is syntactically less complex, and the inverse
predication for the Catalan data, namely that in Catalan, the less complex ÉSSER-
construction will take over ESTAR contexts.

4.3  The data on the acquisition of Romance attributive adjectives

The same multilingual children examined in Section 4.2 are examined again here.
In this case, we aim to illustrate the Romance pattern for adjective placement for
(un)grammatical productions of all bilinguals and trilinguals.11 This is seen in
­Figure 3. The x-axis label should read as follows: the languages in capital letters
are those depicted in each column, the language combination in brackets illus-
trates the multilingual group represented in the column. For example, for CAT
(Sp-­Ger-Cat), Catalan is the language represented, which belongs to the Spanish-
German-Catalan trilinguals.

.  For two trilingual children (Eric and Kilian), there are only Catalan and German
­recordings available. This is why the Spanish language of these children is not represented in
Figure 3.
 Laia Arnaus Gil & Natascha Müller

100% 9
0 2 1 24
90%
80%
70%
60%
50% 114 55 121 184
194
40%
30%
20%
10%
0%
CAT (Sp- SP (Sp-Ger-Cat) FR (Fr-Sp) SP (Fr-Sp) SP (Sp-Ger)
Ger-Cat)

target-like target-deviant

Figure 3.  (Un)grammatical adjective placement for all Romance languages of the multilingual
children

100% 5
19
90% 23

80% 1 62

70% 63
98
4
60%
50%
5
40% 98
30% 122

20% 96 51

10%
0%
French (Fr-Sp) Spanish (Fr-Sp) Spanish (Sp-Ger) Catalan (Sp-Ger-Cat)

ONLY pre-/postnominal target-like ONLY pre-/postnominal target-deviant

BOTH positions target-like BOTH positions target-deviant

Figure 4.  Adjective placement in Romance depending on (non)categorical placement; bilin-

gual and trilingual children

As Figure 3 shows, Spanish-German bilingual children produce the highest

number of observed target-deviant positions of the Spanish adjective, followed by
the Spanish-French child Syca-Inès, who acquired French and Spanish adjective
 The acquisition of Spanish in a bilingual and a trilingual L1 setting 

order in a nearly error-free way. The trilinguals always produce Catalan adjectives
according to the adult system, and they perform at ceiling in Spanish, too.
Let us have a closer look at the ungrammatical placements of Spanish adjec-
tives. Remember from Section 2.2.1 that a group of Spanish adjectives (i.e. scalar
adjectives) can be positioned pre- and postnominally, although the most frequent
position is postnominal, and a second group of adjectives selects exclusively for
one of these positions. Figure 4 shows how (un)grammaticality in adjective place-
ment in the child data depends on these two adjectival groups. The language
combination in brackets represents the multilingual group depicted in the corre-
sponding column. From left to right: French-Spanish bilinguals, Spanish-German
bilinguals, Spanish-German-Catalan trilinguals. The same conventions apply to
Figure 5 below.
Figure 4 clearly shows that inaccurate word orders with adjectives that only
permit one adjectival position are few (9 cases), whereas Spanish adjectives that
allow both positions experience the highest number of target-deviant productions.
Still, these ungrammatical productions represent less than 10% of the data and are
not statistically significant between the groups of Spanish-French and Spanish-
German bilinguals (X2 (1) = 2.17, p > .05).
From the description of the syntactic derivation for pre- and postnominal
adjective order, we hypothesized in Section 3 that prenominal adjectives should be
overgeneralized in the child’s data since the derivation of prenominal adjectives is
less complex than that of postnominal adjectives.

100% 0 0 0
90%
80%

70% 5

60% 19
50% 1
40%
30%
20% 4

10% 5
0%
French (Fr-Sp) Spanish (Sp-Ger) Spanish (Fr-Sp)

*A+N categorical *A+N non-categorical *N+A categorical *N+A non-categorical

Figure 5.  Ungrammatical productions of adjectives in Spanish, pre- and postnominal

 Laia Arnaus Gil & Natascha Müller

Figure 5 above illustrates that this is exactly what happens. We observe that
non-categorical adjectives are associated with most of the target-deviant produc-
tions of adjective placement. Moreover, prenominal placement is overgeneralized
in both bilingual groups, i.e. in the Spanish-German children and the Spanish-
French child. The same was observed in the study by Rizzi et al. (2013) for other
language combinations, such as Italian-German, French-German and French-
Italian. For example, in the DP la gentile volpe ‘the kind fox’, the adjective gentile
‘kind’ should occur postnominally from the perspective of the adult language (cf.
Rizzi et al. 2013, p. 137).

5.  Discussion and conclusions

This study has provided data concerning the early acquisition of the Spanish copula
verbs SER and ESTAR. The children’s productions have shown that monolinguals
and multilinguals behave differently with respect to copula assignment to perma-
nent/temporal properties of Spanish adjectives. We have observed that whereas
monolingual Spanish children demonstrate relatively few problematic instances
of SER and ESTAR assignment (cf. Figure 1), multilinguals, and ­Spanish-German
bilingual children in particular show a high rate of target-deviant productions,
such as copula omissions and commissions (i.e. overgeneralizations). In this
respect, bilinguals seem to have more difficulties in assigning a permanent rather
than a temporal property, which is seen in the high rate of target-deviant pro-
ductions in SER contexts. Following the approach outlined in Section 2 on the
syntactic complexity of Spanish copulas SER and ESTAR, these results indicate
that it is more costly to assign the (permanent) property in the head-complement
relation in v and the later movement of the copula SER to T than doing so in the
head-complement relation in T, where ESTAR is directly inserted. This pattern is
found not only with those (lexical-syntactic) adjectives of the guapo type in which
both copulas compete, but also with (lexical) adjectives such as constante or desi-
erta where only one copula verb can be selected. Since this behavior is observed
for the bilingual group but not for the Spanish monolinguals, we propose that this
linguistic phenomenon shows evidence of cross-linguistic influence from German
to Spanish, as several studies for the language pair English-Spanish have claimed
(e.g. Silva-Corvalán & Montanari 2008). For predicative adjectives, it has been
argued that Spanish copula verbs enter the syntactic derivation in two different
ways: whereas ESTAR is inserted directly under T and does not require a vP, SER
must be inserted closer (and earlier than ESTAR) to the adjective in v in order
to assign the permanent property to the adjective. In a later stage, SER moves
from v to T. For German (and also for French), it has been proposed that property
 The acquisition of Spanish in a bilingual and a trilingual L1 setting 

assignment to the predicative adjective also takes place syntactically, that is, SEIN
(or ÊTRE) is inserted either in T or v to assign the corresponding characteristic
to the adjective (a temporal or permanent property, respectively). Following this
idea, condition 1 for cross-linguistic influence (Hulk & Müller 2000; Müller &
Hulk 2001) is fulfilled, namely that the linguistic phenomenon of SER and ESTAR
is an interface property, here between semantics and syntax, in Sorace and Ser-
ratrice’s (2009) words, an internal interface property. Following these authors, no
vulnerability for cross-linguistic influence within internal interfaces is expected.
However, our study shows that the reverse is true. Moreover, we have also assumed
that condition 2 for cross-linguistic influence (Müller & Patuto 2009) is satisfied:
Spanish and German not only share a less complex syntactic derivation, namely
insertion in T of ESTAR(>temporalAdj) and in T of SEIN(>permanentAdj), but also a more
complex syntactic derivation which involves the insertion of SER(>permanentAdj) and
SEIN(>temporalAdj) in v (and later movement to T). Therefore, the fact that both lan-
guages share the less complex analysis promotes the use of the less complex syn-
tactic structure in Spanish when the more complex structure would be needed.
This has been outlined in Section 3 where we predicted that, for Spanish and
Catalan, children will employ the less complex analysis and overuse the Spanish
copula verb ESTAR in SER contexts and that, for Catalan, ÉSSER will take over
ESTAR contexts since it too is less complex. On the basis of the data analysed in
Section 4.2, we can confirm our hypothesis, namely that the less complex analysis
tends to take over contexts in which the more complex analysis would be expected.
This seems to be the case for Spanish as well as for Catalan.
Turning to the trilingual group’ results regarding the acquisition of the Spanish
copula verbs, the data in relation to the property assignment of predicative adjec-
tives showed that trilinguals are also sensitive to the distinctions between syntactic
(e.g. guapo) and lexical property assignment (e.g. constante, desierta) in Spanish,
as bilinguals are. However, results indicated differences between both groups:
whereas multilinguals do not show divergence for adjectives of the guapo-type,
they do for adjectives of the constante- and desierta-type. In particular, bilinguals
were shown to have difficulty specifying a permanent property, while trilinguals
had more difficulties with the assignment of a temporal property. These results
seem to indicate that trilinguals benefit from the syntactic ways of assigning prop-
erties to adjectives in German and Catalan for Spanish (cf. Table 4) by employ-
ing the less complex syntactic derivation, namely overuse of SER and omission of
ESTAR, in particular for those Spanish adjectives which enter the derivation fully
specified with a permanent or temporal property.
Regarding the second prediction, adjective placement in Spanish surely involves
the syntax-semantics interface. Furthermore, the syntactic derivations for prenom-
inal adjectives in the four languages studied, Spanish, Catalan, French and German,
 Laia Arnaus Gil & Natascha Müller

converge. Prediction (2) in Section 3 proposed that bilingual children have a less
(A + N) and a more (N + A) complex analysis available in Romance. As such, it was
predicted that the bilingual child would make use of the less complex analysis for
contexts in which the more complex, postnominal analysis was expected.
Against our expectations, we have observed only a few target-deviant adjec-
tive placements in bilingual and trilingual Spanish. Since studies on monolingual
Spanish children are scarce, it is hard to tell whether the few target-deviant adjec-
tive orders in bilingual and trilingual Spanish are due to the acquisition of more
than one language. Notice, however, the interesting result that target-deviant
placement mainly concerns adjectives that are allowed in two positions in S­ panish.
Together with the result that prenominal ordering is overgeneralized by all chil-
dren, we argue that all children find postnominal adjectives more difficult than
prenominal ones, if and only if they compete with each other in the language in
question. This is even more so in children who acquire a second L1 like German
that only allows the less complex prenominal order. Following this idea, it would
be very interesting to examine a language which has exclusively the more complex
derivation available, that is to say, that has only postnominal attributive adjec-
tives. For the simultaneous acquisition of Spanish with at least one language which
exhibits only postnominal adjective order, we hypothesize that the multilingual
child will benefit for the language that exhibits a complex and a less complex deri-
vation like Spanish. We leave this to future research.
Our investigation of SER and ESTAR and adjective placement in Spanish bilin-
guals and trilinguals allows us to modify condition 2 for cross-linguistic influence:

2. If language A exhibits a complex syntactic derivation (SynD+) and a less com-

plex syntactic derivation (SynD), and language B exclusively provides the less
complex syntactic derivation (SynD), the multilingual child is more likely to
use SynD in language A in contexts in which SynD+ is expected.

Notice that the trilinguals had a small advantage over the bilinguals regarding
adjective placement. They seem to have captured the syntactic differences between
pre- and postnominal adjective placement for both groups of adjectives (what we
have called categorical and non-categorical adjectives). Very tentatively, we would
like to deduce the following generalization from this observation which has to be
verified in future research:

3. If language A exhibits a complex (SynD+) and a less complex syntactic deriva-

tion (SynD) and language B is similar to language A in providing SynD and
SynD+, the multilingual child is more likely to use SynD+ as required by the
target language. In other words, more experience with the more complex deri-
vation, especially if it comes from two different languages, can speed up the
acquisition process.
 The acquisition of Spanish in a bilingual and a trilingual L1 setting 

If this tentative generalization is on the right track, it would thus predict a faster
development of Catalan and Spanish copula verbs and adjective placement by
bilingual Catalan-Spanish children, than what is expected for their monolingual
peers, since in these two languages both the less and the more complex syntactic
derivation are available.
A final note with respect to the trilingual children is in order here: these chil-
dren are trilinguals from birth and they outperform the bilinguals. This observa-
tion is surprising compared to the view shared by many people in society that the
acquisition of more than one language is effortful and leads to semi-lingualism.
Our investigation shows that properties of the acquired grammatical systems are
much more important for the developmental path and the success of multilingual
children than the number of languages which they acquire.

References

Abney, S. (1987). The English Noun Phrase in its Sentential Aspect. Unpublished Ph.D. disserta-
tion. MIT.
Alexiadou, A. (2001). Functional Structure in Nominals: Nominalization and Ergativity.
­Amsterdam: John Benjamins. DOI: 10.1017/s0022226702211986
Arche, M.J. (2006). Individuals in Time. Amsterdam: John Benjamins.
DOI: 10.1017/s0022226708005215
Arnaus Gil, L. (2013). La selección copulativa y auxiliar: Las lenguas romances (español –
­italiano  – catalán – francés) y el alemán en contacto. Su adquisición en niños bilingües y
trilingües.Tübingen: Narr.
Arnaus Gil, L. (To appear). Acquisitional advantages of simultaneous 3L1 trilingual children:
The Spanish copulas SER and ESTAR. Educational Linguistics.
Arnaus Gil, L., Eichler, N., Jansen, V., Patuto, M., & Müller, N. (2012). The syntax of code switch-
ing and the left periphery of DPs: Mixed DPs containing an adjective. Evidence from bilin-
gual child language. In K. L. Geeslin & M. Díaz-Campos (Eds.), Proceedings from the His-
panic Linguistics Symposium 2010. (pp. 242–257). Somerville, MA: Cascadilla Proceedings.
Austin, J. (2001). Language Differentiation and the Development of Morphosyntax in Bilingual
Children Acquiring Basque and Spanish. Unpublished Ph.D. dissertation. Cornell Univer-
sity, Ithaca, NY.
Barnes, J. (2011). The influence of child-directed speech in early trilingualism. International
Journal of Multilingualism, 8(1), 42–62. DOI: 10.1080/14790711003671861
Barreña, A. (1997). Desarrollo diferenciado de sistemas gramaticales en un niño vasco-español
bilingüe. In A. T. Pérez-Leroux & W. Glass (Eds.), Contemporary Perspectives on the Acqui-
sition of Spanish (pp. 55–74). Somerville, MA: Cascadilla Press.
Becker, M. (2000). The Development of the Copula in Child English: The Lightness of be. Unpub-
lished Ph.D. dissertation. University of California, Los Angeles.
Bernstein, J. (2001). The DP hypothesis: Identifying clausal properties in the nominal domain.
In M. Baltin & C. Collins (Eds.), The Handbook of  Contemporary Syntactic Theory
(pp. 536–561). Malden, MA: Blackwell. DOI: 10.1002/9780470756416.ch17
 Laia Arnaus Gil & Natascha Müller

Bouchard, D. (1998). The distribution and interpretation of adjectives in French: a consequence

of bare phrase structure. Probus, 10, 139–183. DOI: 10.1515/prbs.1998.10.2.139
Bouchard, D. (2002). Adjectives, Number and Interfaces: Why Languages Vary. Oxford: Elsevier
Science. DOI: 10.9793/elsj1984.21.467
Brucart, J.M. (2012). Copular alternation in Spanish and Catalan attributive sentences. Lin-
guística. Revista de Estudos Linguísticos da Universidade do Porto, 7, 9–43.
Bruhn de Garavito, J. (2009). Eventive and stative passives: The role of transfer in the acquisition
of ser and estar by German and English L1 speakers. In J. Collentine, M. Garcia, B. Lefford,
& F. Marcos Marin (Ed.), Selected Proceedings of the 11th Hispanic Linguistics Symposium
(pp. 27–38). Somerville, MA: Cascadilla Press.
Camacho, J. (2012). Ser and estar: The individual/stage level distinction and aspectual predica-
tion. In J. I. Hualde, A. Olarrea, & E. O’Rourke (Eds.), The Handbook of Spanish Linguistics
(pp. 453–476). Oxford: Blackwell. DOI: 10.1002/9781118228098.ch22
Cantone, K.F., Kupisch, T., Müller, N., & Schmitz, K. (2008). Rethinking language dominance in
bilingual children. Linguistische Berichte, 215, 307–343.
Carlson, G. (1977). Reference to kinds in English. Unpublished Ph.D. dissertation. University of
Massachusetts, Amherst.
Castro, D., & Gavruseva, E. (2003). Finiteness and aspect in Spanish/English bilingual acquisi-
tion. First Language, 23, 171–192. DOI: 10.1177/01427237030232002
Cinque, G. (1995). Posizione del soggetto nel DP italiano. In C. Lupu & L. Renzi (Eds.), Studi
Rumeni e Romanzi (pp. 436–46). Padova: Unipress.
Cinque, G. (2010). The Syntax of Adjectives. A Comparative Study. Linguistic Inquiry ­Monographs.
Cambridge, MA: The MIT Press. DOI: 10.7551/mitpress/9780262014168.001.0001
Corpas, R. (2007). ‘Ser’ and ‘estar’ in early child Spanish. Paper presented in Ontario Dialogues
on the Acquisition of Spanish. University of Ottawa, March 2 2007.
Crain, S., & Thornton, R. (1998). Investigations in Universal Grammar: A Guide to Experiments
on the Acquisition of Syntax and Semantics. Cambridge, MA: The MIT Press.
DOI: 10.1017/s0022226701238651
de Miguel, E. (1999). El Aspecto Léxico. In I. Bosque & V. Demonte (Eds.), Gramática Descrip-
tiva de la Lengua Española, Vol. 2 (pp. 2977–3060). Madrid: Espasa Calpe.
de Miguel, E. (2004). Qué significan aspectualmente algunos verbos y qué pueden llegar a signifi-
car. In J. L. Cifuentes & C. Marimón (Eds.), Estudios de Lingüística: El verbo (pp. 167–206).
Alicante: Estudios de Lingüística de la Universidad de Alicante.
Demonte, V. (2008). Meaning-form correlations and adjective position in Spanish. In
C. ­Kennedy & L. McNally (Eds.), The Semantics of Adjectives and Adverbs (pp. 71–100).
Oxford: OUP.
Ezeizabarrena, M.J. (1996). Adquisición de la morfología verbal en euskera y castellano por niños
bilingües. Vitoria: Servicio Editorial del País Vasco.
Ezeizabarrena, M.J. (1997a). Infinitivos verbales en el lenguaje de los niños bilingües. In L. Díaz &
C. Pérez (Eds.), Views on the Acquisition and Use of a Second Language (pp.  177–190).
­Barcelona: Universidad Pompeu Fabra.
Ezeizabarrena, M.J. (1997b). Morfemas de concordancia con el sujeto y con los objetos del cas-
tellano infantil. In A. T. Pérez-Leroux & W. Glass (Eds.), Contemporary Perspectives on the
Acquisition of Spanish (pp. 21–63). Somerville, MA: Cascadilla Press.
Ezeizabarrena, M.J. (2002). Root infinitives in two pro-drop languages. In A. T. Pérez-­Leroux
& J. Liceras (Eds.), The Acquisition of Spanish Morphosyntax. The L1/L2 Connection
(pp. 35–66). Dordrecht: Kluwer. DOI: 10.1007/978-94-010-0291-2_2
 The acquisition of Spanish in a bilingual and a trilingual L1 setting 

Fernández Fuertes, R., & J. Liceras. (2008). Copula omission in the English developing gram-
mar of English/Spanish bilingual children. Paper presented in International Conference on
Models of Interaction in Bilinguals. Bangor, Wales, UK.
DOI: 10.1080/13670050.2010.488285
Fernández Fuertes, R., & Liceras, J. (2010). Subject and copulas: The directionality of interlin-
guistic influence. Paper presented in 16. Wuppertaler Linguistisches Kolloquium. Languages
in Contact: Cross-linguistic Influence on Different Domains in Bilingual First Language
Acquisition. Bergische Universität Wuppertal, Germany.
Fernández Leborans, M.J. (1999). La predicación: Las oraciones copulativas. In I. Bosque &
V. Demonte (Eds.), Gramática descriptiva de la Lengua Española (pp. 2358–2460). Madrid:
Espasa Calpe.
Gallego, A., & Uriagereka, J. (2011). The Lexical Syntax of Ser and Estar. Ms. Universitat
Autónoma de Barcelona & University of Maryland.
Gavarró, A. (2003). Economy and word order patterns in English-Dutch acquisition. Bilingual-
ism: Language and Cognition, 6(1), 69–79. DOI: 10.1017/s1366728903001044
Gumiel-Molina, S., & Pérez-Jiménez, I. (2012). Aspectual composition in ser/estar + adjec-
tive structures: Adjectival scalarity and verbal aspect in copular constructions. Borealis,
1, 33–62. DOI: 10.7557/1.1.1.2321
Hauser-Grüdl, N., Arencibia Guerra, L., Witzmann, F., Leray, E., & Müller, N. (2010). Cross-­
linguistic influence in bilingual children: Can input frequency account for it? Lingua,
120(11), 2638–2650. DOI: 10.1016/j.lingua.2010.06.008
Hernández Pina, F. (1979). Etapas en la adquisición del lenguaje infantil. Infancia y Aprendizaje,
8, 23–32. DOI: 10.1080/02103702.1979.10821759
Hernández Pina, F. (1984). Teorías psicosociolingüísticas y su aplicación a la adquisición del espa-
ñol como lengua materna. Siglo XXI, Madrid. DOI: 10.2307/342474
Holtheuer, C. (2009). Learning Ser and Estar: A Study of Production and Comprehension in Chil-
ean Spanish. Unpublished Ph.D. dissertation. Australian National University, Canberra.
Hulk, A., & Müller, N. (2000). Crosslinguistic influence at the interface between syntax and
pragmatics. Bilingualism: Language and Cognition, 3(3), 227–244.
DOI: 10.1017/s1366728900000353
Hulk, A., & Zuckerman, S. (2000). The interaction between input and economy: Acquiring
optionality in French wh-questions. In S. C. Howell, S.A. Fish & T. Keith-Lucas (Eds.),
Proceedings of the 24th Annual Boston University Conference on Language Development
(pp. 438–449). Somerville, MA: Cascadilla Press.
Hyams, N. (2001). Now you hear it, now you don’t: The nature of optionality in child language.
In A. H.-J. Do, L. Domínguez, & A. Johansen (Eds.), Proceedings of the 25th Annual Boston
University Conference on Language Development (pp.34–58). Somerville, MA: Cascadilla
Press.
Idiazábal, I. (1996). First stages in the acquisition of noun phrase determiners by a Basque-
Spanish bilingual child. In C. Silva-Corvalán (Ed.), Spanish in Four Continents: Studies in
Language Contact and Bilingualism (pp. 243–259). Washington DC: Georgetown Univer-
sity Press. DOI: 10.1017/s0047404500019631
Kayne, R. (1994). The Antisymmetry of Syntax.Cambridge, MA: The MIT Press.
Krasinski, E. (2005). The acquisition of ser and estar in a bilingual child. In L. A. Ortiz López &
M. Lacorte (Eds.), Contactos y contextos lingüísticos. El español en los Estados Unidos y en
contacto con otras lenguas (pp. 217–226). Madrid & Frankfurt: Iberoamericana & Vervuert.
DOI: 10.1075/sic.4.1.07gar
 Laia Arnaus Gil & Natascha Müller

Larrañaga, M.P. (2000). Ergative Sprachen, akkusative Sprachen. Der Erwerb des Kasus bei bilin-
gualen Kindern. Frankfurt: Vervuert. DOI: 10.1515/zrph.2004.757a
Lema, J. (1995). Distinguishing copular and aspectual auxiliaries: Spanish ser and estar. In
J. Amastae, G. Goodall, M. Montalbetti, & M. Phinney (Eds.), Contemporary Research in
Romance Linguistics: Papers from the XXII Linguistic Symposium on Romance Languages, El
Paso/Juárez, February 22–24, 1992. (pp. 257–274). Amsterdam: John Benjamins.
DOI: 10.1075/cilt.123.19lem
Liceras, J., Fernández Fuertes, R., de la Fuente, A.A., & Tercedor Sánchez, M. (2010). Lexi-
cally-based interlinguistic influence at the syntax-semantic interface: Copula omission
in the English grammar of English-Spanish bilinguals. In C. Borgonovo, M. Español-­
Echevarría, & P. Prévost (Eds.), Selected Proceedings of the 12th Hispanic Linguistics Sympo-
sium (pp. 183–193). Somerville, MA: Cascadilla Proceedings Project.
Lleó, C. (2001a). The interface of phonology and syntax. The emergence of the article in the early
acquisition of Spanish and German. In J. Weissenborn & B. Höhle (Eds.), Approaches to
Bootstrapping. Phonological, Lexical, Syntactic and Neurophysiological Aspects of Early Lan-
guage Acquisition (pp. 23–44). Amsterdam: John Benjamins. DOI: 10.1075/lald.24.03lle
Lleó, C. (2001b). The transition from prenominal fillers to articles in German and Spanish.
In M.  Almgren, A. Barreña, M.J. Ezeizabarrena, I. Idiazábal, & B. McWhinney (Eds.),
Research on Child Language Acquistion. Proceedings of the 8th Conference of the Inter-
national Association for the Study of Child Language (pp. 713–737). Somerville. MA:
­Cascadilla Press.
MacWhinney, B. 2000. The CHILDES Project: Tools for Analyzing Talk. Mahwah, NJ: Lawrence
Erlbaum Associates. DOI: 10.1177/014272370002006006
Marín, R. (2004). Entre ser y estar. Madrid, Arco Libros.
Montanari, S. (2009). Multi-word combinations and the emergence of differentiated ordering
patterns in early trilingual development. Bilingualism: Language and Cognition, 12(4),
503–519. DOI: 10.1017/s1366728909990265
Müller, N., & Hulk, A. (2001). Crosslinguistic influence in bilingual language acquisition: Italian
and French as recipient languages. Bilingualism: Language and Cognition, 4(1), 1–22.
DOI: 10.1017/s1366728901000116
Müller, N., & Patuto, M. (2009). Really competence-driven cross-linguistic influence in bilin-
gual first language acquisiton? The role of the language combination. In P. Bernardini,
V.  ­Egerland, & J. Granfeld (Eds.), Mélanges plurilingues offerts à  Suzanne Schlyter à 
l′occasion de son 65ème anniversaire (pp. 299–319), Études Romanes de Lund 85. Lund:
Lunds Universitet, Språch- ochlitteraturcentrum Romanska.
Müller, N., Kupisch, T., Schmitz, K., & Cantone, K. (2011). Einführung in die Mehrsprachigkeits-
forschung. Deutsch, Französisch, Italienisch. Tübingen: Gunter Narr.
DOI: 10.1515/zrs.2011.052
Müller, N. (Ed.) (in prep.). Code-Switching: Spanisch, Französisch, Italienisch. Eine Einführung.
Tübingen: Gunter Narr.
Patuto, M. (2013). Der Erwerb des Subjekts in (Nicht-) Nullsubjektsprachen. Die Rolle des Sprache-
neinflusses und der Sprachdominanz bei bilingual deutsch-italienisch, deutsch-spanisch und
französisch-italienisch aufwachsenden Kindern. Tübingen: Gunter Narr.
Perpiñán, S. (2015). The locative paradigm in the L2 Spanish of Catalan native speakers. In T.
Judy & S. Perpiñán (Eds.), The Acquisition of Spanish in Understudied Language Pairings
(pp. 105–132). Amsterdam: John Benjamins.
 The acquisition of Spanish in a bilingual and a trilingual L1 setting 

Picallo, C. (2002). L’adjectiu i el sintagma adjectival. In J. Solà, M–R. Lloret, J. Mascaró, &
M.  Pérez-Saldanya(Eds.), Gramàtica del català contemporani, Vol. 2 (pp. 1643–1688).
­Barcelona: Empúries.
Ramos Alfarín, J.R. (2008). Sintaxi. L’atribució. In J. Solà, M.R. Lloret, J. Mascaró, & M. Pérez
Saldanya (Eds.) Gramàtica del català contemporani, Vol. 2. Barcelona: Editorial Empúries.
Rizzi, L. (1993/1994). Some notes on linguistic theory and language development: The case of
root infinitives. Language Acquisition, 3, 371–393. DOI: 10.1207/s15327817la0304_2
Rizzi, S., Arnaus Gil, L., Repetto, V., Geveler, J., & Müller, N. (2013). Adjective placement in
bilingual Romance-Romance and Romance-German children. Studia Linguistica, 67(1),
127–147. DOI: 10.1111/stul.12009
Schmeißer, A., Hager, M., Arnaus Gil, L., Jansen, V., Geveler, J., Eichler, N., Patuto. M., &
­Müller, N. (To appear). Related but different: The two concepts of language dominance and
language proficiency. In C. Silva-Corvalán & J. Treffers-Daller (Eds.), Language Dominance
in Bilinguals: Issues of Operationalization and Measurement. Cambridge: CUP.
Schmitt, C. (1992). Ser and estar: A matter of aspect. NELS 22 Proceedings, 411–426. Amherst,
MA: GLSA.
Schmitt, C., & Miller, K. (2007). Making discourse dependent decisions: The case of the copulas
ser and estar in Spanish. Lingua, 117(11), 1907–1929. DOI: 10.1016/j.lingua.2006.11.007
Schmitt, C., Holtheuer, C., & Miller, K. (2004). Acquisition of ser and estar in Spanish: Learning
lexico-semantics, syntax and discourse. In A. Brugos, L. Micciulla, and C. E. Smith, Pro-
ceedings of the 28th annual Boston University Conference on Language Development, Online
Supplement. Somerville, MA: Cascadilla Press.
Sera, M.D. (1992). To be or to be. Journal of Memory and Language, 31, 408–427.
DOI: 10.1016/0749-596x%2892%2990021-o
Sera, M.D., Bales, D.W., & del Castillo Pintado, J. (1997). Ser helps speakers identify “real” prop-
erties. Child Development, 68, 820–831. DOI: 10.2307/1132035
Serratrice, L. & Sorace, A. (2003). Overt and null subjects in monolingual and bilingual lan-
guage acquisition. In B. Beachley, A. Brown, & F. Conlin (Eds.), Proceedings of the 27th
Annual Boston University Conference on Language Development (pp. 739–750). Somerville,
MA: Cascadilla Press.
Serratrice, L., Sorace, A., & Paoli, S. (2004). Crosslinguistic influence at the syntax-pragmatics
interface: Subjects and objects in Italian-English bilingual and monolingual acquisition. 
Bilingualism Language and Cognition, 7, 183–205. DOI: 10.1017/s1366728904001610
Silva-Corvalán, C. (2003). Linguistic consequences in reduced input in bilingual first language
acquisition. In S. Montrul & F. Ordóñez (Eds.), Linguistic Theory and Language Develop-
ment in Hispanic Languages (pp. 375–397). Somerville, MA: Cascadilla Press.
Silva-Corvalán, C., & Montanari, S. (2008). The acquisition of ser, estar (and be) by a
­Spanish-English bilingual child: The early stages. Bilingualism: Language and Cognition,
11, 341–360. DOI: 10.1017/s136672890800357x
Sorrace, A., & Serratrice, L. (2009). Internal and external interfaces in bilingual language devel-
opment: Beyond structural overlap. International Journal of Bilingualism, 13(2), 195–210.
DOI: 10.1177/1367006909339810
Vila, I., & Cortés, M. (1991). Aspectos relativos al desarrollo lexical y morfosintáctico de los
bilingües familiares. In I. Idiazábal (Ed.), Adquisición del lenguaje en niños bilingües y
monolingües – Hizkuntz Jabekuntza Haur Elebidun eta Elebakarretan (pp. 109–127). San
Sebastián: Servicio Editorial de la U.P.V.
 Laia Arnaus Gil & Natascha Müller

Wexler, K. (1994). Optional infinitives, head movement and the economy of derivations. In
D. Lightfoot & N. Hornstein (Eds.), Verb Movement (pp. 305–350). Cambridge: CUP.
DOI: 10.1017/cbo9780511627705.016
Wexler, K. (1998). Very early parameter setting and the unique checking constraint: A new
explanation of the optional infinitive stage. Lingua, 106, 23–79.
DOI: 10.1016/s0024-3841%2898%2900029-1
Zagona, K. (2009). Ser and estar: Phrase structure and aspect. Proceedings of 2008. Chronos y
Cahiers Chronos (pp. 1–25). Austin, TX: University of Texas.
Zuckerman, S. (2001). The Acquisition of Optional Movement. Unpublished Ph.D. dissertation.
University of Groningen, The Netherlands.
 Knowledge and processing of subject-related
discourse properties in L2 near-native speakers
of Spanish, L1 Farsi

Tiffany Judy
Wake Forest University

This chapter investigates the discourse-constrained distribution of referential

subject pronouns (RSP) by native Farsi-speaking, adult second language (L2)
Spanish speakers. Results from offline and online tasks measuring participants’
knowledge and processing of overt and null subjects in Contrastive Focus,
Topic Shift and Topic Maintenance contexts were mixed, showing divergence
and convergence. The results are discussed in light of the Interface Hypothesis
(IH; Sorace 2011, 2012), which predicts divergence on the processing of
external interface-conditioned properties like RSP distribution. While some
unexpected group results obtained, the results are ultimately interpreted as
not fully supporting the IH since native-like processing of Topic Shift tokens
obtained. While RSPs have been extensively examined, this study makes new
contributions by examining near-native L2 speakers of languages sharing
analogous RSP distribution for the contexts tested using both online and offline
methodology.

1.  Introduction

Much generative research conducted in previous decades focused on grammatical

representation, as this was the assumed point of divergence in adult second lan-
guage (L2) acquisition (e.g. Hawkins & Chan 1997; Tsimpli & D ­ imitrakopoulou
2007; Vainikka & Young-Scholten 1994, 1996). Yet, the amount of research
demonstrating restructuring (in the sense of parameter resetting, new feature
acquisition, feature reassembly, etc.) has forced the field to look elsewhere in
its quest to identify the locus of divergence (see White 2003, 2009). The Inter-
face Hypothesis (IH; Sorace 2006, 2011, 2012; Sorace & Filiaci 2006; Sorace &
­Serratrice 2009) has proven an influential account attempting to reconcile some

doi 10.1075/ihll.3.07jud
© 2015 John Benjamins Publishing Company
 Tiffany Judy

s­ pecific differences between adult L2 and native (L1) speakers, on one hand, with
convergence on certain domains of language, on the other. In some ways, the IH
has spurred a shift from examining non-native language representation to exam-
ining the integration of linguistic and cognitive information (in early instantia-
tions) and processing (currently) of non-native languages. This is so because the
IH assumes that convergence on syntax is possible (though not guaranteed), but
predicts divergence for interface-conditioned properties. Specifically, proper-
ties requiring integration of linguistic and cognitive information (i.e. external
interface-conditioned) are predicted points of divergence (Sorace 2011, 2012),
as they are deemed costly for processing. Results from early IH-related work,
like other accounts listed above, were interpreted as evidence that representa-
tional differences were the culprit of divergence, pointing to language pairing as
a deterministic factor in convergence (e.g. Belletti, Bennati, & Sorace 2007 for L2
acquisition; Tsimpli, Sorace, Heycock, & Filiaci 2004 for L1 attrition). However,
given that differences have obtained for both child and adult bilinguals whose
languages are substantially similar for the property tested in offline tasks (e.g.
Bini 1993; Lozano 2006; Margaza & Bel 2006; Sorace & ­Serratrice 2009; Sorace,
Serratrice, Filiaci, & Baldo 2009), the IH currently claims that divergence results
from bilingualism itself and not (only) L1-L2 differences (Sorace 2011): since
bilinguals must inhibit the non-relevant language (Bialystok 2009) while activat-
ing the other, finite cognitive resources are divided and processing differences
obtain.
Determining whether processing differences obtain for online tasks in bilin-
guals whose languages coincide for the property tested may inform the repre-
sentation vs. processing debate: holding other variables constant, if processing
differences obtain only for L1-L2 pairings that differ, and not those that coincide,
representation is implicated. Conversely, if processing differences obtain irrespec-
tive of language pairing, evidence is found supporting the IH’s claim that the very
nature of bilingualism and its effect on processing is implicated. As follows, this
study examines near-native adult L2 Spanish speakers (L1 Farsi) who were tested
for convergence on referential subject pronoun (RSP) distribution, an external-
interface conditioned property. As described in Section 2, Spanish and Farsi share
similar discourse distributions of RSPs, making it possible to discount grammati-
cal representation of subjects as a potential source of divergence should it obtain.
Section 3 describes relevant findings from previous child and adult acquisition
studies examining subject distribution. Section 4 outlines the research questions,
participant groups and tasks while Section 5 describes the offline and online
results. Finally, Section 6 discusses the results in light of the research questions
and the IH’s predictions.
 Knowledge and processing of subject-related discourse properties 

2.  Subject Distribution in Spanish and Farsi

Spanish and Farsi are null-subject languages, meaning that pro is licensed and
identified per Rizzi’s (1982) requirements. Along with Alexiadou and Anagnos-
topoulou (1998) and Goodall (2001, 2002), I assume that the Extended Project
Principle (EPP, which states that all clauses must have a subject) is a D(eterminer)
feature of Tense that is checked via movement in Spec, TP in Spanish. For Farsi,
Karimi (2005) argues that the EPP is satisfied morphologically without movement
as all phrasal elements may remain in the vP. Nevertheless, the particular syn-
tactic analysis assumed of the languages has little bearing on the property exam-
ined herein: the discourse-constrained distribution of RSPs. Given principles of
economy such as the Avoid Pronoun Principle (Chomsky 1981, 1995, 2000, 2001),
which restricts overt pronouns from being used unless absolutely necessary, we
may ask why overt subject pronouns are ever employed in Spanish and Farsi, espe-
cially considering the rich person and number morphology in these languages1
that allow for subject identification. However, the availability of the null subject
pronoun, pro, does not necessitate its use. Rather, overt and null subject pronouns
are distributed according to discursive factors, with some contexts requiring overt
subjects and others null.
Based on considerable previous research (see Alonso-Ovalle & D’Introno
2000; Fernández-Soriano 1989, 1993; Luján 1987, 1999; Picallo 1994, 1998; Rizzi
1997 and sources cited within), Rothman (2009) describes the discourse-related
restrictions on RSP distribution in Spanish. Relevant examples for this study
(Contrastive Focus, Topic Shift and Topic Maintenance)2 are described below for
Spanish (see Rothman (2009) for original examples) and Farsi. Table 1 provides a
summary.
Examples (1) and (2) show subject distribution in Contrastive Focus contexts,
which are contexts that create a contrast between two entities. In both Spanish
and Farsi, overt subjects are preferred over null. Although verbal morphology is

.  In Spanish, present, preterite and future tense morphology for first and second person sin-
gular and plural subjects is distinct; for the imperfect and conditional tenses, some compound
tenses and subjunctive moods, first and third person singular are conflated. In Farsi, all person
and number morphology is distinct.
.  Regarding Topic Shift and Topic Maintenance contexts, see Alonso-Ovalle, Fernández-
Solera, Frazier and Clifton (2002) and Filiaci (2011) for Spanish, Filiaci (2011) for Italian and
Bel and García-Alcaraz (this volume) for Moroccan Arabic. In general, these studies found
that null subjects were often interpreted as co-referential with subjects (and thus an instance
of Topic Maintenance), while more varied results obtained for overt subjects.
 Tiffany Judy

distinct for the second and third person singular subjects (digas versus diga in
S­ panish and beguyi versus beguyam in Farsi), the absence of the overt subject ren-
ders the sentence infelicitous.
(1) O te lo digo yo o te lo
or you-dat it-acc tell-1.sg.pres I or you.dat it-acc
dice ella.
tell-3.sg.pres she
“Either I will tell you or she will tell you.”
a. Quiero que me lo digas tú.
b. #Quiero que me lo digas pro
  want-1.sg.pres that me-dat it-acc tell-2.sg.pres-subj (you)
“I want you to tell it to me [and not her].”
(2) Man be to beguyam ya u be to beguyad.
I to you tell-1.sg.pres or u to you tell-3.sg.pres
“Either I tell you or she tells you.”
a. Man mixam ke to be man beguyi.
b. #Man mixam ke pro be man beguyi.
 I dur-want.1.sg that you to me tell-2.sg.pres
“I want you to tell me.”
Next, Examples (3) and (4) show Topic Shift contexts in which the topic (here, the
subjects) changes within the discourse. Specifically, the topic changes from yo/
man “I” to ellas/unha “they,” respectively. Consequently, the overt subject ellas/
unha “they” is preferred over pro to identify the matrix subject. Again, this is true
despite the recoverability of the subject via verbal morphology.
(3)   No almorcé hoy.
pro not eat-1.sg-pret today
“I did not eat lunch today.”
a. Ellas piensan que tengo hambre ahora.
b. #pro3 piensan que tengo hambre ahora.
 (they) think-3.pl-pres that pro have-1.sg-pres hunger now
“They think I am hungry.”
(4) Man em ruz nahar nakhordam.
I today lunch neg-eat-1.sg.pret
“I did not eat lunch today.”

.  In Rothman (2009), the judgment of this sentence is listed as ungrammatical (*). Since
this sentence is entirely grammatical, the judgment has been changed to # to represent the
infelicity of the null subject in this context. The same is true of (4).
 Knowledge and processing of subject-related discourse properties 

a. Unha fekr mikonand ke goshneh basham.

b. #pro fekr mikonand ke goshneh basham.
 (they) thought dur-do.3.pl that pro hungry be-1.sg.pres
“They think that I am hungry.”

Finally, the subjects of the contexts and follow-up sentences in Examples (5)
and (6) below both refer to the same women: María and Hilda (5) and Shabnam
and Leila (6). Thus, the topic is maintained across the discourse rendering the
overt subject ellas/unha “they” less preferable than pro.
(5) María e Hilda no almorzaron hoy.
Mary and Hilda not eat-3.pl.pret today
“Mary and Hilda did not eat lunch today.”
a. #Ellas tendrán mucha hambre.
b. pro tendrán mucha hambre.
(they) have-3.pl.fut much hunger
“They must be hungry.”
(6) Shabnam va Leila em ruz nahar nakhordand.
Shabnam and Leila today lunch neg-eat-3.pl.pret
“Shabnam and Leila did not eat lunch today.”
a. #Unha bayad goshneh bashand.
b. pro bayad goshneh bashand.
(they) must hungry be-3.pl.pres
“They must be hungry.”

Table 1 below summarizes the discourse-constrained distribution of RSPs for

Spanish and Farsi.

Table 1.  Subject distribution preferences in Spanish and Farsi

Spanish Farsi

Contrastive Focus Overt Overt

Topic Shift Overt Overt
Topic Maintenance Null Null

That overt and null subject distribution is similar in Spanish and Farsi, two
null-subject languages, is not surprising.4 Nonetheless, since recent research

.  But, see Kissock (1995) for Telagu and Duarte (1993, 1995) for Brazilian Portuguese for
evidence showing rather distinct distributions of overt and null subject pronouns in these
null-subject languages.
 Tiffany Judy

­emonstrated that even typologically similar null-subject languages such as

d
­Catalan, Italian and Spanish may differ subtly with respect to overt subject resolu-
tion and subject distribution (Filiaci, Sorace, & Carreiras 2013; Prada Pérez 2009),
judgments of overt and null subjects were obtained from 39 native Farsi speakers
for the three contexts above. The data from Judy and Feizmohammadpour (2012)
demonstrate a similar distributional pattern in that the Farsi speakers (1) rated far
more overt than null subjects as good in Contrastive Focus contexts (5.82 vs. 0.92
of 6), (2) rated more overt than null subjects in Topic Shift contexts as good (5.69 vs.
4.31 of 6) and (3) rated far more null subjects in Topic Maintenance contexts than
overt as good (5.77 vs. 2 of 6). Demonstrating this similarity between the languages
is important for the goals of this study since the IH (Sorace 2011) claims that inter-
face vulnerability obtains from the division of finite cognitive resources caused by
bilingualism (specifically, the presence of two grammars). The testable prediction
that falls out from this claim is that similarities between languages do not safe-
guard L2 speakers from processing differences. Consequently, the IH predicts that
processing differences regarding RSP distribution will obtain in the near-native L2
Spanish speakers tested herein despite similar L1 distribution.

3.  Previous acquisition studies

Subject distribution is perhaps the most widely studied linguistic phenomena

in child monolingual and bilingual studies and adult L2 studies. Select, yet rel-
evant, studies from this research are briefly examined below. First, Grinstead
(2004) found that monolingual Catalan and Spanish children (1;6–2;6 years) used
overt subjects later than children acquiring non-null-subject languages. In line
with the IH, these differences are attributed to difficulties at the interface between
syntax and discourse-pragmatics, not grammatical deficits. In older monolingual
Spanish-speaking children (mean age 6;8 to 14;7 years), Lapidus Shin and Cairns
(2011) examined sensitivity to Topic Shift and Topic Maintenance with third per-
son singular subjects. Younger participants (~7 years) did not statistically distin-
guish between use of overt subject pronouns in Topic Shift and Topic Maintenance
contexts (51% vs. 41% suppliance of overt subjects), while slightly older children
(~9 years) and adults did (64% vs. 51% and 83% vs. 27% suppliance, respectively).
Studies examining bilingual children have also found divergent uses com-
pared to child monolinguals. Paradis and Navarro (2003) found overuse5 (~35%)

.  But see Liceras, Fernández Fuertes and Alba de la Fuerte (2012) for results suggesting that
the overuse seen in Paradis and Navarro (2003) results from the input.
 Knowledge and processing of subject-related discourse properties 

of overt subjects (including subject pronouns) in an English-Spanish bilingual

(1;8–2;7 years) as compared to two child monolinguals (~17% and 20%). Simi-
larly, in a study testing subject distribution in Topic Shift contexts, Sorace et al.
(2009) found overuse of overt subjects in two groups of English-Italian and one
group of Spanish-Italian bilingual children (6–7 years and 8–10 years). More overt
subjects were accepted in contexts preferring null subjects by both the younger
monolingual Italian (~31% overt) and bilingual groups (~35% Italy English-
Italian, ~60% UK English-Italian and ~27% Spanish-Italian) as compared to the
older monolinguals (~14%), bilinguals (~27% Italy English-Italian, and ~22%
UK English-Italian) and adults (~12%). The same finding obtained for the older
Spanish-Italian bilingual children.6 Since overuse of overt subjects obtained in
monolingual Italian and bilingual Spanish-Italian children, Sorace et al. (2009)
concluded that the overuse is part of development and that overt subjects may be
the default. Taken together, the child monolingual and bilingual studies demon-
strate that convergence on subject distribution, an interface-conditioned property,
is somewhat delayed even when children have converged on the related syntactic
property. Therefore, and connected to the adult acquisition studies that follow,
that adults should evidence a similar developmental sequence is not surprising. As
Sorace argues (2011, 2012), near-native speakers must therefore be tested to avoid
this developmental delay.
Research on adult bilinguals has shown both divergence and convergence
for subject distribution and related properties. For example, Belletti and Leonini
(2004) examined convergence on VS word order and null subject use in adult L2
Italian speakers  of varying L1s (i.e. non-null and null-subject languages). The
authors found that proficiency and L1 affected subject distribution: lower profi-
ciency speakers and those with non-null subject L1s used more overt subjects.7
Proficiency was also found to play a role in overt subject production in the L1
English learners of L2 Spanish in Montrul and Louro-Rodríguez (2006), the L1
Greek learners of L2 Spanish in Margaza and Bel (2006) and the heritage Spanish
speakers of Montrul (2004). That is, like the younger monolinguals and bilinguals
discussed above, lower proficiency speakers overproduced overt subjects in con-
texts preferring null subjects. Nevertheless, Belletti et al. (2007) also found overt
subject overuse in near-native L2 Italian speakers (L1 English). Interpretive differ-
ences regarding overt subjects have also been found as in Sorace and Filiaci (2006)

.  See Bini (1993) for similar results in Spanish-Italian adult bilinguals.
.  Differently from the majority of studies examining subject use, Tsimpli and Sorace (2006)
found overuse of null subjects across three proficiency groups of L1 Russian naturalistic
learners of L2 Greek.
 Tiffany Judy

where the L1 English, near-native Italian speakers interpreted overt subjects to a

significantly higher degree than the control group as coreferencing with the matrix
clause subject, but matched the control group regarding null subjects.
Proficiency effects were also found in studies finding convergence, like
­Rothman (2008) and Rothman (2009), which investigated L1 English speakers of
L2 Spanish. The intermediate proficiency group of Rothman (2008) demonstrated
syntactic knowledge of subjects, but did not converge on subject distribution:
they rated overt subjects favorably where null subjects were preferred in the judg-
ment task while they underused overt subjects in the production task. Advanced
proficiency learners, however, demonstrated convergence on the syntactic and
­discourse-distribution of subjects. Similar findings obtained in Rothman (2009) in
that only the advanced group showed knowledge of related syntax-semantics and
syntax-discourse properties. While 28 of 38 intermediate proficiency participants
of the study showed knowledge of the syntax-semantics property tested therein,
they accepted and used more overt and null subjects than the control group, indi-
cating that L2 speakers converge on external interface-conditioned properties
later, but that convergence is possible. Finally, Gürel (2006) showed that L1 English
learners of L2 Turkish (presumed near-native speakers since they lived in ­Turkey
for at least 10 years) converged on the pragmatic use of overt and null subjects.
Overall, the experimental studies examined above provide some evidence sup-
porting the IH in that external interface-conditioned properties were subject to
divergence in child language acquisition and in some adult L2 studies. However, the
three studies outlined in the previous paragraph reported convergence on external
interface-conditioned properties, providing counterevidence. Importantly, both
bodies of research indicate that syntax is not the source of delay or divergence on
external interface-conditioned properties. This finding is in line with the current
state of the IH which claims that syntax is not inevitably vulnerable. The studies
also point to a positive correlation between proficiency and convergence on exter-
nal interface-conditioned properties. Still, none of the studies described above is
able to test the IH’s most recent claims for two reasons. First, the claims are applica-
ble to near-native speakers only, yet only two studies (Gürel 2006; Sorace & Filiaci
2006) tested near-native speakers. Second, the IH predicts processing differences,
yet only offline methods were used. The current methodological design remedies
these limitations in addition to making it possible to test the claim that processing
differences arise despite parametric similarities between the grammars.

4.  Experimental methodology

This study examines the explicit judgments and processing of the discourse-­
constrained distribution of RSPs in Spanish. As described in Section 2, while both
 Knowledge and processing of subject-related discourse properties 

overt and null subjects are permissible in Spanish, their distribution is conditioned
upon discursive factors such as focus and topic shift/maintenance. The methodol-
ogy was designed with the following IH-inspired research questions in mind:

1. Do near-native L2 speakers converge on the discourse-constrained distribu-

tion of RSPs in Spanish in the offline task?
2. Do near-native L2 speakers exhibit processing differences as compared to the
native speakers in the online task?

Recall from the introduction that the IH claims that processing differences will
obtain between native and near-native speakers due to the division of finite cogni-
tive resources that results from simultaneous inhibition and activation of bilin-
guals’ grammars. Thus, in response to research question (2), the IH predicts
that processing differences will obtain between the groups. However, regarding
research question (1), since offline tasks do not directly measure processing, the
IH allows for convergence on these same properties when tested offline (although
it does not preclude differences).

4.1  P
 articipants
Twenty-four native Spanish speakers and 8 near-native8 L2 Spanish speakers par-
ticipated in this study. The native speakers were born in Argentina to native River
Plate Spanish-speaking parents and were exposed to this dialect since birth, thus
controlling for dialectal variation. Most native speakers reported knowledge of an
L2 (largely English or French), yet none reported neither being particularly profi-
cient nor using the L2 frequently. All native speakers received formal instruction9
about Spanish. At the time of testing, the average age of the native speakers was
29.3 years (range 20–62 years). Differently, the native Farsi-speaking adults were
born in Iran to native Farsi-speaking parents. They were first exposed to Spanish
upon immigrating to Argentina in the late 1970’s, the average age of first expo-
sure being 20.5 years (range 17–26 years). Only three participants reported receiv-
ing formal instruction regarding Spanish, while the other five reported learning
naturalistically. Those that attended Spanish classes reported learning more
from their environment than from classes as the instruction was short-lived and

.  Classification as a near-native speaker and participation in the study was conditioned
upon scoring a minimum of 45/50 on an abbreviated version of the Diploma del español como
lengua extranjera (DELE 2002) that is widely used in the US (see work by Montrul, Slabakova
and White, for example) and is comprised of a vocabulary and a cloze section.
.  All received formal instruction in school and some continued to study the Spanish lan-
guage at university.
 Tiffany Judy

­ uantitatively insignificant. While all bilingual participants are fully integrated

q
into Argentine society and use Spanish for most communication, they maintain
the L1. In fact, 5 participants self-reported being Farsi-dominant, despite having
spent over 30 years in Argentina, a non-Farsi-speaking environment. These facts
highlight the constant activation and inhibition of the bilingual’s languages, an
important point. The near-native speakers’ average age at the time of testing was
51.8 years (range 43–58 years). Finally, all participants had normal or corrected-
to-normal vision and no linguistic or reading impairment.

4.2  Context-matching felicitousness task (CMFT)

The purpose of the CMFT was to examine the judgments of overt and null subjects
in three discourse-constrained contexts. This task contained 36 tokens (n  =  12
Contrastive Focus, n = 12 Topic Shift, n = 12 Topic Maintenance). Participants
read brief contexts followed by a short target sentence that they judged based on
how well it was expressed given the context. Participants rated the target sentence
from 1 (100% bien “good”) to 4 (100% mal “bad”). The token types, with examples,
are described below.
The first token type, Contrastive Focus, made a contrast between two sub-
jects. Thus, all Contrastive Focus tokens necessarily introduced two subjects in
the context. Each context began with an adverbial phrase such as Cuando salimos
a cenar “When we go out for dinner” and the two subjects employed were always
a third person singular feminine DP, such as mi novia “my girlfriend” and the first
person singular pronoun yo “I.” Half the Contrastive Focus tokens contained overt
pronouns (7a) and half contained null pronouns (7b). Recall from Section 2 that
overt pronouns are preferred over null pronouns in Contrastive Focus contexts as
indicated by # in (7b).
(7) Cuando salimos a cenar, mi novia prefiere comer platos livianos, pero yo
prefiero comer algo sustancioso.
“When we go out to eat, my girlfriend prefers to eat light dishes, but I prefer
to eat something of substance.”
a. Así que ella come ensaladas y yo como milanesas en los restaurantes.
b. #Así que pro come ensaladas y pro como milanesas en los restaurantes.10
“So, she eats salads and I eat breaded meats in restaurants.”

Similarly, all Topic Shift contexts introduced two subjects such that a shift in ref-
erence could be made. The third person singular feminine subject was always

.  pro appears for expository purposes only and was not present in the actual tokens. This
is true of all other tokens.
 Knowledge and processing of subject-related discourse properties 

i­ ntroduced in the first sentence of the context. Then, the first person singular sub-
ject pronoun was introduced in the second and any subsequent sentences. Mini-
mally, three verbs conjugated for first person singular separated the first sentence
of the context and the target sentence, reinforcing the shift. Half the Topic Shift
tokens contained overt pronouns (8a) while the other half contained null pro-
nouns (8b). Like Contrastive Focus contexts, overt pronouns are preferred11 in
Topic Shift contexts.

(8) Mi hija quiere ser autora y no tiene otros intereses. Yo creo que es mejor
tener varios intereses y sugiero otras actividades, pero no importa lo que
diga yo.
“My daughter wants to be an author and she has no other interests. I think
that it is best to have various interests and I suggest other activities, but it
doesn’t matter what I say.”
a. Finalmente ella escribe cuentos y pasa todo el día en su cuarto.
b. #Finalmente escribe cuentos y pasa todo el día en su cuarto.
“In the end, she writes stories and she spends the whole day in her
room.”

Differently from the two previous contexts, Topic Maintenance contexts introduced
only one third person singular DP feminine subject (i.e. mi hija “my ­daughter”)
which was initially introduced in the first sentence of the context. From there, the
context contained a minimum of three verbs conjugated for third person singular
with null subjects. Half the Topic Shift tokens contained overt pronouns (9a) and
half contained null pronouns (9b). Recall from Section 2 that null pronouns are
preferred in Topic Maintenance contexts.

(9) Mi cuñada es muy sociable. Tiene muchos amigos y por eso va a muchas
cenas a la canasta donde tiene que contribuir con algo.
“My daughter-in-law is very social. She has a lot of friends and for that
­reasons, she goes to a lot of potluck dinners where she has to share
s­ omething.”
a. #Así que ella lleva postres y comparte todo con sus amigos.
b. Así que lleva postres y comparte todo con sus amigos.
“So, she takes desserts and shares everything with her friends.”

.  While preferred, more variation has been shown with Topic Shift and Topic Maintenance
than Contrastive Focus (e.g. Bentivolgio 1987; Enríquez 1984; Filiaci et al., 2013; Lapidus
Shin & Cairns 2011; Lubbers Quesada & Blackwell 2009; Otheguy, Zentella, & Livert 2007;
Prada Pérez 2009; Silva-Corvalán 1994).
 Tiffany Judy

In addition to the care taken with the creation of each context type described
above, all subjects, matrix clause verbs and direct objects (DO) were counter-
balanced across token type and subject type to allow for maximal structural
similarity. This is especially important for the online version of this task in that
statistical comparisons of the Reaction Times (RT) to the regions of interest
(matrix clause verb and DOs) are made. Twelve matrix clause transitive verbs
(come “eat,” hace “make,” prepara “prepare,” toma “drink,” ofrece “offer,” lleva
“take,” vende “sell,” escribe “write,” lee “read,” compra “buy,” pinta “paint,” dibuja
“draw”) and 12 matrix clause DOs (ensaladas “salads,” alfajores “cookies with
dulce de leche,” empanadas “savory pastries,” agua “water,” café “coffee,” postres
“desserts,” revistas “magazines,” cuentos “stories,” poemas “poems,” esculturas
“sculptures,” paisajes “landscapes,” figuras “figures”) were distributed evenly
across the token and subject types. Finally, the offline task was taken after the
online task to avoid priming.

4.3  Self-paced reading task (SPRT)

The SPRT (Aaronson & Scarborough 1976; Mitchell & Green 1978) tested par-
ticipants’ processing of RSP distribution in the same discourse-constrained con-
texts described in Section 4.2.12 The purpose of this task, which also contained
36 tokens (n=12 Contrastive Focus, n = 12 Topic Shift, n = 12 Topic Maintenance)
and employed E-prime software (Schneider, Eschman & Zuccolotto 2002), was to
determine if the L2 group’s processing was different from that of the native group.
A non-cumulative moving window technique13 (Just, Woolley & C ­ arpenter 1982)
was employed to mimic normal reading as closely as possible and to m ­ easure

.  The 36 contexts employed in the offline task were employed in the online task. They were
counterbalanced such that participants saw each context once with a target sentence with an
overt subject and once with a null subject.
.  This presentational format begins with a blank screen containing several underscored
sentence items whose content becomes visible as the participant presses a button to advance
along word-by-word or region-by-region (i.e. in a non-cumulative and linear fashion as
opposed to the centered technique). Example (i) provides a short English example of this tech-
nique. Each line represents the visual field the participant would see as they progress through
the sentence, which starts with all regions hidden and ends with the final word.
(i) a. ___ ___ ___ _ ____
b. The ___ ___ _ ____
c. ___ little ___ _ ____
d. ___ ___ boy _ ____
e. ___ ___ ___ is ____
f. ___ ___ ___ _ tired.
 Knowledge and processing of subject-related discourse properties 

RT for each region of interest. After having read the entire target s­entence,
­participants answered a short meaning-based yes/no comprehension question
designed to focus their attention on the content of the experimental stimuli and
not its felicitousness.
As in the CMFT, participants read short contexts shown in their entirety fol-
lowed by a target sentence presented via the moving window task described in
footnote 13 and then answered a comprehension question. They were instructed
to read the contexts and target sentences at a normal rate and respond to the com-
prehension questions as quickly and accurately as possible. Prior to the experi-
mental task, participants completed a training session. Then, the actual task began
with two additional non-experimental tokens, after which the 36 experimental
tokens were presented randomly. The token types and their respective division
according to regions of interest are described below.
All token types were purposefully and consistently divided to control for pre-
sentation in E-prime and to allow for statistical comparisons across tokens. Par-
ticipants first read the context in its entirety, as in (10), and then read the target
sentence (either (10a) or (10b)) region by region. The forward slashes (/) indicate
a division in the presentation of the token in E-prime and are used for expository
purposes only. The regions of interest for both Contrastive Focus with overt sub-
ject Examples (10a) and null subject Examples (10b) that are analyzed in Section 5
are the verb come “eats” and the DO ensaladas “salads.”
(10) Cuando salimos a cenar, mi novia prefiere comer platos livianos, pero yo
prefiero comer algo sustancioso.
“When we go out to eat, my girlfriend prefers to eat light dishes, but I prefer
to eat something of substance.”
a. Así que/ella/come/ensaladas/y/yo/como/milanesas/en los restaurantes.
b. #Así que/come/ensaladas/y/como/milanesas/en los restaurantes.
“So, she eats salads and I eat breaded meats in restaurants.”
(Comprehension Question) ¿Mi novia come ensaladas?
 “Does my girlfriend eat salads?”

Example (11) shows a Topic Shift token. The regions of interest for overt subject
(11a) and null subject tokens (11b) alike are the verb escribe “writes” and the DO
cuentos “stories.”

(11) Mi hija quiere ser autora y no tiene otros intereses. Yo creo que es mejor
tener varios intereses y sugiero otras actividades, pero no importa lo que
diga yo.
“My daughter wants to be an author and she has no other interests. I think
that it is best to have various interests and I suggest other activities, but it
doesn’t matter what I say.”
 Tiffany Judy

a. Finalmente/ella/escribe/cuentos/y/pasa/todo/el día/en su cuarto.

b. #Finalmente/escribe/cuentos/y/pasa/todo el día/en su cuarto.
“In the end, she writes stories and she spends the whole day in her
room.”
(Comprehension Question) ¿Mi hija pasa todo el día en la biblioteca?
 “Does my daughter spend the whole day in the library?”

Lastly, the regions of interest examined in Topic Maintenance overt subject (12a)
and null subject tokens (12b) are the verb lleva “takes” and the DO postres “des-
serts” of the matrix clause.

(12) Mi cuñada es muy sociable. Tiene muchos amigos y por eso va a muchas
cenas a la canasta donde tiene que contribuir con algo.
“My daughter-in-law is very social. She has a lot of friends and for that
reasons, she goes to a lot of potluck dinners where she has to share
­something.”
a. #Así que/ella/lleva/postres/y/comparte/todo/con sus amigos.
b. Así que/lleva/postres/y/comparte/todo/con sus amigos.
“So, she takes desserts and shares everything with her friends.”
(Comprehension Question) ¿Mi cuñada tiene muchos amigos?
 “Does my daughter-in-law have a lot of friends?”

As described at the close of Section 4.2, all subjects, matrix clause verbs and DOs
were counterbalanced across token and subject type. The online task was com-
pleted before the offline task to avoid potential priming.

5.  Results

5.1  Context-matching felicitousness task

The contingency table below shows that the participants’ ratings from the offline
task largely corresponded to the ends of the scale (either 1 (=100% good) or 4
(=100% bad)). As such, ratings of 1 and 2 have been combined as “good” and
ratings of 3 and 4 have been combined as “bad” for all token types. The statistical
analyses described below are based on this binary distinction.
The group average “good” ratings for the Contrastive Focus with overt (CFO)
and Contrastive Focus with null (CFN) tokens are shown in Figure 1 below. Both
the native and L2 groups rated a high number of CFO tokens as “good” (5.92 and
5.75 out of 6, respectively) while also rating considerably fewer CFN tokens as
“good” (0.71 and 0.25, respectively). Importantly, all participants rated more CFO
 Knowledge and processing of subject-related discourse properties 

Table 2.  Distribution of ratings

Rating

1 = 100% good 2 = more or less good 3 = more or less bad 4 = 100% bad

NS 60.4% 9.4%    9.6%  20.6%

L2 70.5% 6.3% 10.1% 13.2%

than CFN tokens as “good”. These results are analyzed and interpreted at the con-
clusion of Section 5.1.

6
Group average "good" rating (N=6)

0
CFO #CFN

NS 5.92 0.71

L2 5.75 0.25

Figure 1.  CMFT group results: Contrastive Focus. CFO = Contrastive Focus token with overt
subject; CFN = Contrastive Focus token with null subject

Figure 2 below shows the group averages for Topic Shift tokens. For both the
native and L2 groups, the average number of TSO tokens rated as “good” was quite
high (5.54 and 5.75 out of 6, respectively). A higher number of TSN tokens, as
compared to the CFN tokens, was rated as “good” across both groups (2.75 and
4.38, respectively). All but one native and two L2 speakers rated more TSO than
TSN tokens as “good”.
Lastly, Figure 3 below shows the group averages for Topic Maintenance tokens.
Both the native and L2 group’s average number of “good” ratings was rather high
for TMO (4.58 and 5.50 out of 6, respectively) and TMN tokens (5.54 and 6.00,
 Tiffany Judy

Group average "good" rating (N=6) 6

0
TSO #TSN

NS 5.54 2.75

L2 5.75 4.38

Figure 2.  CMFT group results: Topic Shift. TSO = Topic Shift token with overt subject;
TSN = Topic Shift token with null subject

6
Group average "good" rating (N=6)

0
#TMO TMN

NS 4.58 5.54

L2 5.50 6.00

Figure 3.  CMFT group results: Topic Maintenance. TMO = Topic Maintenance token with
overt subject; TMN = Topic Maintenance token with null subject
 Knowledge and processing of subject-related discourse properties 

respectively). Only 11 of 24 native and 3 of 8 L2 speakers rated more TMN than

TMO tokens as “good”. Of the 13 native speakers that did not make the expected
distinction, 11 rated an equal number of TMO and TMN tokens as “good”. The five
L2 speakers that did not make the expected distinction all rated an equal number
of TMO and TMN tokens as “good.”
A Repeated Measures ANOVA with Bonferroni post-hoc analyses (p = .05)
was conducted for the CMFT with Subject Type (overt or null), Context Type
(Contrastive Focus, Topic Shift and Topic Maintenance) and Group (native or
L2 speaker) as factors. Significant main effects for Subject Type (F(1, 1140)
5.397, p = .020), Context Type (F(2, 1140) 89.487, p < .001) and Group (F(1, 25)
37.641, p  <  .001) were found. Interactions between Subject Type and Con-
text Type (F(2, 1140) 181.232, p < .001), Subject Type and Group (F(1, 1140)
35.774, p < .001) and Context Type and Group (F(2, 1140) 58.946, p < .001)
were found. Finally, a three-way interaction between Subject Type, Context
Type and Group obtained (F(2, 1140) 32.776, p < .001). For reasons of space
and the goals of this chapter, the interaction between Subject Type and Group
will not be discussed.
Regarding the interaction between Subject Type and Context Type, the
­B onferroni post-hoc pairwise t-tests showed that both the native and the L2
group rated CFO tokens significantly better (i.e. more “good” ratings) than CFN
tokens (p < .001). Similarly, both groups rated significantly more TSO than TSN
tokens as “good” (p < .001), while rating significantly more TMN than TMO
tokens as “good” (p = .012 and p = .044). These results are expected since overt
subjects are preferred in Contrastive Focus and Topic Shift contexts, while null
subjects are preferred in Topic Maintenance contexts. For the Context Type
and Group interaction, the Bonferroni post-hoc pairwise t-tests showed that
the native and L2 groups rated both CFO and CFN tokens to the same degree
(p = .312 and p = .148, respectively). These findings are expected if the L2 group
has converged on the distribution of subject pronouns in Contrastive Focus con-
texts in Spanish. With respect to the Topic Shift tokens, no difference was found
between the native and L2 group’s rating of TSO tokens (p = .492). However,
the L2 group rated significantly more TSN tokens as “good” as compared to the
native group (p  =  .011). Finally, L2 group rated significantly more TMO and
TMN tokens as “good” than the native group (p = .037 and p = .009, respec-
tively), which is unexpected if they have converged on the distribution of sub-
jects in Topic Maintenance contexts.
In summary, both the native and L2 group treated overt and null subjects
differently in each context type. While the native and L2 groups rated CFO, CFN
and TSO tokens to the same degree, they differed with respect to TSN, TMO and
TMN tokens.
 Tiffany Judy

5.2  Self-paced reading task (SPRT)

RT data from the SPRT was treated as follows. First, data points corresponding to
incorrect answers to comprehension questions were excluded. Second, RT data
was trimmed such that data points slower than 2 seconds14 were excluded. Finally,
the mean RT and standard deviation for each region of interest were calculated
across the SPRT. Data points higher than 2 standard deviations from the mean
were replaced with the cutoff value (mean RT + (standard deviation X 2)).15 For
Regions 2 and 3, respectively, Tables 4 and 5 below provide the average RTs for
each token type per participant, where S and F in the first column represent native
Spanish and native Farsi speakers, respectively. Per participant, the average RT to
each token type is provided under its label (#CFN, CFO, etc.) and the differential
RT (calculated by subtracting each participant’s average RT to the felicitous token
(for Contrastive Focus contexts, CFO tokens) from the average RT to the infe-
licitous token (#CFN)) is provided thereafter in the column labeled “Dif.”. Since
the RT to infelicitous tokens is expected to be slower, positive numbers should
obtain if participants detect infelicity. Concerning both regions and each con-
text, it is clear that some variation is observed. While 16 native speakers make the
expected distinction between #TSN and TSO tokens in Region 2, less than half
do so for #CFN and CFO (10 participants) or #TMO and TMN tokens (7 par-
ticipants). Nonetheless, upon examining both regions, it is clear that the majority
of participants appropriately distinguish between the tokens in at least one of the
two regions: 15 participants for #CFN-CFO, 18 for #TSN-TSO and 17 for #TMO-
TMN. Moreover, 9 native speakers distinguished between the tokens in all three
contexts in at least one region. Together, these findings point to the validity of
the experimental tokens. The same is true of the L2 participants: 5 distinguished
between the tokens in all three contexts in at least one of the regions.16
Another observation is that, across all contexts and regions of interest, the
native group average RT was faster than that of the L2 group. In fact, a simple
analysis of overall RT between the groups across the contexts and regions of inter-
est revealed many significant differences.

.  This data, which constituted a mere 0.313%, was excluded on the grounds that such high
RTs (i.e. slower than 2 seconds) likely represent external distractions.
.  See Hopp (2007, p. 224) for similar data treatment. 3.4% and 3.7% of data was replaced
for Regions 2 and 3, respectively.
.  Participants F2, F3 and F7 did not distinguish between #TMO and TMN tokens in either
region, but did so in the other context types.
 Knowledge and processing of subject-related discourse properties 

Table 4.  SPRT individual RTs to Region 2 (verb)17

Part. #CFN CFO Dif. #TSN TSO Dif. #TMO TMN Dif.

S3 278 281 -3 273 272 1 267 258 8

S4 254 259 -5 264 255 10 234 259 -25
S5 284 422 -13717 419 297 122 324 360 -36
S6 286 329 -44 329 311 19 362 288 74
S7 323 341 -18 379 269 110 259 237 22
S8 280 352 -72 302 317 -15 270 276 -5
S9 350 288 62 303 285 18 322 389 -67
S10 366 341 25 382 320 63 302 318 -15
S11 286 302 -16 325 302 23 277 397 -120
S12 258 229 29 237 228 9 232 257 -26
S13 370 351 19 401 268 132 484 362 122
S14 329 346 -17 293 411 -118 258 342 -84
S15 345 503 -157 519 537 -18 428 456 -27
S16 308 312 -4 286 315 -29 304 315 -11
S17 345 367 -22 488 463 26 394 418 -24
S18 335 321 14 347 327 20 320 327 -8
S19 191 191 0 273 274 -1 281 242 39
S20 354 374 -21 367 377 -9 378 340 38
S21 350 312 38 367 365 3 326 340 -14
S22 323 306 17 354 395 -41 339 344 -5
S23 398 395 4 410 405 4 377 395 -17
S24 239 234 5 379 231 148 208 215 -7
S25 323 307 17 370 270 100 265 300 -35
S26 215 216 -1 261 323 -63 367 292 75
Ave. 308 320 -7 347 326 21 316 322 -6

F1 609 648 -40 667 610 57 565 621 56

F2 480 494 -15 526 537 11 541 471 -70
F3 571 487 85 530 484 46 418 617 -199
F4 400 590 -191 645 599 46 609 505 104

(Continued)

.  All values in Tables 4 and 5 are rounded to whole numbers (e.g. original values for native
speaker #5 were 284.25ms (#CFN)-421.60ms (CFO) = 137.35ms.)
 Tiffany Judy

Table 4.  SPRT individual RTs to Region 2 (verb)  (Continued)

Part. #CFN CFO Dif. #TSN TSO Dif. #TMO TMN Dif.
F5 464 390 74 495 357 138 482 444 38
F6 487 417 70 434 416 19 425 416 9
F7 466 618 -152 607 398 209 500 559 -59
F8 662 529 133 387 674 -287 636 639 -3
Ave. 517 522 -5 536 509 30 520 534 -16

However, this comparison only shows that the native group average RT is faster
overall. It does not provide information regarding the groups’ detection of infelic-
ity, which is demonstrated by slower RTs to infelicitous versus felicitous tokens as
described above concerning the “Dif.” column. Thus, a comparison between the
RTs to overt and null tokens within each context type was performed to determine
if RTs to one subject type was comparatively slower than the other.

Table 5.  SPRT individual RTs to Region 3 (DO)

Part. #CFN CFO Dif. #TSN TSO Dif. #TMO TMN Dif.

S3 271 285 -14 268 282 -14 269 248 21

S4 256 347 -91 333 292 41 305 325 -20
S5 367 480 -113 317 353 -35 282 451 -169
S6 310 441 -131 350 354 -4 329 355 -26
S7 451 378 74 424 502 -78 416 539 -123
S8 279 292 -12 317 344 -27 303 270 33
S9 277 282 -6 298 339 -42 322 286 36
S10 351 326 25 389 354 35 413 335 78
S11 304 431 -127 310 338 -27 304 371 -67
S12 264 284 -20 249 229 20 229 240 -11
S13 457 530 -73 351 311 40 519 359 160
S14 348 343 5 273 368 -95 274 303 -30
S15 494 501 -7 439 573 -135 612 459 153
S16 386 345 41 313 319 -6 332 324 8
S17 393 404 -11 444 516 -71 491 482 9
S18 346 373 -27 340 339 2 403 332 71
S19 273 253 21 278 251 27 220 257 -37
S20 379 373 7 372 387 -15 359 361 -2

(Continued)
 Knowledge and processing of subject-related discourse properties 

Table 5.  (Continued)

Part. #CFN CFO Dif. #TSN TSO Dif. #TMO TMN Dif.
S21 564 465 99 336 435 -99 446 307 139
S22 351 360 -9 423 391 31 346 346 0
S23 449 460 -11 414 394 20 388 541 -153
S24 197 245 -48 270 279 -9 214 284 -70
S25 343 332 10 398 420 -22 328 306 22
S26 396 462 -65 349 476 -128 420 432 -12
Ave. 354 375 -20 344 369 -25 355 355 0

F1 1065 949 116 995 828 167 892 1081 -189

F2 680 644 35 563 588 -25 527 567 -39
F3 804 784 21 778 579 199 531 590 -59
F4 1171 969 202 852 842 10 776 967 -191
F5 651 589 63 446 463 -16 713 396 317
F6 512 558 -46 645 542 103 642 405 237
F7 741 649 92 641 542 98 663 725 -62
F8 718 672 46 831 522 309 849 560 289
Ave. 793 727 66 719 613 106 699 661 38

This is referred to as the average differential RT, which is the difference

between the average RT to infelicitous tokens minus the average RT to felicitous
tokens within each context type. The average differential RT is computed for each
group for Regions 2 and 3.18

5.2.1  Contrastive focus

Figure 4 below shows the average differential RT for Regions 2 and 3 for the CFN
versus CFO tokens repeated below in (13). The regions of interest are bolded.
(13) Cuando salimos a cenar, mi novia prefiere comer platos livianos, pero yo
prefiero comer algo sustancioso.
“When we go out to eat, my girlfriend prefers to eat light dishes, but I prefer
to eat something of substance.”
a. Así que/ella/come/ensaladas/y/yo/como/milanesas/en los restaurantes.
b. #Así que/come/ensaladas/y/como/milanesas/en los restaurantes.
“So, she eats salads and I eat breaded meats in restaurants.”

.  No comparison can be made for the first region of interest, the subject ella “she,” since it is
necessarily present in the overt subject tokens but necessarily absent in the null subject tokens.
 Tiffany Judy

120
100
80
60
Difference in 40
Average RT
(milliseconds) 20
0
–20
–40
–60
Region 2 Region 3
#CFN-CFO
NS –12 –20
L2 –4 66

Figure 4.  SPRT comparative group results (verb and DO) with S.E. bars. CFN-CFO = group
average RT to CFN tokens minus group average RT to CFO tokens

The average differential RT was calculated by subtracting the group average

RT to the felicitous CFO from the group average RT to the infelicitous CFN token
type. For Region 2 (verb), it is clear that this equation resulted in a negative aver-
age differential RT for both the native and L2 groups. The native group difference
in average RT for CFN-CFO was greater than that of the L2 group (-11.94ms
versus -4.37ms, respectively). The negative result indicates that, on average, the
effect of detecting infelicitousness (i.e. slower RTs) is not evidenced in Region 2. If
it were, the result would be positive since the first element of the equation, CFN,
would produce a slower RT compared to the second element, CFO. Regarding
Region 3 (DO), an interesting result obtains in that the average differential RT for
the native group is negative (-20.24ms), but positive (66.10ms) for the L2 group
(see Figure 4). The positive result is expected if the participants detect the infelici-
tous nature of null subjects in Contrastive Focus contexts; thus, it seems that L2
group detects the infelicity of the null subject.
The independent samples t-test showed no statistically significant differ-
ence for Region 2 (t(30) = -.177, p = .864). This suggests that the L2 group’s
treatment of the Contrastive Focus tokens matched the native group’s treat-
ment. However, a statistically significant difference was found for Region  3
(t(30) = -3.431, p = .002).

5.2.2  Topic shift

Next, the average differential RTs to Topic Shift tokens is examined. An example is
repeated below in (14) with the regions of interest bolded.
 Knowledge and processing of subject-related discourse properties 

(14) Mi hija quiere ser autora y no tiene otros intereses. Yo creo que es mejor
tener varios intereses y sugiero otras actividades, pero no importa lo que
diga yo.
“My daughter wants to be an author and she has no other interests. I think
that it is best to have various interests and I suggest other activities, but it
doesn’t matter what I say.”
a. Finalmente/ella/escribe/cuentos/y/ella/pasa/todo/el día/en su cuarto.
b. #Finalmente/escribe/cuentos/y/pasa/todo el día/en su cuarto.
“In the end, she writes stories and she spends the whole day in her
room.”

160
140
120
100
Difference in 80
60
Average RT
40
(milliseconds)
20
0
-20
-40
-60
Region 2 Region 3
# TSN - TSO
NS 21 –25
L2 27 106

Figure 5.  SPRT comparative group results (verb and DO) with S.E. bars. TSN-TSO = group
average RT to TSN tokens minus group average RT to TSO tokens

Again, the average differential RT was calculated by subtracting the group

average RT to the felicitous TSO token from the group average RT to the infelici-
tous TSN token type for Regions 2 and 3. Figure 5 shows that for Region 2 (verb), a
positive average differential RT for both the native and L2 group obtained (21.39ms
and 27.04ms, respectively), which indicates that, on average, the effect of detecting
infelicitousness (i.e. slower RTs) is evidenced here. Figure 5 also shows the average
differential RT to Region 3 (DO). Similarly to what was seen for Region 3 of the
Contrastive Focus tokens, the average differential RT for the native group is nega-
tive (-24.59ms), but it is positive for the L2 group (105.58ms). This latter finding
is expected if the participants detect the infelicitous nature of the TSN tokens,
indicating that the L2 group demonstrates knowledge of the infelicitousness of the
null subject in Region 3.
 Tiffany Judy

The independent samples t-test revealed no statistically significant difference

in average differential RT between the two groups for Region 2 (t(30) = -.155,
p = .878), indicating that the distinction the L2 group made between the TSO and
TSN tokens matched that of the native group. For Region 3, however, a statistically
significant difference obtained (t(30) = -3.059, p = .016).

5.2.3  Topic maintenance

Finally, this section presents a comparison of the RTs to felicitous and infelicitous
Topic Maintenance tokens, an example of which is repeated below in (15) with the
second and third region of interests bolded.
(15) Mi cuñada es muy sociable. Tiene muchos amigos y por eso va a muchas
cenas a la canasta donde tiene que contribuir con algo.
“My daughter-in-law is very social. She has a lot of friends and for that
­reasons, she goes to a lot of potluck dinners where she has to share
s­ omething.”
a. #Así que/ella/lleva/postres/y/comparte/todo/con sus amigos.
b. Así que/lleva/postres/y/comparte/todo/con sus amigos.
“So, she takes desserts and shares everything with her friends.”

80
60
40
Difference in
Average RT 20
(milliseconds) 0
-20
-40
-60
Region 2 Region 3
#TMO-TMN
NS -6 0
L2 -16 38

Figure 6.  SPRT comparative group results (verb and DO). TMO-TMN = group average RT to
TMO tokens minus group average RT to TMN tokens

Across both comparisons, slower RTs are expected for the overt subjects
since these are less preferred than null subjects in Topic Maintenance contexts.
­Figure 6 shows that the average differential RT for Region 2 (verb) is negative in
both groups (-6.20ms and -15.57ms, respectively). The negative result indicates
that, on average, the infelicitousness of TMO tokens was not evidenced in this
region. For Region 3 (DO), both groups demonstrated a positive group average
 Knowledge and processing of subject-related discourse properties 

differential RT (0.46ms for the native group and 37.84ms for the L2 group). This
finding is expected if the participants detect the infelicitous nature of the TMO
tokens. Thus, the positive result evidenced in both groups demonstrates that they
have detected this infelicity.
The independent samples t-test revealed no statistically significant differ-
ence in average differential RT for Region 2 (t(30) = -.360, p = .721). This finding
suggests that the distinction the L2 group made between TMN and TMO tokens
equals that of the native group. Likewise, no statistically significant difference
obtained for Region 3 (t(30) = -.490, p = .638).

6.  Discussion & conclusion

We now return to the research questions reproduced below as (3) and (4):

3. Do near-native L2 speakers converge on the discourse-constrained distribu-

tion of RSPs in Spanish in the offline task?
4. Do near-native L2 speakers exhibit processing differences with respect to the
discourse-constrained distribution of RSPs in the online task?

Regarding the first research questions, two statistical analyses were conducted
on the offline task data. The interaction between Subject Type and Context Type
described in Section 5.1 showed that both groups distinguish between overt versus
null subjects according to context. This is expected and provides evidence suggest-
ing that the L2 group demonstrates knowledge of discourse-constrained distribu-
tion of RSPs in Spanish. Nevertheless, somewhat mixed results were found via
the three-way interaction. No statistically significant differences obtained between
the groups for CFO, CFN or TSO tokens, but significant differences were found
for the TSN, TMO and TMN tokens, a finding in line with Sorace (2011) where it
is claimed that focus- and topic-related properties do not present equal difficulty
(topic-related properties are deemed more difficult). Returning to the L1 Farsi data
(Judy & Feizmohammadpour 2012) provided in Section 2, a potential L1-based
explanation regarding the higher number of “good” ratings for TSN tokens is
found: TSN tokens were accepted at higher rates by L1 Farsi speakers than the
current study’s L1 Spanish speakers in similar offline tasks. This is not to say that
the L2 group does not distinguish between TSO and TSN tokens, as Section 5.1
demonstrated that they do, but rather that the offline tolerance of null subjects in
Topic Shift contexts may be higher in Farsi. This echoes Filiaci et al. (2013) in that
subtle differences may obtain between null-subject languages, which could explain
this offline difference. Still, the high number of “good” ratings for the TMO tokens
 Tiffany Judy

does not have an L1-based explanation as the L1 Farsi speakers accepted only 2
of 6 TMO tokens (compared to 5.77 of 6 TMN tokens in ­Spanish). Nonetheless,
although significant differences obtained between the groups for half the token
types, these findings are not entirely problematic for the claims made herein.
Recall that null subjects are preferred in Topic Maintenance contexts and that they
constitute the default subject (i.e. are used when overt subjects are not required
for contrast or when referencing previously mentioned topics). Thus, using overt
subjects in Topic Maintenance contexts is perhaps less infelicitous than using null
subjects when overt subjects are required for contrastive focus of topic shift (i.e.
a question of redundancy versus lack of crucial information). If so, we should not
expect to see similar ratings of overt and null subjects in Contrastive Focus or
Topic Shift contexts. In fact, Figures 1–3 show that both the native and L2 groups’
ratings are more polarized in Contrastive Focus and Topic Shift contexts compared
to Topic Maintenance contexts. Further support for this claim is found in corpus
data which has shown subject use variation in Topic Maintenance contexts (see
Lapidus Shin & Carirns 2011 for review and sources cited therein). Finally, while
a significant difference obtained between the groups for TMN tokens, both groups
rated them as “good” (5.54 vs. 6.00, respectively) and both made the appropriate
distinction between TMN and TMO tokens, statistically preferring TMN. The sta-
tistical difference could be a product of the lack of variation in the L2 group: all L2
speakers rated 6 of 6 TMN tokens as “good”. While the offline data does not speak
to the L2 speakers’ processing of RSP distribution, they do provide some evidence
demonstrating convergence.
Regarding the second research question, the average differential reaction time
(RT) for the token types in the Self-Paced Reading Task (SPRT) were examined
to determine if processing differences arose between the groups. An unexpected
result regarding the native group obtained from the SPRT. Recall that, as a group,
the native speakers did not make the appropriate distinction between counter-
balanced CFO and CFN tokens or TMN and TMO tokens in Region 2 or 3 or in
Region 3 for TSO and TSN tokens. Since a distinction is predicted where infelicity
is detected, the fact that no group distinction obtained in 5 of 6 regions examined
might lead one to suggest that these tokens are not appropriate measures. How-
ever, two important points suggest otherwise. First, returning to the individual
data provided in Tables 4 and 5, across Regions 2 and 3, the majority of native
speaker participants were shown to appropriately distinguish between counter-
balanced tokens within each context type in Region 2 or in Region 3: 10 and 8
of 24, respectively, for Contrastive Focus, 17 and 8 of 24, respectively, for Topic
Shift and 7 and 11 of 24, respectively, for Topic Maintenance. Second, both groups
distinguished between the counterbalanced tokens in each context in the offline
task, pointing to their differential treatment of overt versus null subjects in each
 Knowledge and processing of subject-related discourse properties 

c­ ontext. Taken together, these points suggest that the tokens were successful in
eliciting the expected responses, but that the differences regarding which region
the distinction was evidenced gives the impression that few distinctions were
made in the online version.
Concerning the L2 group data, group distinctions between the counterbal-
anced felicitous and infelicitous tokens were made in all but two regions: Region 2
of Contrastive Focus and Topic Maintenance. Disregarding for the moment the lack
of distinction at the group level, that the average differential RTs for the counterbal-
anced Contrastive Focus tokens do not statistically differ between the groups is not
problematic for the IH under its current claims since Sorace (2011) argues that dif-
ferences exist between interface properties like topic and focus. Specifically, Sorace
claims that mixed results from prior studies (compare Belletti et al. (2007), Sorace
and Filiaci (2006) and Sorace et al. (2009) with Rothman (2008, 2009)) obtain from
the linguistic properties tested therein, essentially claiming that convergence on
Contrastive Focus is less problematic than Topic Shift and that the latter is a more
appropriate syntax-discourse property to test the IH. Regardless of whether one
agrees with this claim or not, the online processing data for Topic Shift contexts,
precisely where Sorace predicts differences, provides evidence against the IH. First,
both groups evidenced the appropriate distinction in Region 2 at the group level
and, second, no statistically significant difference obtained between the average dif-
ferential RTs of the groups for Region 2 (a difference did obtain for Region 3). This
indicates that the L2 group processed the Topic Shift target sentences as the native
speaker group did, showing native-like processing.
In conclusion, while the Topic Shift data presented herein are interpreted as
providing some counterevidence to the IH, the experimental group, while novel,
was small. It is clear that definitive claims regarding the IH’s tenability can only be
made based on a wide body of research examining many properties across varied
language pairings. The results of this study simply suggest that the claims of the
IH may not hold in all scenarios. This experiment provides a small piece of a large
puzzle that researchers examining near-native language acquisition at the external
interfaces must continue to investigate. Specifically, and in order to better determine
whether representation or processing is implicated in RSP distribution, near-native
speakers with a non-null subject L1 should be tested using similar methodology.

References

Aaronson, D., & Scarborough, H.S. (1976). Performance theories for sentence coding: Some
quantitative evidence. Journal of Experimental Psychology: Human Perception and Perfor-
mance, 2, 56–70. DOI: 10.1037/0096-1523.2.1.56
 Tiffany Judy

Alexiadou, A., & Anagnostopoulou, E. (1998). Parametrizing AGR: Word Order, V movement
and EPP-checking. Natural Language and Linguistic Science, 16(3), 491–539.
DOI: 10.1023/a:1006090432389
Alonso-Ovalle, L., & D’Introno, F. (2000). Full and null pronouns in Spanish: the zero pronoun
hypothesis. In H. Campos, E. Herburger, A. Morales-Front, & T.J. Walsh (Eds.), Hispanic
Linguistics at the Turn of the Millennium (pp. 400–414). Somerville, MA: Cascadilla Press.
Alonso-Ovalle, L., Fernández-Solera, S., Frazier, L., & Clifton, C. (2002). Null vs. overt pronouns
and the topic-focus articulation in Spanish. Rivista Di Linguistica, 14, 151–169.
Bel, A., & García-Alcaraz, E. (2015). Subject pronouns in the L2 Spanish of Moroccan Arabic
speakers: Evidence from bilingual and second language learners. In T. Judy & S. Perpiñán
(Eds.), The Acquisition of Spanish in Adult and Child Bilingualism. Data from Understudied
Language Pairings (pp. 201–232). Amsterdam: John Benjamins.
Belletti, A., & Leonini, C. (2004). Subject inversion in L2 Italian. EUROSLA Yearbook, 4, 95–118.
DOI: 10.1075/eurosla.4.06bel
Belletti, A. Bennati, E., & Sorace, A. (2007). Theoretical and developmental issues in the syntax
of subjects: Evidence from near-native Italian. Natural Language and Linguistic Theory, 25,
657–689. DOI: 10.1007/s11049-007-9026-9
Bentivoglio, P. (1987). Los sujetos pronominales de primera persona en el habla de Caracas.
­Caracas: Universidad Central de Venezuela. DOI: 10.1017/s0047404500014214
Bialystok, E. (2009). Bilingualism: The good, the bad, and the indifferent. Bilingualism: Lan-
guage and Cognition, 12(1), 3–11. DOI: 10.1017/s1366728908003477
Bini, M. (1993). La adquisición del italiano: Más allá de las propiedades sintácticas del pará-
metro pro-drop. In J.M. Liceras (Ed.), La linguistica y el analisis de los sistemas no nativos
(pp. 126–139). Ottawa: Dovehouse.
Chomsky, N. (1981). Lectures on Government and Binding. Dordrecht: Foris.
DOI: 10.2307/2273965
Chomsky, N. (1995). The Minimalist Program. Cambridge, MA: The MIT Press.
DOI: 10.1017/s0272263197241070
Chomsky, N. (2000). On Nature and Language. Cambridge: CUP.
Chomsky, N. (2001). Derivation by phase. In M. Kenstowicz (Ed.), Ken Hale: A Life in Language
(pp. 1–52). Cambridge, MA: The MIT Press. DOI: 10.1017/s0022226704322747
DELE (2002). Spanish Embassy, Washington, DC.
Duarte, M.E. (1993). Do pronome nulo ao pronome pleno: A trajetória do sujeito no português
do brasil. In I. Roberts & M.A. Kato (Eds.), Do pronome nulo ao pronome pleno: A trajetória
do sujeito no português do Brasil (pp. 107–128). Campinas: Editora da UNICAMP.
Duarte, M.E. (1995). A perda do princípio “Evite pronome” no português brasileiro. Unpublished
Ph.D. dissertation. UNICAMP.
Enríquez, E.-V. (1984). El pronombre personal sujeto en la lengua española hablada en
Madrid (4). Madrid: Consejo Superior de Investigaciones Científicas, Instituto Miguel
de C
­ ervantes [The subject personal pronoun in the Spanish language spoken in Madrid].
Fernández-Soriano, O. (1989). Strong pronouns in null subject languages and the Avoid Pro-
noun Principle. MIT Working Papers in Linguistics, 11, 228–239.
Fernández-Soriano, O. (1993). La visión paramétrica del lenguaje: Más sobre los sujetos y
objetos nulos. In J.M. Liceras (Ed.), La lingüística y el análisis de los sistemas no nativos
(pp. 161–176). Ottawa: Dovehouse.
Filiaci, F. (2011). Anaphoric Preferences of Null and Overt Subjects in Italian and Spanish:
A Cross-linguistic Comparison. Unpublished Ph.D. dissertation. University of Edinburgh.
DOI: 10.1080/01690965.2013.801502
 Knowledge and processing of subject-related discourse properties 

Filiaci, F., Sorace, A., & Carreiras, M. (2013). Anaphoric biases of Null and Overt Subjects in Italian
and Spanish: A cross-linguistic comparison. Language and Cognitive Processes, 29(7), 825–843.
DOI: 10.1080/01690965.2013.801502
Goodall, G. (2001). The EPP in Spanish. In W. Davies & S. Dubinsky (Eds.), Objects and
Other Subjects: Grammatical Functions, Functional Categories and Configurationality
(pp. 193–223). Dordrecht: Kluwer.
Goodall, G. (2002). On preverbal subjects in Spanish. In T. Satterfield, C. Tortora, & D. Cresti
(Eds.), Current Issues in Romance Languages: Selected Papers from the 29th Linguistics Sym-
posium on Romance Languages (LSRL) (pp. 95–109). Amsterdam: John Benjamins.
DOI: 10.1075/cilt.220.08goo
Grinstead, J. (2004). Subjects and interface delay in child Spanish and Catalan. Language, 80(1),
40–72. DOI: 10.1353/lan.2004.0024
Gürel, A. (2006). L2 acquisition of pragmatic and syntactic constraints in the use of overt and
null subject pronouns. In R. Slabakova, S. Montrul, & P. Prévost (Eds.), Inquiries in Lin-
guistic Development: In Honor of Lydia White (pp. 259–282). Amsterdam: John Benjamins.
DOI: 10.1075/z.133.15gur
Hawkins, R., & Chan, Y–H. (1997). The partial availability of universal grammar in second lan-
guage acquisition: The “failed functional features hypothesis”. Second Language Research,
13, 187–226. DOI: 10.1191/026765897671476153
Hopp, H. (2007). Ultimate Attainment at the Interfaces in Second Language Acquisition: Gram-
mar and Processing. Groningen: Grodil Press.
Just, M.A., Woolley, J.D., & Carpenter, P.A. (1982). Paradigms and processes in reading compre-
hension. Journal of Experimental Psychology: General, 111(2), 228–238.
Judy, T., & Feizmohammadpour, A. (2012). When parsing failures fail: Referential subjects
and anaphora resolution in Farsi learners of L2 English. Poster presented at: Gen-
erative Approaches to Language Acquisition North America 5. October 11–12, 2012;
Lawrence, KS.
Karimi, S. (2005). A Minimalist Approach to Scrambling: Evidence from Persian. Berlin: de
Gruyter. DOI: 10.1515/9783110199796
Kissock, M. (1995). Reflexive-middle Constructions and Verb Raising in Telugu. Unpublished
Ph.D. dissertation. Concordia University.
Lapidus Shin, N., & Cairns, H.S. (2011). The development of NP selection in school-age chil-
dren: Reference and Spanish subject pronouns. Language Acquisition, 19(1), 3–38.
DOI: 10.1080/10489223.2012.633846
Liceras, J.M., Fernández Fuertes, R., & Alba de la Fuente, A. (2012). Overt subjects and copula
omission in the Spanish and the English grammar of English-Spanish bilinguals: On the
locus and directionality of interlinguistic influence. First Language, 32(1–2), 88–115.
DOI: 10.1177/0142723711403980
Lozano, C. (2006). Focus and split-intransitivity: The acquisition of word order alternations in
non-native Spanish. Second Language Research, 22, 145–187.
DOI: 10.1191/0267658306sr264oa
Lubbers Quesada, M., & Blackwell, S. (2009). The L2 acquisition of null and overt Spanish sub-
ject pronouns: A pragmatic approach. In J. Collentine, M. García, B. Lafford, & F.M. Marín
(Eds.), Selected Proceedings of the 11th Hispanic Linguistics Symposium (pp. 117–130).
Somerville, MA: Cascadilla Proceedings Project.
Luján, M. (1987). Los pronombres implícitos y explícitos del español. Revista Argentina de
Lingüística, 3, 19–54.
 Tiffany Judy

Luján, M. (1999). Expresión y omisión del pronombre personal. In I. Bosque & V. Demonte,
(Eds.), Gramática descriptiva de la lengua española (pp. 1275–1315). Madrid: Espasa Calpe.
Margaza, P., & Bel, A. (2006). Null subjects at the syntax–pragmatics interface: Evidence from
Spanish interlanguage of Greek speakers. In M.G. O’Brien, C. Shea, & J. Archibald (Eds.),
Proceedings of GASLA 2006. (pp. 88–97). Somerville, MA: Cascadilla Press.
Mitchell, D.C., & Green, D.W. (1978). The effects of context and content on immediate process-
ing in reading. Quarterly Journal of Experimental Psychology, 30(4), 609–636.
DOI: 10.1080/14640747808400689
Montrul, S. (2004). Subject and object expression in Spanish heritage speakers: A case of mor-
phosyntactic convergence. Bilingualism: Language and Cognition, 7(2), 125–142.
DOI: 10.1017/s1366728904001464
Montrul, S., & Louro-Rodríguez, C. (2006). Beyond the syntax of the null subject param-
eter. In V. Torrens & L. Escobar (Eds.), The Acquisition of Syntax in Romance Languages
(pp. 401–418). Amsterdam: John Benjamins. DOI: 10.1075/lald.41.19mon
Otheguy, R., Zentella, A.C., & Livert, D. (2007). Language and dialect contact in Spanish in
New York: Toward the formation of a speech community. Language, 83(4), 770–802.
DOI: 10.1353/lan.2008.0019
Paradis, J., & Navarro, S. (2003). Subject realization and crosslinguistic interference in the bilin-
gual acquisition of Spanish and English: what is the role of input? Journal of Child Lan-
guage, 30, 371–393. DOI: 10.1017/s0305000903005609
Picallo, M.C. (1994). Catalan possessive pronouns: The Avoid Pronoun Principle revisited.
­Natural Language and Linguistic Theory, 12, 259–299. DOI: 10.1007/bf00993146
Picallo, M.C. (1998). On the extended projection principle and null expletive subjects. Probus,
10, 219–241. DOI: 10.1515/prbs.1998.10.2.219
de Prada Pérez, A. (2009). Subject Expression in Minorcan Spanish: Consequences of Contact with
Catalan. Unpublished Ph.D. dissertation. Pennsylvania State University.
Rizzi, L. (1982). Issues in Italian Syntax. Dordrecht: Reidel. DOI: 10.1515/9783110883718
Rizzi, L. (1997). A parametric approach to comparative syntax: Properties of the pronomi-
nal system. In L. Haegeman (Ed.), The New Comparative Syntax (pp. 268–285). London:
­Longman. DOI: 10.9793/elsj1984.10.1
Rothman, J. (2008). How pragmatically odd! Interface delays and pronominal subject distribu-
tion in L2 Spanish. Studies in Hispanic and Lusophone Linguistics, 1(2), 317–339.
Rothman, J. (2009). Pragmatic deficits with syntactic consequences?: L2 pronominal subjects
and the syntax-pragmatics interface. Journal of Pragmatics, 41(5), 951–973.
DOI: 10.1016/j.pragma.2008.07.007
Schneider, W., Eschmann, A., & Zuccolotto, A. (2002). E-Prime [Computer software]. ­Pittsburgh,
PA: Psychology Software Tools Inc.
Silva-Corvalán, C. (1994). Language Contact and Change: Spanish in Los Angeles. Oxford: OUP.
DOI: 10.1017/s0047404500020753
Sorace, A. (2006). Gradedness and optionality in mature and developing grammars. In G. F ­ anselow,
C. Féry, R. Vogel, & M. Schlesewsky (Eds.), Gradience in Grammar (pp. 106–123). Oxford:
OUP. DOI: 10.1093/acprof:oso/9780199274796.003.0006
Sorace, A. (2011). Pinning down the concept of “interface” in bilingualism. Linguistic Approaches
to Bilingualism, 1(1), 1–33. DOI: 10.1075/lab.1.1.01sor
Sorace, A. (2012). Pinning down the concept of “interface” in bilingualism: A reply to peer com-
mentaries. Linguistic Approaches to Bilingualism, 2, 209–216. DOI: 10.1075/lab.2.2.04sor
 Knowledge and processing of subject-related discourse properties 

Sorace, A., & Filiaci, F. (2006). Anaphora resolution in near-native speakers of Italian. Second
Language Research, 22(3), 339–368. DOI: 10.1191/0267658306sr271oa
Sorace, A., & Serratrice, L. (2009). Internal and external interfaces in bilingual language devel-
opment: Beyond structural overlap. International Journal of Bilingualism, 13, 195–210.
DOI: 10.1177/1367006909339810
Sorace, A., Serratrice, L. Filiaci, F., & Baldo, M. (2009). Discourse conditions on subject pro-
noun realization: Testing the linguistic intuitions of older bilingual children. Lingua, 119,
460–477. DOI: 10.1016/j.lingua.2008.09.008
Tsimpli, I.-M., & Dimitrakopoulou, M. (2007). The Interpretability Hypothesis: Evidence from
wh-interrogatives in second language acquisition. Second Language Research, 23, 215–242.
DOI: 10.1177/0267658307076546
Tsimpli, I.-M., & Sorace, A. (2006). Differentiating interfaces: L2 performance in syntax-­
semantics and syntax-discourse phenomena. In D. Bamman, T. Magnitskaia, & C. Zaller
(Eds.), Proceedings of the 30th Annual Boston University Conference on Language Develop-
ment (pp. 653–664). Somerville, MA: Cascadilla Press.
Tsimpli, I.-M., 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, 257–277. DOI: 10.1177/13670069040080030601
Vainikka, A., & Young-Scholten, M. (1994). Direct access to X′-theory: evidence from Korean
and Turkish adults learning German. In T. Hoekstra & B. Schwartz (Eds.) Language Acqui-
sition Studies in Generative Grammar: Papers in Honor of Kenneth Wexler from the 1991.
GLOW Workshops (pp. 265–316). Amsterdam: John Benjamins.
DOI: 10.1075/lald.8.13vai
Vainikka, A., & Young-Scholten, M. (1996). Gradual development of L2 phrase structure. ­Second
Language Research, 12, 7–39. DOI: 10.1177/026765839601200102
White, L. (2003). Second language acquisition and Universal Grammar. Cambridge: CUP.
DOI: 10.1017/cbo9780511815065.002
White, L. (2009). Grammatical theory: interfaces and L2 knowledge. In E. Ritchie & T. B ­ hatia
(Eds.), The New Handbook of Second Language Acquisition (pp. 49–68). Bingham, UK:
Emerald.
 Subject pronouns in the L2 Spanish
of Moroccan Arabic speakers
Evidence from bilingual and second language learners

Aurora Bel & Estela García-Alcaraz

Universitat Pompeu Fabra

The goal of this study is to shed light on the conditions that regulate the
interpretation of null and overt third person subject pronouns in ambiguous
intra-sentential contexts in bilingual and L2 Spanish. An acceptability judgment
task was administered to 26 Moroccan Arabic (MA)/Spanish bilinguals, 26 MA
learners of L2 Spanish and 34 native controls. Three variables were manipulated:
type of pronoun (null vs. overt), antecedent choice (subject vs. object) and clause
order (main-subordinate vs. subordinate-main). The results reveal an effect of
pronoun and clause order. Overall, both experimental groups show native-like
pronominal preferences but L2 learners differ from native speakers specifically
in subordinate-main order as they do not seem to benefit from microvariation
between languages in the domain of pronoun interpretation.

1.  Introduction

The variety of referring expressions and the study of their linking properties
have been examined in different frameworks such as topicality hierarchy (Givon
1984), accessibility theory (Ariel 1990, 2001) and centering theory (Gordon,
Grosz & Gilliom 1993). Two key referring expressions are nominal and (null and
overt) pronominal elements, which contribute to reference organization in dis-
course. In broad terms, a pronoun is likely to be interpreted as denoting an entity
that has already been introduced by a salient category in the discourse (e.g. a
­nominal); conversely, nominal categories are used to refer to less salient or acces-
sible referents.
Psycholinguistic research has also examined the factors that might influ-
ence the choice of an antecedent by the different types of pronominal a­ naphoric
expressions and have found that speakers tend to use pronouns to refer to a
prominent or salient discourse entity (Garvey, Caramazza & Yates 1975). In

doi 10.1075/ihll.3.08bel
© 2015 John Benjamins Publishing Company
 Aurora Bel & Estela García-Alcaraz

ambiguous referential contexts, it has also been observed that null subject pro-
nouns refer back to topic antecedents and overt pronouns tend to express topic
change (Belletti, Bennati & Sorace 2007). Additionally, recent research has
shown that anaphoric elements are sensitive to different factors such as gender,
order of presentation (Arnold, Eisenband, Brown-Schmidt & Trueswell 2000;
Trueswell, Papafragou & Choi 2011) or the syntactic function of the antecedent
(Carminati 2002; Filiaci 2011).
As far as the acquisition of reference forms is concerned, the most influen-
tial research on pronominal reference has been conducted under the Interface
Hypothesis (IH) framework (Sorace & Filiaci 2006; Sorace, Serratrice, Filiaci  &
Baldo 2009; see Sorace 2011 for an overview). The hypothesis implies that in
linguistic domains in which morphosyntax interacts with discourse-pragmatic
factors, such as the use and interpretation of anaphoric expressions, L2 learners
and bilingual speakers do not reach native-like performance. Sorace et al. (2009)
proposed that this linguistic behaviour, characterized by instability and optional-
ity in production, as well as in comprehension, is derived from a compensatory
strategy to supersede online processing demands that speakers encounter when
integrating information from different linguistic levels in a non-primary language
in line with the shallow processing among L2 speakers suggested by Clahsen and
Felser (2006). For instance Sorace et al. (2009), among others, reported that Italian
L2 speakers and bilinguals with high proficiency in the target language admit-
ted and produced overt subjects in the target null-subject language in contexts
where monolinguals would not have admitted or produced them, sometimes lead-
ing to redundancy. This redundancy is assumed to reflect the instability of overt
pronouns in the production and comprehension of non-native speakers: overt
pronouns would imply a less costly processing operation to implement discourse
interface operations efficiently even when their use entails redundancy.
The predictions derived from the IH diminish the impact of cross-linguistic
influence. When the language pairing differs in the realization of the relevant fea-
ture, for example English L2 learners of Spanish (a non-null subject language and
a null one), the logical reasoning is to ascribe the overextension of overt pronouns
mainly to the L1 influence at a syntactic level (Pérez-Leroux & Glass 1999) or
at a discourse strategy level (Jegerski, VanPatten & Keating 2011). However, it is
known that this overextension persists in the acquisition of pronouns between
two null subject languages (Margaza & Bel 2006; Sorace et al. 2009, for Greek-
Spanish and Italian-Spanish, respectively); this behaviour cannot be attributed to
cross-linguistic influence since, due to a substantial overlap in pronoun realiza-
tion patterns of the two languages, the result would be one of positive influence
leaving no room for an overuse of overt pronouns. In addition, as pointed out by
 Subject pronouns in the L2 Spanish of Moroccan Arabic speakers 

developmental studies, the mastery of the pragmatic conditions that regulate the
distribution and interpretation of pronouns is a difficult and challenging task that
comes with age and exposure even in L1 (see Shin & Cairns 2012, for Spanish, and
Arnold, Brown-Schmidt & Trueswell 2007, for English). Taking all these aspects
into consideration, in the present work we examine the acquisition of these fea-
tures in Arabic/Spanish bilinguals and native speakers of Moroccan Arabic who
are learning Spanish as a L2. Like Spanish, Arabic is a null subject language and
thus we would expect a similar effect as that observed in L1/L2 null subject pair-
ings mentioned above. Here we compare populations differing in their exposure
to the target language but analogous with respect to their L2 knowledge. To do so,
we will compare the performance of early sequential bilinguals living in Spain and
that of L2 learners learning Spanish in a foreign language setting.

2.  Theoretical background

2.1  A
 naphora resolution in null subject languages: Spanish and
Moroccan Arabic
Spanish, as a null subject language with rich agreement marking on verbs, has
two types of pronouns: an overt pronoun and a phonologically null pronoun
(pro). Their use, however, is not fully optional: they carry information that con-
tributes to anaphora resolution. The distribution and interpretation of subject
pronouns at the syntax-discourse interface is regulated by the [±topic shift] prag-
matic feature. As a general rule, it can be said that a noun phrase introduces
a referent into discourse, an overt pronoun signals a shift in topic, and a null
pronoun expresses topic continuity. Additionally, an overt pronoun may convey
contrastive meaning, an issue that we do not address here. This is displayed in the
following examples:

(1) a. Luisi telefoneaba a su madre cuando proi llegaba tarde. [–topic shift]
‘Luis telephoned his mother when pro was late.’
b. Ana vio a su vecinai mientras ellai se bañaba en el mar. [+topic shift]
‘Ana saw her neighbour while she was bathing in the sea.’

It is worth mentioning that the use of a null pronoun instead of the overt one (ella)
in (1b) would lead to an interpretation of topic continuity (Ana was bathing) or
would create ambiguity (either Ana or her neighbour would be bathing). The same
is true for (1a) where, in strict terms, pro can refer as well to the preceding object
‘su madre’ (‘her mother’). Furthermore, in a topic continuity context the overt
 Aurora Bel & Estela García-Alcaraz

third person pronoun may refer directly to an unambiguous antecedent generat-

ing redundancy as in (2) below:
(2) Juliai dice que ellai confesó la verdad.
‘Julia says that she confessed the truth.’

In the last decade, some studies (Montrul 2004; Sorace & Filiaci 2006; Sorace
et al. 2009) have revealed that speakers have antecedent preferences for pronouns
when the context is potentially ambiguous. The most influential work has been
Carminati’s (2002) study in which she formulated a processing account for the
interpretation of null and overt pronouns in Italian, the ‘Position of Antecedent
Hypothesis’ (PAH), according to which a null pronoun prefers an antecedent in
a subject position, while an overt pronoun prefers an antecedent in a position
other than subject. The following example from Carminati (2002, p. 222) illus-
trates these biases:
(3) Roberto ha insultato Ugo quando lui/pro era ubriaco.
‘Roberto insulted Ugo when he/pro was drunk.’
a. Roberto era ubriaco
‘Roberto was drunk.’
b. Ugo era ubriaco
‘Ugo was drunk.’

In (3) there are two potential antecedents for the overt pronouns (lui) and for the
null pronoun (pro). The results showed that Italian speakers are significantly more
likely to choose the null pronoun when referring to the subject (3a) and the overt
pronoun when referring to the object (3b): overt pronouns were interpreted as
referring to non-subject antecedents in 83.3% of the cases whereas null pronouns
were taken as referring back to subject antecedents 80.7% of the time, clearly giv-
ing support to the PAH. This generalization holds mainly for gender-ambiguous
contexts as the one in (3); on the contrary, when gender clearly identifies anteced-
ents, speakers seem to be more flexible, accepting different antecedent positions
for overt pronouns. The violation of the PAH does not generate an ungrammati-
cal but an infelicitous sequence, and the speakers rely on it to prevent misunder-
standing. Translating these findings into discourse terms, it can be said that null
pronouns express topic continuity, they refer to the topic subject of the previous
clause, while overt pronouns convey topic change since they occupy the subject
topic position of the second clause and refer back to a prior element, the object of
the previous clause that is not in topic position.
Alonso-Ovalle, Fernández-Solera, Frazier and Clifton (2002) investigated pro-
nominal preferences in two sentence discourses in Spanish. Following C ­ arminati’s
(2002) PAH, they predicted that, while a null pronoun would prefer a subject
 Subject pronouns in the L2 Spanish of Moroccan Arabic speakers 

a­ ntecedent, the overt pronoun would not. Unlike the experimental sentences of
Carminati’s (2002) work, these authors used inter-sentential contexts in experi-
mental sentences. An example from Alonso-Ovalle et al. (2002, p. 166) is given
below:
(4) a. Sara abrazó a Teresa. pro está emocionada.
‘Sara embraced Teresa. pro is excited.’
b. Sara abrazó a Teresa. Ella está emocionada.
‘Sara embraced Teresa. She is excited.’

In contexts as the one in (4a), in 73.2% of the cases the null pronoun was linked to
the subject, Sara, while in examples such as (4b) the overt pronoun was assigned
to either Sara or Teresa (50.2% of the total cases were linked to the subject). Even
though Alonso-Ovalle et al. (2002) argued that their results supported the PAH in
Spanish across sentences, this can only be maintained for null pronouns which,
similarly to Italian, prefer a subject position as their antecedent; conversely, overt
pronouns do not prefer a particular antecedent, in contrast with Carminati’s
(2002) findings.
More recently, Filiaci (2011) has also looked into the division of labor between
null and overt subject pronouns in Spanish. She replicated the original experi-
ment from Carminati (2002) and the sentences were translated and adapted into
­Spanish. However, unlike the preceding research, the second sentence was manip-
ulated so as to control the semantics of the predicate and thus favor either topic
continuity (as in 5a) or topic change (as in 5b), disambiguating towards the pre-
ceding subject or object, respectively:
Cuando Ana visitó a María en el hospital, ella/pro le llevó un ramo
(5) a. 
de rosas.
 ‘When Ana visited Mary in the hospital, she/pro brought her a bunch
of roses.’
Cuando Ana visitó a María en el hospital, ella/pro ya estaba fuera
b. 
de peligro.
 ‘When Ana visited Mary in the hospital, she/pro was already out of
danger.’

Filiaci (2011) measured reading times on these sentences and found that, in
the  null subject condition, participants were faster in topic continuity contexts
than in topic change contexts, which suggests that null pronouns are associ-
ated with the [−topic shift] feature, thus supporting the PAH. On the contrary,
in the overt pronoun condition reading times were similar across the two con-
texts, a result which is in line with what Alonso-Ovalle et al. (2002) found for
inter-sentential anaphora. Interestingly, the experiment was also conducted in
 Aurora Bel & Estela García-Alcaraz

among Italian speakers and results supported Carminati’s (2002) findings, hence
confirming the attested division of labor of Italian pronouns. As for clause order,
although ­Carminati (2002) did not find differences between main-subordinate
and subordinate-­main clause orders, Filiaci (2011) did. Our study also attempts to
shed some light on this question.
To sum up, previous research on the antecedent preferences of null and overt
pronouns in Romance languages is far from showing a definite picture. On the
one hand, null pronouns seem to have a specific role: they are clearly specialized
in establishing co-reference with elements in subject position, giving continuity
to the element (the subject) that initiated the discourse and that is the current
topic. And this is so irrespective of the sentence type: subordinate intra-senten-
tial contexts as well as inter-sentential contexts. On the other hand, the use of
overt pronouns does not offer a clear picture since varying degrees of anteced-
ent ­preferences are shown across languages. More precisely, while in Italian overt
pronouns visibly signal a switch in subject reference, this connection seems to be
much weaker in Spanish, thus suggesting the existence of microvariation among
null subject languages.
Given the observation that there is cross-linguistic variation in the usage and
interpretation of pronouns among null-subject languages, we asked ourselves what
the preferences are in Moroccan Arabic (MA), or Darija, the Arabic vernacular
spoken in Morocco. The country is linguistically heterogeneous: MA coexists with
Amazigh (or Berber) languages, and Modern Standard Arabic (MSA) and French,
the languages of education, administration and the media.
MA, as described by Aoun, Benmamoun, and Choueiri (2010), is a language
with rich verbal morphology that marks person, number and gender on verbs.
As a null subject language, it has two types of pronouns: null and overt. Null
pronouns are licensed by rich phi-features on the verb and it is predicted that
they can be identified or recovered when verbs display person marking since
in MA, unlike other Arabic dialects, agreement is obligatory regardless of word
order variation (Benmamoun 2000). Overt pronouns are used in pleonastic and
emphatic contexts as well as in copulative sentences, since MA lacks the present-
tense copular form.
MA, like other Arabic dialects and MSA, has relatively free canonical word
order. It is predominantly VSO, though SVO is a possible alternative basic
word order. VOS is a marked word order used in specific pragmatic contexts
where the subject conveys new information, that is generally undetermined,
or contrast (this meaning can be expressed by a Determiner Phrase, DP, or an
overt pronoun). The following examples, from Kortobi (2002, p. 218), reflect
these contrasts (note that the determined DP in (c) renders the sentence
ungrammatical):
 Subject pronouns in the L2 Spanish of Moroccan Arabic speakers 

(6) a. I-wəld qra ktab.

The-boy read book
b. qra I-wəld ktab.
read the-boy book
c. *qra ktab I-wəld.
 read book the-boy
‘The boy read a book.’

As for the interpretative properties of pronouns in discourse, there is a lack of

studies based on empirical findings that address this issue in MA and other ­Arabic
dialects. Only a few studies have examined the topic in Egyptian Arabic, EA
(Farghaly 1982; Schulte-Nafeh 2004). Looking at subordinate contexts, such as
those analyzed in the present research study, Farghaly (1982) proposed that third
person overt pronouns convey contrast and emphasis functions; he also addressed
the interpretation of overt pronouns in complex main-subordinate sentences and
claimed that overt pronouns tend to be interpreted as referring to the preceding
subject as antecedent whereas null pronouns tend to be interpreted as referring to
the object (i.e. the closest NP). This seems counterintuitive and goes against the
PAH, as null pronouns would signal topic change whereas overt pronouns would
express topic continuity. More recently, in a very detailed study based on both
spoken data and acceptability judgments in different sentence types and person
contexts, Schulte-Nafeh (2004) noticed that “the motivations to use an overt pro-
noun are not strong” (p. 161). It was found that there is a rather opposite role for
overt pronouns in subordinate clauses in EA since they tend to express subject
switch, particularly with ambiguous verbal predicates, while null pronouns convey
topic or subject continuity (p. 184–185). Returning to the current study of L1 MA,
two native informants were asked to evaluate some main-subordinate sentences
that contained a null or an overt pronoun and judged them in this same exact
way. Consequently, an experiment was conducted to provide empirical data and
to examine the existence of a relationship between pronouns and discourse func-
tions in MA; in other words, we aimed to determine to what extent (if any) MA
pronouns also fit into the patterns constrained by the PAH. The procedure and
findings of this experiment are summarized later in Section 4.

2.2  Studies on the acquisition of pronominal anaphora

Sorace and collaborators initiated the current growing body of research on the
acquisition of pronominal anaphora. Based on intra-sentential contexts, their
work focused in how Italian [null-subject language]-English [non-null-subject lan-
guage] bilinguals and L2 learners managed this phenomenon in Italian according to
­Carminati’s (2002) PAH. Among their work, the most influential and widely cited in
 Aurora Bel & Estela García-Alcaraz

the bilingual acquisition literature is perhaps Sorace et al. (2009). The main objective
of this study was to assess the acceptability of Italian (null and overt) and English
pronominal subjects in [±topic shift] contexts. The performance of ­English-Italian
and Spanish-Italian children aged 6–7 and 8–10 was compared to that observed
among age-matched monolingual children and monolingual adults. The authors
considered the potential impact of language exposure and, consequently, two sub-
groups were included: (1) E ­ nglish-Italian bilinguals living in Italy and (2) English-
Italian bilinguals living in the UK. An acceptability judgement task that followed a
short video clip was devised. The experiment dealt with two conditions each having
two levels: type of pronoun (null vs. overt) and discourse function (+topic shift
vs. −topic shift). For ­Italian, the patterns of results yielded more pragmatically inap-
propriate overt subject pronouns among younger monolingual and bilingual chil-
dren than among older monolingual children and adults. Moreover, in line with
previous findings (Montrul & Rodríguez Louro 2006; ­Rothman 2009), the trouble
experienced by bilingual children was not solely restricted to the interpretation of
overt pronouns. The results showed that bilingual children also accepted more null
subject pronouns in [+topic shift] than monolingual Italian children, although they
committed proportionally more errors involving redundancy than ambiguity. In
addition, 6–7 year old English-Italian bilinguals living in the UK over-accepted sig-
nificantly more redundant overt subject pronouns in [−topic shift] than the rest of
child groups, while in the 8–10 age group it was the Spanish-Italian bilingual group
that performed less accurately than all other groups of children. The fact that the
overacceptance of overt pronouns increases with age in Spanish-Italian bilinguals
may be pointing to a high sensitivity to Spanish pronoun interpretation, since, as
shown by previous work, the scope of the overt pronoun in ­Spanish seems to be
wider than in ­Italian. The overuse of overt pronouns in bilingual children has also
been attested in Dutch-Italian bilinguals (Pinto 2006), German-Italian bilinguals
(­Müller, Kupisch, Schmitz & ­Cantone 2006), English-Italian bilinguals (­Serratrice
2007; Serratrice, Sorace & Paoli 2004), Spanish-English bilinguals (­Paradis &
Navarro 2003), and Turkish-English bilinguals (Haznedar 2006).
As far as L2 is concerned, many studies also reported a notable proportion
of pragmatically infelicitous overt subject pronoun (Belletti et al. 2007; Liceras
1988; Lozano 2006; Sorace & Filiaci 2006), and null subject pronouns (­Keating,
­VanPatten & Jegerski 2011; Lafond, Hayes & Bhatt 2001; Montrul 2004; M ­ ontrul &
Rodriguez Louro 2006; Rothman 2009).
Rothman (2009) examined the syntax-before-discourse hypothesis through
null vs. overt subject pronoun interpretation in L2 Spanish. Spanish learners at
intermediate and advanced levels and two groups of native English and Spanish
speakers were analyzed. The results showed, in line with Montrul and Rodriguez
Louro (2006), but against Sorace and Filiaci’s (2006) predictions, an overuse of
both overt and null subject pronouns.
 Subject pronouns in the L2 Spanish of Moroccan Arabic speakers 

Also, more recently, Jegerski et al. (2011) have attempted to ascertain whether
cross-linguistic influence is a major factor in explaining divergences in anaphora
resolution between L2 and native speakers of Spanish. Offline interpretation of
ambiguous subject pronouns in temporal intra-sentential contexts was examined
in English and Spanish. Two experimental groups were analyzed: intermediate
and advanced English learners of Spanish. The results revealed that both groups
failed to show native-like PAH strategy when interpreting Spanish sentences; on
the contrary, they resorted to English discourse structure as a cue when interpret-
ing ambiguous pronouns in Spanish. As the authors point out, this cross-linguistic
influence does not seem to be unique among English dominant speakers; it is sup-
posed to be a rather universal strategy and, thus, applies to any language, be it
pro-drop or non-pro-drop.
Finally, a reference must be made to a recent study on the acquisition of null
subjects in L2 Arabic by Spanish and English speakers conducted by Alhawary
(2007, 2009). Their results show that L2 learners produce subjectless sentences
early on in the acquisition process, that English speakers are able to switch the
value from a non-null subject language to a null-subject one and that there is
some evidence of positive transfer according to the performance of Spanish
learners.
In sum, from the previous literature it seems that the overuse and overaccep-
tance of overt pronouns in [−topic shift] contexts, i.e. in contexts in which they
take a subject as their antecedent, is a very common strategy among bilingual and
L2 speakers but not among natives. On the contrary, the null pronoun’s occur-
rence appears to be consistent across different types of speakers and learners and
different language pairs given that they are rarely linked to [+topic shift] contexts.
The present research aims to contribute to a better understanding of these issues
through the study of two null subject languages -L1 Moroccan Arabic/L2 Spanish,
an unexplored linguistic combination so far.

3.  Research questions

In this study, our goal is to better understand the distinction between null and
overt pronouns in ambiguous intra-sentential contexts in bilingual Spanish and
in the acquisition of L2 Spanish by MA speakers, whose L1 (as it will be examined
in Section 4) also distinguish between null and overt pronouns. In particular, we
want to investigate the following questions:

1. Is the PAH operative in non-biased ambiguous intra-sentential contexts

in Spanish? Does sentence clause order have an effect on the resolution of
pronouns?
 Aurora Bel & Estela García-Alcaraz

2. Do MA speakers of Spanish know the differences in the interpretative proper-

ties of null and overt third person subject pronouns in Spanish?
3. Is there any cross-linguistic influence with respect to this issue due to cross-
linguistic microvariation?
4. Do MA-Spanish bilinguals (i.e. early sequential bilinguals) differ from L2
learners in their mastery of these properties? Furthermore, have the different
amounts of language exposure that bilinguals and L2 learners receive had an
impact on their knowledge of null and overt pronouns?
5. Additionally, in order to perform an effective comparison and discover poten-
tial cross-linguistic effects, what are the main factors that affect pronoun reso-
lution preferences in ambiguous intra-sentential contexts in MA?

We first address question 5 and report the corresponding results for MA in the
next section.

4.  Experiment 1: Anaphora resolution pre-test in Moroccan Arabic

4.1  Design and procedure

To determine how pronominal anaphora resolution works in ambiguous contexts
in MA, we recruited 45 students from the Instituto Cervantes of Marrakech, native
speakers of MA who did not participate in the main study reported on Section 5.
The participants completed a sentence interpretation task in order to ascertain the
antecedent preferences of subject pronouns in two sentence discourses in MA.
The task was administered orally in order to avoid artificiality. MA language
is rarely written; it is associated with spoken registers because MSA, as well as
French, is the language used in written, administrative and academic settings. The
participants also completed a sociolinguistic background questionnaire. The mean
age was 25.7 (range 17–42). All had a minimum of a secondary level education;
some of them were studying at the university and the rest were workers and pro-
fessionals. The students reported MA as being their familial language, but also
reported knowledge of MSA, French and some Spanish (they were recruited from
the beginners’ levels in the Instituto, corresponding to A.2 level of the Common
European Framework of Reference for Languages (CEFR). A number of them also
reported (some) knowledge of English.
A two-option forced-choice preference task was designed. The participants
heard two-sentence ambiguous discourses. The first sentence contained a transi-
tive verb with two characters, designated by proper names of the same gender, in
subject and object position; this sentence was followed by a second sentence con-
taining a null or an overt pronoun in subject position that could refer to one of the
 Subject pronouns in the L2 Spanish of Moroccan Arabic speakers 

two previous referents. Participants were asked to answer who was the ‘doer’ of the
action expressed in the second sentence, thus choosing between competing ante-
cedents in subject or object position; recall that, given that the antecedents were of
the same gender, the (null or overt) pronoun was fully ambiguous. Although some
sentences may sound rather strange or unnatural, we tried to lessen the influence
of semantic and discourse cues – as these play an important role during anaphoric
inferencing in real discourse – by keeping contexts completely ambiguous in order
to determine if pronouns have clearly identifiable biases.
In total there were 50 items, which included 20 experimental items (5 items
per condition) and 30 fillers. Two experimental conditions with two levels each
were tested: sentence order (main-subordinate vs. subordinate-main) and type of
pronoun (null vs. overt). The sentence was followed by a comprehension question
that asked for the preferred antecedent for the pronoun. Each item was recorded
by a male native speaker of MA, who produced them with neutral intonation and
in loud, clear voice. An example of each condition is given in (7):
(7) a. Maria faʒʡt ħanan melli tzewʒt.
Maria surprised-3s-F ħanan when pro married-3s-F
¿ʃkun li tzeweʒ?
  Who asp married?
A. ħanan B. Maria
A. ħanan B. Maria
‘Maria surprised ħanan when she was getting married.
Who was getting married? A. ħanan B. Maria’
b. Hiʃam ʢ̰allem ʢ̰abdelwaħed melli howa tʢ̰afa
Hiʃam taught -3s-M ʢ̰abdelawħed when he asp
men l-ksida. ¿ʃkun li tʢ̰afa men l-ksida?
recovered-3s-M the accident Who asp recovered the accident?
A. ʢ̰abdelwaħed  B. Hiʃam
A. ʢ̰abdelwaħed  B. Hiʃam
‘Hiʃam taught ʢ̰abdelwaħed when he was recovered from the accident.
Who was recovered from the accident? A. ʢ̰abdelwaħed B. Hiʃam’
c. Melli Iħsan ddat ѵizlan, bdat katebki.
When Iħsan picked up-3s-F ѵizlan, pro started-3s-F crying
¿ʃkun li bda kibki?
Who asp started crying?
A. Iħsan  B. ѵizlan
A. Iħsan  B. ѵizlan
‘When Iħsan picked up ѵizlan, she started crying.
Who started crying? A. Iħsan B. ѵizlan’
 Aurora Bel & Estela García-Alcaraz

d. Melli Fatiħa gatʢ̰et Nadia, hiya bdat katedħak.

When Fatiħa interrupted-3s-F Nadia, she started-3s-F laughing
¿ʃkun li bda kidħak?
Who asp started laughing?
A. Nadia  B. Fatiħa
A. Nadia  B. Fatiħa
‘When Fatiħa interrupted Nadia, she started laughing.
Who started laughing? A. Nadia B. Fatiħa’

4.2  M
 ain findings
Proportions of responses were evaluated. Comparing mean scores (see Figures 1
and 2), different proportions of responses were attested depending on the type of
pronoun. Subjects were more frequently chosen as antecedents for null pronouns.
This preference was even more pronounced in the subordinate-main order (66.89%
subordinate-main vs. 58.37% main-subordinate). As for overt pronouns, subjects
and objects were selected as antecedents, regardless of sentence order (44.33% of
subjects in main-subordinate vs. 46.67% of subjects in subordinate-main).

70,00

60,00

50,00

40,00

30,00

20,00

10,00

0,00
Subject antecedent Object antecedent Subject antecedent Object antecedent
NULL PRONOUN OVERT PRONOUN
MAIN + SUBORDINATE

Figure 1.  Percentage of antecedent preferences of pronouns in MA (native speakers);

main-subordinate order

The data were submitted to a 2 × 2 repeated measures ANOVA with type

of pronoun (null vs. overt) × sentence order (main-subordinate vs. subordinate-
main) as within-subject factors on the basis of mean proportions. The analysis
 Subject pronouns in the L2 Spanish of Moroccan Arabic speakers 

70,00

60,00

50,00

40,00

30,00

20,00

10,00

0,00
Subject antecedent Object antecedent Subject antecedent Object antecedent
NULL PRONOUN OVERT PRONOUN
SUBORDINATE + MAIN

Figure 2.  Percentage of antecedent preferences of pronouns in MA (native speakers);

subordinate-main order

yielded a main effect of type of pronoun, (F(1, 44) = 18.983, p = .000), no main
effect of sentence order, and no interaction between the two factors. This reflects
the stronger preference to associate null pronouns with subject antecedents in both
sentence orders though there were a number of them that referred back to objects.
On the other hand, overt pronouns showed no significant preferences since they
were assigned freely to either subject or object antecedents. In short, these results
reveal that the interpretation of null pronouns is highly biased toward the subject,
while overt pronouns display a more varied pattern.
To further confirm these results, paired-samples t-tests – one for each type
of pronoun – were conducted on the proportions to identify differences between
pronouns regarding antecedent preferences for the two orders; the results were
significant for null pronouns (t (44) = 2.369; p = .022; t (44) = 5.135; p = .000) but
not for overt pronouns.
These results reveal that the interpretation strategy for null and overt pro-
nouns in MA intra-sentential ambiguous contexts partially fits in with Carminati’s
PAH in that a different behaviour has clearly been discovered between the two pro-
nouns: null pronouns show a strong tendency towards topic continuity since they
select subjects as antecedents whereas overt pronouns remain neutral about topic
continuity or change and select antecedents by chance. Moreover, the order of
presentation of clauses does not influence the choice of antecedent as suggested by
Carminati (2002) but in contrast to Filiaci (2011). However, although i­ nterpretive
 Aurora Bel & Estela García-Alcaraz

behaviour regarding null pronouns in MA does not differ from other null subject
languages that have been examined within the framework of the PAH (Italian,
Spanish, etc.), overt pronouns are more unpredictable. The pattern observed for
MA overt pronouns is similar to that observed in Spanish by Alonso Ovalle et al.
(2002) and Filiaci (2011) but differs substantially from the results observed by
Carminati for Italian; this leads us to conclude that the PAH can only describe a
portion of the MA data, as in the case of Spanish. Having obtained this informa-
tion, it is easier to address the role of transfer of L1 MA to L2 Spanish.

5.  E
 xperiment 2: The interpretation of subject pronouns in bilingual
and L2 Spanish

5.1  P
 articipants
Twenty-six early sequential MA-Spanish bilinguals and 26 MA L2 learners of
Spanish participated in the experiment, together with a control group of 34 native
speakers of Spanish. All the participants were voluntarily. They were asked to fill
out a language background questionnaire in which they provided linguistic and
sociolinguistic background information.
The sequential MA-Spanish bilinguals were teenagers recruited from second-
ary schools (from 7th to 10th grade). They had early contact with Spanish: some
of them were born in Spain and others started the contact before age 6 in their
pre-school period. Their familial language is MA and the environmental language
is Spanish, together with Catalan. All of them report having a good knowledge of
their first language as well as Spanish (see Table 1). The participants of this group
were selected from a large pool of bilinguals who produced two texts in oral and
written modalities. A judge specialized in the evaluation of Spanish as a foreign
language assessed the two production tasks. Only those with a Spanish proficiency
level of B2 or higher on the CEFR scale (independent user: vantage or upper inter-
mediate level) were selected.

Table 1.  Early sequential MA-Spanish bilinguals’ sociolinguistic background

Speakers information Mean Range SD

Age (years) 13.88 12–17 1.39

Arrival age (years) 1.19 0–5.5 2.16
Self-reported Spanish level (out of 100) 91 75–100 0.12
Self-reported MA level (out of 100) 82 50–100 0.17
Average daily use of Spanish (out of 100) 75 58–91 0.09
 Subject pronouns in the L2 Spanish of Moroccan Arabic speakers 

The L2 learners were students of Spanish at the Instituto Cervantes of

­Marrakech (Morocco) (see Table  2). Their first language is MA and all partici-
pants were placed in general Spanish courses for a B2 level. As we mentioned ear-
lier, Morocco is a multilingual country so our learners had knowledge of various
languages: in addition to French, MSA, Spanish and MA (their first language),
they also report different levels of knowledge of other languages such as English,
among others. Importantly, none of them reported high proficiency in a different
null subject language (besides MA and Spanish).

Table 2.  L2 learners’ sociolinguistic background

Age (mean in years) 28.08
Onset age of acquisition of Spanish (mean in years) 24.19
Language(s) spoken at home in early childhood MA
Language of schooling at elementary level MSA-French
Language of schooling in middle and high school MSA-French-English
Travelled to Spanish-speaking country Yes (34.61%)
Length of stay in Spanish-speaking country 1 week-1 month

The control group was formed by 34 university students from Barcelona

(average age 22.9; range 20–25), native speakers of Spanish; all with knowledge
of Catalan, a language that shares many similarities with Spanish. In fact, Spanish
and Catalan have similar discourse realization of pronouns because of the close-
ness of the two languages; studies on the comprehension and production of null
and overt pronouns in Catalan confirm this closeness and do not reveal a very
different picture from that of Spanish (Bel, Perera & Salas 2010; Mayol & Clarck
2010). In addition, according to de Prada Pérez (2009) contact language contexts
favor convergent outcomes. Therefore, these results do not seem to suggest that
the knowledge of Catalan greatly interferes with the knowledge of Spanish in this
specific domain.

5.2  Procedure and materials

An acceptability judgment task (AJT) was completed by all participants. The AJT
was an untimed pencil-and-paper task in which each token consisted of two dis-
course sentence items. The first of the pair contained two characters of the same
gender introduced by proper names while the second contained either a null or
overt pronoun in subject position. Each item was followed by a third sentence
continuation conveying a possible interpretation favoring either the character in
subject or object position. The participants were asked to judge the plausibility of
 Aurora Bel & Estela García-Alcaraz

the continuation according to a four-value Likert scale, ranging from 1 (totally

unacceptable) to 4 (perfectly acceptable).
The experimental sentences were completely ambiguous and there were no
semantic clues that helped to disambiguate the pronoun, a feature that distin-
guishes our work from previous research reported above. To ensure this, the first
sentence verbs were selected taking into account the implicit causality of transi-
tive interpersonal verbs in Spanish according to Goikoetxea, Pascual, and Acha’s
(2008) normative study. In this way we avoid obtaining results with a preference
toward causally implicated referents. We selected 16 verbs showing a neutral or
almost neutral bias towards either the subject or the object. The list of verbs is
shown in Table 3.

Table 3.  Implicit causality of experimental verbs (bias towards subject in percentages;
from Goikoetxea et al. 2008)
Verb Bias towards Verb Bias towards
subject subject

Abandonar 48.1 Investigar 47.1

(to abandon) (to investigate)
Aguantar 52.0 Recoger 37.5
(to withstand) (to pick up)
Asustar 54.7 Saludar 54.3
(to scare) (to greet)
Desmentir 58.3 Seguir 51.1
(to deny) (to follow)
Evitar 59.3 Servir 39.2
(to avoid) (to serve)
Formar 56.9 Sorprender 52.5
(to prepare) (to surprise)
Hablar 62.3 Soportar 52.9
(to speak) (to stand)
Interrumpir 53.7 Ver 46.9
(to interrupt) (to see)

Three conditions, with two levels each, were tested: type of pronoun (null vs.
overt), position of the antecedent (subject vs. object) and order of sentence (main-
subordinate sentence vs. subordinate-main sentence). Sixty-four experimental
items were constructed (8 for each condition level); they were afterwards com-
bined with 80 fillers. The characters’ gender was also balanced across conditions.
The subordinate clause was introduced by a temporal conjunction (half by cuando,
‘when’, and half by mientras, ‘while’). The resulting items were counterbalanced
 Subject pronouns in the L2 Spanish of Moroccan Arabic speakers 

and randomized across two lists. Participants were randomly assigned either to
list 1 or list 2. Table 4 provides an example of each experimental condition.

Table 4.  Sample set of stimulus items

Main-subordinate clause. Null
pronoun. Subject antecedent
Main-subordinate clause. Null
pronoun. Object antecedent
Main-subordinate clause. Overt
pronoun. Subject antecedent
Main-subordinate clause. Overt
pronoun. Object antecedent
Subordinate-main clause. Null
pronoun. Subject antecedent
Subordinate-main clause. Null
pronoun. Object antecedent
Subordinate-main clause. Overt
pronoun. Subject antecedent
Subordinate-main clause. Overt
pronoun. Object antecedent
Iker evita a Iván cuando tiene problemas. Iker tiene
problemas.
‘Iker avoids Ivan when pro has problems. Iker has
problems’
Ángel asustó a Héctor mientras entraba en la habitación.
Héctor entraba en la habitación.
‘Angel scared Héctor while pro came in the room. Hector
came in the room’
Fernando investigó a Antonio cuando él trabajaba para el
gobierno. Fernando trabajaba para el gobierno.
‘Fernando investigated Antonio when he worked for the
government. Fernando worked for the government’
Sebastián sorprendió a Marcos mientras él se casaba.
Marcos se casaba.
‘Sebastián surprised Marcos while he was getting married.
Marcos was getting married’
Mientras José interrumpía a Miguel, se empezó a reír. José
se empezó a reír.
‘While José interrupted Miguel, pro started laughing. José
started laughing’
Cuando Sheila vio a Natalia, estaba nerviosa. Natalia
estaba nerviosa.
‘When Sheila saw Natalia, pro was nervous. Natalia was
nervous’
Mientras Leo servía a Lorenzo, él tiró la bebida. Leo tiró la
bebida.
‘While Leo was serving Lorenzo, he threw the drink. Leo
threw the drink’
Mientras Susana seguía a Sandra, ella se encontró una
moneda. Sandra se encontró una moneda.
‘While Susana was following Sandra, she found a coin.
Sandra found a coin’

5.3  R
 esults
In this section we report the general results of our study. We first describe the
results of the whole set of participants and then we refer separately to our control
group. Finally, comparisons between groups are reported.
Table 5 shows mean rates and standard deviations for the three groups anal-
ysed (natives, bilinguals, L2 learners) and the three conditions tested (order, type
of pronoun and antecedent choice).
 Aurora Bel & Estela García-Alcaraz

Table 5.  Mean rates (and SDs) in the acceptability preference of pronouns in Spanish
(maximum score = 4)
Group Main-subordinate

Null pronoun Overt pronoun

Subject Object Subject Object

Antecedent Antecedent Antecedent Antecedent

Spanish native (control) 2.75 (0.28) 2.96 (0.46) 2.14 (0.42) 3.03 (0.47)
Early sequential bilinguals 2.85 (0.30) 3.00 (0.50) 2.61 (0.44) 3.03 (0.61)
Spanish L2 learners 2.74 (0.38) 2.54 (0.48) 2.16 (0.48) 2.81 (0.63)

Subordinate-main

Null pronoun Overt pronoun

Subject Object Subject Object

Antecedent Antecedent Antecedent Antecedent

Spanish native (control) 3.46 (0.41) 2.17 (0.49) 2.40 (0.43) 3.22 (0.46)
Early sequential bilinguals 3.25 (0.53) 2.46 (0.75) 2.74 (0.72) 2.78 (0.49)
Spanish L2 learners 2.64 (0.70) 2.39 (0.39) 2.40 (0.69) 2.63 (0.74)

Starting with main-subordinate sentence order, Spanish native speaker con-

trols rated subject and object antecedents as equally probable; from Table 5, a near
neutral difference of –0.21 between the choice of an antecedent in subject and
object position arises, with a very subtle tendency towards objects as the negative
value indicates (the differences are calculated subtracting the mean rate of object
and subject antecedent columns). As for overt pronouns, the difference between
the two options is almost 1 negative bias point (–0.89), which reveals a non-trivial
preference to refer back to an object antecedent. Turning to subordinate-main sen-
tence order, the results are skewed in the opposite direction: the native speakers’
interpretation of null pronouns is clearly biased toward referents in the preceding
subject position as this strong difference shows (more than 1 positive point, 1.29,
the most pronounced discrepancy) and L2 speakers, once again, show a predispo-
sition to assign overt pronouns object antecedents (a negative difference of –0.82,
similarly to what has been attested for the opposite clause order).
The two experimental groups behave in similar ways: the patterns for bilin-
guals and L2 learners generally follow those of native speakers described above;
the most striking discrepancies between the experimental groups and the con-
trol group are attested in the case of overt pronouns in subordinate-main clause
order: while controls show a clear bias towards object antecedents (nearly a -0.80
 Subject pronouns in the L2 Spanish of Moroccan Arabic speakers 

­ istinction), the two experimental groups remain rather neutral in their anteced-
d
ent choices in this specific condition (a mere -0.08 distinction for bilinguals and
-0.23 for L2 speakers).
We performed a repeated measure ANOVA on the overall 2 × 2 × 2 × 3 design
with sentence order (main-subordinate vs. subordinate-main) × type of pronoun
(null pronoun vs. overt pronoun) × antecedent choice (subject vs. object) and group
(natives, bilinguals and L2 learners) as between-subjects factor. The analyses gave a
significant main effect of pronoun (F(1, 40) = 7.157, p = .011) and the correspond-
ing interactions: order*pronoun (F(1, 40) = 4.430, p = .042); order*antecedent
(F(1, 40) = 29.137, p = .000); pronoun*antecedent (F(1, 40) = 74.430, p = .000);
pronoun*antecedent*group (F(1, 40) = 7.374, p = .002); order*pronoun*antecedent
(F(1, 40) = 12.867, p = .001); order*pronoun*antecedent*group (F(1, 40) = 12.224,
p = .000). No significant main effect for antecedent alone was detected, nor was
there an effect found for order. Finally, the effect of group was not found to be
significant. The main effect of pronoun confirms the previously reported differen-
tiation of behaviour between null and overt pronouns.
To further interpret the main effects and the interactions found, we con-
ducted separate analyses. First, a repeated measures ANOVA was conducted for
our native control group (see Figures 3 and 4). The analysis yielded a main effect of
pronoun (F(1, 16) = 7.900, p = .013); no main effect of sentence order or anteced-
ent was found. There were also significant interactions between order*antecedent
(F(1, 16) = 82.105, p = .000), pronoun*antecedent (F(1, 16) = 64.514, p = .000) and
pronoun*antecedent*order (F(1, 16) = 33.008, p = .000).

3,5

2,5

1,5

0,5

0
NULL/SUBJECT NULL/OBJECT OVERT/SUBJECT OVERT/OBJECT

Figure 3.  Antecedent preferences for main-subordinate order (native speakers)

 Aurora Bel & Estela García-Alcaraz

3,5

2,5

1,5

0,5

0
NULL/SUBJECT NULL/OBJECT OVERT/SUBJECT OVERT/OBJECT

Figure 4.  Antecedent preferences for subordinate-main order (native speakers)

Independent analyses inside each order revealed an effect both for pronoun
(F(1, 16) = 10.108, p = .006) and antecedent (F(1, 16) = 32.859, p = .000) in main-
subordinate order, whereas only antecedent (F(1, 16) = 7.095, p = .017) – as well
as the interaction between antecedent and pronoun (F(1, 16) = 65.935, p = .000) –
was found to be significant in subordinate-main order. As the results in Figure 3
and 4 reveal, overt pronouns show a constant pattern across different sentence
orders while null pronouns present a different behaviour depending on sentence
order: only a null subject pronoun in a main clause that follows the subordinate
clause seems to straightforwardly fit in the PAH as proposed for null subject lan-
guages, i.e. null pronouns display a strong bias towards subject antecedents only
in this order.
To confirm these findings in detail, we calculated paired t-tests for null and
overt pronoun antecedent preferences in the two different orders separately. While
in main-subordinate order the results were significant only for overt pronouns
(t(16) = −6.735, p = .000), in subordinate-main order the results were significant for
both null pronouns (t(16) = 9.488, p = .000) and overt pronouns (t(16) = −4.616,
p = .000).
In a nutshell, these results indicate that native speakers of Spanish display
a strong preference toward associating overt pronouns with object antecedents
regardless of sentence order. On the other hand, null pronouns only show a pre­
ference pattern when the subordinate clause precedes the main clause. Surpris-
ingly, this is not in accordance with previous findings for Spanish (Alonso-Ovalle
et al. 2002; Filiaci 2011). We can attribute this divergence to the different nature of
 Subject pronouns in the L2 Spanish of Moroccan Arabic speakers 

our task; recall that our participants are not forced to make a choice as in Alonso
Ovalle et al.’s experiment: they are asked to judge how probable both pronouns
are in different syntactic but semantically neutral contexts. Moreover, in their task
the experimental sentences were not linked by a subordinate conjunction; instead
they were juxtaposed.
Once the native speaker results were obtained, two one-way between-groups
ANOVA, one for each order (main-subordinate vs. subordinate-main), were run
on the data to compare the group’s mean scores for pronoun and antecedent con-
ditions. For main-subordinate order the results revealed a main effect of group for
null pronouns referring to an element in object position (F(2, 40) = 3.811, p = .031)
and for overt pronouns referring to an element in subject position (F(2,  40) =
4.907, p = .012). Post hoc Tukey tests confirmed that the bilingual group signifi-
cantly differed from the L2 learners group in the null pronoun-object anteced-
ent condition (p = .049) and in the overt pronoun-subject antecedent condition
(p = .036). This last condition was found to be also significant between bilinguals
and natives (p = .017). In Figure 5 bilinguals’ higher mean ratings in relation with
L2 learners and native speakers demonstrate these divergences.

4 *
3.5 *
*
3
2.5
2
1.5
1
0.5
0
BILINGUALS

BILINGUALS

BILINGUALS

BILINGUALS
L2 LEARNERS

L2 LEARNERS

L2 LEARNERS

L2 LEARNERS
L1 NATIVES

L1 NATIVES

L1 NATIVES

L1 NATIVES

NULL/SUBJECT NULL/OBJECT OVERT/SUBJECT OVERT/OBJECT

Figure 5.  Antecedent preferences for main-subordinate order

On the other hand, in subordinate-main order a main effect of group was

found in the two remaining conditions. Significant results were found for null pro-
nouns referring to subject antecedents (F(2, 40) = 8.628, p = .001) and for overt
pronouns referring to object antecedents (F(2, 40) = 4.417, p = .018). Post hoc
 Aurora Bel & Estela García-Alcaraz

Tukey tests confirmed that L2 learners differed from both the natives (p = .001)
and the bilinguals (p = .019) in the null pronoun-subject antecedent condition
and from natives in the overt-pronoun-object antecedent condition (p = .020) (see
Figure 6).

4 *
*
3.5
3
2.5
2
1.5
1
0.5
0
BILINGUALS

BILINGUALS

BILINGUALS

BILINGUALS
L2 LEARNERS

L2 LEARNERS

L2 LEARNERS

L2 LEARNERS
L1 NATIVES

L1 NATIVES

L1 NATIVES

L1 NATIVES
NULL/SUBJECT NULL/OBJECT OVERT/SUBJECT OVERT/OBJECT

Figure 6.  Antecedent preferences for subordinate-main order

In summary, the effect of pronoun type is significant for all groups; this effect is
stronger in subordinate-main clause order where null and overt pronouns display
a marked division of labour. In this specific clause order, some differences have
been found between the experimental and control groups. Finally, some discrep-
ancies have also been attested between bilinguals and L2 speakers in both clause
orders but in complementary conditions: in the former case, null pronouns refer
back to object antecedents and overt pronouns to subject antecedents in main-
subordinate order and, in the latter case, null pronouns refer back to subjects and
overt pronouns to objects in subordinate-main order.

6.  Discussion and conclusions

We first address research question 1 concerning the results of our native speak-
ers, who offer a picture of the factors governing the interpretation of pronouns
in totally ambiguous contexts in Spanish. When native speakers were asked to
judge the plausibility of the continuation given for a (null or overt) pronoun,
their a­ verage answer rating displayed a clearer pattern when the sentence order
 Subject pronouns in the L2 Spanish of Moroccan Arabic speakers 

was subordinate-main. In this case, their answers were remarkably biased as they
interpreted null pronouns as referring to the subject of the previous subordinate
sentence and overt pronouns to the object. These results are congruent with the
PAH. Nevertheless, the main predictions of the PAH for Spanish focused on
the opposite sentence order (Filiaci 2011) or on two main juxtaposed sentences
(Alonso-Ovalle et al. 2002). However, our results for main-subordinate order do
not straightforwardly fit into the PAH approach. Let us try to shed some light on
these findings.
Overall, the strongest preferences we found are the following. On the one
extreme, null pronouns in subordinate-main order are clearly interpreted as refer-
ring to subjects. On the other extreme, overt pronouns in main-subordinate order
show a visible tendency to be linked to referents in object position. In between
these two poles, two intermediate patterns are attested: a slight or almost neu-
tral bias in the null pronoun-subject antecedent condition in main-subordinate
order, and a subtle bias in the overt pronoun-object antecedent condition in
s­ ubordinate-main order.
These findings point to a non-categorical division between null and overt
pronouns. As reported in the literature review section, the pronominal choices
that null subject languages display are far from consistent across languages. This
is particularly true as far as overt pronouns are concerned since a great variability
has been observed. Furthermore, the more contextual conditions are included, the
more variability seems to occur. Carminati (2002) argued in similar fashion. In
discussing different degrees of flexibility in pronoun patterns, mainly in ambigu-
ous contexts, she suggested that the flexible behavior of overt pronouns could be
better explained by adopting a continuum in which a certain degree of variability
in the uses of the overt pronouns must be allowed.
One salient contrast between our sentence stimuli and the ones used in the
studies reported is that we have carefully controlled the implicit causality of the
verb in order to ensure that the (null or overt) pronoun is assessed in a completely
ambiguous environment. This could arguably explain some discrepancies among
the experiments. Carminati for Italian only tested globally ambiguous complex
sentences in the main-subordinate order. For the opposite order (subordinate-
main), the sentences were constructed allowing semantic-pragmatic disambigua-
tion so that the lack of global ambiguity can serve to explain the division attested
between the two pronouns. On the contrary, when ambiguity is at play at all levels
(as is the case in our experiment) the distinctions may be attenuated leading to
non-categorical gradient patterns.
Likewise, our results differ from Jegerski et al. (2011) and Alonso-Ovalle et al.
(2002). In the former study, complex subordinate sentences with preposed main
clauses were tested and, in the latter, juxtaposed inter-sentential utterances were
 Aurora Bel & Estela García-Alcaraz

evaluated. Even though all these studies deal with ambiguous anaphora, none of
them controlled for implicit causality of verbs. This could be a source of the dis-
crepancies found between our experimental results.
Having maintained ambiguity constant through all conditions, one could ask
why null pronouns display the most solid preference when the subordinate clause
precedes the main clause showing a clear association with the subject antecedent
(see Figure 4). From a processing point of view, it has been widely shown that
the speaker has access to the structural representation of a sentence and that the
processor is sensitive to main and subordinate clause order. The classic work of
Bever and Townsend (1979) and Garnham, Oakhill, and Cain (1998) supported
the idea of a basic processing difference between main and subordinate clauses.
Given that the computation of a preposed subordinate clause consumes many cog-
nitive resources, when in the following main clause a null pronoun occupies the
subject position, the same reference reading is favored because it is the neutral and
less demanding reading; hence the null pronoun is linked to the preceding subject.
As for overt pronouns, due to the large resources employed in computing a first
subordinate clause, when the overt pronoun comes the information remains active
and consequently the two preceding referents are equally potential candidates to
be recovered by the overt pronoun. Thus the interpretation of the overt pronoun is
more likely due to chance albeit there is some bias to refer back to object anteced-
ents, in accordance with the general trend for overt pronouns.
Finally, a note on the nature of the task should be made. A forced choice task
was used in the majority of offline experiments reported in the literature. Obvi-
ously, this task is simpler for participants than an acceptability judgment task.
What is more, an AJT may obscure the results and lead to wider variability in
answers and often make them more difficult to interpret; nevertheless judging all
options by means of a numerical scale can capture finer-grained contrasts and help
uncover subtle distinctions: it is possible that the participants considered interpre-
tations that would not be considered otherwise. Thus, the interpretive optionality
reflected in our results may strongly depend on the task and on our effort to hold
all factors as ambiguous as possible. For these reasons we believe that our experi-
mental proposal complements previous findings in Spanish.
While we have discussed anaphora resolution in Spanish, we have not
yet addressed the pronominal choices in native Moroccan Arabic (research
­question 5). The pre-test on anaphora resolution conducted for MA provided new
empirical data and revealed that the PAH is only partially applicable in this lan-
guage. Null pronouns show a strong tendency towards topic continuity since they
prefer to establish co-reference with elements in subject position regardless of sen-
tence order. These results confirm a relatively steady interpretative behavior with
regard to null pronouns among the different null subject languages that have been
 Subject pronouns in the L2 Spanish of Moroccan Arabic speakers 

approached from the perspective of the PAH (Italian, Spanish and Catalan). On
the other hand, overt pronouns show a more unpredictable behavior since they
were assigned to either subject or object antecedents. Thus, the pattern observed
for MA is more in line with Alonso-Ovalle et al. (2002) and Filiaci’s (2011) find-
ings for Spanish, as well as with our results for Spanish, than those reported by
­Carminati (2002), for Italian, and Mayol and Clarck (2010) for Catalan, which
showed evidence of a division of labor between null and overt pronouns. Never-
theless, if we observe preferences by clause order, the results for MA do not entirely
align with our findings for Spanish: while MA shows a parallel pattern in both
clause orders, Spanish only shows a clear preference pattern in subordinate-main
clause order and particularly for null pronouns. It is worth remembering that the
two tasks differ: a forced-choice task for MA and an AJT for Spanish. Summing up,
the preference of null pronouns in same-reference intra-sentential contexts seems
to be a robust and steady phenomenon among different null-­subject languages,
even typologically distinct (Romance and Arabic), whereas overt pronouns are
more unpredictable.
In light of the IH (Sorace 2011), pronominal anaphora resolution, as a phe-
nomenon at the syntax-discourse interface, is a vulnerable domain that could also
be affected by cross-linguistic influence according to the proposal by Hulk and
Müller (2000), as there is structural overlap in the two languages involved. Given
that pronoun preferences seem to work in similar directions in MA and ­Spanish
(with the mentioned exception of null pronouns in main-subordinate clause
order), one would expect a positive transfer, thus reinforcing the favorite options
of the target language.
We turn now to research question 2. Based on the outcomes identified above
for our control group, how do our bilingual speakers and L2 learners perform
when solving ambiguous pronouns in Spanish?
As reported in Section 5.3, no differences among groups were attested. The
lack of significant differences suggests that both bilinguals and L2 learners have
attained native-like processing strategies when solving anaphora ambiguity in
Spanish. Nevertheless, some discrepancies are found in some conditions, but this
is not unexpected if we recall that some residual indeterminacy has been observed
even in bilinguals and in near-natives when interpreting pronominal forms
(­Sorace & Filiaci 2006; Sorace et al. 2009, among others). To be more specific,
the clearest division of labor between null pronouns (selecting a subject anteced-
ent) and overt pronouns (selecting an object antecedent) has been observed in
subordinate-main order. And it is precisely here where the experimental groups
(early bilinguals and L2 learners) differ from native controls. While bilinguals
seem to have discovered the reported associations, L2 learners do not seem to
show in-depth pragmatic knowledge of the properties that regulate the use of
 Aurora Bel & Estela García-Alcaraz

both null and overt pronouns. For the opposite clause order, main-subordinate,
the three groups display similar profiles. Comparing the patterns between native
MA and L2 S­ panish in order to explore a potential transfer effect, the L2 learners
do not seem to take advantage of their first language: if they would have projected
the choices of MA for null pronouns – the element that shows most consistent
choices –, higher rates of acceptance should have been attested in L2 Spanish. Null
pronouns in the subordinate-main order are particularly striking and illuminat-
ing: null pronouns display the most extreme choice towards subject antecedents
both in L1 MA (almost 70%; see Figure 2) and in L1 Spanish (3.5 mean rate out
of 4; see Figure 6) but L2 learners behave differently from natives (see Figure 6);
thus they do not seem to apply the L1 MA rules to L2 Spanish. Bilinguals, on the
other hand, pattern typically with natives, except in the case of overt pronouns. We
will return to this issue shortly.
As for research question 3, the difficulties that L2 learners seem to encounter
in facing a linguistic domain where syntax and pragmatics converge are not new
and have also been reported in the literature. For English learners of L2 Spanish,
Keating et al. (2011) found that there was nearly free variation between null and
overt pronouns in continuity of reference contexts. This could be explained by the
residual indeterminacy in the linguistic phenomena at the interfaces shown by L2
learners or by cross-linguistic influence from L1. In our particular linguistic com-
bination, as discussed in the previous paragraph, the impact of positive transfer is
ruled out because learners seem not to benefit from microvariation. Consequently,
the optionality in pronoun usage, manifested by a wide tolerance in the acceptance
of different pronouns for different pragmatic roles, is best explained by the IH
framework.
Nevertheless, to confirm the minor weight of cross-linguistic influence, evi-
dence of Spanish learners whose L1 is not a null subject language should be tested.
In this respect, Jegerski et al. (2011) discovered an English-like strategy in the
acquisition of antecedent choices in temporal subordinated sentences in Spanish
by English speakers but not in coordinate sentences. Similarly, Roberts, Gullberg,
and Indefrey (2008) detected patterns of L1 effects in offline anaphora resolution
data but not in online data in the L2 Dutch of German and Turkish learners. These
results, together with the results of the present research, seem to point to a possible
effect of negative cross-linguistic transfer in interpreting pronouns, as seen for
English-Spanish and Dutch-Turkish combinations, but not to a positive transfer
as Spanish-Moroccan Arabic and Dutch-German combinations evidence. Sorace
(2004) also points out that cross-linguistic influence seems to be unidirectional:
from the less complex to the most complex grammar. As a partial conclusion, the
role of L1 influence is still far from clear-cut and further research is needed in
order to understand its scope.
 Subject pronouns in the L2 Spanish of Moroccan Arabic speakers 

Another important issue to highlight is the overuse of overt pronouns in

[−topic shift] contexts detected in bilinguals in main-subordinate order. This find-
ing has been broadly attested among bilinguals in previous literature (Sorace 2004;
Serratrice 2007), but goes against Rothman (2009) and Montrul and ­Rodriguez
Louro (2006), who found that both null and overt pronouns were affected by
overproduction/overacceptance in advanced developing grammars. No over­
acceptance of either null or overt pronouns was found for L2 learners in the pres-
ent study. The question of why overextention of overt pronouns especially affects
bilinguals still remains an open question.
Next, we address research question 4: Is there an effect of the different amount
of exposure on the acquisition of pronominal anaphora resolution in Spanish?
Even though the between-subjects analysis did not detect significant differences,
some specific differences appeared between sequential bilinguals and L2 learners
in the interpretation of pronouns. Since they belong, as noted above, to similar lev-
els of proficiency in Spanish, the discrepancies detected in some conditions might
be attributed to different input exposure rates. It is known from previous research
(Montrul 2008; Pires & Rothman 2009; Unsworth 2013, among others) that one
of the sources of variability outcomes in bilinguals is in the quantity and quality
of the language they are exposed to. Together with language proximity – a topic
that will soon be addressed in this section –, the amount of exposure has been
proposed as a facilitator, particularly for those linguistic domains at the syntax-
pragmatic interface (Sorace 2011), as it is the case for the one studied here. The
amount of exposure is clearly variable in our participants. Our bilingual teenagers
are regularly engaged in casual and everyday conversations whereas L2 learners’
Spanish practice is confined to the language classroom space and time; only a third
of them have visited a Spanish-speaking country but only for an average period of
one month.
This set of observations suggests that, like in the case of children (Shin  &
Cairns 2012), development has an impact on the acquisition of pronominal inter-
pretation properties and that practice is crucial when learners are faced with com-
plex form-function mappings that depend on a variety of factors that the speaker
has to integrate together, especially when ambiguity is involved.
There is also another factor that differentiates our two experimental groups
and can bring some light to the divergences observed. It is known that starting
to learn an L2 at an earlier age, which correlates with earlier and more extensive
exposure, leads to superior outcomes. This has often been used to argue in favor
of the critical period hypothesis (see Bialystok 2001 for an overview). It is obvious
that our groups differ considerably in this respect; recall that the mean age of onset
for the L2 speakers sample is 24 years, clearly after whichever age is considered
‘critical’.
 Aurora Bel & Estela García-Alcaraz

Of course, it is impossible to identify the amount of input ‘exemplars’ needed

for a successful acquisition. However, as pointed out above, if the appropriate use
of pronouns is constrained by a number of complex features, then it appears rea-
sonable to think that a notable frequency of occurrences will be necessary. What is
more, it is well attested that overt third person pronouns are relatively infrequent
in natural input whereas null pronouns, on the other hand, are numerous. The
scarcity of pronouns in spontaneous data has been invoked to explain the delay in
the L1 acquisition of overt pronouns (Shin & Cairns 2012). Thus, in the case of our
L2 learners, the underrepresentation of a linguistic element whose acquisition is
highly dependent on input practice must definitely have some cost. As a final note,
it should be added that neither formal instruction nor instructional input address
this issue explicitly.
Summing up our findings, we have provided evidence that the interpretation
of ambiguous pronouns in Spanish seems to be very sensitive to different contex-
tual conditions particularly when the semantic and pragmatic contexts are unbi-
ased in which case it is likely that all options remain opened to interpretation. The
experimental design devised in the present piece of research provides a comple-
mentary perspective to the previous research on pronoun resolution in Spanish.
We have also provided some empirical evidence that MA exhibits remarkable
parallels with antecedent preferences of pronouns in Spanish neutral non-biased
intra-sentential contexts.
With respect to comparison of groups, we did not find widely diverging pat-
terns of pronoun interpretation overall, although deviations in some conditions
were observed. Despite the parallelisms in subject pronoun interpretation in the
two languages involved, L2 learners did not take advantage of cross-linguistic
influence from L1. This finding is consistent with the predictions derived from
the IH as well as the indeterminacy of overt pronouns attested among bilinguals.
To close our conclusions, it deserves mentioning that the data and methods
afforded here cannot address the distinction of topichood and subjecthood. Two
factors that are in competition with one another and are different in nature have
been shown to influence antecedent choices of pronouns. Subjects and topics can
differ structurally since not all preverbal subjects are topics in Spanish but they
appear in the same linear order, in first sentence position. Our experimental sen-
tences, as well as the ones of the literature reported, conflate subject and topic in
placing one of the potential antecedents of the pronoun in first position. A pos-
sible way to overcome this limitation is to use presentational or topicless construc-
tions in which the postverbal subjects are introduced or presented into existence
in discourse, thus separating subjects from topics. This is not included in the scope
of the current paper and should certainly be considered for future research.
 Subject pronouns in the L2 Spanish of Moroccan Arabic speakers 

Acknowledgments

This research received financial support through  FFI2009-09349 and  FFI2012-

35058  grants (Aurora Bel) and the FPU program pre-doctoral grant (Estela
­García-Alcaraz) from the Spanish Ministry of Education. Special thanks are due
to A. R. Lara for her linguistic advice and to two anonymous reviewers for their
useful comments on an earlier version of this chapter. We would like to sincerely
thank the two editors for their suggestions and valuable comments while editing
this book. Finally, our deep gratitude goes to the students who participated in the
experiments and to the Instituto Cervantes of Marrakech for its kind and collab-
orative support.

References

Alhawary, M. (2007). Null subjects use by English and Spanish learners of Arabic as an L2. In
E. Benmamoun (Ed.), Perspectives on Arabic Linguistics XIX (pp. 217–245). Amsterdam:
John Benjamins. DOI: 10.1075/cilt.289.16alh
Alhawary, M. (2009). Arabic Second Language Acquisition of Morphosyntax. New Haven, CT:
Yale University Press.
Alonso-Ovalle, L., Fernández-Solera, S., Frazier, L., & Clifton, C. (2002). Null vs. overt pronouns
and the topic-focus articulation in Spanish. Rivista di Linguistica, 14, 151–169.
Aoun, J., Benmamoun, E., & Choueiri, L. (2010). The Syntax of Arabic. Cambridge: CUP.
DOI: 10.1017/cbo9780511691775
Ariel, M. (1990). Accessing Noun Phrase Antecedents. London: Routledge.
DOI: 10.1017/s0022226700015504
Ariel, M. (2001). Accessibility Theory: An overview. In T. Sanders, J. Schilperoord & W. Spooren
(Eds.), Text Representation: Linguistic and Psycholinguistic Aspects (pp. 29–87). ­Amsterdam:
John Benjamins.
Arnold, J., Brown-Schmidt, S., & Trueswell, J. (2007). Children’s use of gender and order-of-­
mention during pronoun comprehension. Language and Cognitive Processes, 22(4), 527–565.
DOI: 10.1080/01690960600845950
Arnold, J., Eisenband, J., Brown-Schmidt, S., & Trueswell, J. (2000). The rapid use of gender
information: Evidence of the time course of pronoun resolution from eyetracking. Cogni-
tion, 76(1), B13-B26. DOI: 10.1016/s0010-0277%2800%2900073-1
Bel, A., Perera, J., & Salas, N. (2010). Anaphoric devices in written and spoken narrative
­discourse: Data from Catalan. Written Language and Literacy, 13(2), 236–259.
DOI: 10.1075/wll.13.2.03bel
Belletti, A., Bennati, E., & Sorace, A. (2007). Theoretical and developmental issues in the syntax
of subjects: Evidence from near-native Italian. Natural Language and Linguistic Theory, 25,
657–689. DOI: 10.1007/s11049-007-9026-9
Benmamoun, E. (2000). The Feature Structure of Functional Categories. A Comparative Study of
Arabic Dialects. Oxford: OUP. DOI: 10.1017/s0022226701221367
 Aurora Bel & Estela García-Alcaraz

Bever, T., & Townsend, D. (1979). Perceptual mechanisms and formal properties of main and
subordinate clauses. In W. Cooper & E. C. T. Walker (Eds.), Sentence Processing: Psycholin-
guistic Studies Presented to Merrill Garrett (pp. 159–226). Hillsdale, NJ: Lawrence Erlbaum
Associates.
Bialystok, E. (2001). Bilingualism in Development: Language, Literacy and Cognition. ­Cambridge:
CUP. DOI: 10.1017/cbo9780511605963
Carminati, M.N. (2002). The Processing of Italian Subject Pronouns. Unpublished Ph.D. disserta-
tion. University of Massachusetts at Amherst.
Clahsen, H., & Felser, C. (2006). Grammatical processing in language learners. Applied Psycho-
linguistics, 27, 3–42. DOI: 10.1017/s0142716406060024
de Prada Pérez, A. (2009). Subject Expression in Minorcan Spanish: Consequences of Contact with
Catalan. Unpublished Ph.D. dissertation. Pennsylvania State University.
Farghaly, A. (1982). Subject pronoun deletion rule in Egyptian Arabic. In S. Gamal & R. ­Bowers
(Eds.), Discourse Analysis: Theory and Application Proceedings of the Second National Sym-
posium on Linguistics and English Language Teaching (pp. 60–69). Cairo: Center for Devel-
oping English Language Teaching, Ain Shams University.
Filiaci, F. (2011). Anaphoric Preferences of Null and Overt Subjects in Italian and Spanish: a Cross-
linguistic Comparison. Unpublished Ph.D. dissertation. University of Edinburgh.
DOI: 10.1080/01690965.2013.801502
Garnham, A., Oakhill J., & Cain K., (1998). Selective retention of information about the superfi-
cial form of text: Ellipses with antecedents in main and subordinate clauses. The Quarterly
Journal of Experimental Psychology, 51(1), 19–39. DOI: 10.1080/713755747
Garvey, C., Caramazza, A., & Yates, J. (1975). Factors influencing assignment of pronouns and
antecedents. Cognition, 3, 227–243. DOI: 10.1016/0010-0277%2874%2990010-9
Givón, T. (1984). Syntax: A Functional-Typological Introduction, Vol. I. Amsterdam: John
­Benjamins. DOI: 10.1017/s0022226700010434
Goikoetxea, E., Pascual, G., & Acha, J. (2008). Normative study of the implicit causality of 100
interpersonal verbs in Spanish. Behavior Research Methods, 40(3), 760–772.
DOI: 10.3758/brm.40.3.760
Gordon, P., Grosz, B., & Gilliom, L. (1993). Pronouns, Names, and the Centering of Attention in
Discourse. Cognitive Science, 17(3), 311–347. DOI: 10.1207/s15516709cog1703_1
Haznedar, B. (2006). Cross-linguistic interference in the bilingual acquisition of Turkish and
English: The overuse of subjects in bilingual Turkish. Paper presented at GALANA 2,
Montreal.
Hulk, A., & Müller, N. (2000). Bilingual first language acquisition at the interface between syn-
tax and pragmatics. Bilingualism: Language and Cognition, 3(3), 227–244.
DOI: 10.1017/s1366728900000353
Jegerski, J., VanPatten, B., & Keating, G. (2011). Cross-linguistic variation and the acquisition of
pronominal reference in L2 Spanish. Second Language Research, 27(4), 481–507.
DOI: 10.1177/0267658311406033
Keating, G., VanPatten, B., & Jegerski, J. (2011). Who was walking on the beach? Anaphora reso-
lution in Spanish heritage speakers and adult second language learners. Studies in Second
Language Acquisition, 33, 193–221. DOI: 10.1017/s0272263110000732
Kortobi, I. (2002). Gapping and VP-deletion in Moroccan Arabic. In J. Ouhalla & U. Shlonsky
(Eds.), Themes in Arabic and Hebrew Syntax (pp. 217–240). Dordrecht: Kluwer.
DOI: 10.1007/978-94-010-0351-3_8
 Subject pronouns in the L2 Spanish of Moroccan Arabic speakers 

Lafond, L., Hayes, R., & Bhatt, R. (2001). Constraint demotion and null subjects in Spanish L2
acquisition. In J. Camps & C. Wiltshire (Eds.), Romance Syntax, Semantics and L2 Acquisi-
tion (pp. 121–135). Amsterdam: John Benjamins. DOI: 10.1075/cilt.216.11laf
Liceras, J.M. (1988). Syntax and stylistics: More on the pro-drop parameter. In J. Pankhurst,
M. Sharwood Smith & P. Van Buren (Eds.), Learnability and Second Languages: A Book of
Readings (pp. 71–93). Dordrecht: Foris. DOI: 10.1177/026765839100700104
Lozano, C. (2006). Focus and split intransitivity: The acquisition of word order alternations in
non-native Spanish. Second Language Research, 22, 1–43.
DOI: 10.1191/0267658306sr264oa
Margaza, P., & Bel, A. (2006). Null subjects at the syntax-pragmatics interface: Evidence from
Spanish interlanguage of Greek speakers. In M. Graham O’Brien, C. Shea & J. Archibald
(Eds.), Proceedings of the 8th Generative Approaches to Second Language Acquisition Con-
ference (GASLA 2006) (pp. 88–97). Somerville, MA: Cascadilla Proceedings Project.
Mayol, L., & Clark, R. (2010). Pronouns in Catalan: Games of partial information and the use of
linguistic resources. Journal of Pragmatics, 42, 781–799.
DOI: 10.1016/j.pragma.2009.07.004
Montrul, S. (2004). Subject and object expression in Spanish heritage speakers: A case of
­morpho-syntactic convergence. Bilingualism: Language and Cognition, 7(2), 125–142.
DOI: 10.1017/s1366728904001464
Montrul, S. (2008). Incomplete Acquisition in Bilingualism: Re-examining the Age Factor.
­Amsterdam: John Benjamins. DOI: 10.1075/sibil.39
Montrul, S., & Rodríguez Louro, C. (2006). Beyond the syntax of the null subject parameter:
A look at the discourse-pragmatic distribution of null and overt subjects by L2 learners of
Spanish. In V. Torrens & L. Escobar (Eds.), The Acquisition of Syntax in Romance Languages
(pp. 401–418). Amsterdam: John Benjamins. DOI: 10.1075/lald.41.19mon
Müller, N., Kupisch, T., Schmitz, K., & Cantone, K. (2006). Einführung in die Mehrsprachigkeits-
forschung. Tübingen: Gunter Narr. DOI: 10.1515/zrs.2011.052
Paradis, J., & Navarro, S. (2003). Subject realization and cross-linguistic interference in the bilin-
gual acquisition of Spanish and English: what is the role of input? Journal of Child Lan-
guage, 30, 1–23. DOI: 10.1017/s0305000903005609
Pérez-Leroux, A.T., & Glass, W.R. (1999). Null anaphora in Spanish second language acqui-
sition: Probabilistic versus generative approaches. Second Language Research, 15,
220–249. DOI: 10.1191/026765899676722648
Pinto, M. (2006). Subject pronouns in bilinguals: Interference or maturation. In L. ­Escobar &
V.  Torrens (Eds.), The Acquisition of Syntax in Romance languages (pp. 331–352).
­Amsterdam: John Benjamins. DOI: 10.1075/lald.41.16pin
Pires, A., & Rothman, J. (2009). Acquisition of Brazilian Portuguese in late childhood: Impli-
cations for syntactic theory and language change. In A. Pires & J. Rothman (Eds.), Mini-
malist Inquiries into Child and Adult Language Acquisition: Case Studies Across Portuguese
(pp. 129–154). Berlin: Mouton De Gruyter. DOI: 10.1515/9783110215359.1.129
Roberts, L., Gullberg, M., & Indefrey, P. (2008). Online pronoun resolution in L2 discourse:
L1 influence and general learner effects. Studies in Second Language Acquisition, 30(3),
333–357. DOI: 10.1017/s0272263108080480
Rothman, J. (2009). Pragmatic deficits with syntactic consequences? L2 pronominal subjects
and the syntax-pragmatics interface. Journal of Pragmatics, 41, 951–973.
DOI: 10.1016/j.pragma.2008.07.007
 Aurora Bel & Estela García-Alcaraz

Schulte-Nafeh, M. (2004). Overt Subject Pronouns in Cairene Arabic: Pragmatic and Syntactic
Functions. Unpublished Ph.D. dissertation. The University of Arizona.
Serratrice, L. (2007). Cross-linguistic influence in the interpretation of anaphoric and cataphoric
pronouns in English-Italian bilingual children. Bilingualism: Language and Cognition,
10, 225–238. DOI: 10.1017/s1366728907003045
Serratrice, L., Sorace, A., & Paoli, S. (2004). Subjects and objects in Italian-English bilingual and
monolingual acquisition. Bilingualism: Language and Cognition, 7, 183–206.
DOI: 10.1017/s1366728904001610
Shin, N.L., & Cairns, H.S. (2012). The development of NP selection in school-age children:
reference and Spanish subject pronouns. Language Acquisition: A Journal of Developmental
Linguistics, 19(1), 3–38. DOI: 10.1080/10489223.2012.633846
Sorace, A. (2004). Native language attrition and developmental instability at the syntax-­
discourse interface: Data, interpretations and methods. Bilingualism: Language and Cogni-
tion, 7, 143–145. DOI: 10.1017/s1366728904001543
Sorace, A. (2011). Pinning down the concept of “interface” in bilingualism. Linguistic Approaches
to Bilingualism, 1, 1–33. DOI: 10.1075/lab.1.1.01sor
Sorace, A., & Filiaci, F. (2006). Anaphora resolution in near-native speakers of Italian. Second
Language Research, 22, 339–368. DOI: 10.1191/0267658306sr271oa
Sorace, A., Serratrice, L., Filiaci, F., & Baldo, M. (2009). Discourse conditions on subject pro-
noun realization: Testing the linguistic intuitions of older bilingual children. Lingua, 119,
460–477. DOI: 10.1016/j.lingua.2008.09.008
Trueswell, J., Papafragou, A., & Choi, Y. (2011). Syntactic and referential processes: What devel-
ops? In E. Gibson & N. Pearlmutter (Eds.), The Processing and Acquisition of Reference
(pp. 65–108). Cambridge, MA: The MIT Press.
DOI: 10.7551/mitpress/9780262015127.003.0004
Unsworth, S. (2013). Assessing the role of current and  cumulative  exposure in simultaneous
bilingual acquisition: The case of Dutch gender.’  Bilingualism: Language and Cognition,
16, 86–110. DOI: 10.1017/s1366728912000284
 The construal of goal-oriented motion events
by Swedish speakers of L2 Spanish
Encoding of motion endpoints and Manner
of motion

Alejandra Donoso1 & Emanuel Bylund1,2

1Stockholm University & 2Linnaeus University

The current study investigates motion event construal in Swedish speakers of

L2 Spanish. In particular, the study examines the encoding of motion endpoints
and manner of motion through elicited video clip descriptions of everyday
motion event situations. The results show that Swedish learners of Spanish
exhibit the same, high endpoint frequencies as their monolingual Swedish peers,
thus deviating from the Spanish native pattern. Moreover, the learners used the
same amount of manner verbs as Spanish natives, but were more prone to give
additional manner information in periphrastic constructions. These findings are
interpreted in relation to previous literature on the construal of motion events in
L2 learners and the notion of conceptual transfer (Cadierno & Ruiz 2006; Jarvis &
Pavlenko 2008; von Stutterheim 2003).

1.  Introduction

In the past two decades, there has been a steady increase in crosslinguistic research
on the selection and organisation of information in discourse. An ever-growing
body of research shows that speakers of different languages differ in the way
they structure information about, for example, time, space, and causation (e.g.
­Berman & Slobin 1994; Hasselgård, Johansson, Behrens & Fabricius-Hansen 2002;
Strömqvist & Verhoeven 2004). One domain that has received focused attention
in this line of research is motion events. In a generic definition, a motion event is
a situation that involves physical displacement whereby an entity occupies differ-
ent spatial positions at different temporal intervals. Current research on motion
events has shown that languages across the world are very selective in the way
they choose and structure (or construe) information about motion. Whereas some

doi 10.1075/ihll.3.09don
© 2015 John Benjamins Publishing Company
 Alejandra Donoso & Emanuel Bylund

languages regularly provide information about the Manner in which motion is

carried out (e.g. running, cycling etc.), other languages leave this information to
be inferred by the context. Moreover, in some languages it is preferred to mention
an alleged endpoint of the motion (e.g. walk to a house), whereas others leave such
information out and instead zoom in on the ongoing motion (e.g. walking).
These crosslinguistic differences constitute the main topics of two distinct
approaches to the study of motion. The approach concerned with the encod­­­­­­­
ing of motion endpoints will be labelled ‘the Grammatical Aspect approach’,
as its underlying assumption is that crosslinguistic differences in the encoding
of grammatical aspect induce crosslinguistic differences in endpoint encoding
(e.g. Athanasopoulos & Bylund 2013; Bylund 2008; Carroll & von Stutterheim
2006; Schmiedtová, von Stutterheim & Carroll 2011; von Stutterheim & Nüse
2003). The other approach is concerned with how information regarding Man-
ner and Path (the trajectory of motion) is encoded in different languages (e.g.
Donoso 2013; I­barretxe-Antuñano 2004, 2009; Jovanovic & Martinovic-Zic
2004; Naigles & T ­ errazas 1998; Slobin 1996; Talmy 1985, 2000). We will call this
approach to motion ‘the Manner and Path approach’. The Grammatical Aspect
approach and the Manner and Path approach are not necessarily competitors in
the sense that they seek to explain the same phenomenon from different theoreti-
cal stances. Instead, they focus on different aspects of motion events.
The fact that there are crosslinguistic differences in the selection and organisa-
tion of motion information has implications for second language (L2) learning. In
short, such crosslinguistic differences imply that the learner has to master aspects
of the target language that go beyond purely formal features (such as number and
gender agreement). That is, the learner has to acquire the rules and principles that
guide information structure in the L2. Research on this matter has shown that
such knowledge is not easily attainable: studies within the Grammatical Aspect
approach demonstrate that learners are prone to transferring the endpoint encod-
ing principles of their first language (L1) when describing goal-oriented motion
in the L2 (e.g. Schmiedtová & Flecken 2008; von Stutterheim 2003). Similarly,
findings from studies on the encoding of Manner and Path among L2 learners
show that L1-specific patterns may prevail even at advanced stages of learning (e.g.
Cadierno 2004; Hendricks & Hickmann 2011; Hohenstein, Eisenberg & Naigles
2006; Navarro & Nicoladis 2005).
The general aim of the present study is to investigate the construal of goal-
oriented motion events in Swedish users of L2 Spanish. Swedish and Spanish is a
language pair that so far has remained unexplored in this regard. As it turns out,
Swedish and Spanish differ both with regards to endpoint encoding preferences
as well as the encoding of Manner and Path. For this reason, the study seeks to
 The construal of goal-oriented motion events by Swedish speakers of L2 Spanish 

combine the frameworks offered by the Grammatical Aspect approach and the
­Manner and Path approach in order to gain a more nuanced picture of the vari-
ables that govern L2 learning in this domain. Even though it is not uncommon that
languages exhibit such ‘double’ contrasts, there is, to the best of our knowledge, no
study to date that has attempted to explain learner behaviour in the domain of
motion by integrating the Aspect approach with the Manner and Path approach.

2.  Background

2.1  Motion endpoints and grammatical aspect in L1 and L2 speakers

Human beings often ‘engage’ in motion events with the intention to reach a dif-
ferent location than the one we started from. There is, in other words, often a spe-
cific goal or an endpoint to our motion. Research within the Grammatical Aspect
approach has shown that speakers of different languages differ with regards to
the extent to which they actually mention endpoints when talking about motion.
For example, when showing a scene with two people walking towards a house,
Athanasopoulos and Bylund (2013) found that Swedish speakers were more prone
to mentioning the endpoint than were English speakers. The following examples
were typical of the two language groups’ descriptions of the scene:
(1) Two people are walking
(2) Två människor går till ett hus
two humans walk-pres to a house
‘two humans walk to a house’

Apart from the obvious difference concerning endpoint encoding, the two exam-
ples above also differ with respect to the selection of the time span during which
the event unfolds: in the English example, the predicate is marked with progres-
sive aspect, and thus the ongoing phase of the event is conveyed. In the Swedish
sentence, in contrast, there is no aspectual morphology.
Evidence from a number of crosslinguistic investigations show that speakers
of English, Modern Standard Arabic, Russian, and Spanish are less prone to men-
tioning, allocating visual attention to, and remembering motion endpoints than
are speakers of Afrikaans, Dutch, Norwegian, and Swedish. These clusters of end-
point behaviour are consistent with the clusters given on the basis of grammatical
aspect in these languages. Even though the aspectual systems of Arabic, English,
Russian, and Spanish may differ with regards to the exact semantic boundaries
between subcategories (i.e. the Spanish imperfective may include features that
 Alejandra Donoso & Emanuel Bylund

are not present in the Arabic imperfective, such as habituality), their common
denominator is that they mark the predicate for ongoingness (imperfective and/
or progressive aspect) in a productive and obligatory fashion. Afrikaans, Dutch,
Norwegian, and Swedish, in contrast, do not have grammatical aspect. This means
that these languages do not have the same structural means nor the obligation to
encode ongoingness. Of course, ongoingness may be optionally coded through
lexical circumlocution or serial verbs, but the use of such means in motion event
descriptions is not obligatory and has proven scarce (Bylund, Athanasopoulos &
Oostendorp 2013; Carroll, von Stutterheim & Nüse 2004). Crucially then, ‘aspect
languages’ and ‘non-aspect languages’ do not necessarily differ in what they are
able to express, but in the relative ease and frequency with which they can express
a concept such as ongoingness.1 In aspect languages, the concept of ongoingness
has a higher degree of codability than in non-aspect languages.
Building on the notion that grammatical categories comprise a set of sche-
mata for talking and thinking about reality (e.g. Langacker 2000; Slobin 2003), it
has been suggested that the fact that a language has grammatical aspect implies
that the ongoing phase of events is treated as prominent in the conceptualisa-
tion of states of affairs (Carroll et al. 2004, p. 185). The implication of this is that
speakers of aspect languages acquire a greater sensitivity towards ongoingness,
and therefore, when watching an unfolding motion event these speakers are more
prone to depict it as ongoing, while omitting reference to the possible endpoint
of the motion. The same logic applies to the behaviour of speakers of non-aspect
languages: in the absence of a grammatical feature that directs the attention to
ongoingness, these speakers are more inclined to adopt a holistic perspective of an
unfolding motion event, according to which the event is presented in its entirety
with the possible endpoint included (Bylund & Jarvis 2011).
At present, a handful of studies have investigated whether speakers, whose L1
is an aspect language, transfer the native endpoint patterns when learning a non-
aspect-language, and vice-versa (e.g. Schmiedtová et al. 2011). von ­Stutterheim
(2003), for example, examined endpoint encoding patterns among advanced
English speakers of L2 German, and advanced German speakers of L2 English.
No information was given about the learners’ current place of residence (L1 or
L2 setting). The participants were shown short video clips depicting everyday
single-event scenes, such as a man walking, a person cycling, a car driving, etc.
The analyses of the event descriptions revealed that the L2 speakers’ endpoint

.  Henceforth, we will use the term “aspect language” to refer to languages in which the
aspectual category of ongoingness is grammaticised, and “non-aspect language” to refer to
languages in which this notion is not grammaticised.
 The construal of goal-oriented motion events by Swedish speakers of L2 Spanish 

frequencies were significantly different from the target language frequencies, but
not from those given by the monolingual speakers of their L1s. In other words,
­German learners of English were prone to mentioning endpoints to the same
extent as ­German monolinguals, and English learners of German behaved like
English monolinguals. In view of their findings, von Stutterheim and colleagues
have suggested that L1 conceptual patterns are highly resistant to restructur-
ing (e.g. Carroll & von Stutterheim 2006) (see however Bylund 2009; Bylund &
­Athanasopoulos 2014).
The type of transfer documented in these studies is commonly referred to as
conceptual transfer (Jarvis & Pavlenko 2008). The kind of knowledge that is being
transferred is, in other words, language-specific patterns of selecting and structur-
ing information relative to motion. These patterns of event construal have been
acquired in the course of L1 acquisition, whereby the speaker has learnt to attend
to and encode certain aspects of reality in accordance with the lexical and gram-
matical categories available in the L1 (Slobin 1996). When acquiring a L2 that has
distinct event construal patterns from the L1, the learner has to overcome the reli-
ance on L1 patterns when presenting information and instead attain the new pat-
terns specific to the target language. If the learner relies (in part or wholly) on the
construal patterns of the L1 then, the behaviour is labelled L1-based conceptual
transfer (Jarvis & Pavlenko 2008).

2.2  Path and Manner of motion in L1 and L2 speakers

Inspired by the work of Talmy (1985, 1991, 2000), the second main approach to the
study of motion events is the “Manner and Path approach” (e.g. Berman & ­Slobin
1994; Ibarretxe-Antuñano 2004, 2009; Naigles, Eisenberg, Kako, Highter  &
McGraw 1998; Ragnarsdóttir & Strömqvist 2004; Slobin 1996, 2003; Zlatev &
Y
­ angklang 2004). This approach investigates how languages encode information
about the fashion (Manner) in which different objects move in space and how their
trajectories (Path) are conveyed by linguistic means. Path and Manner do not have,
however, equal importance in the expression of motion events. While Path is the
core schema in a motion event, Manner is an optional semantic component. This
typology divides the languages of the world into two main groups depending on
how they encode Path. The first group is Satellite-framed languages (S-languages)
and the second Verb-framed languages (V-languages).2 In the first group, Path is
expressed in a satellite (“a grammatical category of any constituent other than a

.  A third group, equipollently-framed languages (e.g. Thai) has also been identified. See
Slobin (2004) for further discussion.
 Alejandra Donoso & Emanuel Bylund

noun-phrase or prepositional phrase complement that is in a sister relation to the

verb root”, Talmy 2000, p. 102); in the latter, into the main verb.
Several scholars have proposed that Swedish belongs to the group of S-lan-
guages (e.g. Ragnarsdóttir & Strömqvist 2004) in which Path is typically expressed
in satellites, for example ut (out) or ner (down), and Manner is encoded in the
main verb. Spanish, on the other hand, would belong to the second group, in
which Motion and Path conflate in the main verb, whereas Manner tends to be
conveyed separately, frequently by an adjunct, a gerund, a noun phrase or another
kind of expression (Stam 2010; Talmy 1985, 2000).
(3) Flaskan flöt ut från grottan.
(Flöt: Manner and Motion; ut: Path)
‘The bottle floated out from the cave.’
(4) La botella salió flotando de la cueva.
(Salió: Path and Motion; flotando: Manner)
‘The bottle exited floating from the cave.’
In the Swedish example, Manner is expressed in the main verb, i.e. Flaskan flöt ut
från grottan (‘the bottle floated out of the cave’), and Path is conveyed in a satellite
(ut, ‘out’). Spanish, on the other hand, lexicalizes Path in the main verb, salir (‘to
go out’), and Manner is encoded in a gerund (flotando, ‘floating’).
Although Manner in Spanish may be expressed in a gerund, it is not a core
element (as is Path), and can therefore be left out in speech. However, when cog-
nitively relevant, Manner can be expressed by different means, even in the main
verb, though this encoding option may be limited to situations in which the trajec-
tory of the moving entity does not include a telic Path phrase (see below about the
“boundary-crossing constraint”).
Still, Manner verbs tend to be less numerous in V-languages than in S­ -languages
(Jovanovic & Martinovic-Zic 2004). Even more, and especially regarding ­Spanish,
explicit Manner descriptions are scarce compared to those of S-languages and
other V-languages, such as Basque (Ibarretxe-Antuñano 2004). This scarcity of
Manner descriptions in Spanish resides in the fact that Manner is an optional
component but also in the difficulty to readily express it in speech (Slobin 2004).
Besides the differences regarding the expression of Path and Manner between
V-languages and S-languages, the ‘boundary-crossing constraint’ (Aske 1989;
­Cadierno 2010; Pedersen 2012; Stam 2010) represents a central factor that differenti-
ates both language groups. This restriction, originally proposed by Aske (1989), states
that V-languages do “not generally allow Manner-of-motion verbs to take a direc-
tional complement and to function as bounded predicates” (­Zubizarreta & Oh 2007).
(5) ?La bailarina bailó en la habitación. (With directional meaning)
‘The dancer danced in the room.’
 The construal of goal-oriented motion events by Swedish speakers of L2 Spanish 

Because of this constraint, V-languages would be less flexible to generate phrases

that both conflate Manner and Path in the main verb (bailar, ‘to dance’) and that
are directly tied to telic endpoints (en la habitación, ‘in the room’, with directional
meaning). In other words, when crossing a boundary, the usual way to describe
it in Spanish would be by using a bare Path verb (without a satellite) plus a telic
prepositional phrase, as in la mujer entró en la habitación (‘the woman entered the
room’). However, it is important to take into consideration that in some cases it
is actually possible to express a bounded event (a motion situation containing a
direction and an endpoint) by using a Manner verb (e.g. Cadierno 2010; Pedersen
2012).
To summarize, Swedish and Spanish differ considerably in how they encode
Manner and Path. Overall, Swedish presents fairly complex Path descriptions with
an extensive use of directional satellites, endpoint encoding and a wider range
of Manner verbs than Spanish (Bylund 2008; Donoso 2013; Sjöström 1990).
Moreover, Swedish L1 speakers seem to follow a highly schematic pattern for
motion encoding, which is usually represented by the structure of [SUBJ V OBL]
(­Pedersen 2012). This schema would apparently facilitate motion descriptions
containing information not only about Manner and Path, but also possibly about
the Ground (i.e. a reference point in space to which the Figure moves), in the form
of a bounded predicate.
Several studies on L2 Spanish have investigated how differences in the L1 may
have an impact in learning in both speech (e.g. Cadierno 2004; Cadierno & Ruiz
2006; Hijazo-Gascón 2011; Hohenstein, Esisenberg & Naigles 2006; ­Larrañaga,
Treffers-Daller, Tidball & Ortega 2012) and gesture (Negueruela, Lantolf, J­ ordan &
Gelabert 2004). The general results speak both for and against a possible cross-
linguistic influence, which can take two different shapes that may be interrelated
or act separately. The first mode involves transfer that originates at the level of lin-
guistic knowledge and implies that structures from the L1 are simply transferred
to the L2; the second one originates at the conceptual level and implies that the
way speakers conceptualize and encode motion events in their L1 affects how they
express the same type of events in their L2.
As earlier SLA research has shown (Ijaz 1986; Jarvis & Odlin 2000; Odlin
1989, 2003), L2 learners are prone to transfer different aspects of their L1, includ-
ing linguistic means and patterns of thought. For instance, Cadierno’s work (2004)
on written narratives of the book Frog, Where Are You? (Mayer 1969) found that
Danish learners of Spanish used fewer motion verb types than the Spanish L1
speakers; second, some of the learners (in particular, those at the intermediate
level) exhibited a “satellization” of the Spanish locative constructions, i.e. the use
of redundant and anomalous Path particles not found in the Spanish speakers’
data; third, the learners added more Ground adjuncts to the motion verbs (which
 Alejandra Donoso & Emanuel Bylund

could include the source, the medium, and the goal of movement). Consequently,
Cadierno assumes that the L1 patterns for motion encoding may still play a role in
adult L2 learning, and that proficiency levels seem to interact with the impact of
crosslinguistic influence. In sum, her results show that the learners’ L1 seems to be
present in some traits of their L2 use but not in others: the Danish learners proved
to be able to encode both Manner and event conflation in the same fashion as the
Spanish natives but they were still influenced by their L1 when encoding Path and
Ground (Cadierno 2004, p. 42).
In a similar vein, Navarro and Nicoladis (2005) using oral narratives of a
Pink Panther cartoon among proficient English speakers of L2 Spanish (most
of them students at University level), found that learners and native Spanish
speakers use Path verbs to the same extent. Secondly, they found that L2 speak-
ers produce more Path-intransitive verbs followed by a post-verbal phrase,
which could contain information about goal, source and location. This latter
result was interpreted as L1 influence given that the L2 narratives resembled
the patterns of the English narratives included in their study. The authors also
found that L2 speakers produced fewer post-verbal Manner expressions than
the group of native Spanish speakers, which the learners compensated for by
using somewhat more Manner verbs. At the opposite end, the native S­ panish
speakers resorted more to bare Path intransitive verbs and produced fewer
Manner verbs, which they balanced by including a larger number of post-verbal
adverbials and gerunds. In sum, the authors conclude that L2 Spanish speakers
had almost fully achieved the L1 Spanish pattern to describe motion events,
which is consistent with the above mentioned findings of Cadierno (2004).
Similar results have been obtained by Hijazo-Gascón (2011) on German speak-
ers of L2 Spanish, and Cadierno and Ruiz (2006) on Danish learners of Spanish.
Surprisingly, and contrary to these results, Larrañaga et al. (2012) studying oral
descriptions of a group of English learners of Spanish, found that the encoding
of Manner, in particular in boundary-crossing situations, actually carries seri-
ous difficulties, even among learners who had been living in a S­ panish-speaking
country prior to data collection. These subjects were prone to use M ­ anner verbs
in combination with a Path or locative preposition (e.g. the preposition en,
which in Spanish does not entail a directional meaning) thus evading the use of
the gerund and conflating Manner and Motion when confronted to a boundary
crossing situation. Larrañaga et al. (2012) attribute the difficulties to the scar-
city of manner verbs in Spanish.
The reported findings, in conclusion, differ in some points, especially regard-
ing how L2 speakers provide information about Manner and goal, source and
location. Although learners of Spanish seem to be able to encode Manner and
Path in a target-like way, the extent to which they do so depends in part on factors
such as L1-background.
 The construal of goal-oriented motion events by Swedish speakers of L2 Spanish 

3.  Aims and scope of the present study

Against this background, the overall aim of the current study is to investigate
how Swedish speakers of L2 Spanish construe goal-oriented motion. As seen
above, Swedish and Spanish differ with regards to the encoding of motion events.
­Swedish, being a non-aspect language, presents a higher tendency than Spanish to
encode endpoints for goal-oriented motion events. Moreover, since Swedish is an
S-language, it has a greater repertoire of Manner verbs than Spanish, and ­Swedish
speakers allegedly use Manner verbs more frequently and in slightly different con-
texts than Spanish speakers.
With these crosslinguistic differences in mind, the study seeks to address
whether and to what extent these contrasting patterns of event construal pose a
challenge to Swedish learners of Spanish. The current study design differs some-
what from previous studies on Spanish L2 acquisition: first, from a more general
viewpoint, Swedish learners of Spanish are a population that, to the best of our
knowledge, is very scarcely studied. Another novelty is the fact that the partici-
pants reside in the L2 environment. As seen above, most studies on the acquisition
of motion construal in Spanish have dealt with foreign language learners who are
not immersed in a Spanish-speaking setting, and therefore exhibit different pat-
terns of input and use of the target language.
The main research questions pursued in the study are as follows:
1. Do Swedish speakers of L2 Spanish encode motion endpoints more often than
Spanish L1 speakers?
In view of the previous findings on endpoint encoding patterns in L2 speak-
ers (e.g. von Stutterheim 2003), it could be expected that the Swedish learners
will be more prone than Spanish L1 speakers to taking a holistic perspective
of goal-oriented motion, according to which endpoints are included.
2. Do Swedish speakers of L2 Spanish encode information about Manner of
motion to the same extent as Spanish L1 speakers?
If transfer from the L1 occurs in this regard, the high frequency with which
Manner of motion is encoded in Swedish could be expected to prompt the
learners to encode Manner information to a higher degree than Spanish L1
speakers.
3. Does the encoding of information relative to Manner of motion vary as a
function of endpoint encoding?
Spanish and Swedish exhibit different typological traits with regard to the
encoding of Manner in combination with an event endpoint (as seen above).
If the Swedish learners combine endpoints and Manner information in the
same way as Spanish L1 speakers, then they can be said to have assimilated the
target language principles.
 Alejandra Donoso & Emanuel Bylund

4.  Method

4.1  P
 articipants
Seventeen Swedish speakers of L2 Spanish participated in the study. These indi-
viduals were living in Chile at the time of testing, and Spanish was their primary
language of communication. Their mean age was 35 years and their mean length
of residence in Chile was 8 years. Seven of these individuals had received some
kind of formal instruction in Spanish, either in Sweden prior to departure or in
Chile. All of them had learnt Spanish as adults.
Two groups of L1 Spanish speakers and L1 Swedish speakers were included
for comparative purposes. The Spanish speakers (N = 12) were originally from
Argentina and Chile, and were short-term visitors in Sweden at the time of testing.
The Swedish speakers (N = 12) were living in Sweden at the time of testing. Both
native speaker groups were in their mid-thirties.

4.2  Material
A set of video clips projecting goal-oriented motion events was used to elicit
motion event descriptions. The clips had been compiled by the research team
of Mary Carroll and Christiane von Stutterheim, and have been widely used in
other studies (e.g. Athanasopoulos & Bylund 2013; Bylund 2008; Carroll et  al.
2004; Schmiedtová et al. 2011; von Stutterheim 2003). The video clips showed
an entity (a vehicle or a person) moving along a trajectory at the end of which
there was a possible endpoint (e.g. a building). The current set contained a total
of 41 video clips. These were distributed into the following categories: 8 clips with
a high degree of goal-orientation (category I), 10 with an intermediate degree of
goal-orientation (category II) and 8 clips with a low degree of goal-orientation
(­category III). In addition, there were 14 distractor clips depicting people or ani-
mals involved in activities void of trajectory-based motion, such as a man fly-
fishing by a river. The distractors were not included in the analyses.
The first category (I), consisting of scenes with a high degree of orientation,
showed how the endpoint of the trajectory was reached, for example, a seagull
landing on a perch. Given that the arrival at the endpoint of these events was
overtly shown, speakers of different languages are not expected to differ sig-
nificantly in the degree to which they mentioned endpoints in these clips (von
­Stutterheim, Andermann, Carroll, Flecken & Schmiedtová 2012). Half of the clips
in this category contained boundary-crossing motion (e.g. a dog walking into a
greenhouse).
The second category (II), with an intermediate degree of goal-orientation,
contained scenes in which there was a visible, possible endpoint for the motion,
 The construal of goal-oriented motion events by Swedish speakers of L2 Spanish 

but the arrival at it was not overtly shown. The subjects’ references to these events
did not need to include a mention of the goal of the motion. An example of this
is the situation of a man on a horse carrier on the countryside: on this item, the
subjects could say either a man riding on a horse-drawn carriage or a man riding
on a horse-drawn carriage to a village.
Finally, category III with a low degree of goal-orientation, consisted of scenes
depicting the motion of an entity along a trajectory without a clearly discernable
endpoint. An example of this is the clip with a jeep driving in the desert: in this
scene, there is no conceivable endpoint except for perhaps a mountain range at
the horizon.

4.3  P
 rocedure
The subjects were administered the tests individually by a native speaker of the rel-
evant language. The participants were shown the clips on a computer monitor and
were asked to describe what was happening in each scene, following the prompt
¿Qué pasa en la escena? (‘What happens in the scene’) or Vad händer i scenen?
(‘What happens in the scene?’) as soon as they recognised the type of situation.
Moreover, they were told to focus on the event and not to describe the scene in
terms of, for example, weather conditions or the protagonist’s clothing.
Each participant’s descriptions of the video clips were audio-recorded and
transcribed. The descriptions were then classified according to endpoint encoding
and perspectivation. Locative expressions containing the moving entity’s arrival at
or intention of arriving at a goal were counted as endpoint verbalisations. Exam-
ples of such expressions in Spanish are ir a una parada de autobuses (‘to go to a
bus stop’), caminar hacia un convento (‘to walk towards a convent’), meterse en
un invernadero (approx. ‘to go into a greenhouse’). The descriptions were also
coded with respect to the use of the Spanish progressive form (i.e. the gerund):
verb phrases containing a gerund form (either sole gerund or combined with an
auxiliary) were classified gerund occurrences.
The scene descriptions were further classified according to whether they con-
tained Manner verbs or not. Examples of Swedish verbs of Manner are gå (‘to
walk’), köra (‘to drive’), and springa (‘to run’). Examples of Spanish Manner verbs
are caminar (‘to walk), conducir (‘to drive’), and correr (‘to run’). In the analy-
ses, it was also taken into account whether the Spanish descriptions (both L1 and
L2) contained information about Manner of motion outside the main verb, such
as una mujer en bicicleta yendo por la calle (‘a woman on a bicycle going on the
street’). In this example, the motion verb ir (yendo) is void of any information as
to the Manner of motion, but the complement en bicicleta (‘on a bicycle’) conveys
information about the way in which the woman travels. Another example of how
 Alejandra Donoso & Emanuel Bylund

Manner can be conveyed is through a verbal periphrasis, such as in the sentence

un perro entra corriendo (‘a dog enters running’, i.e. ‘a dog runs in’), where cor-
riendo is the gerund of the verb correr (‘to run’). This type of information about
Manner of motion was labelled ‘periphrastic Manner’.

5.  Results

The findings will be presented in the order of the research questions posed in
the section above. Firstly, the results on endpoint encodings will be presented.
­Secondly, the patterns of use of Manner verbs and periphrastic encoding of
­Manner are analysed. Finally, the relationship between the encoding of endpoints
and Manner of motion is assessed.

5.1  Endpoint encoding in L2 Spanish

An analysis of the participants’ scene descriptions showed that the Spanish L1
speakers encoded event endpoints 27.2 % (SD 10.2) of the time, whereas the
­Swedish L1 speakers did so 50 % (SD 16.4) of the time. The L2 Spanish speakers
were found to exhibit an endpoint frequency that fell in between the two mono-
lingual groups, 42.5 % (SD 18.0). A one-way Analysis of Variance (ANOVA)
indicated that there was a significant difference between groups, F (2, 38) = 6,51,
p < .01. Post-hoc tests (Tukey’s) revealed that the Swedish L1 speakers encoded
significantly more endpoints than the L1 Spanish speakers (p < 01), thus con-
firming previous research on endpoint encoding frequencies in aspect and non-
aspect languages in general (Athanasopoulos & Bylund 2013; von Stutterheim
et al. 2012), and in Spanish and Swedish in particular (Bylund 2008). Interestingly,
the L2 speakers were found to mention significantly more motion endpoints than
the L1 Spanish speakers (p < .05), but not than the L1 Swedish speakers (p > .05).
When confronted with a scene showing, for example, two women walking towards
a building, the L2 speakers were more prone than the Spanish L1 group to describe
the event as dos mujeres caminando hacia un edificio (‘two women walking towards
a building’).
The endpoint encodings were further analysed according to category of end-
point orientation (i.e. I = high endpoint orientation; II = medium endpoint ori-
entation; III = low endpoint orientation), with the intention to see whether the
group differences in endpoint frequencies were found across all three categories
or not. These results are presented in Table 1. Three separate ANOVAs revealed
significant differences in all three categories, F (2, 38) = 4.76, p < .02 (Category I);
F (2,  38) = 4.68, p < .02 (Category II); F (2, 38) = 4.56, p < .02 (Category III).
 The construal of goal-oriented motion events by Swedish speakers of L2 Spanish 

Post-hoc tests (Tukey’s) revealed that the monolingual L1 Spanish and L1 ­Swedish
speakers differed significantly from each other in each category (p < .02). As for
the L2 speakers, it was found that they encoded significantly more endpoints than
the Spanish L1 speakers only in category II (p < .02) but not in categories I or III
(p > .05). This means that the L2 endpoint frequencies diverged with the L1 fre-
quencies in the clip category where there was a movement towards a goal, but the
goal was not reached.

Table 1.  Endpoint frequencies (%) according to endpoint orientation category

Category I Category II Category III

Spanish L1 55.2 (20.9) 22.5 (13.6)  5.2 (8.4)

Spanish L2 72.3 (19.4) 42.9 (27.3) 11.8 (16.2)
Swedish L1 79.2 (19.5) 48.3 (20.4) 22.9 (16.7)

In order to see whether the L2 speakers used the gerund verb form to the
same extent as the Spanish L1 speakers, the scene descriptions were further ana-
lysed with respect to the use of this form. Overall, it turned out that both the L1
group and the L2 group used the gerund to a similar extent, 72 % (SD 16) and 71
% (SD 17), respectively. The L2 speakers’ use of the gerund was further analysed
in the descriptions that contained endpoints, with the intention to see whether
their behaviour was consistent with the native speakers’ in this particular regard
too. The results showed no substantial differences between the two groups’, as they
exhibited similar frequencies: 68 % (SD 22) (Spanish L1); 71 % (SD 23) (Spanish
L2). This means that even when encoding event endpoints, the L2 speakers used
the gerund form to a similar extent as the native speakers, such that both groups
were equally likely to describe the scene above as dos mujeres caminando (‘two
women walking’), as opposed to dos mujeres caminan (‘two women walk’).

5.2  Manner of motion in L2 Spanish

The next main step in the analysis involved examining the scene descriptions with
regards to the encoding of Manner of motion. To this end, the amount of Manner
verbs was calculated across all three groups. The Swedish L1 speakers were shown
to use Manner verbs in 79.2 % (SD 14.0) of their motion descriptions, whereas the
Spanish L1 speakers did so in 29.0 % (SD 7.4) of theirs. For example, when con-
fronted with a scene of a car driving on a country road, the Swedish speakers were
more prone to say en bil kör på landet (‘a car drives on the countryside’), whereas
the Spanish speakers would say un coche yendo por el campo (‘a car going on the
countryside’). This result aligns with previous findings showing that speakers of
 Alejandra Donoso & Emanuel Bylund

S-languages are more prone to encode Manner information into the main verb
than are speakers of V-languages. The L2 speakers indeed adhered to the Spanish
pattern, using Manner verbs in only 26.5 % (SD 8.8) of their motion descriptions.
An ANOVA revealed statistically significant differences between groups, F (2, 38)
= 108.3, p < .001, with the L1 and L2 Spanish-speaking groups differing signifi-
cantly from the Swedish L1 group (p < .001), but not from each other (p > .05)
(Tukey’s).
Besides these results concerning Manner verb frequency in the three groups,
a closer look at the verbs used by the Spanish-speaking groups showed that the
L1 speakers on average used 5.1 (SD 1.4) different Manner verbs, whereas the L2
speakers on average used 5.2 (SD 1.7). There was, in other words, no visible dif-
ference between the groups in this respect with regard to the number of different
Manner verbs the groups used. However, a closer look at the event descriptions
showed that even though the L2 speakers used the same number of Manner verbs
and to the same extent as the L1 Spanish speakers, the actual verbs they used were
slightly different. This is seen, for example, in a scene showing two men climbing
a rock wall. Here, the verbs trepar (‘to climb’) and escalar (‘to climb’) are found in
the L2 speaker group, whereas only escalar is found in the L1 speaker group. In this
particular example, the use of trepar to describe the kind of rock-climbing shown
in the scene is not very felicitous, since this verb is more often used when referring
to climbing something which is not as high as a hill or a mountain. This deviant
use of some Spanish manner verbs might be related to difficulties with Manner/
Path; however, we do have to acknowledge that an alternative explanation could
be related to lexical difficulties.
Apart from looking at Manner information as conveyed through the verb, we
also examined periphrastic Manner encoding in the L1 and L2 Spanish-speaking
groups. Whereas the native speakers encoded Manner periphrastically in 9.6 %
(SD 6.2) of their scene descriptions, the L2 speakers did so 14.9 % (SD 10.3) of the
time. This difference did not, however, reach statistical significance, t(29) = 1.58,
p  > .05. On average, then, the two groups would produce descriptions like un
hombre en bicicleta yendo por la calle (‘a man on bicycle going on the street’) in
approximately 9–15 % of the times, whereas the rest of the time they would simply
say un hombre yendo por la calle (‘a man going on the street’).

5.3  Manner information in endpoint encodings

The last main step in the analysis consisted in exploring whether there was a
relationship between the expression of Manner and endpoint encoding. In this
regard, we first looked at whether there were any group differences with respect to
the proportion of Manner verbs in those scene descriptions containing ­endpoint
 The construal of goal-oriented motion events by Swedish speakers of L2 Spanish 

e­ ncodings. It turned out that 77.3% (SD 16.2) of the Swedish L1 endpoint encod-
ings contained Manner verbs, whereas the corresponding number for the L1
­Spanish encodings was 33.5 % (SD 26.2). As for the L2 Spanish speakers, 26.7 %
(SD 12) of their endpoint encodings contained Manner verbs. These distributions
are, in other words, very similar to the ones documented above on general use of
Manner verbs in the three groups. An ANOVA indicated that these differences
were statistically significant, F(2, 38) = 29.4, p < .001, such that the L1 and L2
Spanish patterns were significantly different from the L1 Swedish pattern (p < .01)
but not from each other.
Next, the Spanish L1 and L2 groups were examined with regards to their use
of periphrastic Manner in scene descriptions that contained endpoint encodings.
The native speaker group was found to convey this kind of Manner information in
9.7 % (SD 5.9) of their endpoint encodings, whereas the L2 speaker group did so
in 21.2 % (SD 8.6) of theirs. An independent samples t-test confirmed that these
differences were statistically significant, t(29) = 2.2, p < .05. This means that the L2
speakers were more likely to produce endpoint encodings in which information
about Manner was given periphrastically (e.g. en bicicleta ‘on bicycle’, en auto ‘in
car’, etc.).
Summarising the results, it was found that the L2 speakers encoded signifi-
cantly more motion endpoints than the L1 speakers. Their use of the gerund, on
the other hand, did not differ from that of the native speakers, neither at a general
level (across all event descriptions), nor at a more specific level (in the descrip-
tions containing event endpoints). As for Manner of motion, it was found that
the L2 speakers used Manner verbs to the same extent as the L1 Spanish speakers,
and they also encoded Manner periphrastically to the same extent. However, it
was documented that the L2 speakers used slightly different verbs in some scenes,
which produced anomalous event descriptions. Finally, it was found that the L2
speakers’ encoding of Manner information in scene descriptions with endpoints
was similar to that of L1 speakers in terms of the extent to which they used M ­ anner
verbs. However, the L2 speakers tended to encode periphrastic Manner more often
in combination with endpoints than did the Spanish L1 speakers.

6.  Discussion

This paper asked three questions with regards to motion event construal in ­Swedish
speakers of L2 Spanish: First, do Swedish speakers of L2 Spanish encode motion
endpoints more often than Spanish L1 speakers? Second, do Swedish speakers of
L2 Spanish encode information about Manner to the same extent as Spanish L1
speakers? And third, does the encoding of information relative to Manner vary as
 Alejandra Donoso & Emanuel Bylund

a function of endpoint encoding? In what follows, we will discuss the results with
regards to each of these questions.
Starting with the issue of endpoint behaviour, the finding that the L2 speakers
encoded motion endpoints to the same elevated extent as monolingual Swedish L1
speakers is taken as evidence of L1-based conceptual transfer. That is to say, when
speaking Spanish, the L2 speakers are more prone to adopting an event perspec-
tive by which motion endpoints are included. This they do to the same extent as
their Swedish monolingual peers. Interestingly, however, it was also found that
even though the L2 speakers in general differed from the Spanish L1 speakers, a
more nuanced picture of the L2 endpoint behaviour emerged when the endpoint
encoding were analysed according to category of endpoint orientation. Here, it
was found that the learners really only differed from the L1 Spanish speakers in
the category that shows movement towards a goal, but the goal is not reached
(category II). In categories I and III, the L2 speakers converged with the target lan-
guage patterns. This finding suggests that the L2 speakers did not indiscriminately
produce more endpoints, but they did so only in certain types of motion scenes.
The finding that L2 speakers transfer endpoint encoding patterns from the L1
is consistent with previous literature on the topic. For example, von Stutterheim
(2003) found that advanced learners of German and English seemed to rely on
L1 patterns when encoding endpoints in their L2. Interestingly, von Stutterheim
noted that the German speakers of L2 English were nonetheless closer to the target
language endpoint frequencies than were the English learners of German. A pos-
sible explanation to this differential behaviour, von Stutterheim suggests, is the
fact that the English progressive form increases the German learners’ sensitivity
towards ‘ongoingness perspectives’ of events, according to which event endpoints
are excluded. In German, in contrast, there is no verbal form that has an equally
salient function, but instead a simple present tense that is used to denote both
ongoing and non-ongoing events.
Applying this reasoning to the current study, one could expect that the pres-
ence of a verb form in Spanish that denotes ongoingness (i.e. the gerund, and also
the imperfective) would lead learners of Spanish to also attain increased sensi-
tivity to ongoingness perspectives, and therefore they would become less prone
to encoding endpoints. The current results on the learners’ frequency of use of
the gerund furthermore seems to suggest that they were using the gerund to the
same extent as the Spanish L1 speakers, even when describing motion endpoints.
The fact remains, however, that the L2 speakers’ endpoint frequencies were equal
to those of the monolingual Swedish speakers. There is an important difference
between the English progressive form and the Spanish gerund that may explain
the apparent discrepancy between the current finding and that of von Stutterheim
(2003): whereas the English progressive is obligatorily used to refer to ongoing
 The construal of goal-oriented motion events by Swedish speakers of L2 Spanish 

behaviour, there is in Spanish a greater degree of freedom to choose between

the gerund and the simple present tense. If this is the reason why the current L2
­Spanish speakers exhibited elevated endpoint frequencies in spite of their profi-
ciency with the gerund, then it would suggest that it is not the mere presence of
a form in the target language that encodes ongoingness that drives mastery with
endpoint patterns, but rather the degree to which that form is obligatorily used.
Regarding Manner of motion, our research questions intended to delve fur-
ther into the extent to which Swedish speakers of L2 Spanish encode information
about Manner, especially in combination with endpoint encoding. In other words,
are the learners more prone than the Spanish L1 speakers to express Manner in
the main verb following the conflation patterns of Swedish as a Satellite-framed
language? Or will the encoding of information relative to Manner vary as a func-
tion of endpoint encoding?
Our results revealed that both the L1 and the L2 Spanish speaking groups
used Manner verbs to the same extent, which is in line with the results of C­ adierno
(2004), Cadierno and Ruiz (2006) and Hijazo-Gascón (2011), but not with the
outcomes presented in the study of Navarro and Nicoladis (2005), who found
that English learners of Spanish use somewhat more Manner verbs than ­Spanish
natives, and the ones of Larrañaga et al. (2012), who found various cases of Motion
and Manner conflation particularly in boundary-crossing situations. In other
words, our subjects do exhibit a similar behaviour as the natives concerning Path
and Manner encoding: they do not overtly conflate Motion and Manner in the
main verb; they rather use Path verbs. As Cadierno (2004) has put it, the range
of Manner verbs in Spanish is not as abundant as it is in S-languages, hence the
amount of Manner verbs to acquire in the target language may not result prob-
lematic. We need to stress, nevertheless, that our participants were immersed in
a Spanish speaking setting at the moment of data collection, which may be an
explaining factor. Interestingly, however, our results show that even though the
frequency of use of Manner verbs is similar in both L1 and L2 Spanish, the type of
verbs used differed somewhat and quite often resulted in uncommon or anoma-
lous descriptions.
We also asked whether Manner of motion would appear in separate expres-
sions (i.e. outside the verb) in the Spanish L2 data. To investigate this, Spanish L1
and L2 groups were examined with regards to their use of periphrastic Manner.
We found significant differences in Category II regarding endpoint encoding but
not when the groups described Categories I (with an arrival at an endpoint) and III
(with undetermined endpoint). In these latter categories, the learners behaved like
the Spanish L1 speakers: they resorted to Path verbs with no significant differences
regarding the use of Manner verbs and periphrastic Manner, but when rendering
situations that contained a visible endpoint (i.e. Category II) the learners mixed
 Alejandra Donoso & Emanuel Bylund

the encoding patterns of both Spanish L1 and Swedish L1. This was observed first
in the extent to which they encoded an endpoint, following the Swedish L1 typical
construal pattern of motion events; second, in the extent to which they used ger-
und forms to express ongoingness, adapting to the Spanish pattern; and third, in
the significant use of periphrastic Manner, which is a typical feature of the Spanish
language but which was not overtly used by the Spanish L1 speakers in this study.
Regarding the third finding, and as earlier explained, Spanish presents a syn-
tactic restriction which limits the use of Manner verbs associated with telic comple-
ments and therefore the range of Manner verbs and their use might be reduced both
in the Spanish L1 and L2 data in these type of scenes. However, the reason behind
the differences regarding the use of periphrastic Manner between the groups when
encoding endpoints may be the fact that the learners are still influenced by the con-
ceptualization patterns of their L1, revealing a case of conceptual transfer. Manner
is a salient semantic component of motion which often (and almost obligatorily)
appears codified in L1 Swedish, which may lead Swedish speakers into using all the
available linguistic resources to express it when learning a V-language.
On a more general level, the finding that the learners converged with L1
Spanish speakers in certain aspects of Manner encoding, but differed in terms
of endpoint encodings, is consistent with the results of Carroll, Weimar, Flecken,
Lambert, and von Stutterheim (2012). In their study, they found that whereas
French learners of English and German acquired the use of manner verbs, they
did not use the same spatial concepts when grounding events (such as motion
endpoints) as did native speakers of these languages. Taken together, these find-
ings thus contribute to the emerging picture that learners may selectively acquire
target-like motion event construal patterns in the L2.

7.  C
 onclusions

The current study set out to investigate the construal of goal-oriented motion in
Spanish L2 speakers with Swedish as an L1. In particular, we were interested in
the frequency with which the L2 speakers encoded motion endpoints, Manner
of motion, and how these two aspects were combined. It was found that Spanish
L2 speakers encode more endpoints than Spanish L1 speakers when referring to
motion events that contain movement towards a goal, but the goal is not reached.
It was also found that the L2 speakers encoded Manner periphrastically more
often than the Spanish L1 speakers when an endpoint was at play. Both results
prove that the patterns for motion encoding of the L1 may have an impact on L2
learning and that conceptual transfer occurs even when other aspects of the target
language are mastered.
 The construal of goal-oriented motion events by Swedish speakers of L2 Spanish 

References

Aske, J. (1989). Path predicates in English and Spanish: A closer look. Proceedings of the Berkeley
Linguistics Society, 15, 1–14.
Athanasopoulos, P., & Bylund, E. (2013). Does grammatical aspect affect motion event cogni-
tion? A cross-linguistic comparison of English and Swedish speakers. Cognitive Science,
37, 286–309. DOI: 10.1111/cogs.12006
Berman, R.A., & Slobin, D.I. (1994). Development of linguistic forms: English. In R.A. Berman
& D.I. Slobin (Eds.), Relating Events in Narrative: A Crosslinguistic Developmental Study.
­Hillsdale, NJ: Lawrence Erlbaum Associates. DOI: 10.1017/s0272263100014443
Bylund, E. (2008). Procesos de conceptualización de eventos en sueco y en español: Diferencias
translingüísticas [Event conceptualization processes in Swedish and Spanish: Crosslinguis-
tic differences]. Revue Romane, 43(1), 1–24. DOI: 10.1075/rro.43.1.02byl
Bylund, E. (2009). The effects of age of L2 acquisition on L1 event conceptualization principles.
Bilingualism: Language and Cognition, 12, 305–322. DOI: 10.1017/s1366728909990137
Bylund, E., & Athanasopoulos, P. (2014). Language and thought in a multilingual context: The
case of isiXhosa. Bilingualism: Language and Cognition, 17, 431–441.
DOI: 10.1017/s1366728913000503
Bylund, E., & Jarvis, S. (2011). L2 effects on L1 event conceptualization. Bilingualism: Language
and Cognition, 14, 47–59. DOI: 10.1017/s1366728910000180
Bylund, E., Athanasopoulos, P., & Oostendorp, M. (2013). Motion event cognition and gram-
matical aspect. Evidence from Afrikaans. Linguistics, 51, 929–955.
DOI: 10.1515/ling-2013-0033
Cadierno, T. (2004). Expressing motion events in a second language: A cognitive typological
perspective. In M. Achard & S. Niemeier (Eds.), Cognitive Linguistics, Second Language
Acquisition, and Foreign Language Teaching (pp. 13–49). SOLA Studies on Language
Acquisition. Berlin: Mouton de Gruyter. DOI: 10.1515/9783110199857.13
Cadierno, T. (2010). Motion in Danish as a second language: Does the learner’s L1 make a dif-
ference? In Z. Han & T. Cadierno (Eds.), Linguistic Relativity in SLA: Thinking for Speaking
(pp. 1–33). Bristol: Multilingual Matters. DOI: 10.1017/s0272263111000428
Cadierno, T., & Ruiz, L. (2006), Motion events in Spanish L2 acquisition. Annual Review of
­Cognitive Linguistics, 4, 183–216. DOI: 10.1075/arcl.4.08cad
Carroll, M., & von Stutterheim, C. (2006). The impact of grammaticalized temporal catego-
ries on ultimate attainment in advanced L2 acquisition. In H. Byrnes (Ed.), Educating for
Advanced Foreign Language Capacities (pp. 40–53). Washington DC: Georgetown Univer-
sity Press.
Carroll, M., von Stutterheim, C., & Nüse, R. (2004). The language and thought debate: A psy-
cholinguistic approach. In C. Habel & T. Pechmann (Eds.), Multidisciplinary Approaches to
Language Production (pp. 183–218). Berlin: Mouton De Gruyter.
DOI: 10.1515/9783110894028.183
Carroll, M., Weimar, K., Flecken, M., Lambert, M., & von Stutterheim, C. (2012). Tracing tra-
jectories: Motion event construal by advanced L2 French-English and L2 French-German
speakers. Language, Interaction and Acquisition, 3(2), 202–230.
DOI: 10.1075/lia.3.2.03car
Donoso, A. (2013). Camino, base y manera en español y sueco. Un estudio contrastivo [Path,
Ground and Manner in Spanish and Swedish. A contrastive study]. Revue Romane, 48(1),
1–31. DOI: 10.1075/rro.48.1.01don
 Alejandra Donoso & Emanuel Bylund

Hasselgård, H., Johansson, S., Behrens, B., & Fabricius-Hansen, C. (Eds.). (2002). Information
Structure in a Cross-linguistic Perspective. Amsterdam: Rodopi. DOI: 10.1075/lic.7.2
Hendriks, H., & Hickmann, M. (2011). Space in second language acquisition. In V. Cook &
B. Bassetti (Eds.). Language and Bilingual Cognition (pp. 315–339). Hove, UK: Psychology
Press.
Hijazo-Gascón, A. (2011). La expresión de eventos de movimiento y su adquisición en segundas
lenguas. Unpublishd Ph.D. dissertation. Universidad de Zaragoza.
Hohenstein, J., Eisenberg, A., & Naigles, L. (2006). Is he floating across or crossing afloat? Cross-
influence of L1 and L2 in Spanish–English bilingual adults. Bilingualism: Language and
Cognition, 9, 249–261. DOI: 10.1017/s1366728906002616
Ibarretxe-Antuñano, I. (2004). Dicotomías frente a continuos en la lexicalización de los eventos
de movimiento. Revista Española de Lingüística, 34(2), 481–510.
Ibarretxe-Antuñano, I. (2009). Path salience in motion events. In J. Guo, E. Lieven, N. B ­ udwig,
S. Ervin-Tripp, K. Nakamura, & S. Özçalişkan (Eds.), Crosslinguistic Approaches to the
­Psychology of Language: Research in the Tradition of Dan Isaac Slobin (pp. 403–414).
New York, NY: Psychology Press.
Ijaz, H. (1986). Linguistic and cognitive determinants of lexical acquisition in a second lan-
guage. Language Learning, 36, 401–451. DOI: 10.1111/j.1467-1770.1986.tb01034.x
Jarvis, S., & Odlin, T. (2000). Morphological type, spatial reference, and language transfer. Stud-
ies in Second Language Acquisition, 22, 535–556. DOI: 10.1017/s0272263100004034
Jarvis, S., & Pavlenko, A. (2008). Crosslinguistic Influence in Language and Cognition. New York,
NY: Routledge. DOI: 10.1558/sols.v6i3.603
Jovanovic, J., & Martinovic-Zic, A. (2004). Why manner matters. Contrasting English and
Serbo-Croatian typology in motion description. In C.L. Moder & A. Martinovic-Zic (Eds.),
­Discourse across Languages and Cultures (pp. 211–226). Amsterdam: John Benjamins.
DOI: 10.1075/slcs.68.12jov
Langacker, R. (2000). Grammar and Conceptualization. Berlin: De Gruyter.
DOI: 10.1515/9783110800524
Larrañaga, P., Treffers-Daller, J., Tidball, F., & Ortega, M. (2012). L1 transfer in the acquisition
of manner and path in Spanish by native speakers of English. International Journal of Bilin-
gualism, 16(1), 117–138. DOI: 10.1177/1367006911405577
Mayer, M. (1969). Frog: Where are you? New York, NY: Dial Press.
Naigles, L., & Terrazas, P. (1998). Motion-verb generalizations in English and Spanish: The
influence of language and syntax. Psychological Science, 9(5), 363–369.
DOI: 10.1111/1467-9280.00069
Naigles, L., Eisenberg, A., Kako, E., Highter, M., & McGraw, N. (1998). Speaking of motion:
Verb use in English and Spanish. Language and Cognitive Processes, 13(5), 521–549.
DOI: 10.1080/016909698386429
Navarro, S., & Nicoladis, E. (2005). Describing motion events in adult L2 Spanish narratives.
In D. Eddington (Ed.), Proceedings of the 6th Conference on Acquisition of Spanish and
Portuguese as First and Second languages (pp. 102–107). Somerville, MA: Cascadilla
Press.
Negueruela, E., Lantolf, J.P., Jordan, S., & Gelabert, J. (2004). The “private function” of gesture in
second language speaking activity: A study of motion verbs and gesturing in English and
Spanish. International Journal of Applied Linguistics, 14(1), 113–147.
DOI: 10.1111/j.1473-4192.2004.00056.x
Odlin, T. (1989). Language Transfer: Cross-linguistic Influence in Language Learning. Cambridge:
CUP. DOI: 10.1017/cbo9781139524537
 The construal of goal-oriented motion events by Swedish speakers of L2 Spanish 

Odlin, T. (2003). Crosslinguistic influence. In C. Doughty & M. Long (Eds.), The Handbook of
Second Language Acquisition (pp. 436–486). Oxford, UK: Blackwell.
DOI: 10.1002/9781405198431.wbeal0292
Pedersen, J. (2012). Spanish expressions of directed motion revisited – variable type framing. In
H. Boas & F. Gonzálvez-García (Eds.), Constructional Approaches to Romance Languages
(pp. 269–304). Amsterdam: John Benjamins.
Ragnarsdóttir, H., & Strömqvist, S. (2004). Time, space and manner in Icelandic and Swedish. In
S. Strömqvist & L. Verhoeven (Eds.), Relating Events in Narrative: Typological and Contex-
tual Perspectives (pp. 113–141). Hillsdale, NJ: Lawrence Erlbaum Associates.
Schmiedtová, B., & Flecken, M. (2008). The role of aspectual distinctions in event encoding:
Implications for second language acquisition. In S. de Knop & T. de Rycker (Eds.), Cogni-
tive Approaches to Pedagogical Grammar (pp. 357–384). Berlin: De Gruyter.
DOI: 10.1017/s0272263109990428
Schmiedtová, B., von Stutterheim, C., & Carroll, M. (2011). Language-specific patterns in event
construal of advanced second language speakers. In A. Pavlenko (Ed.), Thinking and Speak-
ing in Two Languages (pp. 112–141). Clevedon: Multilingual Matters.
Sjöström, S. (1990). Spatial Relations. Towards a Theory of Spatial Verbs, Prepositions and Pro-
nominal Adverbs in Swedish. Unpublished Ph.D. dissertation. Gothenburg University,
Sweden.
Slobin, D. (1996). Two ways to travel: Verbs of motion in English and Spanish. In M. Shibatani &
S. Thompson (Eds.), Grammatical Constructions. Their Form and Meaning (pp. 195–219).
Oxford: Clarendon Press.
Slobin, D. (2003). Language and thought online: Cognitive consequences of linguistic relativity.
In D. Gentner & S. Goldin-Meadow (Eds.), Language in Mind (pp. 157–191). Cambridge,
MA: The MIT Press.
Slobin, D. (2004). How people move. In C.L. Moder & A. Martinovic-Zic (Eds.), Discourse across
Languages and Cultures (pp. 195–210). Amsterdam: John Benjamins.
DOI: 10.1075/slcs.68.11slo
Stam, G. (2010). Can an L2 speaker’s patterns of thinking for speaking change? In Z. Han &
T. Cadierno (Eds.), Linguistic Relativity in SLA: Thinking for Speaking (pp. 59–83). Bristol:
Multilingual Matters. DOI: 10.1017/s0272263111000428
Strömqvist, S., & Verhoeven, L. (2004). Relating Events in Narrative, Vol. 2: Typological and Con-
textual Perspectives. London: Taylor & Francis.
Talmy, L. (1985). Lexicalization patterns: Semantic structure in lexical forms. In T. Shopen (Ed.),
Language Typology and Semantic Description, Vol. 3: Grammatical categories and the lexi-
con (pp. 36–149). Cambridge: CUP.
Talmy, L. (1991). Path to realization: A typology of event conflation. In Proceedings of the 17th
Annual Meeting of the Berkeley Linguistics Society (pp. 480–519). Berkeley, CA: Berkeley
Linguistics Society.
Talmy, L. (2000). Toward a Cognitive Semantics. Cambridge, MA: The MIT Press.
DOI: 10.1162/coli.2000.27.2.309
von Stutterheim, C. (2003). Linguistic structure and information organisation. The case of very
advanced learners. In S. Foster-Cohen & S. Pekarek Dohler (Eds.), EuroSLA Yearbook
(pp. 183–206). Amsterdam: John Benjamins. DOI: 10.1075/eurosla.3.11stu
von Stutterheim, C., & Nüse, R. (2003). Processes of conceptualization in language production:
Language-specific perspectives and event construal. Linguistics, 41(5), 851–881.
DOI: 10.1515/ling.2003.028
 Alejandra Donoso & Emanuel Bylund

von Stutterheim, C., Andermann, M., Carroll, M., Flecken, M., & Schmiedtová, B. (2012). How
grammaticized concepts shape event conceptualization in language production: Insights
from linguistic analysis, eye tracking data, and memory performance. Linguistics, 50(4),
833–867. DOI: 10.1515/ling-2012-0026
Zlatev, J., & Yangklang, P. (2004). A third way to travel: The place of Thai in motion event typol-
ogy. In S. Strömqvist & L. Verhoeven (Eds.), Relating Events in Narrative. Typological and
Contextual Perspectives (pp. 159–190). Hillsdale, NJ: Lawrence Erlbaum Associates.
Zubizarreta, M.L., & Oh, E. (2007). On the Syntactic Composition of Manner and Motion.
­Cambridge, MA: The MIT Press.
part iii

Spanish as an L2 in an instructional context

 Object drop in L2 Spanish, (complex) feature
reassembly, and L1 pre-emption
Comparing English, Chinese, European and
Brazilian Portuguese learners

Michael Iverson & Jason Rothman

Macquarie University / University of Reading

L2 learners from different L1s may differ in L2 development and ultimate

attainment for the same target property. Examining object drop in L2 Spanish,
we maintain that performances of distinct L1 groups (English, Chinese, Brazilian
and European Portuguese) are not as contradictory as they first appear. We argue
that L1 influences are sources of observed L2 behaviours, and suggest that feature
reassembly (e.g. Lardiere 2009) and L1 pre-emption (e.g. Rothman & Iverson
2011, 2013; Trahey & White 1993) are not mutually exclusive; they may apply
simultaneously and differentially depending on the L1/L2 pairs and target L2
property, creating distinct learning tasks for different L1s.

1.  Introduction

In spite of the unsettled debate on how to characterize adult L2 acquisition, some

empirical facts remain uncontested. Among them is the fact that adult learners
frequently do not achieve a native-like mastery of the L2, at least in a holistic
sense. This is not to say that the acquisition of new target L2 properties is impos-
sible, as studies show L2 successes across various domains (for review, among
others see Rothman 2008; Slabakova 2006, 2008; White 2003). Another observ-
able fact of adult L2 learning is that the outcomes are highly variable, both across
individuals for the same domain as well as across domains within the same indi-
viduals. In sum, adults typically do not show the consistency in developmental
sequence or ultimate attainment that characterizes normal L1 acquisition. Given
these observations, one conundrum L2 acquisitionists seek to address is deter-
mining the extent to which these observations should lead us to maintain that L2
acquisition is indeed fundamentally different from L1 acquisition or, conversely,

doi 10.1075/ihll.3.10ive
© 2015 John Benjamins Publishing Company
 Michael Iverson & Jason Rothman

if it is reasonable to account for these differences without hypothesizing that the

mechanisms responsible for successful acquisition in childhood are inaccessible
to adults. Attempting to do so seems warranted in light of ample acquisition evi-
dence that points to successful acquisition of subtleties of the L2 grammars that
are not transferrable from previous L1 linguistic knowledge and are characterised
by the same logical problem of acquisition as is the case for the L1 (i.e. so-called
poverty of the stimuli and/or bankruptcy of the stimuli properties; for discussion
see Rothman 2008; Schwartz 1998; Schwartz & Sprouse 2000).
At first glance, the claim that the process of language acquisition is funda-
mentally different for adults and children can explain the observed differences in
a seemingly straightforward way (e.g. Bley-Vroman 1990, 2009; Clahsen & Hong
1995; Meisel 1997, 2011). Adults show high levels of variation and various degrees
of divergence in production because L2 grammars are destined to be acquired via
domain general learning mechanisms as opposed to domain specific ones in child-
hood. If adults do not have access to the same genetic linguistic endowment (i.e.
universal grammar (UG)) that children do, then it is predicted that their mental
representations of an L2 grammar will be different and, as a result, they will inevi-
tably show this in their enactment of L2 use (empirical testing or otherwise). As
alluded to above, however, such theories fail to account for instances of successful
acquisition of highly nuanced properties of the L2 (e.g. Rothman & Iverson 2008;
Schwartz & Song 2009; Slabakova & Montrul 2003; among many others). Since the
goal of any comprehensive theory of L2 representation is to be able to account for
as much of the available data as possible, it is reasonable to claim that adults must
have access to at least some aspects of UG and that other variables must conspire
to explain the general lack of L2 success.

2.  Accounting for lack of success in L2 acquisition

Although it is not the case that L1 transfer alone could explain L1-L2 asymmetries,
it is the case that L1 transfer can explain and even predict some developmental
trends in L2 acquisition and failure in ultimate attainment. Successful acquisition
of the L2 in a domain-by-domain sense is in principle possible, but not inevitably
so, a claim first made in the UG framework by White (1989) and subsequently
developed into the Full Transfer/Full Access (FT/FA) model of the initial state in
line with its developmental predictions (Schwartz & Sprouse 1996). For exam-
ple, if the transferred L1 value constitutes a superset to the target subset gram-
mar, this L1 transfer as the initial hypothesis of each morpho-syntactic domain
of the L2 might itself block parsing failures necessary for linguistic development
that would otherwise obtain straightforwardly. Building on the seminal work of
 Object drop in L2 Spanish, (complex) feature reassembly, and L1 pre-emption 

Schwartz and Sprouse’s FT/FA model, more recent work within generative SLA
has acknowledged the limitations of assuming solely full transfer (e.g. Schwartz
2013) and has taken seriously the task of trying to articulate more precisely the
dynamic nature of the causes behind protracted delays and failures in adult second
language acquisition.
Herein, we focus on two of several efforts in the literature attempting to
account for some L2 differences while at the same time problematizing simplis-
tic notions of L1 transfer: (1) L1 pre-emption effects and (2) feature reassembly.
Although initial L1 transfer does not always block new L2 acquisition given access
to UG, it is also the case that new L2 syntactic acquisition does not always result in
the exclusion of the L1 syntax from the L2 interlanguage once the target has been
acquired. Trahey and White (1993) first labelled this as the L1 pre-emption effect,
which results in non-target variation of a seemingly L1-constrained behaviour
coexisting with L2 target behaviour precisely because the developing L2 gram-
mar has multiple syntactic options. Environments which are ripe for pre-emption
effects are ones in which L2 input can be parsed using the syntax of the L1, but
where the L1/L2 interpretative (i.e. semantic or pragmatic) subtleties do not nec-
essarily coincide.1 In such an environment the L2 input provides positive evidence
for the correct instantiation of a particular L2 property, but the parsing failures
traditionally considered necessary for grammatical restructuring will not obtain.
L2 learners may recognize that their L1 option is the unfavourable one (e.g. from
a statistical or probabilistic standpoint) without ruling it out as ungrammatical
or infelicitous, allowing it to surface in variation with the target form. The persis-
tent difficulty of unlearning an L1 option in the absence of explicit evidence has
been well documented (e.g. Gabriele 2009; Westergaard 2003), but has not been
shown to be an unsurmountable problem in all situations (e.g. Slabakova 2006;
Yuan 2001).
Lardiere’s (2009) discussion of feature reassembly, which has been of growing
importance in the past few years, is also of particular interest given the research we
will review in the remainder of this chapter. Feature re-assembly highlights the fact
that new feature acquisition is not the only or even most important part of the L2
learning task, but rather acquisition proper is further complicated by the p ­ resence
of the L1 due to the complexities of re-mapping procedures. Indeed, these are
independent tasks. In the case where no new L2 features need to be acquired, the

.  With this understanding of pre-emption, L2 acquisition scenarios susceptible to the

Subset Principle (Manzini & Wexler 1987) would serve as potential cases of a pre-emption
problem. Distinct from the Subset Principle, however, pre-emption problems need not neces-
sarily stem from contrasting L1/L2 settings of the same linguistic parameter.
 Michael Iverson & Jason Rothman

relevant features must still be combined and mapped to the appropriate lexical
items and learned to be used in felicitous discourse contexts. Considering that
linguistic features may be instantiated overtly (i.e. morphologically) or not, and
may be expressed syncretically (i.e. on one morpheme) or not, even acquiring or
redeploying only a few L2 features necessitates that the L2 learner entertain a host
of possibilities.
We will take the position that, depending on the L1-L2 language pairing,
L1 pre-emption and feature reassembly – separately, or at times in combination
with one another (and possibly other factors) – are in fact at the centre of an ulti-
mate explanation for why L2 learners from various L1 backgrounds acquiring
the same target properties may differ in developmental sequence and ultimate
attainment. Bringing together previous data sets from four distinct L1 learner
groups of L2 Spanish (English, Mandarin Chinese, Brazilian and European
­Portuguese) examining the acquisition of syntactic and semantic properties
related to the licensing of Spanish object drop, we will try to offer a principled
account of why these learners at comparable levels of Spanish proficiency dis-
play very different patterns of (successful) acquisition towards the target. We
will suggest that a simple view of L1 transfer alone does not predictively capture
these differences nor account for them, but rather a nuanced account that looks
at the learning task from both a feature reassembly and an L1 pre-emption at the
level of syntax does.
Looking at these L1s, acquiring Spanish is of particular interest for both
the remit and scope of this book as well as general theoretical interest for rea-
sons we have laid out above. While English does not have object drop, the other
three L1s examined in this review do. As we will see, however, none of these
languages has the semantic restrictions that Spanish imposes and thus all groups
have to acquire the semantics. Two of the four, European Portuguese (EP) and
Mandarin Chinese (MC), share with Spanish the same syntactic representation
of dropped objects, whereas Brazilian Portuguese (BP) does not. One might sus-
pect then that EP and MC should pattern together in L2 Spanish and perhaps be
different than native BP speakers, who have a seemingly more complex learning
task based on simple L1 transfer alone. Anticipating the results a bit, we will see
that this is in fact not the case. In fact, the BP and EP learners will show success-
ful acquisition of the semantics whereas MC learners do not. EP and BP will also
show differences as well. We will show how framing the learning tasks in terms
of not only L1 transfer, but also in terms of what features are available (or not)
elsewhere in the respective L1s and the subsequent (re)assembly/(re)mapping
tasks, as well as in terms of possible L1 pre-emption effects, combine to explain
the whole of the data. Our attempt to do this is significant beyond explaining
across Spanish L2 data sets; it is also an acknowledgement that we must account
 Object drop in L2 Spanish, (complex) feature reassembly, and L1 pre-emption 

for all data and not only the data sets that emerge from individual experimenta-
tion. After all, at the level of generative SLA theory we are interested in explain-
ing L2 acquisition in general, not only the acquisition of X in language Y for L1
group Z. Of course, this is a statement that should not come as controversial;
however, mired in individual projects, as we often are as researchers, this greater
picture sometime eludes us.

3.  Object expression in Spanish

Spanish employs a paradigm of object clitics for pronominal object expression.

Accusative object clitics are inflected for person, number, and additionally for gen-
der in the 3rd person. While not generally considered a true null object language,
standard Spanish also allows for phonetically empty objects in certain circum-
stances (see Campos 1986; Sánchez 2004) based on definiteness and specificity.2
Dropped objects are limited to indefinite referents, as seen in the question-answer
pairs (1) vs. (2):

(1) Q: ¿Trajiste la cámara?

   brought-you the camera
‘Did you bring the camera?’
A: Sí, *(la) traje.
yes,    it brought-I
‘Yes, I brought it.’

(2) Q: ¿Trajiste galletas?

   brought-you cookies
‘Did you bring cookies?’
A: Sí, (*las) traje.
yes,    it brought-I
‘Yes, I brought them/some.’
 (Bruhn de Garavito & Guijarro-Fuentes 2002, p. 60)

.  Definiteness and specificity are properties that refer to the state of knowledge of both
the speaker and hearer, and to the saliency or noteworthiness of the entity (as determined
by the speaker), respectively (see e.g. Ionin 2006). Briefly and informally, definite entities are
those identifiable by or known to both the speaker and hearer, and specific entities are those
which the speaker considers to have some noteworthy quality (regardless of the hearer’s
knowledge of this entity).
 Michael Iverson & Jason Rothman

In (1), a definite, specific3 object may be replaced by an object clitic, but cannot be
omitted in the answer. In contrast, as seen in (2), an indefinite, nonspecific object
must be dropped in the response.4 Similarly, this alternation is seen in Clitic Left
Dislocation and topicalization structures, as in (3) and (4) respectively:
(3) Estos zapatos, *(los) compré la semana pasada.
These shoes,    them bought-I the week last
‘These shoes, I bought last week.’
(4) Café, (*lo) tomo todas las mañanas.
Coffee, (*it) drink-I every the mornings
‘Coffee, I drink every morning.’
 (Cuza, Pérez-Leroux & Sánchez 2013, p. 97)
Campos (1986), following Huang’s (1984) analysis for Chinese and Raposo’s
(1986) for European Portuguese, claimed that the dropped object in cases like (2)
is an operator that has moved to the left periphery. By invoking movement, this
claim makes further predictions that dropped objects in Spanish are subject to
subjacency constraints. This is borne out in the data: while indefinite object clitics
can be dropped in embedded clauses, as in (5a), they cannot be dropped from DP
islands (5b), CP islands (5c), or adjunct islands (5d).
(5) Q: ¿Juan trajo cerveza a la fiesta?
‘Did Juan bring beer to the party?’
a. A: Su novia me dijo que (*la) trajo.
‘His girlfriend told me that he brought (some).’
b. A: Existe el rumor de que *(la) trajo.
‘There exists the rumor that he brought (some).’
c. A: Que *(la) trajo es obvio.
‘That he brought (some) is obvious.’
d. A: Sí, todos nos emborrachamos porque *(la) trajo.
‘Yes, we all got drunk because he brought (some).’
 (adapted from Campos 1986, p. 355)
So, in addition to the semantic restriction on definiteness, the realization of
dropped objects in Spanish also depends on the syntactic environment. In ­general,

.  In addition to definiteness, dropped objects in Spanish also depend on the specificity of
the referent, and are limited to nonspecific objects. A full treatment of specificity, including
the distinction between [-definite, +specific] vs. [-definite, -specific] objects, will not be given
here because it is not relevant for the studies under discussion. While these studies did take
specificity into account in their respective designs, it was not a variable under examination.
.  In contrast to Campos’ (1986) claims, the clitic in examples like (2) is optionally available
to some speakers. We note this here, but as the consequences are minimal for the studies we
detail, we continue to assume Campos’ analysis and intuitions.
 Object drop in L2 Spanish, (complex) feature reassembly, and L1 pre-emption 

for native-like knowledge of dropped objects, learners of Spanish must associate

the object clitic with the definiteness of the referent, learn that only indefinite
objects may be dropped (the semantic restriction), and learn that dropped objects
are a result of movement in the grammar, and subject to subjacency (the syntactic
restriction).

4.  Acquisition of object drop in L2 Spanish by speakers of various L1s

4.1  L
 1 English: Bruhn de Garavito & Guijarro-Fuentes (2002)
Bruhn de Garavito and Guijarro-Fuentes (2002) examined the acquisition of
dropped objects by L1 English learners of L2 Spanish (as well as L1 European
Portuguese learners of L2 Spanish, discussed below). English is distinct from
Spanish in that it does not have object clitics and does not allow indefinite object
drop. English learners of L2 Spanish must develop a representation of the Spanish
clitic paradigm in addition to learning the semantic and syntactic restrictions on
dropped objects.
Participants in this study were 18 L1 English speakers at an intermediate level
of L2 Spanish. They completed a scalar grammaticality judgment task that exam-
ined knowledge of both the syntactic and semantic restrictions of object drop in
Spanish. Items were presented in question-answer pairs, similar to (1) and (2)
above, and participants were asked to rate the answer in light of the question on a
scale from 1 (ungrammatical) to 5 (grammatical). Items varied based on definite-
ness of the referent, presence of the clitic, and syntactic structure of the answer to
be judged (here, we refer to structures that are not potential syntactic islands as
“simple syntax”). There were nine item types, seen in Table 1:

Table 1.  Item types in Bruhn de Garavito and Guijarro-Fuentes (2002) and their
grammaticality in Spanish
Item type Referent Structure Clitic Grammaticality

1 Definite Simple Overt √

2 Definite Simple Null *
3 Indefinite Simple Null √
4 Definite Complex DP Overt √
5 Indefinite Complex DP Null *
6 Definite Embedded CP Null √
7 Indefinite Sentential CP Null *
8 Definite Adjunct Overt √
9 Indefinite Adjunct Null *
 Michael Iverson & Jason Rothman

Post-hoc tests following up a significant ANOVA revealed that the L1 ­English

learners distinguished significantly between grammatical and ungrammatical
items. Furthermore, for cases of simple syntax, they rated cases of grammatical
object drop significantly higher than cases of ungrammatical object drop. For the
various island constraints, the learners made a significant distinction between
grammatical and ungrammatical items testing the Complex DP and Sentential
CP constraint, but not for those items testing for knowledge of object drop in
adjuncts.
The authors argue that, on the whole, L1 English learners were successful
in acquiring knowledge of object drop, as they reliably distinguished between
dropping definite and indefinite objects, as well as showed sensitivity to subja-
cency constraints. However, this claim had to be reconciled with the empirical
performance of these learners with respect to the constraint on extraction out of
adjuncts – a condition in which they should have, but did not, distinguish between
grammatical and ungrammatical items. The authors offer some tentative explana-
tions, appealing to the fact that these are only intermediate learners, and speculat-
ing that adjuncts may be more difficult to process than arguments (represented by
the other two island types).

4.2  L1 European Portuguese: Bruhn de Garavito & Guijarro-Fuentes (2002)

In the same study described in the previous section, Bruhn de Garavito and
­Guijarro-Fuentes also tested L1 European Portuguese learners of L2 Spanish. Sim-
ilar to Spanish, European Portuguese has object clitics marked for person, number
and gender (in the 3rd person). However, unlike Spanish, European Portuguese
allows for definite null objects, as seen in (6):
(6) A Joana viu Ø na TV ontem.
The Joana saw Ø on TV yesterday
‘Joana saw them/him/her/it on TV yesterday.’ (Raposo 1986, p. 373)

While the semantic restriction on dropped objects differs from Spanish, the syn-
tactic restrictions are the same. Dropped objects are subject to subjacency con-
straints, and European Portuguese examples analogous to the Spanish examples in
(5b–d) would be ungrammatical. Because dropped objects show the same restric-
tions on movement as those in Spanish, they are claimed to have the same under-
lying syntactic representation (Raposo 1986). Given that L1 European Portuguese
learners of Spanish already have knowledge of clitics and the syntactic represen-
tation of dropped objects, and assuming L1 transfer (in the vein of Schwartz &
Sprouse 1996), the learning task for these learners would consist only of narrowing
the licit semantic space, limiting dropped objects to exclusively those with indefi-
nite antecedents.
 Object drop in L2 Spanish, (complex) feature reassembly, and L1 pre-emption 

The European Portuguese group (N = 11) in this study were at an advanced

level of Spanish, and, as may be expected, fared even better than the English
natives. Unsurprisingly, the European Portuguese group respected all of the syn-
tactic restrictions, which are the same in their L1, and correctly and significantly
distinguished between grammatical and ungrammatical items testing for knowl-
edge of the Complex DP, Sentential CP and adjunct constraints. Additionally, they
showed knowledge of the semantic constraint, rating indefinite dropped objects
significantly higher than definite dropped objects. The authors note that this is
an important and perhaps unexpected result, as it required that the European
Portuguese learners constrained their grammar from a superset (allowing drop-
ping of both definite and indefinite objects) to a subset (allowing only indefinites).
This result is contrary to the claims of the Subset Principle, which predicts that
such constriction should be impossible because the superset grammar can always
accommodate the subset grammar, and with only positive evidence from the
input, the parsing failures necessary for grammatical restructuring will not occur.
In light of these data, however, we must consider that, at least for the semantic
portion of this property, learners are sensitive to some other characteristic of the
input.

4.3  L1 Brazilian Portuguese: Rothman & Iverson (2013)

Rothman and Iverson (2013) tested L1 Brazilian Portuguese learners of L2 ­Spanish.
While Brazilian Portuguese (BP) has a full paradigm of accusative clitics like those
in Spanish, it is distinct from Spanish with respect to dropped objects. First, there
is no restriction on definiteness,5 as seen in question-answer pairs (7) vs. (8):
(7) A Carmen trouxe o sorvete ao jantar?
The Carmen brought-3.sg the ice cream to-the dinner?
‘Did Carmen bring ice cream to the dinner?’
A1: #Ela o trouxe, sim. A2: Ø Trouxe, sim.
   She it brought-3.sg, yes. Ø brought-3.sg, yes.
‘Yes, she brought it.’ ‘Yes, she brought (it).’
(8) O Pablo serviu cerveja na festa?
the Pablo serve-3.sg beer in-the party
‘Did Pablo serve beer at the party?’

.  We note here that we are restricting our discussion to inanimate objects. Animate objects
are subject to different constrains (see Bianchi & Figueiredo 1993; Schwenter 2006 for details).
Experimental tasks used by Bruhn de Garavito & Guijarro-Fuentes and Rothman & Iverson
consisted of only inanimate objects.
 Michael Iverson & Jason Rothman

A1: #Ele a serviu, sim A2: Ø Serviu, sim.

   He it served-3.sg, yes. Ø served-3.sg, yes.
‘Yes, he served it.’ ‘Yes, he served (it).’
 (Rothman & Iverson 2013, p. 596)

Whether the antecedent is definite (7) or indefinite (8), the object clitic can be
omitted (the A2 responses). Additionally, the subjacency effects seen in Spanish
are not present in BP, seen in (9):
(9) Q: Carlos trouxe vinho para a festa?
‘Did Carlos bring beer to the party?’
a. A: A sua namorada disse que (#o) trouxe.
‘His girlfriend said that he brought (some).’
b. A: Não conheço a pessoa que (#o) trouxe.
‘I don’t know the person that brought (it).’
c. A: Que (#o) trouxe é claro.
‘That he brought (some) is obvious.’
d. A: Sim, todos nós ficamos bêbados porque (#o) trouxe.
‘Yes, we all got drunk because he brought (some).’
 (adapted from Rothman & Iverson 2013, p. 597)

In contrast to Spanish, objects can be omitted in DP islands (9b), CP islands (9c),

and adjunct islands (9d). Farrell (1990) observed that because these examples lack
subjacency effects, syntactic movement must not be involved in the derivation of
omitted objects in BP, and claimed that they are true null objects (i.e. pro). This
syntactic representation of null objects in BP and its counterpart in Spanish and
European Portuguese offer a plausible explanation of the asymmetry seen in the
distribution of object omission in these languages.
The distinct representation of null objects in BP also makes the learning
task distinct from the L1s already examined. BP learners of Spanish already have
knowledge of a comparable clitic paradigm. But, for successful representation of
dropped objects in Spanish, they must both acquire a new syntactic representa-
tion (Huang’s (1984) operator variable, instead of pro) and also restrict the avail-
ability of dropped objects to only those with indefinite antecedents. In light of the
European Portuguese data from Bruhn de Garavito and Guijarro-Fuentes (2002),
Rothman and Iverson hypothesized that the semantic constraint may be inde-
pendent of the syntax, and predicted that BP learners of Spanish will successfully
acquire the restriction. These same learners, however, were expected to experi-
ence (perhaps insurmountable) difficulty with converging on the correct syntactic
representation.
Rothman and Iverson (2013) adapted the methodology of Bruhn de ­Garavito
and Guijarro-Fuentes, expanding it to more thoroughly examine the potential
 Object drop in L2 Spanish, (complex) feature reassembly, and L1 pre-emption 

acquisition of the definiteness restriction. The participants, L1 BP learners of L2

Spanish (N = 25), completed a scalar grammaticality judgment task in which they
were presented a question-answer pair and asked to rate the answer on a scale
from 1 (totally unnatural) to 5 (totally natural). Target test items were constructed
on the basis of definiteness of the antecedent, presence of the clitic, and syntac-
tic environment (simple, DP island, CP island, and adjunct island), for a total of
15 item types, seen in Table 2 below:

Table 2.  Item types in Rothman and Iverson (2013) and their grammaticality in Spanish
Item type Referent Structure Clitic Grammaticality

1 Definite Simple Overt √

2 Definite Simple Null *
3 Indefinite Simple Null √
4 Definite Complex DP Overt √
5 Definite Complex DP Null *
6 Indefinite Complex DP Overt √
7 Indefinite Complex DP Null *
8 Definite Sentential CP Overt √
9 Definite Sentential CP Null *
10 Indefinite Sentential CP Overt √
11 Indefinite Sentential CP Null *
12 Definite Adjunct Overt √
13 Definite Adjunct Null *
14 Indefinite Adjunct Overt √
15 Indefinite Adjunct Null *

For simple syntactic structures, BP learners showed sensitivity to definiteness,

rating the grammatical items (realization of a definite object clitic and omission
of an indefinite object) significantly higher than the ungrammatical item (omis-
sion of a definite object). They were less successful with items testing the syntac-
tic restriction in various island types. They did rate indefinite dropped objects in
complex DPs significantly lower than indefinite dropped objects in simple syntac-
tic environments, but made no distinction between those in simple clauses and
those in sentential CP and adjunct islands. They did, however, continue to respect
the definiteness constraint in these more complex syntactic environments. While
they tended to reject any dropped object in complex DP islands, dropped definite
objects in these contexts were rated significantly lower than dropped indefinite
objects. In the other two contexts, sentential CP and adjunct islands, dropped
 Michael Iverson & Jason Rothman

­ efinite objects were similarly rejected, but dropped indefinite objects were rated
d
as acceptable.
Rothman and Iverson claim that these results show robust knowledge of the
semantic restriction, and, given a lack of comparable performance demonstrat-
ing knowledge of the syntactic restriction, suggest that the semantic and syntac-
tic portions of this property can be acquired independently of each other. The
results here also corroborate Bruhn de Garavito and Guijarro-Fuentes’ conclusion
that the definiteness constraint can be acquired even by those learners whose L1
instantiates a superset of this property (i.e. Brazilian and European Portuguese
learners of Spanish).
The Brazilian Portuguese learners of Spanish performed asymmetrically with
respect to the syntactic constraint on dropped objects. On the one hand, these
learners correctly rejected any instances of object drop in DP islands; on the other,
they allowed dropped indefinite objects in sentential CP and adjunct islands.
Rothman and Iverson consider the possibility that these learners do have the cor-
rect, Spanish-like representation for dropped objects, as evidenced by their per-
formance with DP islands, but that they continue to entertain the syntactic option
from their L1, as evidence by their acceptance of certain dropped objects in island
contexts. In short, they have failed to pre-empt their L1 grammar.
If it is the case that these learners have both Spanish-like and BP-like represen-
tations available for dropped objects, it must be explained why the target-like rep-
resentation is only employed in DP island contexts. Rothman and Iverson suggest
that perhaps this is due to the relative unacceptability of the different island types,
with DP islands the least unacceptable of the three examined because it poses fewer
obstacles for extraction (in the sense of the Barriers framework, Chomsky 1986).
If the complexity of syntactic representations has any ramifications for language
processing, this idea is compatible with the performance of the English learners
from Bruhn de Garavito and Guijarro-Fuentes, who failed to show sensitivity to
adjunct islands: the less complex the syntactic structure, the easier it is for devel-
oping learners to process it with their L2 grammar. In cases where the structure
is more complex, learners may show indeterminacy (like the English group) or
resort to a competing L1 option (like the Brazilian Portuguese Group).

4.4  L
 1 Chinese: Cuza, Pérez-Leroux & Sánchez (2013)
Cuza et al. (2013) studied the acquisition of dropped objects in L2 Spanish by
speakers of L1 Chinese. Null objects in Chinese are not subject to the same restric-
tions found in Spanish, and definite null objects are licit, as in (10):
(10) Zhangsani shuo Lisi bu reshi Ø*i/j
Zhangsan say Lisi not know ec
‘Zhangsan said that Lisi doesn’t know (him).’ (Huang 1984, p. 541)
 Object drop in L2 Spanish, (complex) feature reassembly, and L1 pre-emption 

Huang (1984) notes that the null object in Chinese examples like (10) cannot refer
to the matrix subject (here, Zhangsan), but rather refers to some external indi-
vidual. It can, however, be interpreted as coreferent with a topic DP, as in (11):
(11) Neige reni, Zhangsan shuo Lisi bu reshi Øi
That man Zhangsan say Lisi not know ec
‘That man, Zhangsan said that Lisi doesn’t know (him).’
 (Huang 1984, p. 541)

Based on these observations, Huang (1984) proposed that null objects in Chinese
can be analysed as a variable bound by a (null) topic. This, then, is similar to the
case of Spanish (and indeed his analysis was the starting point for the analysis of
dropped objects in Spanish), with the difference being that null objects are not
restricted to indefinites in Chinese, similar to European Portuguese.
According to Cuza et al. the learning task of L1 Chinese learners of L2 Spanish
is twofold. These learners must acquire the appropriate representation for accusa-
tive clitics, but must also learn the correct combination of features on these clitics
(namely, [–definite, –specific]) that would allow for them to be realized as null
elements. Leaving the syntactic representation of these Spanish clitics and null
objects aside, Chinese learners of Spanish must restrict the semantic domain in
which null objects are available from both definite and indefinite objects (their L1
setting) to exclusively indefinite objects (the target L2 setting).
The researchers surveyed four groups for their knowledge of clitics and null
objects: a control group of (Peruvian) Spanish monolinguals (N = 15), simulta-
neous bilinguals of Spanish and Chinese (N = 12), L1 Chinese childhood immi-
grants to Peru (N = 13), and L1 Chinese adult immigrants to Peru (N = 13). It
was expected that although the L2 learners of Spanish may produce and accept
clitics, they would not have native-like knowledge of their alternation with the
null object, and would continue to produce and accept null objects in contexts
that are ungrammatical in Spanish. Furthermore, it was predicted that there may
be some disparity between the three groups of bilinguals due to age-of-acquisi-
tion effects.
Participants completed four tasks: an elicited production task, a truth-value
judgment task (TVJT), a sentence-completion task and an acceptability judgment
task. Here we detail the acceptability judgment task, as it is the most comparable
to the tasks used in the studies detailed above. This task tested for the acceptability
of null and overt clitics in CLLD and topicalization structures. Participants were
presented with a sentence and asked to judge it on a scale from 1 (odd) to 5 (fine).
Three different contexts were examined: definite, specific antecedents; definite,
nonspecific antecedents; and indefinite, nonspecific antecedents. For each con-
text, half of the target sentences contained a clitic, while the other half did not,
resulting in an equal number of grammatical and ungrammatical items. The only
 Michael Iverson & Jason Rothman

contexts in which the clitic could be omitted were those in which the antecedent
was indefinite and nonspecific.
We focus here on the results from the adult L1 Chinese learners of L2 Span-
ish for comparability with the other studies. On the acceptability judgment task
these learners tended to accept all item types, with group average ratings greater
than 4 out of 5. As expected, this group allowed for dropped objects in all con-
texts. Contrary to the predictions of the authors, however, the adult Chinese
learners also accepted an object clitic in indefinite contexts. Results from other
tasks were consistent with those of the acceptability judgment task. Adult learn-
ers reliably used null objects in favour of clitics in elicited production, did not
have significantly different interpretations of null and overt clitics in the TVJT,
and overwhelmingly employed null objects in favour of clitics in the sentence
completion task.
The authors interpret these results as evidence that the adult L1 Chinese
learners of L2 Spanish do not have full knowledge of object drop in Spanish.
Specifically, they claim that these learners remain insensitive to the definiteness
and specificity features required for successful knowledge of the alternation of
null and overt object clitics. This is hypothesized to be a result of the transfer of
the L1, which has fewer restrictions on null objects, at least in the semantic sense,
and in which definiteness is not grammaticalized in a way similar to Spanish.
As this experiment was not designed to test knowledge of the subjacency effects
associated with dropped objects in Spanish, it cannot empirically address these
learners’ syntactic representations for this property.

5.  Discussion

The three studies involving L2 Spanish learners from four different L1s detailed above
investigated knowledge of object drop in L2 Spanish using comparable methodolo-
gies, but the L1(s) of the participants were distinct across studies. Bringing them
together allows us to compare the participants from each language background,
gauge their relative success, and speculate about the cause(s) of any asymmetries
observed with respect to the properties associated with object drop in Spanish.
First, we will summarize the results from the empirical studies. Recall that
these studies tested for at least one of two constraints associated with object drop,
a semantic constraint and a syntactic constraint. Semantically, dropped objects
in Spanish are restricted to indefinite non-specific objects. Syntactically, given
that they are not an instantiation of the empty category pro, they show subjacency
effects, and are ungrammatical in complex DP, sentential CP and adjunct islands.
The results for each L1 group are summarized in Table 3:
 Object drop in L2 Spanish, (complex) feature reassembly, and L1 pre-emption 

Table 3.  Results from each L1 Group

L1 Knowledge of restrictions?

Semantic Complex DP Sentential CP Adjunct

English Yes Yes Yes No

EP Yes Yes Yes Yes
BP Yes Yes No No
Chinese No Not tested Not tested Not tested

The English, European Portuguese and Brazilian Portuguese groups all showed
knowledge of the semantic restriction on definiteness, while the Chinese group did
not. Which groups showed knowledge of the syntactic restrictions was less clear,
perhaps due to the more fine-grained methodology used to test them. The ­European
Portuguese group respected subjacency constraints in all conditions, the English
group did so with both Complex DP islands and Sentential CP islands (but not with
adjunct islands), and the Brazilian Portuguese group did so with only Complex DP
islands. Cuza et al. (2013) did not test for the syntactic restrictions, and therefore we
can make no observations about the L1 Chinese learners. For our purposes of com-
parison, this is unfortunate. One could speculate that since ­Chinese and ­Spanish
have the same syntactic instantiation (i.e. an operator), the group would show no
problems with the syntax, akin to what was shown for the European Portuguese
group, but this would need to be empirically confirmed.
Although each study does a good job at attempting to explain the perfor-
mance of the group on which it focuses, the observable discrepancy across differ-
ent L1s acquiring the same properties in L2 Spanish is worthy of consideration. L1
transfer seems a likely possibility to explain the differences, however, the question
remains as to what exactly from the L1 influences the differences noted. Recall
that knowledge of dropped objects in Spanish relies on a target-like representation
of clitics, including the features necessary for agreement with an antecedent (per-
son, number, gender, definiteness, specificity), in addition to the proper syntactic
representation of dropped objects (i.e. an operator) and the associated semantic
constraint on definiteness. With this in mind, similar or competing properties in
the various L1s are obvious candidates for cross-linguistic influence. European
Portuguese and Brazilian Portuguese both have accusative object clitics, which
function like those in Spanish, while English and Chinese do not.6 English does

.  We do not ignore the fact that Brazilian Portuguese does not regularly use third person
object clitics, although they have them (and these learners are competent speakers of the stan-
dard variety). We refer the interested reader to Rothman and Iverson (2013) for discussion.
 Michael Iverson & Jason Rothman

not instantiate zero-objects, while the other three L1s do. However, in none of
these languages do the zero-objects have the same distribution as they do in Span-
ish. In all three languages, they are not restricted to only indefinite antecedents,
but also allow definite antecedents. Both Chinese and European Portuguese show
subjacency effects (and thus are assumed to have the same syntactic representation
as in Spanish), but Brazilian Portuguese does not.
In addition to these properties that seem straightforwardly related to Spanish
clitics and dropped objects, we may also want to consider a property that is less
directly related – the instantiation of definiteness in these languages. Definiteness
can be defined as the property of being able to be identified by both the speaker
and the hearer (e.g. Ionin, Ko & Wexler 2004). According to Lyons (1999), lan-
guages express this “identifiability” in a variety of ways. Spanish, English, E
­ uropean
­Portuguese and Brazilian Portuguese, which all have definite/­indefinite article
systems, serve as prototypical cases of languages in which definiteness has been
grammaticalized. In languages like these, definiteness is obligatorily and unambig-
uously marked using unique functional morphology. Chinese, on the other hand,
may be considered a language in which definiteness has not been grammaticalized
and is expressed through more semantic or pragmatic means. Chinese indicates
definiteness through (an interaction of) context, syntactic position or modifi-
cation of the nominal with a classifier or plural morpheme (Cheng  & Sybesma
1999). Furthermore, marking a noun as definite or indefinite is not obligatory, and
certain marking strategies may not always result in unambiguity (for details and
discussion, see Chen 2004). Having prior knowledge of a language in which defi-
niteness is grammaticalized (in the sense of Lyons 1999) may serve as a catalyst in
acquiring the semantic constraint on object drop in Spanish. We mention this in
light of feature reassembly; that is, only in these languages does definiteness take
the form a grammatical (syntactic) feature that is then available for repurposing
(reassembly) in the L2.
A summary of the existence of relevant properties in the various L1s – gram-
matical definiteness, accusative object clitics, zero-objects and their syntactic type
(Operator or pro), the restriction of zero-objects to indefinite antecedents, and
subjacency effects associated with zero-objects – is given in Table 4 below:
Here, each language brings a unique array of linguistic properties to the table,
and therefore the learning task for each group is slightly different. The European
Portuguese group has the “simplest” task, having to only restrict the scope of
dropped objects to indefinites. The Brazilian Portuguese group must also do this,
as well as acquire a new syntactic representation for dropped object and addition-
ally inhibit their L1 syntactic representation of truly null objects to show sensitivity
to islands. The L1 English group must acquire clitics, and the appropriate syntactic
representation and semantic constraints of dropped objects. The L1 ­Chinese have
 Object drop in L2 Spanish, (complex) feature reassembly, and L1 pre-emption 

Table 4.  L1 Properties relevant for L2 Spanish

L1 L1 Properties

Grammatical Clitics Ø-Objects [–definite] Subjacency

definiteness (Type) only

Span Yes Yes Yes (Op) Yes Yes

Eng Yes No No (n/a) n/a n/a
EP Yes Yes Yes (Op) No Yes
BP Yes Yes Yes (pro) No No
Ch No No Yes (Op) No Yes

perhaps the greatest challenge: they must acquire the Spanish clitic system, restrict
the semantic space in which dropped objects are available, and, if a grammatical-
ization of definiteness is necessary for the acquisition of object drop, they must
also acquire this feature.
Let’s examine the syntactic and semantic restrictions separately to see if there
is any correlation between successful acquisition and the properties of the L1.
Considering the semantic constraint first, we note that only the Chinese group
was unable to acquire this; all other groups were successful. This asymmetry is
paralleled with respect to the degree of grammaticalization of definiteness in the
respective L1s: it is not grammaticalized in Chinese, while it is in the other lan-
guages. English, European Portuguese and Brazilian Portuguese, like Spanish,
have determiner systems which differentiate forms based on definiteness. This
observation may suggest, then, that knowledge of the determiner system and/or its
relevant features may be a prerequisite for the acquisition of the semantic restric-
tions on object drop in Spanish. This is supported by observations in first language
acquisition, which has shown that the acquisition of object clitics is delayed with
respect to articles in languages that exhibit both (e.g. Hamann 2003; Kupisch &
Müller 2009; Marinis 2003), and similarly for L2 (Dimitrakopoulou, Kalaitzidou,
Roussou & Tsimpli 2004). If successful use of the object clitic system depends on
knowledge of article systems (or definiteness grammaticalized elsewhere in the
L1), and knowledge of null and overt object expression in Spanish depends on
the acquisition of object clitics, then knowledge of dropped objects is indirectly
dependent on the acquisition of articles. Lack of a grammaticalized notion of defi-
niteness seems to put the L1 Chinese group at a disadvantage. This is particularly
evident in comparison to the L1 English group, who have an article system similar
to Spanish, but like the L1 Chinese group, lack clitics. In contrast to the Chinese
group, the English group shows knowledge of the semantic restrictions of object
drop.
 Michael Iverson & Jason Rothman

The case of the Brazilian and European Portuguese groups is distinct from that
of the Chinese and English groups, because both varieties of Portuguese instanti-
ate object clitics, similar to Spanish. However, the task for all Portuguese learners
with respect to the semantic component is to restrict the availability of the null
variant of the object clitic to exclusively indefinites. Since their L1 grammars allow
for definite and indefinite null objects, a superset of what the Spanish grammar
allows, this would seem to impose a learnability problem on these groups. How-
ever, results from these learners show that this problem is not insurmountable.
Similar to the situation of the semantic constraint, there was also variable suc-
cess in the acquisition of the syntactic constraint. We first note that the Chinese
group studied in Cuza et al. (2013) was not tested for knowledge of subjacency
restrictions, so here we could at best predict how this group might have performed
on such a task and speculate why this performance would have been expected. For
now, though, we leave such speculation aside and turn to the results that are avail-
able. The English, European and Brazilian Portuguese groups were examined in
three different island conditions: complex DP, sentential CP and adjunct islands.
The European Portuguese group were successful in all three conditions, and were
the only group that performed like native Spanish speakers across the board. The
English group was successful in two conditions (complex DP and sentential CP
islands), and the Brazilian Portuguese group was successful in only the complex
DP condition. It is interesting to note that both groups did not perform natively on
adjunct islands, suggesting that there is something special about this condition.7
These asymmetric results may not be surprising given that the learning task
with respect to the syntax of object drop in Spanish is distinct for the three groups.
European Portuguese is hypothesized to have the same syntactic representation of
object drop as Spanish, so it is perhaps expected that this group show knowledge
of the property. Neither English nor Brazilian Portuguese have a Spanish-like rep-
resentation of dropped objects, but differ crucially in that Brazilian Portuguese
does instantiate what might be considered a competing form (i.e. null object pro).
When confronted with dropped objects in Spanish linguistic input, English speak-
ers have no recourse to systematically interpret them, resulting in a straightforward
parsing failure and subsequent grammatical restructuring. ­Brazilian ­Portuguese
speakers, however, can interpret all Spanish dropped objects with their L1 gram-
mar without further development. In light of this contrast, the acquisition task
of the English group is perhaps less complex. While both groups must acquire
the proper syntactic representation of dropped objects, the Brazilian Portuguese
group must additionally pre-empt the competing option from their L1 grammar.

.  See Rothman and Iverson (2013) for further explanation.

 Object drop in L2 Spanish, (complex) feature reassembly, and L1 pre-emption 

As Rothman and Iverson (2013) argue the Brazilian Portuguese performance with
DP islands suggests that the Spanish syntax has been acquired, yet their perfor-
mance in the other conditions also suggests that they do not apply it across all
islands failing to pre-empt their L1 syntax in those contexts. This alone is suf-
ficient to account for why the English group performs better than the Brazilian
Portuguese group.
These results, considered as a whole, are problematic for the view that the
acquisition of a second language grammar is the superficial transfer of the L1
grammar in addition to the application of general rules gleaned from frequency
of particular constructions in the primary linguistic data. Both the Chinese and
European Portuguese groups have L1 grammars that instantiate syntactic repre-
sentations of dropped objects equivalent to what is found in Spanish, and like-
wise have L1 grammars that allow both definite and indefinite dropped objects
(as does Brazilian Portuguese). In short, when considering only this property, the
acquisition task appears to be the same. It could even be claimed that only the
Chinese group in Cuza et al. (2013) was at an advantage, as they were residents in a
Spanish-­speaking country and had unrestricted access to abundant input that was
both native and natural. As was seen though, with respect to the semantic con-
straint, the European Portuguese group (as well as the Brazilian Portuguese group)
succeeded while the Chinese group did not. Furthermore, the English group, who
were unaided by linguistic transfer, also outperformed the Chinese group. Appeal-
ing to L1 transfer for this domain alone, these discrepancies are not predicted,
especially the difference between the European Portuguese and Chinese groups.
We must, then, resort to other explanations to account for the differing results
across groups. Lardiere (2009) suggests that second language acquisition may be
better conceptualized as a task that not only involves transfer of the first language
grammar and the acquisition of new features, but also involves redeployment of
features. Newly acquired and already extant features must be both bundled prop-
erly and mapped onto the L2 lexical items. Features may be bundled or orga-
nized differently in different languages, and furthermore certain features may be
expressed morphosyntactically on a (dedicated) lexical item, or contextually (via
varying syntactic position, prosody, etc.). Slabakova (2009) follows Ramchand and
Svenonius (2008) to develop this point, claiming that not only does the second
language learning task involve transfer, acquisition and redeployment of features,
but also that there is a spectrum of difficulty of acquisition. Some properties may
be harder to acquire than others because of the way they are expressed in the
respective L1 and L2. An easy task is acquiring a property that is morphosyntacti-
cally expressed in both languages with the same bundling of features (i.e. with no
feature reassembly required). More difficult is the acquisition of a property that is
morphosyntactically realized in both the L1 and L2, but requires the r­ eassembly
 Michael Iverson & Jason Rothman

of features. Harder still is the acquisition of a property which is contextually

expressed in the L1, but morphosyntactically expressed in the L2. Combining this
with Lyons’ (1999) idea that languages differ in how they mark definiteness, we
may be able to paint a more satisfactory picture of the acquisition task and its
relative difficulty for the learner groups examined here. To acquire object drop
in Spanish, all groups must first acquire clitics and their associated features (e.g.
definiteness and specificity). Both European and Brazilian Portuguese have clit-
ics, while English and Chinese do not. In spite of the similarity in this respect
between the English and Chinese groups, the Chinese group was the only group
that was unsuccessful in showing knowledge of the semantic constraint on object
drop in Spanish. The difference may be that Chinese learners are also the only
group whose L1 does not have definite and indefinite articles – definiteness is not
grammaticalized, and is expressed through more semantic or pragmatic means.
This changes the relative difficulty of the acquisition task for the Chinese group as
compared to the others whilst keeping in mind Slabakova’s (2009) footnote to Lar-
diere’s Feature Reassembly Hypothesis (2009). While the English group may have
to re-map features such as definiteness from L1 morphemes to new morphemes
(i.e. from definite articles to object clitics), the Chinese group must, minimally,
map a contextually expressed feature onto a lexical item. This, as Slabakova (2009)
claims, is a harder task, a claim which is consistent with the Chinese group’s rela-
tively poor performance when compared to speakers of other L1s.
Symmetry of morphologically realized features between the L1 and L2 can-
not be the whole story, however; if it were, we would expect symmetrical per-
formances from the non-Chinese groups, since all grammaticalize definiteness in
their article systems. Symmetry is perhaps especially expected between the Euro-
pean and Brazilian Portuguese groups, who, in addition to definite and indefinite
articles, instantiate object clitics and object drop in their respective L1s. The dif-
ference between these two groups is not at the semantic level – both allow definite
and indefinite dropped objects, and both successfully acquire the correct ­Spanish
constraint – but rather at the syntactic level. In this respect, European ­Portuguese
speakers have nothing new to acquire, as their syntactic representation of dropped
objects is what is found in Spanish. The Brazilian Portuguese speakers must
acquire this representation, which is distinct from that of their L1. However, this
cannot be the entire story either, because the English group, who similarly must
acquire a representation for object drop that they do not have in their L1 gram-
mar, perform better than the Brazilian Portuguese group on the experimental task.
The difference between the English and Brazilian Portuguese groups is that while
both groups must acquire a new syntactic representation, the Brazilian Portuguese
group must additionally eliminate a competing L1 option, a problem of L1 pre-
emption. Successful acquisition of an L2 property does not necessarily guarantee
 Object drop in L2 Spanish, (complex) feature reassembly, and L1 pre-emption 

that a comparable L1 property will no longer be entertained as a possibility in

the L2 grammar, thus leading to apparent variation (Rothman & Iverson 2011,
2013; Trahey & White 1993). In the following section, we bring the discussion of
these data sets to a larger level by addressing its implications for L2 theory more
generally.

6.  C
 onclusion

We have provided a macro-analysis that accounts for the whole of the existent data
on the acquisition of object drop in L2 Spanish in an effort to address a greater
concern of L2 theory, that of overall explanatory adequacy of the processes of
acquisition and not just explaining individual data sets. As is traditional in genera-
tive approaches to L2 acquisition, we relied heavily on formal syntactic theory to
find a basis for the specific explanation of the data for this domain. In this conclu-
sion, we move beyond the explanation of these data sets by highlighting its impli-
cations for formal L2 theorizing, which we believe is especially relevant in light of
the goal of the present volume which seeks to show how looking at the acquisition
of the same L2 from the perspective of various L1s brings much to bear for general
L2 theory. In the introduction, we took the position that the theoretical constructs
of L1 pre-emption and feature reassembly are important, separately and in com-
bination, for an ultimate explanation of seemingly disparate data sets across L2
studies and thus making strides towards the goal of greater explanatory adequacy
in L2 theorizing. We started with the assumption, following others, that a simplis-
tic view of L1 transfer alone is not able to account for all L2 facts. Alternatively, L1
influences at various levels as one variable (to be sure there are others, but we lim-
ited ourselves to this herein), when understood in contemporary formal linguistic
terms that address its complexity in a sophisticated manner (e.g. Lardiere 2009),
provide a tenable path for current generative SLA. Following Slabakova (2009),
we contended that examining not only feature reassembly, but understanding that
not all features are of equal complexity for the acquisition process (L1, L2 or other)
also aids in further understanding differences across L1 groups of a similar L2.
Bringing together previous data sets from four distinct L1 learner groups of L2
Spanish (English, Mandarin Chinese, Brazilian and European Portuguese) exam-
ining the acquisition of syntactic and semantic properties related to the licensing
of object drop, we showed how considering L1 transfer in simplistic terms would
not account for the data as a whole. However, coupling together notions of L1 pre-
emption effects and a formulation of feature reassembly that considers both what
is available (or not) from the L1 as well as how features are represented (gram-
maticalized or not), not only explain the data but at the same time paint a more
 Michael Iverson & Jason Rothman

accurate picture of the dynamics of L2 acquisition and the detailed intricacies L2

acquisitionists must attend to when trying to explain L2 processes. We leave this
epistemological exercise encouraged that greater, more generalizable explanatory
adequacy in L2 acquisition will ultimately be possible as we come to approach the
task of L2 theory with greater sophistication and armed with more detail oriented
hypotheses that seek to understand things beyond the surface.

Acknowledgements

We would like to thank Roumyana Slabakova and two anonymous reviewers for
their careful reading of this paper and insightful comments. We are grateful to the
editors of this volume for their invitation to contribute as well as their patience and
help. Any and all errors are inadvertent and entirely our own.

References

Bianchi, V., & Figueiredo Silva, M.C. (1993). On some properties of agreement-object in Italian
and in Brazilian Portuguese. Ms., Scuola Normale Superiore di Pisa/Université de Genève.
Bley-Vroman, R. (1990). The logical problem of foreign language learning. Linguistic Analysis,
20, 3–49. DOI: 10.1017/cbo9781139524544.005
Bley-Vroman, R. (2009). The evolving context of the Fundamental Difference Hypothesis. Stud-
ies in Second Language Acquisition, 31, 175–198. DOI: 10.1017/s0272263109090275
Bruhn de Garavito, J., & Guijarro-Fuentes, P. (2002). L2 acquisition of indefinite object drop
in Spanish. In J. Costa & M.J. Freitas (Eds.), GALA 2001. Proceedings (pp. 60–67). Lisbon:
Associação Portuguesa de Linguística.
Campos, H. (1986). Indefinite object drop. Linguistic Inquiry, 17, 354–359.
Chen, P. (2004). Identifiability and definiteness in Chinese. Linguistics, 42(6), 1129–1184.
DOI: 10.1515/ling.2004.42.6.1129
Cheng, L.L., & Sybesma, R. (1999). Bare and not-so-bare nouns and the structure of NP. Linguis-
tic Inquiry, 30, 509–542. DOI: 10.1162/002438999554192
Chomsky, N. (1986). Barriers. Cambridge, MA: The MIT Press. DOI: 10.9793/elsj1984.6.213
Clahsen, H., & Hong, U. (1995). Agreement and null subjects in German L2 development: New
evidence from reaction-time experiments. Second Language Research, 11, 57–87.
DOI: 10.1177/026765839501100103
Cuza, A., Pérez-Leroux, T., & Sánchez, L. (2013). The role of semantic transfer in clitic drop
among simultaneous and sequential Chinese-Spanish bilinguals. Studies in Second Lan-
guage Acquisition, 35(1), 93–125. DOI: 10.1017/s0272263112000691
Dimitrakopoulou, M., Kalaitzidou, M., Roussou, A., & Tsimpli, I.M. (2004). Clitics and deter-
miners in the Greek L2 grammar Proceedings of the 6th International Conference on
Greek Linguistics. 〈www.philology.uoc.gr/conferences/6th ICGL〉
Farrell, P. (1990). Null objects in Brazilian Portuguese. Natural Language & Linguistic Theory,
8(3), 325–346. DOI: 10.1007/bf00135617
 Object drop in L2 Spanish, (complex) feature reassembly, and L1 pre-emption 

Gabriele, A. (2009). Transfer and transition in the SLA of aspect: A bidirectional study of learn-
ers of English and Japanese. Studies in Second Language Acquisition, 31, 371–402.
DOI: 10.1017/s0272263109090342
Hamann, C. (2003). Phenomena in French normal and impaired language acquisition and their
implications for hypotheses on language development. Probus, 15, 91–122.
DOI: 10.1515/prbs.2003.006
Huang, J. (1984). On the distribution and reference of empty pronouns. Linguistic Inquiry, 15,
531–575.
Ionin, T. (2006). This is definitely specific: Specificity and definiteness in article systems. Natural
Language Semantics, 14, 175–234. DOI: 10.1007/s11050-005-5255-9
Ionin, T., Ko, H., & Wexler, K. (2004). Article semantics in L2 acquisition: the role of specificity.
Language Acquisition, 12, 3–69. DOI: 10.1207/s15327817la1201_2
Kupisch, T., & Müller, N. (2009). Relating Italian articles and clitic object pronouns in bilingual
children acquiring German and Italian. In C. Dimroth & P. Jordens (Eds.), Functional Cat-
egories in Learner Language (pp. 307–342). Berlin: De Gruyter.
DOI: 10.1515/9783110216172.307
Lardiere, D. (2009). Some thoughts on the contrastive analysis of feature in second language
acquisition. Second Language Research, 25, 171–225.
DOI: 10.1177/0267658308100283
Lyons, C. (1999). Definiteness. Cambridge: CUP. DOI: 10.1017/cbo9780511605789
Manzini, R.M., & Wexler, K. (1987). Parameters, binding theory, and learnability. Linguistic
Inquiry, 18, 413–444. DOI: 10.1007/978-94-009-3727-7_3
Marinis, T. (2003). The Acquisition of the DP in Modern Greek. Amsterdam: John Benjamins.
DOI: 10.1075/lald.31
Meisel, J. (1997). The acquisition of the syntax of negation in French and German: Contrasting
first and second language development. Second Language Research, 13, 227–263.
DOI: 10.1191/026765897666180760
Meisel, J. (2011). First and Second Language Acquisition: Parallels and Differences. Cambridge:
CUP. DOI: 10.1017/cbo9780511862694
Ramchand, G., & Svenonious, P. (2008). Mapping a parochial lexicon onto a universal seman-
tics. In M.T. Biberauer (Ed.), Limits of Syntactic Variation (pp. 219–245). Amsterdam: John
Benjamins. DOI: 10.1075/la.132.08ram
Raposo, E. (1986). On the null object in European Portuguese. In O. Jaeggli & C. Silva-Corvalán
(Eds.), Studies in Romance Linguistics (pp. 373–390). Dordrecht: Foris.
Rothman, J. (2008). Why not all counter-evidence to the Critical Period Hypothesis is equal or
problematic: Implications for SLA. Language and Linguistics Compass, 2(6), 1063–1088.
DOI: 10.1111/j.1749-818x.2008.00098.x
Rothman, J., & Iverson, M. (2008). Poverty-of-the-stimulus and L2 epistemology: Considering
L2 knowledge of aspectual phrasal semantics. Language Acquisition: A Journal of Develop-
mental Linguistics, 15(4), 270–314. DOI: 10.1080/10489220802352206
Rothman, J., & Iverson, M. (2011). L1 preemption and L2 learnability: The case of object drop
in Brazilian Portuguese native learners of L2 Spanish. In N. Danis, K. Mesh, & H. Sung
(Eds.), BUCLD 35: Proceedings of the 35th Annual Boston University Conference on Lan-
guage Development (pp. 296–307). Somerville, MA: Cascadilla Press.
Rothman, J., & Iverson, M. (2013). Strong islands and null objects in L2 Spanish of Brazilian
Portuguese natives: Do you know the learners who drop ____? Studies in Second Language
Acquisition, 35(4), 589–619.
 Michael Iverson & Jason Rothman

Sánchez, L. (2004). Object Agreement and the Checking of Focus in Spanish. Unpublished ms,
Rutgers University.
Schwartz, B. (1998). The second language instinct. Lingua, 106, 133–160.
DOI: 10.1016/s0024-3841%2898%2900032-1
Schwartz, B. (2013, April). At what level, “transfer”? Plenary paper presented at Generative
Approaches to Language Acquisition 12, University of Florida, Gainesville, FL.
Schwartz, B., & Song, H. (2009). Testing the Fundamental Difference Hypothesis: L2 adult, L2
child, and L1 child comparisons in the acquisition of Korean wh-constructions with nega-
tive polarity items. Studies in Second Language Acquisition, 31, 323–361.
Schwartz, B., & Sprouse, R. (1996). L2 cognitive states and the full transfer/full access model.
Second Language Research, 12, 40–72. DOI: 10.1177/026765839601200103
Schwartz, B., & Sprouse, R. (2000). When syntactic theories evolve: Consequences for L2 acqui-
sition research. In J. Archibald (Ed.), Second Language Acquisition and Linguistic Theory
(pp. 156–186). Oxford: Blackwell.
Schwenter, S. (2006). Null objects across South America. In T.L. Face & C. Klee (Eds.), Selected
Proceedings of the 8th Hispanic Linguistics Symposium (pp. 23–36). Somerville, MA:
­Cascadilla Press.
Slabakova, R. (2006). Is there a critical period for semantics? Second Language Research, 22,
302–338. DOI: 10.1191/0267658306sr270oa
Slabakova, R. (2008). Meaning in the Second Language. Berlin: De Gruyter Mouton.
DOI: 10.1017/s0272263110000124
Slabakova, R. (2009). Features or Parameters: Which one makes SLA easier, and more interest-
ing to study? A response to Lardiere. Second Language Research, 25(2), 313–324.
DOI: 10.1177/0267658308100291
Slabakova, R., & Montrul, S. (2003). Genericity and Aspect in L2 Acquisition. Language Acquisi-
tion, 11, 165–196. DOI: 10.1207/s15327817la1103_2
Trahey, M., & White, L. (1993). Positive evidence and preemption in the second language class-
room. Studies in Second Language Acquisition, 15, 181–204.
DOI: 10.1017/s0272263100011955
Westergaard, M. (2003). Unlearning V2: Transfer, markedness and the importance of input cues
in the acquisition of word order in English by Norwegian children. EUROSLA Yearbook,
3, 73–101. DOI: 10.1075/eurosla.3.07wes
White, L. (1989). Universal Grammar and Second Language Acquisition. Amsterdam: John
­Benjamins. DOI: 10.1177/026765839100700306
White, L. (2003). Second Language Acquisition and Universal Grammar. Cambridge: CUP.
DOI: 10.1017/cbo9780511815065.002
Yuan, B. (2001). The status of thematic verbs in second language acquisition of Chinese: Against
inevitability of thematic-verb raising in second language acquisition. Second Language
Research, 17, 248–272. DOI: 10.1191/026765801680191505
 The acquisition of differential object marking
in Spanish by Turkish speakers

Silvina Montrul & Ayʂe Gürel

University of Illinois Urbana-Champaign / Boğaziçi University

The obligatory use of the preposition ‘a’ with animate, specific direct objects
in Spanish (Juan conoce a María ‘Juan knows Maria’) is a well-known instance
of Differential Object Marking (DOM). This study investigates the acquisition
of DOM in Spanish by native speakers of Turkish, a language that exemplifies
DOM on the basis of case marking. Twenty native speakers of Spanish and 32
Turkish-speaking learners of Spanish completed a written production, a written
comprehension, and a bimodal acceptability judgment task. Our results show that
Turkish learners of Spanish are quite successful at acquiring feature specification
and distribution of DOM even at low levels of proficiency.

1.  Introduction

In second and third language acquisition, inflectional morphology (tense, aspect,

agreement, mood, gender, number, etc.) seems to be a stumbling block and a
source of persistent difficulty. L2 learners exhibit errors of omission and sub-
stitution in oral and written production. They may also have difficulty in recep-
tive tasks, recognizing or processing morphological forms in both online and
off-line tasks. Generative approaches to second language (L2) acquisition have
tried to pinpoint the specific locus of this difficulty, and Lardiere (2008, 2009) has
advanced an explanation based on how the components of morphological knowl-
edge may be acquired and computed in language learning. The Chomskyan view
of language acquisition holds that language learners must assemble the lexicon of
a language by associating lexical items (i.e. morphophonological forms) with the
specific morphosyntactic features that the language selects (Chomsky 2001, p .4).
Thus, language learners must learn which morpholexical forms and their allo-
morphic variants express which specific syntactic and semantic features as well
as the conditions under which such forms are obligatory, optional, or ruled out.
Errors with morphology can be due to the fact that the learner has not acquired

doi 10.1075/ihll.3.11mon
© 2015 John Benjamins Publishing Company
 Silvina Montrul & Ayʂe Gürel

the relevant abstract f­ eatures or their values, or they can be due to the fact that the
feature may be part of their grammars but the learner lacks the relevant knowl-
edge of the conditions for expressing it. An additional complication for second
and third language acquisition is previous knowledge of a language. It has been
shown that the first language constitutes the initial state of L2 acquisition (see
Schwartz & Sprouse 1996 Full Transfer/Full Access Hypothesis and the research
it generated). As a result, learners already bring an entrenched system of morpho-
syntactic features already assembled into lexical items. These are the morphemes
associated with the functional categories in their language. If appropriate and
sufficient positive evidence exists, L2 learners may be able to restructure their
grammars toward the target.
Lardiere’s Feature Reassembly Hypothesis (Lardiere 2008, 2009) assumes the
Full Transfer/Full Access Hypothesis and holds that L2 learners look for morpho-
lexical equivalents of L1 features and initially assume that their values are the same
as in their L1. Development proceeds to the extent that learners are able to recon-
figure the feature values in lexical items and functional categories to those of the
L2 in cases where these are different. In this study we examine this hypothesis
with respect to the acquisition of Differential Object Marking (DOM), the overt
morphological marking of some direct objects, by Turkish-speaking learners of
Spanish.
DOM is a widespread phenomenon in many languages of the world, includ-
ing Spanish, Romanian, Hindi, Hebrew, Farsi, Turkish and Bantu languages,
among many others (Bossong 1991). Languages like English, Swedish, French and
Russian, for example, do not exhibit DOM. DOM is the overt marking of argu-
ments that are semantically or pragmatically more salient or prominent than their
non-overtly marked counterparts. The objects that must be distinguished from
subjects on semantic and pragmatic prominence scales are the ones that typically
receive overt marking (Aissen 2003; Laca 2006; Pensado 1995). DOM on direct
objects depends on factors such as animacy and specificity (or definiteness), and
properties of the direct object and lexical semantics of the verb (Torrego 1998; Von
Heusinger 2005, 2008).
Example (1) shows an animate, specific, definite direct object in English,
which is unmarked. Example (2a) gives the same construction in Spanish, and the
animate, specific, definite direct object is obligatorily preceded by the preposition
‘a’. The sentence is ungrammatical without the preposition, as (2b) shows.

(1) John saw Maria/the woman. English (no DOM)

(2) a. Juan vio a María/la mujer. Spanish DOM

‘Juan saw DOM Maria/the woman.’
 The acquisition of differential object marking in Spanish by Turkish speakers 

b. *Juan vio María/la mujer.

‘Juan saw Maria/the woman’
‘Juan saw Maria/the woman.’
In recent years, Spanish DOM has received attention in the acquisition litera-
ture because it appears to be a very vulnerable area in language contact situa-
tions. The phenomenon is acquired early, before age 3 years, by monolingual
children (Rodríguez-Mondoñedo, 2008), but it is very difficult to master in L2
acquisition by speakers of languages that do not have DOM (Bowles & Montrul
2009; ­Guijarro-Fuentes 2011, 2012; Guijarro-Fuentes & Marinis 2007). ­Spanish
DOM has also been the focus of extensive investigation in early bilingualism
(­Guijarro-Fuentes & Marinis 2011; Ticio in press) and in language attrition when
the dominant language (e.g. English, French) does not exhibit DOM (Girard 1995;
Grosjean & Py 1991; Montrul 2004; Montrul & Sánchez-Walker 2013).
In this study, we assume the Full Transfer/Full Access Hypothesis (Schwartz &
Sprouse 1996) subsumed under the Feature Reassembly Hypothesis (Lardiere
2008, 2009) to investigate the acquisition of Spanish DOM by Turkish speakers.
Both Spanish and Turkish have DOM but although they are very similar in many
respects the two languages also differ in the conditions under which the relevant
features operate in each language. There are several existing studies of Differential
Object Marking in Spanish as a L2 especially by speakers whose first language
(L1) does not have DOM. Our study is unique in providing new empirical evi-
dence of the acquisition of this property of Spanish by speakers whose language
instantiate DOM.
Since both Spanish and Turkish exhibit DOM, we expect Turkish speakers
to converge on the target distribution of Spanish DOM early in their inter-
language development, even when the interpretable features instantiated in
Spanish DOM may be bundled in a slightly different manner as instantiated
in Turkish DOM. English speakers have been shown to display significant
difficulty in deciding when to mark or not mark objects in Spanish based on
animacy and specificity even at advanced levels of development (Bowles  &
Montrul 2009; Guijarro-­Fuentes 2011, 2012; Guijarro-Fuentes & Marinis
2007). Although we do not include data from English speakers in this study, we
will show that Turkish speakers learning Spanish do not seem to experience the
same level of difficulty as reported of English speakers acquiring Spanish DOM
in the existing literature. We provide additional support for the hypothesis that
the L1 plays a crucial role in the initial and intermediate stages of additional
language acquisition as assumed in the Feature Reassembly Hypothesis. We
begin by describing in more detail how DOM is manifested in Spanish and in
Turkish.
 Silvina Montrul & Ayʂe Gürel

2.  Differential object marking in Spanish and in Turkish

2.1  S panish
As shown in (2) and below in (3), in Spanish animate and specific (definite) direct
objects are marked with the preposition ‘a’, the same as the marker for dative case,
as in Roberto le dio un libro a Patricia (‘Roberto gave Patricia a book.’). All other
direct objects such as animate, non-specific (4), inanimate-specific (5), and inani-
mate non-specific (6) receive no marking. Definiteness and specificity coincide
in these examples: specific objects are expressed with a definite article and non-­
specific objects are expressed with an indefinite article. Spanish DOM is mor-
phologically instantiated by the preposition ‘a’ which carries the semantic formal
features [+animate, +specific].
(3) Marina vio a la mujer +animate, +specific
Marina saw dom the woman
‘Marina saw the woman.’
(4) Marina vio una mujer. +animate, –specific
‘Marina saw a woman (any woman).’
(5) Marina vio la casa. –animate, +specific
‘Marina saw the house.’
(6) Marina vio una casa. –animate, –specific
‘Marina saw a house.’

According to Torrego (1998) properties of the nominal object and of the predi-
cate, such as definiteness, specificity, aspect, topicality, agentivity, and affectedness,
in addition to other pragmatic notions, determine when objects are marked in
S­ panish.1 However, in this study, we do not consider features related to ­topicality,

.  Although animate and specific/definite direct objects must be marked in Spanish, it is
less clear when other objects are marked. Nonspecific quantifiers like alguien ‘somebody’ and
nadie ‘nobody’ always require ‘a’ (Conocí a alguien ‘I met somebody’, No vi a nadie. ‘I didn’t
see anybody.’). It is also possible to occasionally mark inanimate objects for disambiguation
purposes if the subject is also inanimate (El submarino hundió al barco ‘The submarine sank
the ship.’). Nonhuman (animal) direct objects exhibit variability with ‘a’ marking depending
on the type of animal (Mató el/al mosquito. ‘He/she killed the mosquito.’). Based on a crosslin-
guistic comparative study, Aissen (2003) notes that in many languages DOM is characterized
by a great deal of apparent fuzziness. The exact semantic, syntactic and pragmatic condi-
tions regulating when accusative objects should be marked with the preposition ‘a’ in Spanish,
­especially with human indefinites, are quite complex and not entirely clear in the linguistics
literature (Leonetti 2004; Torrego 1998; Weissenrieder 1990; Zagona 2002).
 The acquisition of differential object marking in Spanish by Turkish speakers 

the telicity of the predicate, or agentivity of the subject: we only focus on the
semantic features of the object [+animate,+specific]. Several analyses within gen-
erative grammar (Lidz 2006; López 2012; Rodríguez-Mondoñedo 2007; Torrego
1998) propose that DOM objects check specificity or accusative case, depending
on the analysis, in a higher functional projection different from the projection for
unmarked objects. Thus, overtly marked objects are structurally more complex
than unmarked objects.
Although a seemingly simple marker, Spanish ‘a’ with direct objects presents
significant challenges for its acquisition. In the first place ‘a’ may be difficult to per-
ceive, especially when the verb ends in the vowel [a] as well, as in the third person
singular present, as in (7), where the two /a/ sounds are practically collapsed in
speech producing a long vowel [a:]. In the preterite, even if the verb ends in [o],
as in (8), the [o] from the verbal ending and the DOM marker [a] diphthongize in
rapid speech to [oa] or [ua]. If the verb is in the plural, this problem does not arise:
the DOM marker [a] is more acoustically salient because the coda consonant [n]
or [s] in the verb becomes the onset of the next syllable, taking the DOM marker
[a] as its nucleus.

(7) Marisa llama a Juana. [a:]

‘Marisa calls Juana.’

(8) Marisa llamó a Juana. [oa], [ua]

‘Marisa called Juana.’

(9) Llamamos/llamaron a Juana. [sa], [na]

‘We/they called Juana.’

Note that in some sentences, the presence of ‘a’ is crucial to understand who is
doing what. Compare the minimal pair in (10).
(10) a. Llamó Juan.
‘Juan called.’
b. Llamó a Juan.
‘He/she called Juan.’

The example in (10a) has a postverbal subject and no object (V–S), whereas (10b)
has the structure S-V-O, except that the subject is a null pronoun (pro V-DOM-O).
The ‘a’ in front of Juan indicates that Juan is the object and not the subject.
In addition to its low acoustic salience in these contexts, ‘a’ is a polyfunctional
preposition and a case marker. It is used as a locative or directional preposition,
as in (11); it is the dative marker appearing with indirect objects (regardless of
animacy), as in (12), and it is also the dative marker appearing with dative experi-
encer subjects in gustar-type verbs, as in (13).
 Silvina Montrul & Ayʂe Gürel

(11) Mario fue a Mexico.

‘Mario went to Mexico.’
(12) Juan (le) pidió direcciones a la señora.
‘Juan asked directions to the lady.’
(13) A Pedro le gustó la película.
Pedro-dat liked the movie.
‘Pedro liked the movie.’

In all these constructions, the marker ‘a’ is obligatory, and not subject to ani-
macy or definiteness effects with indirect objects and dative subjects. However,
­a-marking with direct objects looks “optional” because it is regulated by several
semantic and pragmatic constraints or features (animacy, definiteness, specificity,
topicality, among many others), as discussed previously. Given the semantic com-
plexity and ambiguity with respect to its use, learners need to figure out from the
input how to extract the precise syntactic and semantic conditions that regulate
DOM. DOM presents difficulty to learners possibly because it engages multiple
interfaces in the sense of Sorace (2011); see also discussion in Montrul (2011).

2.2  T
 urkish
Turkish is a pro-drop Altaic language with SOV as the canonical word order
(Erguvanlı 1984). However, six different word order patterns are possible under
certain morphosyntactic, phonological, and pragmatic conditions: OSV, OVS,
SOV, SVO, VOS and VSO. Case plays a large part in this word order flexibility.
Subjects in main clauses are not marked (i.e. they are in nominative) but sub-
jects of the nominalized embedded clauses are marked with genitive case (Kornfilt
2003). Direct objects may appear bare or with the accusative case marker –(y)I
depending on the specificity/definiteness of the noun. Turkish does not have a
definite article, but it has a numeral form, bir ‘one’, which is generally analyzed as
an indefinite article (Kornfilt 1997; Lewis 1967; Underhill 1976). With respect to
marking indirect objects and directions, unlike Spanish which uses ‘a’ for accu-
sative dative and direction, Turkish uses a separate dative case morpheme. For
example, the Turkish sentences corresponding the Spanish sentences in (11) and
(12) require dative case-marked indirect objects. Furthermore, Turkish does not
mark dative experiencer subjects as in (13).
The examples in (14) illustrate how the accusative case morpheme indicates
the referential property of the direct object and accusative case-word order inter-
action (Aygen 2007; Von Heusinger & Kornfilt 2005, p. 5):
(14) a. (Ben) kitap oku-du-m. (incorporated)
   I book read-pst-1sg
‘I did book-reading.’
 The acquisition of differential object marking in Spanish by Turkish speakers 

b. (Ben) kitab-ı oku-du-m. (definite)

   I book-acc read-pst-1sg
‘I read the book.’
c. (Ben) bir kitap oku-du-m. (nonspecific, indefinite)
   I a book read-pst-1sg
‘I read a book.’
d. (Ben) bir kitab-ı oku-du-m. (specific, indefinite)
   I a book-acc read-pst-1sg
‘I read a certain book.’
e. Ben hızlı kitap oku-r-um.
I fast book 1sg-aor-1sg
f. *Ben kitap hızlı oku-r-um.
 I book fast read-aor-1sg
g. Ben kitab-ı hızlı oku-r-um.
I book-acc fast read-aor-1sg
‘I read a book fast.’

As can be seen in the above examples, in Turkish, specificity is marked on object

DPs. All definite DPs are specific (Enç 1991). Thus, an accusative case marked
object DP is definite, hence specific. If a direct object is in a position other than
the immediately preverbal position, it has to carry the accusative case suffix (see
Examples (14e) through (14g)) (Aygen 2007).
Furthermore, all definite NPs such as names, pronouns, definite descriptions,
and demonstrative NPs are obligatorily marked with accusative case irrespective
of animacy (Enç 1991, p. 9), as shown in (15).
(15) Zeynep Ali-yi o-nu adam-ı o masa-yı gör-dü.
Zeynep Ali-acc s/he-acc man-acc that table-acc see-pst
‘Zeynep saw Ali/her-him/ the man/ that table.’

Turkish is considered to be a DOM language on the basis of accusative case

marking on direct objects (Aissen 2003). It has been suggested that in ­Turkish
DOM depends on specificity and information structure (typically expressed
by word order). Unlike in Spanish, animacy, however, does not play a role in
­Turkish DOM.
A topicalized direct object, in most cases, is required to be specific but the
presence of accusative case alone may not be sufficient to make the direct object
specific enough to qualify as a topic. Consider the examples in (16) from Von
Heusinger and Kornfilt (2005, p. 12):
(16) a. *Bir kitab-ı Murat aceleyle oku-yor.
 A book-acc Murat hurriedly read-prog
Intended meaning: ‘Murat is hurriedly reading a /some book.’
 Silvina Montrul & Ayʂe Gürel

b. Mavi kaplı bir kitab-ı Murat aceleyle oku-yor

blue covered a book-acc Murat hurriedly read-prog
Intended meaning: ‘Murat is hurriedly reading a certain blue-covered book.’

With respect to the idea that the function of DOM is to disambiguate subject from
object, the prediction is that atypical subjects (i.e. inanimate, indefinite, nonspecific)
are to be marked as the mirror image of DOM. Turkish, however, marks specific
subjects only, as in (17a), and does not mark nonspecific ones, as in (17b). Note the
genitive-marked embedded subjects below (Von Heusinger & Kornfilt 2005, p. 15):
(17) a. [Köy-ü bir haydut-un bas-tığ-ı]-nı duy-du-m
   Village-acc a robber-gen raid-nom-3sg-acc hear-pst-1sg
‘I heard that a certain robber raided the village.’
b. [Köy-ü haydut bas-tığ-ı]-nı duy- du-m
   Village-acc robber raid-nom-3sg-acc hear- pst-1sg
‘I heard that robbers raided the village.’

In sum, as illustrated in the above examples, DOM in Spanish and in Turkish

marks objects but they do not completely correspond to each other. In ­Spanish,
DOM marks animacy and specificity and carries the features [+animate, +­specific],
whereas in Turkish DOM marks specificity [+specific] and is blind to animacy. In
other words, among the factors that govern DOM, Turkish is said to be sensi-
tive only to referential features (specificity/definiteness) and information structure
(word order and topicality), while animacy does not restrict the use of the case suf-
fix (Von Heusinger & Kornfilt 2005, p. 13). Our study does not include examples
testing topicality, which is relevant in Spanish and Turkish, and for this reason we
do not represent topicality in Table 1 below. The properties of DOM that are the
focus of our study are summarized in Table 1.

Table 1.  Feature specification of Spanish and Turkish DOM

DOM Morpholexical form Formal features

Spanish a +animate, +specific/definite

Turkish –(y)I +specific/definite

Indeed, as noted by Von Heusinger and Kornfilt (2005), the semantic and
morphological factors determining the presence or absence of accusative case in
Turkish are too complex to merely be motivated by DOM. Nevertheless, it is con-
ceivable that the presence of differential marking of direct objects in Turkish can
sensitize L1 Turkish learners to related properties in Spanish. Before presenting
the details of our study we briefly review the existing literature on the acquisition
of DOM in Spanish.
 The acquisition of differential object marking in Spanish by Turkish speakers 

3.  P
 revious studies on the acquisition of differential object marking
in Spanish

The only available published study of Spanish DOM in L1 acquisition is

­Rodríguez-Mondoñedo (2008), who conducted an analysis of the spontaneous
production of four Spanish-speaking children (ages 0;9 – 2;11) from the ­CHILDES
data base. All sentences containing V–O structures were analyzed. From a total of
991 exemplars of animate and inanimate objects the children made a total of 17
errors – 98.38% accuracy before age 3 – as Rodríguez-Mondoñedo (2008) reports.
But when accuracy with DOM is taken into account by object, it is 85% for ani-
mate (45/53) and 99% (929/937) for inanimate objects. This study suggests that
Spanish-speaking children acquire the semantic constraints on DOM very early,
at least with core cases.
Research on early bilingualism, especially of child and adult Spanish heritage
speakers in the United States has shown that DOM is not fully acquired in bilin-
gual school-age children and young adult heritage speakers, and it is even subject
to L1 attrition in first generation immigrants (Montrul 2004; Montrul & Bowles
2009; Montrul & Sánchez-Walker 2013; Ticio in press). Grosjean and Py (1991)
and Girard (1995) (both reported in Grosjean 2008) also found that first and sec-
ond generation Spanish immigrants in French-speaking Switzerland showed ero-
sion of Spanish DOM.
Spanish DOM seems to be even more problematic in L2 acquisition. DOM
takes a long time to be mastered by adult English-speaking L2 learners of Spanish,
whose L1 does not mark DOM, despite its frequency in the L2 input (Bowles &
Montrul 2009; Guijarro-Fuentes & Marinis 2007; Guijarro-Fuentes 2011, 2012).
Guijarro-Fuentes (2011, 2012) tested English-speaking learners of Spanish rang-
ing in proficiency from low to advanced. He administered a written completion
task and a contextualized written grammaticality judgment task, manipulating the
animacy and humanness of the subject and the aspectual properties of the verbs,
following Torrego (1998). He found significant differences between the native
speaker controls and all the proficiency groups, including the advanced speakers.
Although there was some development among the L2 learners in the two tasks, the
advanced speakers were still very inaccurate on both tasks. Indeed the advanced
speakers performed at 55.8% accuracy on sentences with animate, specific directs
objects (cf. native speakers 93%) in the acceptability judgment task, and their
highest score (68.2%) was on sentences with inanimate specific objects, which are
unmarked as in English. These findings suggested that Spanish DOM is subject
to incomplete acquisition and fossilization in advanced interlanguages, especially
if the native language of the learner does not have DOM. Interestingly, Spanish
speakers have been shown to transfer DOM when acquiring languages that do not
 Silvina Montrul & Ayʂe Gürel

have DOM, such as Brazilian Portuguese (Halloran & Giancaspro 2012; Montrul,
Dias & Santos 2011) and Catalan (Perpiñán 2013). What no study has investigated
so far is whether speakers of DOM languages do not experience the same difficulty
as speakers of non-DOM languages when acquiring DOM in a second or third
language. This is the focus of our study. We believe that having L1 Turkish learners
of Spanish would contribute to our understanding of how cross-linguistic transfer
would work in the acquisition of a phenomenon (i.e. DOM) exemplified in both
languages. From a learnability perspective, slight differences in the number/type
of linguistic features that govern DOM in Turkish and Spanish may still make
the learning process difficult for L1 Turkish speakers as they need to reassemble/
reconfigure relevant features by adding a new feature (i.e. ±animate) as a relevant
dimension in the realization of DOM in Spanish although it is underdetermined
in the input (see also Guijarro-Fuentes 2011). Thus, this study can be revealing as
to the extent to which cross-linguistic variation within DOM-languages still leads
to difficulty in L2/L3 acquisition.

4.  The study

Our study investigates the acquisition of Spanish DOM by Turkish speakers, whose
language also manifests DOM in a different way. Assuming the Full Transfer/
Full Access Hypothesis (FT/FA) (Schwartz & Sprouse 1996), we predict that
­Turkish speakers will have little difficulty with DOM in Spanish even at earlier
stages of interlanguage development. We follow Guijarro-Fuentes (2011, 2012) in
assuming that Lardiere’s (2009) Feature Reassembly Hypothesis (FRH) provides a
suitable theoretical framework to generate our hypotheses. The FRH departs from
the assumption that phonological, semantic and syntactic features (both + and –
interpretable) are the primitive elements that make up lexical items. Features such
as [±wh], [± plural], [± definite] are bundled and assembled differently in lexi-
cal and functional categories in each language. Lardiere states that acquiring a
L2 involves determining how to assemble the lexical items of the target language,
and this requires the learner to reconfigure or remap features that already exist
in their L1 in different ways when acquiring the L2 or L3. This, it turns out, is a
more challenging task than merely switching parameters, according to Lardiere
(2009, p. 175). Extending this approach to the L2 acquisition of Spanish DOM,
Guijarro-Fuentes (2012) assumed that when preceding animate and definite direct
objects, DOM spells out the features [+specific] and [+animate] in the functional
projection where marked objects check these features (Torrego 1998). Although
English also has animacy and specificity features instantiated in pronouns (his,
her vs. it, for example), they are not overtly instantiated in a specific marker for
 The acquisition of differential object marking in Spanish by Turkish speakers 

direct objects. In Turkish, the feature [+specific] is bundled in the accusative case
marker (y)I, but the feature [+animate] is not. Thus Spanish DOM bundles two
features in the lexical item ‘a’ [+animate, +specific] whereas Turkish -(y)I bundles
only one [+specific]. However, when ‘a’ appears with indirect objects and dative
experiencer subjects, it carries the feature [+dative], since it is a marker of dative
case. Therefore, L2 and L3 learners need to distinguish between direct objects,
indirect objects and dative experiencers first by examining the predicates and then
realizing that only direct objects that are [+animate, +specific] are marked with ‘a’
in Spanish.
Assuming the FRH and the Full Transfer/Full Access Hypothesis, according
to which at initial stages of development L2 learners are highly constrained by
their L1 than later in development, Turkish speakers learning Spanish of low and
intermediate proficiency in the language will have little difficulty marking definite
and specific direct objects with ‘a’. However, those learners who may not know
that the feature [+animate] is also bundled in Spanish ‘a’, will tend to produce and
accept inanimate specific direct objects marked with ‘a’ significantly more than
the native speakers.

4.1  P
 articipants
The participants in our study were a group of native speakers of Spanish tested in
Mexico (N = 20) and a group of Turkish speakers learning Spanish in Istanbul,
Turkey (N = 32). The Mexican native speakers (7 male, 13 female), ages 18–25
(mean 21.05) were attending the Universidad de Guanajuato. They knew some
English as a L2, with age of acquisition ranging from 12 to 24 years.
The Turkish speakers were university students of similar age tested in ­Istanbul,
Turkey. All of them knew English and three of them knew German as well. This
makes Spanish a third language rather than a second. Because English is an oblig-
atory foreign language taught in Turkish schools (from primary through high
school), it is not possible to find L1-Turkish speakers learning Spanish as a foreign/
second language prior to English.
These Turkish participants all began learning Spanish after puberty (mean
19.2, range 14–21). They have been exposed to Spanish for less than one year to
5 years with a mean length of exposure of 2.43 years. One of the Spanish instruc-
tors reported that the construction under investigation is not explicitly taught in
Spanish classes.
The native speakers and the Turkish speakers were administered a writ-
ten proficiency test in Spanish, consisting of a cloze passage (with three
­multiple-choice response options for each blank) from a version of the Diploma
de ­Español como Lengua Extranjera (DELE) and a multiple choice vocabulary
 Silvina Montrul & Ayʂe Gürel

part from an MLA placement test. The maximum possible score on the profi-
ciency test (both sections combined) was 50. Basic information about the two
groups is shown in Table 2.

Table 2.  Information about the participants

Groups N Mean age/range Mean/range of AoA Spanish proficiency score
Spanish (max = 50)

Spanish native 20 21.05 – 45.9

speakers (18–25) (42–50)
Turkish speakers 32 21.7 19.2 28.62
learning Spanish (19–36) (14–21) (16–42)

The native speakers and the Turkish speakers did not differ in age (t (51) = 0.908,
p = 0.34), but they differed in proficiency (t (51) = 127.8, p < 0.0001). The Turkish
group was divided into two proficiency levels for the data analysis: low intermedi-
ate (N = 18) with scores ranging from 16–30, mean 24.05 (SD 4.16) and intermedi-
ate (N = 14) with scores ranging from 31–42, mean 34.5 (SD 3.58).

4.2  Tasks
Our study included three tasks: a written production task, a written comprehen-
sion task, and a bimodal acceptability judgment task.

4.2.1  The written production task

The written production task (WPT) was used in Montrul and Bowles (2010)
classroom-­based study with heritage speakers and consisted of 25 target sentences.
Participants were given three words – a noun, a verb in the infinitive, and another
noun – and were instructed to write a complete sentence with the three words
given by adding all the grammatical elements they considered necessary (arti-
cles, prepositions, inflections, etc.). Five sentences targeted transitive verbs with
animate objects, as in (18), and another five targeted sentences with inanimate
objects, as in (19). The other 15 sentences were fillers and included indirect objects
(20) and psych verbs with dative experiencers and animate (21) and inanimate
themes (22) (five of each).
(18) animate direct object
Prompt: estudiante/ visitar/ profesora
  student visit professor
grammatical response: El estudiante visitó a la profesora
  ‘The student visited the professor.’
 The acquisition of differential object marking in Spanish by Turkish speakers 

(19) inanimate direct object

Prompt: Patricio/ visitar/ Museo del Prado
  Patricio visit del Prado Museum
grammatical response: Patricio visitó el Museo del Prado
  ‘Patricio visited the del Prado Museum.’
(20) indirect object
Prompt: Armando/ enviar/ amiga/ flores
  Armando send friend flowers
grammatical response: Armando (le) envió flores a su amiga.
  ‘Armando sent flowers to her friend.’
(21) Gustar-type verbs with animate dative experiencer
Prompt: Juan/ gustar/ Patricia
  Juan like Patricia
grammatical response: A Juan le gusta Patricia/ Patricia le gusta a Juan.
  ‘Juan likes Patricia.’
(22) Gustar-type verbs with inanimate dative experiencer
Prompt: Francisco/ gustar/ música rock
  Francisco like rock music
grammatical response: A Francisco le gusta la música rock/La música rock
le gusta a Francisco.
  ‘Francisco likes rock music.’

The objective of this task was to see if the L2 learners incorrectly omit ‘a’ with
human or animate specific direct objects in written production, when they sup-
posedly have time to compose and write their responses. We also wanted to inves-
tigate whether those Turkish speakers who still do not know that DOM applies
only to animate objects would incorrectly add ‘a’ to sentences with inanimate
objects. Finally, we wanted to see whether the marker ‘a’ is similarly omitted in the
other syntactic contexts that require obligatory ‘a’ marking: namely, with indirect
objects and with dative subjects of some psych verbs.

4.2.2  The written comprehension task

The goal of the comprehension task was to test whether L2 learners assign mean-
ing to a-marking in DOM sentences. A picture-sentence matching task was
designed to test minimal pairs as in (23), which vary on the presence or absence of
DOM and in their argument structure. To minimize chance performance, the test
included a third choice (a foil sentence) per minimal pair (e.g. plural: Llamaron a
Juan ‘They called Juan.’).
Participants saw three pictures (A, B and C) on a computer screen. In
­picture A, Juan is calling somebody by phone, in picture B Juan is receiving a call,
 Silvina Montrul & Ayʂe Gürel

and in picture C Mary and Jane are calling Juan on the phone. At the same time,
participants saw one sentence at a time, such as Llamó a Juan ‘He/she called Juan.’)
and had to indicate which picture matched the sentence, by pressing the A, B or
C on the keyboard. In this case, B is correct. After the participant made a choice,
the responses disappeared from the computer screen: the survey was programmed
so that it was not possible for participants to compare sentences and go back and
change answers.
The same pictures appeared two more times, with two other sentences Llamó
Juan ‘Juan called.’ and Llamaron a Juan ‘They called Juan.’ If L2 learners do not
assign meaning to the DOM marker, they will be more accurate on sentences with
V–S order than on V–O order. But if L2 learners also have difficulty with postver-
bal subjects in V–S sentences, then they will be equally inaccurate in V–O and V–S
sentences. (Sentences with postverbal subjects were also tested in the acceptability
judgment task). All participants were expected to do better in the plural sentences.
The picture-sentence matching task included 10 verbs with accusative objects
(invitar invite,’ saludar ‘say hi’, llamar ‘call,’ etc.), where the preposition ‘a’ is a
DOM marker as shown in Figure 1, and 10 verbs (5 for each version) with indirect
objects (cantar ‘sing,’ escribir ‘write,’ disparar ‘shoot,’ etc.), where ‘a’ is the regu-
lar dative case marker for goals, as shown in Figure 2. Each verb and each series
of 3 pictures appeared 3 times, once with each sentence type (A = DOM/dative,
B = V–S, C = foil), for a total of 60 sentence-picture pairs.
(23)

Llamó Juan. (B)
‘Juan called.’
Llamó a Juan. (C)
‘She called Juan.’
Llamaron a Juan. (A)
‘They called Juan.’

Figure 1.  Sample picture and sentences used in the aural/written comprehension task.
­Accusative condition

4.2.3  The acceptability judgment task

The acceptability judgment task presented grammatical and ungrammatical sen-
tences with ‘a’ marking and omission with different predicates. The task included
140 sentences (75 grammatical, 65 ungrammatical) divided into 28 types, with five
 The acquisition of differential object marking in Spanish by Turkish speakers 

Preguntó Susana. (A)
‘Susana asked.’
Preguntó a Susana. (C)
‘She asked Susana.’
Preguntaron a Susana. (B)
‘They asked Susana.’

Figure 2.  Sample picture and sentences used in the aural/written comprehension task. Dative
condition

token sentences per type, depending on the structure. Fourteen sentence types are
the focus of the study and the rest were fillers and control structures that we will
not discuss. We included sentences with animate/inanimate, definite/­indefinite
direct objects, indirect objects and dative subjects, where the presence and omis-
sion of the preposition ‘a’ were manipulated, as in Table 3. In order to see whether
learners distinguish a DOM-marked object and a postverbal subject as tested in
the comprehension task, we also included grammatical sentences with postver-
bal subjects and ungrammatical sentences with pronominal subjects and non-
marked animate objects. All sentences were recorded in a sound proof booth by a
female native speaker of Mexican Spanish. Each sentence was presented in writ-
ten ­Spanish and with an audio player below, with bimodal stimulus presentation.
Participants were instructed to read each sentence and play the sound file before
rating each sentence on a 1–4 scale (1 = completely unacceptable to 4 = perfectly
acceptable). The task was self-paced.

Table 3.  Target sentences included in the Spanish bimodal acceptability judgment task
1.  animate direct objects
     a.  definite with DOM
     b.  definite with no DOM
     c.  indefinite with DOM (optional)
     d.  indefinite with no DOM
Marina vio a Madonna.
‘Marina saw DOM Madonna.’  
*Julia vio Shakira. 
‘Julia saw Shakira.’
Mi abuelo conoció a unos pintores.
‘My grandfather knew DOM some painters.’
Antonia vio una gitana. 
‘Antonia saw a gipsy.’

(Continued)
 Silvina Montrul & Ayʂe Gürel

Table 3.  (Continued)

2.  inanimate direct objects
     a.  definite with DOM
     b.  definite with no DOM
     c.  indefinite with DOM
     d.  indefinite with no DOM
3.  Other sentences
     a.  Postverbal subject
     b.  Pronominal subject and no DOM
*Josefa eligió al auto rojo. 
‘Josefa chose DOM the red car.’
Ignacio visitó la universidad.
‘Ignacio visited the university.’
*Adriana siguió a unas direcciones.
‘Adriana followed DOM some directions.’
Pablo escuchó un concierto.
‘Pablo listened to a concert.’
Llamó Ignacio.
‘Ignacio called.’
*Él invitó Juan. 
‘He invited Juan.”

The three tasks were administered through a web-based program called Sur-
vey Gizmo. The order of administration of the tasks did not follow a strict pre-
sentation order and varied by participants. The 32 Turkish L2 learners of Spanish
completed the acceptability judgment task, but due to technical difficulties that
affected the pictures of the comprehension task, 10 of the Turkish participants
could not complete the two other written tasks. So, the results of the written
production and the written comprehension tasks are based on 23 participants:
8 intermediate, 15 low proficiency. This technical difficulty did not affect the data
collection with the native speakers.

5.  Results

5.1  The Written Production Task

Overall accuracy on the production task was high: the mean for the native speak-
ers was 98.7%, 85.4% for the 8 intermediate speakers and 69.5% for the 15 low
proficiency speakers. Figure 3 shows the results by sentence types.
We ran a repeated measures ANOVA with sentence types as within-subjects
variable and group as between-subjects variable. There was a main effect for sen-
tences (F(4,40) = 5.24, p < .001), a main effect for group (F(2,40) = 25.35, p < .0001),
but no significant sentence by group interaction (F(2,40) = 1.75, p < .12). The min-
imum score obtained by a native speaker in a sentence type on this task was 60% if
they made two errors. There was only one native speaker who obtained 60% with
one sentence type (human DO), so we took 60–100% as falling within the range
 The acquisition of differential object marking in Spanish by Turkish speakers 

97.5 98 100 100 99 99

100
92.5
90 88.9
81.8
80 77.5 75
70.3
70 65 67.3

60 56
Spanish native
50 Turkish interm
Turkish low
40
30
20
10
0
DO

DO

ct

e
at

at
je

im

im
Ob
an

e
at

an

an
m

im

ct

in
Hu

h
re
an

yc

h
di

yc
In

Ps
In

Ps

Figure 3.  Mean accuracy in the written production task by verb and object type

of native speakers. Table 4 shows the number and percentage of Turkish-speaking

participants per proficiency group who performed target-like (above 60% accu-
racy with each sentence type) in this task.

Table 4.  Written Production Task. Individual participants and percentage per group who
scored above 60% with each sentence type
L2 groups Human Inanimate Indirect Psych Psych
DO DO object animate inanimate

Turkish interm (8) 100% (8) 100% (8) 100% (7) 87.5% (7) 87.5%
(N = 8)
Turkish low (10) 66.6% (9) 60% (13) 86.6% (10) 66.6% (11) 73.3%
(N = 15)

We see that all the intermediate speakers (8 of 8 or 100%) knew that human
objects take ‘a’ and inanimate objects do not. More than 60% of the sample of low
intermediate participants (N =15) were accurate with human and inanimate direct
objects, but about 40% omitted ‘a’ with human objects and overgeneralized ‘a’
to inanimate objects. With the exception of 2 low intermediate participants, the
rest did not omit the dative preposition with indirect objects. Nonetheless, one
participant of the intermediate group and 5 of the low intermediate group omitted
obligatory ‘a’ with psych verbs.
 Silvina Montrul & Ayʂe Gürel

5.2  The written comprehension task

Figure 4 shows the results of sentences with direct objects marked with DOM,
postverbal subjects, and plural sentences. We first ran an ANOVA with repeated
measures contrasting the three sentence types with direct objects by group.
There was a main effect for sentences (F(2,78) = 9.98, p < .0001), for group
(F(2,39) = 4.21 p = .022), and a sentence by group interaction (F(2,39) = 3.29,
p = .004). The native speakers did not differ from the intermediate level speakers,
according to Tukey post hoc tests (p > .10), but the group difference was due to
the fact that the low intermediate Turkish speakers were significantly different
from the native speakers (p < .017). All participants were very accurate with sen-
tences with DOM marking, but had difficulty with VS sentences, suggesting that
they know that Spanish is a DOM language but they may not know that Spanish
allows postverbal subjects.

10 9.8 9.7
9.2 9.1 9.1 9.3 9.4
9 8.5
8
7 6.8

6
Spanish native
5 Turkish interm
Turkish low
4
3
2
1
0
VDOM VS Plural

Figure 4.  Mean accuracy on the written comprehension task. Sentences with direct objects

Figure 5 shows the results of sentences with indirect objects. A repeated mea-
sures ANOVA showed no significant main effect for sentences and no sentence by
group interaction, although the difference by group was significant (F(2,40) = 3.9,
p < .028). All the Turkish-speaking learners knew that indirect objects in Spanish
are marked with the preposition ‘a’. The low intermediate learners also had some
difficulty with sentences with postverbal subjects.
There were 7 individuals in the two Turkish-speaking groups who scored
lower than 70% on sentences with postverbal subjects in this task, but all partici-
pants scored above 80% with V-DOM sentences, the target of our study.
 The acquisition of differential object marking in Spanish by Turkish speakers 

10 9.8 9.7
9.6 9.5 9.5 9.5 9.5
9 8.8

8 7.4
7
6
Spanish native
5 Turkish interm
Turkish low
4
3
2
1
0
VIO VS Plural

Figure 5.  Mean Accuracy on the Written Comprehension task. Sentences with indirect objects

3.9 3.9 3.9

4 3.8 3.8

3.5 3.4 3.4

3
2.7
2.5 Spanish native
2.5 Turkish interm
Turkish low
2
1.6 1.6
1.5
1.2
1
With DOM Unmarked With DOM Unmarked
Human definite, specific Do Human indefinite DO

Figure 6.  Mean acceptability ratings on human direct objects by group

5.3  The bimodal acceptability judgment task

In this task we asked participants to express acceptability judgments on a 4-point
scale on a list of grammatical and ungrammatical sentences. A rating of 4 meant
perfectly grammatical and acceptable, a rating of 1 meant completely ungrammati-
cal and unacceptable. The “I don’t know” option, coded as 0, was provided sepa-
rately from the 4-point scale. All “I don’t know” (0) responses were discarded before
submitting the mean acceptability ratings for statistical analysis. Figure 6 shows the
 Silvina Montrul & Ayʂe Gürel

results of sentences with human, definite and indefinite direct objects. Sentences
with human, definite and specific direct objects require DOM human, definite
objects without DOM are ungrammatical. Sentences with indefinite human objects
are optional with DOM depending on whether the object is specific or not. Unlike
Guijarro-Fuentes (2011, 2012), we did not provide contexts to evaluate the option-
ality of the DOM marker with human indefinites. Thus, because the sentences
were presented in isolation, in principle they are both grammatical depending on
context.
A repeated measures ANOVA with sentence types as the within-subjects
variable and groups as the between-subjects variable showed a main effect for
sentences (F(3,49) = 280.06, p < .0001), for groups (F(2, 49) = 14.97, p < .0001),
and a sentence by group interaction (F(2, 49) = 4.38, p < .0001). Tukey post hoc
tests indicated that the native speakers differed from the two groups of Turkish
speakers (p < .0001), but the two Turkish-speaking groups did not differ from
each other (p > .10). We followed up these tests with ANOVAs and pairwise com-
parisons within each group. The native speakers rated ungrammatical sentences
with unmarked human definite objects significantly different from the other three
(F(1,19) = 4169.45, p < .0001). (Note that the two sentences with indefinite objects
were judged grammatical with marked or unmarked objects). The intermediate
level speakers also rated grammatical sentences with DOM significantly higher
than the two sentence types without DOM (F(1,14) = 22.10, p < .0001). Crucially,
the Turkish speakers rated sentences with indefinite human objects without DOM
significantly more acceptable than sentences with human, definite, specific objects
without DOM (t(13) = 5.01, p < .0001), which means that they are sensitive to
‘a’ as a marker of specificity. The same pattern obtained for the low intermediate
speakers, who also rated the four sentence types statistically different (F(3, 15) =
55.85, p < .0001). Despite their low proficiency, they rated unmarked indefinite
objects as more acceptable than definite objects in sentences without DOM (t(18)
= 6.23, p < .0001), showing that they are sensitive to ‘a’ as a marker of specificity
as well. Even though the native speakers assigned lower acceptability ratings to
ungrammatical animate, specific direct objects without DOM than the Turkish
speakers (F(2, 49) = 8.04, p < .001), the Turkish speakers seem to know that DOM
is required to mark specificity in Spanish.
Figure 7 shows the results of sentences with inanimate objects, which do not
require DOM. Because the Turkish accusative DOM marker marks specificity and
not animacy, we expected Turkish-speaking learners to incorrectly accept inani-
mate objects marked with ‘a’ in Spanish. A repeated measures ANOVA showed
a main effect by sentences (F(3,49) = 72.47, p < .0001), no significant main effect
by group (F(3, 49)= 2.85, p = .065), but a significant sentence by group interac-
tion (F(6, 49) = 6.30, p = .004). The interaction was due to the fact that the low
 The acquisition of differential object marking in Spanish by Turkish speakers 

proficiency Turkish speakers rated inanimate objects with DOM more acceptable
than the other two groups (p < .0001). Nonetheless, the Turkish speakers rated
unmarked inanimate objects more acceptable than inanimate objects with DOM,
which means that they know that inanimate objects do not generally have ‘a’.

4 3.9
4
3.6 3.5
3.5 3.4
3.3

3 2.8
2.6 Spanish native
2.5 Turkish interm
2.1 Turkish low
1.9 2
2 1.8

1.5

1
With DOM Unmarked With DOM Unmarked
Inanimate definite, specific Do Inanimate indefinite DO

Figure 7.  Mean acceptability ratings on inanimate direct objects by group

4
3.8

3.5

3 2.9
2.8
Spanish native
2.5 Turkish interm
2.2 Turkish low
2
1.6
1.5
1.2
1
VS Pron VO

Figure 8.  Mean acceptability ratings on postverbal subjects (VS) and sentences with
­pronominal subjects and unmarked human objects (Pron VO)

Figure 8 shows the results of sentences with postverbal subjects (VS) (gram-
matical) and sentences with pronominal subjects and unmarked human objects
(ungrammatical). Sentences with postverbal subjects were prone to errors by
the Turkish speakers in the comprehension task. In this task as well, there were
 Silvina Montrul & Ayʂe Gürel

­ ifferences between the native speakers and the Turkish speakers with VS sentences
d
(F(2, 49) = 13.97, p < .0001) and with pron VO sentences (F(2,49) = 19.43, p < .0001).
The low intermediate and the intermediate Turkish speakers did not differ from each
other on their ratings of VS sentences (p > .5), but the intermediate level speak-
ers, who did not differ from the native speakers, were more accurate at rejecting
ungrammatical sentences without DOM than the low level speakers (p < .0001).
To summarize, the results of our three tasks showed that low and intermedi-
ate proficiency Turkish-speaking learners of Spanish know that ‘a’ is required with
animate, specific direct objects in Spanish, as shown in the production, compre-
hension and acceptability judgment tasks. Some participants had difficulty inter-
preting and judging sentences with postverbal subjects in Spanish, suggesting that
the difficulty might be related to the fact that they may not know that Spanish
allows postverbal subjects when there is no marking on the subject. The Turkish-
speaking learners were also quite accurate at rejecting ungrammatical sentences
with animate and specific direct objects unmarked with DOM. Finally, a few sub-
jects accepted ungrammatical sentences with DOM-marked inanimate objects.
Although the participants tested were not of advanced proficiency in Spanish, they
seem to know that Spanish is a DOM-language.

6.  Discussion and conclusion

Existing studies on the acquisition of differential object marking in Spanish sug-

gest that DOM is very difficult to acquire in a second language when the first
language of the learners does not mark DOM (Bowles & Montrul 2009; Guijarro-
Fuentes 2011, 2012). The purpose of our study was to investigate whether learners
whose L1 instantiates this phenomenon are able to acquire Spanish DOM, thus
adding new empirical evidence and contributing to theoretical discussion of the
role of L1 transfer in different language combinations. We tested whether Turkish
speakers learning Spanish as a third language knew, at early stages of develop-
ment, the conditions under which direct objects must be marked in Spanish with
the marker ‘a’. In Spanish and in many languages (i.e. Hindi) the DOM marker is
also the dative marker of indirect objects and dative experiencers. In Romanian,
­Turkish and Hebrew the dative is not the source of DOM (Aissen 2003).
According to Aissen (2003, p. 437) there is room for language particular varia-
tion among languages that exhibit DOM. That is, even when DOM exists in two
languages, the dimensions that define DOM and the cut off point for DOM may
vary. In Spanish, animacy and specificity are the main dimensions that regulate
DOM whereas Turkish DOM is determined by the definiteness scale only. In
Turkish, DOM distinguishes between specific and nonspecific objects.
 The acquisition of differential object marking in Spanish by Turkish speakers 

We assumed the Feature Reassembly Approach (Lardiere 2009), according

to which L2 learners must figure out how to bundle features in lexical items in
their L2. In Spanish, DOM has two main features [+animate, +specific] whereas
in Turkish DOM has only one feature [+specific]. Animacy and specificity can
be bundled in a DP. Animacy is lexically encoded, whereas specificity is not lexi-
cally encoded in nouns and can be obtained from contextual information and
other syntactic means (e.g. scrambling, functional information in determiners,
case marking, etc.). Turkish speakers learning Spanish must add a feature [+ani-
mate] to their representation of DOM in Spanish and must encode both features
in the lexical item ‘a.’ But Turkish learners may be confused because ‘a’ is also the
dative case marker appearing with indirect objects and with the dative experienc-
ers of psych verbs, irrespective of animacy (although the great majority of indirect
objects are animate and so are dative experiencers).
We tested our hypothesis with three tasks: a written production task, a writ-
ten comprehension task and a bimodal acceptability judgment task. We found
that, overall, Turkish learners of Spanish know that Spanish is a DOM language,
and they know that ‘a’ is required with animate, specific direct objects in ­Spanish.
More than 60% of the participants produced the DOM marker with 100% accu-
racy in the production task. They were also above 80% accurate in the compre-
hension task, where they had to discriminate DOM marked animate and specific
objects from postverbal subjects. Although we did not test topicality in this study,
perhaps Turkish speaking learners of Spanish may have also noticed that the fea-
ture topicality is relevant in Spanish DOM as it is in Turkish, which may have
also contributed to their initial high performance in Spanish. Guijarro-Fuentes
(2012) found that all the English-speaking learners tested in his study gave similar
ratings to grammatical and ungrammatical sentences with DOM, regardless of
proficiency level. There was some evidence in his studies that the English-speaking
learners had developed some sensitivity to Spanish ‘a’ as a marker of animacy
with direct objects, but had not bundled specificity. They correctly accepted gram-
matical sentences but were very inaccurate rejecting ungrammatical sentences
with no DOM. By contrast, we found in our study that our low intermediate and
intermediate level learners of Spanish whose first language is Turkish gave statisti-
cally different ratings to grammatical and ungrammatical sentences with DOM
and animate, specific direct objects. Taken together, our results suggest that many
Turkish-speaking learners of Spanish at low and intermediate proficiency have
correctly bundled the features [+animate] and [+specific] to DOM in Spanish.
At the same time, the Turkish speakers and the native speakers were different
in other respects. First, the Turkish speakers were inaccurate with sentences with
postverbal subjects in the comprehension task and in the acceptability judgment
task. This may be due to the fact that Spanish does not overtly mark n ­ ominative
 Silvina Montrul & Ayʂe Gürel

case, and the Turkish speakers may have difficulty interpreting a subject that is
not preverbal if unmarked. This is, however, unexpected under the FT/FA view
given that Turkish is also a pro-drop language with flexible word order and does
not morphologically mark subjects either preverbally or postverbally. A more
plausible explanation would be that the Turkish speakers had difficulty compre-
hending these sentences and judging their acceptability because they thought
that the NP following the verb was an object but was not marked with case. This
assumption would imply transfer from L1 Turkish that allows postverbal scram-
bling of objects on the basis of case marking (Aygen 2002; Kural 1997). The
fact that there was incorrect acceptance of inanimate objects with DOM in the
acceptability judgment task may support this hypothesis. That is, some ­Turkish
speakers may still analyze Spanish DOM as a marker of specificity only as in
Turkish, and have not yet figured out that an object needs to be both animate
and specific to be marked in Spanish. So, although many learners may have reas-
sembled and recombined the relevant features, some learners, especially those at
the lowest level of proficiency, are still very much constrained by the features of
their L1. All subjects seem to know that ‘a’ is a marker of dative case with indi-
rect objects, but since Turkish does not have dative experiencer subjects, many
participants have not yet learned that dative experiencers have an obligatory ‘a’
in Spanish as well.
It is highly conceivable that all these remaining errors are due to the influence
of Turkish as we suggested. Nevertheless, let us not forget that Spanish was the
third rather than the second language of these learners. It is possible that some of
the errors are due to English (Bardel & Falk 2007). The difficulty with postverbal
subjects, for example, could be explained by influence from English, since English
has very strict word order by comparison. However, influence from English can-
not explain the overgeneralization of DOM to inanimate objects as found in the
production and in the acceptability judgment task, since English does not mark
inanimate objects either.
In general, and even though we do not have results from very advanced speak-
ers for this study, our results show that Turkish speakers learning Spanish are quite
successful from early stages of development at reassembling and rebundling the
features associated with DOM in Spanish. Our initial study suggests that speakers
of DOM languages are quite successful at acquiring DOM in a second or third
language, even when there is some language-specific variation in the expression
of DOM in the two languages. Admittedly, our conclusion would be stronger if
we had also included data from speakers of English or another non-DOM lan-
guage for more direct comparison. Further research currently in progress includ-
ing speakers of DOM and non-DOM languages will help confirm this preliminary
finding.
 The acquisition of differential object marking in Spanish by Turkish speakers 

References

Aissen, J. (2003). Differential object marking: Iconicity vs. economy. Natural Language and Lin-
guistic Theory, 213, 435–448.
Aygen, G. (2002). Finiteness, Case and Clausal Architecture. Unpublished Ph.D. dissertation,
Harvard University.
Aygen, G. (2007). Specificity and subject-object positions/scope interactions in Turkish. Dil ve
Edebiyat Dergisi/Journal of Linguistics and Literature, 4(2), 56–87.
Bardel, C., & Falk, Y. (2007). The role of the second language in third language acquisition: The
case of Germanic syntax. Second Language Research, 23, 459–484.
DOI: 10.1177/0267658307080557
Bossong, G. (1991). Differential object marking in Romance and beyond. In D. Wanner &
D. Kibbee (Eds.), New Analyses in Romance Linguistics. Selected Papers from the XVIII Lin-
guistic Symposium on Romance Languages (pp.143–170). Amsterdam: John Benjamins.
DOI: 10.1075/cilt.69.14bos
Bowles, M., & Montrul, S. (2009). Instructed L2 acquisition of differential object marking in
Spanish. In R. P. Leow, H. Campos, & D. Lardiere (Eds.), Little Words. Their History, Pho-
nology, Syntax, Semantics, Pragmatics and Acquisition (pp. 199–210). Washington DC:
Georgetown University Press. DOI: 10.1017/s0272263110000446
Chomsky, N. (2001). Derivation by phase. In K. L. Hale & M. Kenstowicz (Eds.), Ken Hale: A Life
in Language (pp. 1–54). Cambridge, MA: The MIT Press. DOI: 10.1017/s0022226704322747
Enç, M. (1991). The semantics of specificity. Linguistic Inquiry, 22, 1–25.
Erguvanlı, E. (1984). The Function of Word Order in Turkish Grammar. Berkeley, CA: University
of California Press. DOI: 10.2307/415636
Grosjean, F. (2008). Studying Bilinguals. Oxford: OUP. DOI: 10.1017/s0272263109090111
Grosjean, F., & Py, B. (1991). La restructuration d’une première langue: L’intégration de vari-
antes de contact dans la compétence de migrants bilingues. La Linguistique, 27, 35–60.
Girard, E. (1995). Intégration de variantes de contact dans la compétence de bilingues de deuxième
génération. Master’s thesis, University of Neuchâtel, Switzerland.
Guijarro-Fuentes, P. (2011). Feature composition in differential object marking. In L. Roberts,
G. Pallotti, & C. Bettoni (Eds.), EUROSLA Yearbook 11 (pp. 138–164). Amsterdam: John
Benjamins. DOI: 10.1075/eurosla.11.09gui
Guijarro-Fuentes, P. (2012). The acquisition of interpretable features in L2 Spanish: Personal a.
Bilingualism, Language and Cognition, 15(4), 701–720.
DOI: 10.1017/s1366728912000144
Guijarro-Fuentes, P., & Marinis, T. (2007). Acquiring the syntax/semantic interface in L2
­Spanish: The personal preposition ‘a’. In L. Roberts, A. Gürel, S. Tatar, & L. Martı (Eds.),
Eurosla Yearbook 7 (pp. 67–87). Amsterdam: John Benjamins.
DOI: 10.1075/eurosla.7.06gui
Guijarro-Fuentes, P., & Marinis, T. (2011). Voicing language dominance: The acquisition of
interfaces by English/Spanish bilingual adolescents. In K. Potowski & J. Rothman (Eds.),
Bilingual Youth: Spanish in English-speaking Societies (pp. 227–248). Amsterdam: John
Benjamins. DOI: 10.1075/sibil.42.14gui
Halloran, B., & Giancaspro, D. (2012, October). Examining L3 transfer models: The acquisition
of differential object marking in L3 Brazilian Portuguese. Poster presented at GALANA,
University of Kansas, Lawrence.
Kornfilt, J. (1997). Turkish. Descriptive Grammars Series. London: Routledge.
 Silvina Montrul & Ayʂe Gürel

Kornfilt, J. (2003). Scrambling, subscrambling, and case in Turkish. In S. Karimi (Ed.), Word
Order and Scrambling, (pp. 125–155). Malden, MA: Blackwell.
DOI: 10.1002/9780470758403.ch6
Kural, M. (1997). Postverbal constituents in Turkish and the linear correspondence axiom. Lin-
guistic Inquiry, 28, 498–519.
Laca, B. (2006). El objeto directo. In C. Company (Ed.) Sintaxis histórica del español. Vol I: La
frase verbal. México: Universidad Autónoma de México.
Lardiere, D. (2008). Feature-assembly in second language acquisition. In J. Liceras, H. Zobl, &
H. Goodluck (Eds.), The Role of Formal Features in Second Language Acquisition
(pp. 106–140). Mahwah, NJ: Lawrence Erlbaum Associates.
Lardiere, D. (2009). Some thoughts on the contrastive analysis of features in second language
acquisition. Second Language Research, 25, 173–227. DOI: 10.1177/0267658308100283
Leonetti, M. (2004). Specificity and differential object marking in Spanish. Catalan. Journal of
Linguistics, 3, 75–114.
Lewis, G.L. (1967). Turkish Grammar. Oxford: OUP.
Lidz, J. (2006). The grammar of accusative case in Kannada. Language, 82, 10–32.
DOI: 10.1353/lan.2006.0054
López, L. (2012). Indefinite Objects. Scrambling, Choice Functions and Differential Marking. Lin-
guistic Inquiry Monograph, 63. Cambridge, MA: The MIT Press.
DOI: 10.1017/s0022226713000376
Montrul, S. (2004). Subject and object expression in Spanish heritage speakers: A case of
­morpho-syntactic convergence. Bilingualism: Language and Cognition, 7(2), 125–142.
DOI: 10.1017/s1366728904001464
Montrul, S. (2011). Interfaces and incomplete acquisition. Lingua, Special issue on Interfaces in
Language Acquisition, 212(4,) 591–604. DOI: 10.1016/j.lingua.2010.05.006
Montrul, S., & Bowles, M. (2009). Back to basics: Differential object marking under incomplete
acquisition in Spanish heritage speakers. Bilingualism: Language and Cognition, 12(3),
363–383. DOI: 10.1017/s1366728909990071
Montrul, S., & Bowles, M. (2010). Is grammar instruction beneficial for heritage language learn-
ers? Dative case marking in Spanish. The Heritage Language Journal, 7(1), 47–73.
Montrul, S., & Sánchez-Walker, N. (2013). Differential object marking in child and adult S­ panish
heritage speakers. Language Acquisition, 20, 1–24.
DOI: 10.1080/10489223.2013.766741
Montrul, S., Dias, R., & Santos, H. (2011). Clitics and object expression in the L3 acquisition of
Brazilian Portuguese. Structural similarity matters for transfer. Second Language Research,
27(1), 21–58. DOI: 10.1177/0267658310386649
Pensado, C. (1995). El complemento directo preposicional: Estado de la cuestión y bibliografía
comentada. In C. Pensado (Ed.). El complemento directo preposicional (pp. 11–59). Madrid:
Visor. DOI: 10.2307/474952
Perpiñán, S. (2013, April). Direct object expression and its semantic properties in Catalan-­
Spanish bilingualism. 43rd Linguistics Symposium on Romance Languages, City University
of New York (CUNY Graduate Center, April 19).
Rodríguez-Mondoñedo, M. (2007). The Syntax of Objects: Agree and Differential Object Mark-
ing. Unpublished Ph.D. dissertation. University of Connecticut.
Rodríguez-Mondoñedo, M. (2008). The acquisition of differential object marking in Spanish.
Probus, 20, 111–145. DOI: 10.1515/probus.2008.004
 The acquisition of differential object marking in Spanish by Turkish speakers 

Schwartz, B.D., & Sprouse, R.A. (1996). L2 cognitive states and the full transfer/full access
model. Second Language Research, 12, 40–72. DOI: 10.1177/026765839601200103
Sorace, A. (2011). Pinning down the concept of “interface” in bilingualism. Linguistic Approaches
to Bilingualism, 1(1), 1–33. DOI: 10.1075/lab.1.1.01sor
Ticio. E. (in press). Differential object marking in Spanish-English early bilinguals. Linguistic
Approaches to Bilingualism.
Torrego, E. (1998). The Dependency of Objects. Cambridge, MA: The MIT Press.
DOI: 10.1017/s0022226799298088
Underhill, R. (1976). Turkish Grammar. Cambridge: MA: The MIT Press.
DOI: 10.1017/s0041977x0010864x
Von Heusinger, K. (2005). The evolution of Differential Object Marking in Spanish. In K. Von
Heusinger, G. Kaiser, & E. Stark (Eds.), Proceedings of the Workshop Specificity and the
Evolution/Emergence of Nominal Determination Systems in Romance. Fachbereich Sprach-
wissenschaft der Universität Konstanz, Arbeitspapier Nr. 119. (pp. 33–70). Konstanz:
­University of Konstanz.
Von Heusinger, K. (2008). Verbal semantics and the diachronic development of DOM in
­Spanish. Probus, 20, 1–31. DOI: 10.1515/probus.2008.001
Von Heusinger, K., & Kornfilt, J. (2005). The case of the direct object in Turkish: Semantics,
syntax and morphology. Turkic Languages, 9, 3–44.
Weissenrieder, M. (1990). Variable uses of the direct-object marker A. Hispania, 73, 223–231.
DOI: 10.2307/343010
Zagona, K. (2002). The Syntax of Spanish. Cambridge: CUP.
DOI: 10.1017/cbo9780511613234
 Copula choice in adjectival constructions
in Dutch L1 Spanish L2

Manuela Pinto1 & Alexia Guerra Rivera2

Utrecht University1, 2 / Universidad Austral de Chile2

L2 learners of Spanish have difficulties choosing the right copula in

ser/estar + adjective constructions. Beyond the effect of processing load of
information pertaining to different cognitive modules (Sorace’s (2011)
Interface Hypothesis), this study aims at pinpointing the specific properties
that make these constructions so problematic for L2 learners. A detailed
analysis of the semantics of Dutch and Spanish copulas reveals a difference
between the two languages at the level of inherent aspectual properties.
A Fill-in-the-Gap Task testing the visibility of these properties showed that
Dutch L1/Spanish L2 learners are aware of the different semantics of ser/estar
versus zijn/worden, but failed to assign the target copula form to a subclass
of gradable adjectives. The explanation for this specific case is still tentative,
but the results of this study contribute to understanding what happens at the
interfaces and how the process of L2 acquisition may work.

1.  Introduction

This paper aims to contribute to the debate on the accessibility of inherent

semantic properties in L2 acquisition by examining a specific construction of
Spanish grammar, the predicative construction formed by the copula ser or estar
and an adjective. The majority of Spanish adjectives can take ser as well as estar.
Each copula form, however, implies a different interpretation of the predicate.
Matching between form, meaning (semantics) and interpretation (pragmatics)
is required in order to produce the target construction. Native speakers seem
to have clear intuitions with regard to this copula choice. Simultaneous bilin-
guals, Heritage ­Speakers, and L2 learners with English L1 show instead persistent
optionality in the choice of the right copula (Bruhn de Garavito & Valenzuela
2008; among others). The problem may be ascribed to the interface nature of
these constructions: ­integration of information pertaining to different cogni-
tive domains implies a higher processing load, making these areas of second

doi 10.1075/ihll.3.12pin
© 2015 John Benjamins Publishing Company
 Manuela Pinto & Alexia Guerra Rivera

language acquisition more prone to show target-deviant constructions (Sorace

2011). Although Sorace’s Interface Hypothesis provides a unified explanation for
a number of apparently unrelated phenomena in L2 acquisition, this may not
be sufficiently fine-grained to account for the fact that not all interface proper-
ties are equally difficult to acquire. The present research addresses precisely this
issue. The data we collected support the Interface Hypothesis: Dutch L1-Spanish
L2 speakers’ target-deviant constructions are localized in ambiguous contexts.
However, our data also show that certain constructions are more problematic
than others. Beyond the Interface Hypothesis, the goal of this study is to analyze
in detail Dutch and S­ panish copula + adjective constructions and pinpoint their
vulnerable aspects in the L2 acquisition process. The experimental data presented
here show that Dutch L1-Spanish L2ers do not have problems with all ambiguous
copula + ADJ constructions, but just with a subclass of adjectives, the so-called
irreversible scalar gradable adjectives. These findings suggest that L2ers may have
selective access to some inherent semantic properties of the copula.
This paper is organized as follows: the next section offers a description of
adjectival constructions in Spanish and gives a brief review of how these construc-
tions have been examined in syntactic and in semantic terms. The distribution
of the two Spanish copulas in adjectival constructions is also compared to Dutch
copulas zijn and worden. In Section 3 we comment on some studies on L2 acquisi-
tion of ser/estar, focusing on ambiguous constructions. In Section 4 we present the
results of two tasks, a Grammaticality Judgment Task testing competence and use
of ser/estar by Dutch L1 speakers, and a Fill-in-the-Gap Task, testing the L2 acces-
sibility of inherent semantic properties of adjectival constructions. The results of
both tasks are discussed in Section 5 and the final section contains our concluding
remarks.

2.  Theoretical background

2.1  Ser and estar with adjectives

Spanish has two copular verbs, ser and estar, whereas a language like English, has
only one form: be. In Spanish, copulas link noun phrases to predicate structures,
noun phrases, adjective phrases, and prepositional phrases. This can be seen in
Examples (1) to (5) below.
(1) Miguel es estudiante.
Miguel ser-3sg student
‘Miguel is a student.’
 Copula choice in adjectival constructions in Dutch L1 Spanish L2 

(2) Miguel está enfermo.

Miguel estar-3sg enfermo-mas
‘Miguel is sick.’
(3) Miguel es rubio.
Miguel ser-3sg rubio-mas
‘Miguel is blond.’
(4) Viviana está en Chile.
Viviana estar-3sg in Chile
‘Viviana is in Chile.’
(5) La fiesta es en casa de María.
The party ser-3sg at house of Maria
‘The party is at Maria’s house.’

In Example (1), the referent Miguel is linked to the NP estudiante ‘student’. In

Examples (2) and (3), Miguel is linked to the AP enfermo ‘sick’. In Examples (4)
and in (5), the subjects Viviana and fiesta are linked to the PP ‘in Chile’ and ‘at
María’s house’ respectively. As the examples show, in all these cases English uses
is (from be), whereas Spanish has the choice between es (from ser) and está (from
estar). While the copula construction with NP always requires ser, both the adjec-
tival and the prepositional constructions can appear with either ser or estar. In
what follows we focus strictly on adjectival constructions.
As shown in Examples (2) and (3), adjectival constructions in Spanish can
have both copulas. Although some adjectives allow only one copular verb (e.g.
el hombre está muerto/ ‘the man is dead’), more than 80% of all adjectives allow
alternation between ser and estar (Mesa Alonso, Dominguez Herrera, Padrón
Sanchez & Morales Aguilera, 1993). When this alternation occurs, the interpre-
tation of the clause changes depending on the copula, as illustrated in (6) and (7)
below.
(6) Mi primo es delgado. (with ser)
‘My cousin is slim.’
(7) Mi primo está delgado. (with estar)
‘My cousin is slim (nowadays, or than before).’

In Example (6), ser refers to properties that are relatively stable, and that define
characteristics at an individual level, whereas estar, in Example (7), denotes a more
temporary, changing property. Both sentences translate into English as ‘my cousin
is thin’. However, in (6) the interpretation is that my cousin is a thin person (always
has been) whereas in (7) my cousin may have become slim as a result of having lost
weight. In this case being slim is interpreted as the result of a change.
 Manuela Pinto & Alexia Guerra Rivera

2.2  Ser and estar between syntax and semantics

Spanish adjective constructions with ser/estar have received much attention in lin-
guistic research. An exhaustive review of the literature is far beyond the scope of
this paper. What follows, however, is a brief review of the main lines of research on
the topic in the last decade.
The intuition that ser does not add to the interpretation of the predicate,
whereas estar conveys aspectual information has been formalized in various ways.
From a morpho-syntactic point of view, Uriagereka (2001), Zagona (2012), and
Gallego and Uriagereka (2009) have proposed that estar consists of copula ser
plus an abstract preposition P that provides the relevant context for the estar con-
structions to be licensed. Along a similar vein, Brucart (2012) argues that estar
includes an interpretable feature of terminal coincidence that is processed in syn-
tax, either as an agreement relation between estar and the attributive predication,
or by means of the copula, that introduces an aspectual delimitation in the state
denoted by the adjective.
From a semantics-oriented point of view, Diesing (1990) applies the stage vs.
individual-level distinction to Spanish adjectival predicates, capitalizing on the
intuition that the property denoted by an adjectival predicate formed with estar
has a temporary interpretation, whereas a property denoted by ser, has a perma-
nent interpretation.
Schmitt (1992), Schmitt (2005), and Schmitt and Miller (2007) explore the
inherent aspectual properties of Spanish copulas and propose that the distinc-
tion between ser and estar is aspectual in nature: estar denotes a copula of the
state-type, whereas ser lacks aspectual content. Copula estar is marked with the
aspectual feature [+STATE], indicating a temporary state. Conversely, ser does not
carry any aspectual properties. Possible aspectual interpretations of ser can only
be achieved by adding lexical aspectual markers. Ser is flexible in terms of its event
type properties (state, event) and therefore can appear in various contexts. The
propensity of estar to appear with temporary predicates and of ser with permanent
properties has a pragmatic nature and arises through implicatures with regard to
the aspectual meaning of the adjective in a specific time interval. If, as suggested
by Schmitt and Miller (2007), ser does not encode any aspectual properties, it does
not contribute to the assertion of an eventuality type. In this sense, ser is said to be
neutral with regard to aspect. Schmitt and Miller’s analysis accounts for the intu-
ition that a large class of predicates can in fact have both interpretations, depend-
ing on the context of use.
Ursini (2011) and Camacho (2013) elaborate on the intuition that copula estar
introduces a measure into predication. Ursini (2011), following Kennedy (1999,
2007), argues that adjectives can be seen as denoting scales, partially ordered sets
 Copula choice in adjectival constructions in Dutch L1 Spanish L2 

of referents having a minimum amount of a given property. In Camacho (2013),

the measure is determined by the copula, in the sense that it is estar that entails a
boundary – the begin point – for the predicate. However, both authors agree that the
features of the copula must match the features of the adjective.
This brief review has shown that recent research on the syntax and semantics
of Spanish copulas considers ser as the neutral, featureless form, and estar, the
semantically complex form. It also emerged that matching of features between
copula and adjective involves a ‘measure’ of the degree of a certain denotation or,
alternatively, of its beginning point. This last observation is particularly relevant
for the present research: as will be explained in more detail below, our experimen-
tal data revealed that the choice of copula is made on the basis of an examina-
tion of the inherent aspectual properties of the adjective, more specifically, on the
modality of the change of state denoted by the predicate.

2.3  L2 acquisition of Ser and Estar

A majority of the studies on copula choice in Spanish L2 have been carried out
with native English speakers who were learning Spanish in instructional settings
in an English-speaking country. The initial concern of these studies was that of
determining how the acquisition of the two copulas develops over time and in
relation to L2 proficiency (Geeslin 2002, 2003a, 2003b; Gómez-Soler 2009; Gun-
terman 1992; Ryan & Lafford 1992; VanPatten 1985, 1987). For example, VanPat-
ten (1985, 1987) collected data from six English-speaking beginning learners of
Spanish enrolled in a college in the United States. The students were individually
recorded for thirty minutes, every two weeks, for nine months, while participating
in a conversation and a picture description task. The results showed that whereas
the use of ser was about 90% correct, the use of estar was at or below 50% dur-
ing almost every recording. VanPatten identified five clear stages through which
s­ tudents pass:

a. Absence of copula in learner speech.

b. Selection of ser to perform most copula functions.
c. Appearance of estar with progressive.
d. Appearance of estar with locative
e. Appearance of estar with adjectives of condition.

STUDIES that followed examined two variables, L2 proficiency and L1 interference.

Geeslin (2001) and Geeslin and Guijarro-Fuentes (2006) tested E ­ nglish-speaking
learners of Spanish at different proficiency levels (from beginner to intermediate)
and found that proficiency positively affected the use of estar; although for ser it was
less clear whether proficiency was a relevant factor.
 Manuela Pinto & Alexia Guerra Rivera

In another study Geeslin and Guijarro-Fuentes (2005) examined the effects of

a number of factors on the acquisition of Spanish ser and estar. One of these factors
was the linguistic background of the participants. This variable was considered
in order to determine whether previously acquired languages (French, ­German,
­English) played a role in Spanish L2 copula choice. Crucially, none of the L1 lan-
guages had a similar copula system as the Spanish one. The authors claimed that
the L1 did not play a relevant role in the acquisition of the Spanish copulas, as the
frequency of use of ser/estar and their sample did not vary across the different L1s.
More recently, research on Spanish copulas in L2 acquisition has shifted the
focus of interest to the aspectual properties of copulas, and to how L2 learners
(henceforth L2ers) deal with these properties. Bruhn de Garavito and Valenzuela
(2006) investigated the acquisition of ser and estar in verbal and adjectival passives
and in individual and stage level predicates by two different L2 populations: early
and late bilinguals (English L1 and Spanish L2). The main aim of this study was
to determine if these learners could acquire the syntactic and interpretive differ-
ences between the copulas and whether some of their properties were vulnerable
to fossilization or incomplete acquisition (Bruhn de Garavito & Valenzuela 2006,
p. 106). Participants were given two tasks, a Grammaticality Judgment Task (with
70 sentences and 14 types) and a Picture Selection Task. The findings showed no
significant difference between monolinguals and early bilinguals, since the latters
distinguished between the copulas and also showed sensitivity to the associated
semantic properties. However, late bilinguals showed a different pattern, they
seemed to distinguish between the two Spanish copulas, but did not appear to
have acquired their interpretive properties. The authors interpreted these results
as evidence that the lexical semantic properties may be learned by young bilin-
guals but appear to become a vulnerable construction for later bilinguals.
Finally, this issue was taken up by Gómez-Soler (2009), who tested 48 college
students on the L2 acquisition of ser and estar in the different structures in which
these copulas occur. All participants were native English speakers. Their knowl-
edge of the two Spanish copulas was tested through a guided Picture Description
Task and a Grammaticality Judgment Task. The results showed overgeneralization
of ser, omission of estar, and difficulty with semantic constraints and the location
of events. Gómez-Soler also observed a difference in the participants’ performance
in the two tasks. Participants seemed to perform better on the written task than
on the oral one. The author explains these findings along the lines of Prévost and
White’s (2000) Missing Surface Inflection Hypothesis and claims that the difficul-
ties with ser and estar derive from a persistent problem of English L1 learners in
identifying the aspectual properties of the Spanish copulas.
On the whole, L2 learners with English L1 seem to overgeneralize ser and have
problems incorporating pragmatic cues into the copula choice. The d ­ ifficulties
 Copula choice in adjectival constructions in Dutch L1 Spanish L2 

with the Spanish copulas seem to increase in specific grammatical contexts,

namely, adjectival phrases and locatives. L2 proficiency seems to positively cor-
relate with the use of estar, but gives unclear results for ser. As with regard to
the possible effects of the L1, Geeslin and Guijarro-Fuentes (2005) argued that,
on a whole, L2 acquisition of ser/estar is not influenced by previously learned
languages.

2.4  Ser/estar in Dutch L1 Spanish L2

To the best of our knowledge, Hanegreefs (2004) is the only publication that
addresses the aspectual properties of Dutch copulas. In a contrastive corpus
analysis of the different possible translations in Spanish of the Dutch copula
worden, ­Hanegreefs shows that the choice of the Spanish copula is related to a
different conceptualization of the process of change. Referring to Delbecque,
­Masschelein and Vanden Bulcke (1995), she argues that worden, when used as a
copula, is normally related to a concept of change.1 The difference between wor-
den on the one hand, and zijn on the other, is thus one involving aspect. Worden
“conveys a dynamic, future and processual perspective which distinguishes it
from the more prototypical copula zijn” (Hanegreefs 2004, p. 2–3). This analysis
is reminiscent of Schmitt’s (1992) account of the aspectual differences between
ser and estar. Dutch zijn, like Spanish ser is presented as a neutral copula, one
that has no aspectual properties; worden, like estar, conveys the meaning of
change. A  first comparison between Dutch and Spanish may suggest that zijn
(Dutch) corresponds to ser (Spanish), and worden (Dutch) corresponds to estar
(Spanish). However, the translation of Dutch worden into Spanish appears to
require a more fine grained analysis of the aspectual properties of the concept of
change. Hanegreefs (2004) reports that, depending on the conceptualization of
change, worden is translated in different ways in Spanish. So, when the internal
dynamics of the process of change is emphasized, Dutch worden is translated
into Spanish with a semi-copula, similar to become in English. When the result
and the effect of the change on the experiencer are emphasized, Spanish prefers
a lexical verb (i.e. a verb derived from an adjective) like, for instance enfermar
(to become ill). Finally, ser and estar are used when the context provides suf-
ficient cues for the interpretation of the aspectual properties of the adjectival
construction. However, whereas ser is used for categorization, the combination
of estar  + participle implies that a change of state has happened, whereby the
result is stressed, instead of the process.

.  Worden can also have the function of auxiliary in compound tenses and in passives.
 Manuela Pinto & Alexia Guerra Rivera

Given this more detailed distinction of the aspectual properties of Dutch

copulas, the distribution of ser/estar in Spanish and zijn/worden in Dutch may be
represented as in Table 1:

Table 1.  Aspectual properties of copulas in Dutch and in Spanish

Dutch Spanish

no aspectual properties ZIJN Ser

change of state > process WORDEN (present perfect) Estar
change of state > result Zijn Estar

2.5  Worden vs. estar

As Table 1 shows, both Dutch worden and Spanish estar are assumed to have
inherent aspectual properties: they are both marked for [state]. They both surface
as a copula containing this specific aspectual feature that has to match the seman-
tic and pragmatic features of the adjective. This kind of information is also avail-
able in the input and indeed L2ers do not have problems with ser and estar when
the adjective clearly requires just one of the two copulas. The problems for L2ers
seem to involve constructions in which adjectives may take either ser or estar,
depending on the required interpretation. This suggests that the problem may
be caused by difficulties in mapping meaning into form.2 Table 1 makes some
clear predictions: Dutch L1 learners will identify ser as the aspectually neutral
copula corresponding to Dutch zijn. However, they may have difficulties with
estar. Dutch copula worden conveys an interpretation of change of state whereby
the process is emphasized. Conversely, a change of state focusing on the result
must be constructed with copula zijn plus some lexical elements which anchor
the construction to a time frame. This is not the case with the Spanish copula
estar, which carries the inherent aspectual interpretation of the result of a change,
neglecting the process.
Summarizing, Dutch, like Spanish, has two copulas, zijn and worden, which
at first sight seem to correspond to ser and estar. When considering the inherent
aspectual properties of these elements, however, Dutch worden and Spanish estar
appear to differ in the modality of the change denoted by the predicate which has
taken place. This language pair thus provides the right ingredients to study the L2
accessibility of properties beneath word level.

.  In the sense of Lardiere (2000).

 Copula choice in adjectival constructions in Dutch L1 Spanish L2 

3.  Method

In this section we report of two experiments we ran with Dutch L1 Spanish L2

speakers with an advanced proficiency level. The first experiment, a Grammatical-
ity Judgment Task, was used in order to determine whether the availability of the
two copulas in the L1 may facilitate the acquisition of the two copulas in the L2.
The results showed that even at a high proficiency level copula + ADJ construc-
tions are problematic for Dutch L1 Spanish L2.
The same subjects completed then a Fill-in-the-Gap task in order to establish
which adjective type, might be responsible for the target-deviant results. The results
of this test and a subsequent item analysis revealed that our subjects only had prob-
lems with a specific subclass of adjectives: irreversible scalar gradable adjectives.
A total of 37 (N = 37) participants took part in this study: 14 Dutch L1 speak-
ers and 23 Spanish native speakers who served as a control group. The L2 speakers
completed a linguistic background questionnaire in order to determine demo-
graphic variables such as age of first exposure to the L2, occupation, parents’ L1,
proficiency in other languages, level of education, etc. The most relevant infor-
mation obtained from this questionnaire can be summarized as follows: the L2
participants were Dutch native speakers, majored in Spanish at a Dutch university,
and according to the Common European Framework of Reference for Languages,
their proficiency level corresponded to C1. At the time of testing, their age ranged
between 23 and 30. All of them had travelled to a Spanish-­speaking country on
holidays or to study and live there between 1 and 24 months. None of them had a
Spanish-speaking parent or were raised in a Spanish-Dutch bilingual environment.
All of them had native Dutch-speaking parents. With regard to the participants’
knowledge of other languages, 10 out of 14 students spoke four other languages
besides Dutch; namely, French, German, English and Spanish. Therefore, for all of
them, Spanish was their third or fourth foreign language. Nine of them had had
formal Spanish classes in high school or other language institutes before entering
the Spanish major. This information is presented in Table 2.
The control group was formed by 23 Spanish and Chilean young adults.
Eleven of them were originally from Spain and 12 from Chile. All participants
were contacted by email and performed the tasks on their own time, returning
their responses by email once completed.
In order to evaluate their proficiency level in Spanish, L2 participants were
asked to complete an independent proficiency test.3 The test was adapted from a

.  Following common practice of the sources we relied on, we did not take any independent
measure of the proficiency level of the control group. However, we agree with one of the
­reviewers that this may give a more reliable baseline measure.
 Manuela Pinto & Alexia Guerra Rivera

Table 2.  Linguistic background questionnaire of Dutch L1 Spanish L2 subjects

learning Spanish

abroad (Spain or
South America)
score (max. 50)
Participant ID

Other Foreign
Age in years

Age started
Proficiency

Time spent
languages
CH 21 42 English-German 16 2 months
JJ 24 45 English-German-French 18 6 months
TL 31 46 English-German-French 19 7 months
HB 36 46 English-German-French 23 1 months
DB 25 40 English-Catalan-French 16 9 months
MB 25 45 English-French-German 19 12 months
MO 26 44 English-French 18 14 months
AD 27 44 English-French 18 12 months
EM 24 47 English-German 16 12 months
FH 25 43 English-German-French 18 8 months
NG 26 38 English-German-French 22 8 months
DP 22 37 English 17 24 months
MD 21 38 English-German-French 18 18 months
AT 25 41 English-German-French 18 8 months

version of the Diploma de Español como Lengua Extranjera (DELE), and consisted
of a cloze passage with three multiple choice options for each blank, as well as a
multiple choice vocabulary section, adapted from a Modern Language Association
(MLA) placement test. Following previous research using the same methodology
(Montrul & Bowles 2009; Montrul & Slabakova 2003), scores between 40 and 50
points were considered as the baseline for advanced, scores between 30 and 39
points were considered as the baseline for intermediate proficiency, and scores
between 0 and 29 points were considered as low proficiency. The average score for
the L2 group was 42.5 points, SD 3.29; hence most of them had an advanced level
of proficiency.

4.  Task 1: Grammaticality Judgment Task

The Grammaticality Judgment Task was a pen-and-paper task. The task included
thirty six Spanish sentences. Twelve sentences contained ser and twelve c­ ontained
 Copula choice in adjectival constructions in Dutch L1 Spanish L2 

estar, equally distributed over 12 grammatical sentences and 12 ungrammati-

cal ones. Moreover, 6 grammatical and 6 ungrammatical filler sentences were
included and randomly distributed through the task. The participants were asked
to judge the grammaticality of each sentence on a Likert scale, indicating 1 when
totally unacceptable and 5 when totally acceptable. The test sentences were divided
into four different types. Each type of sentence appeared six times during the task.
The different sentences types used in this task are described below:
a. Ser + AP (grammatical):
El hijo de María es cruel.
‘María’s son is cruel.’
b. Ser + AP (ungrammatical):
*Los nuevos empleados están muy capaces.
‘The new workers are very capable.’
c. Estar +AP (grammatical):
El vaso está lleno de agua.
‘The glass is full of water.’
d. Estar + AP (ungrammatical):
*El entrenador de fútbol es satisfecho.
‘The football trainer is satisfied.’

The adjectives used in this task were of two types. (1) adjectives that unambigu-
ously combine with one specific copula. For example, satisfecho ‘satisfied’, lleno
‘full’ always select estar, capaz ‘handy’, cruel ‘cruel’ always select ser. (2) adjectives
that can be used with both copulas, like rubio ‘blond’, flaco ‘thin’. In this case, how-
ever, we manipulated the context in such a way that only one of the two possible
interpretations was possible.

4.1  Results of the GJT

A Multivariate Analysis of Variance (MANOVA) statistical test conducted on the
average number of responses per condition shows that a significant difference in
the performance of the two groups is only present in one of the tested conditions,
namely, ‘estar ungrammatical’. The L1 group performed similarly to the L2 group
in the three other conditions and only significantly different in this specific condi-
tion. In addition, the L2 group had more troubles in judging ungrammatical test
sentences than grammatical ones; see Figure 1 below.
As can be seen in Figure 1, the main difference between the L1 and the L2
group occurs in the “estar ungrammatical” condition. This shows that Dutch L1
Spanish L2 speakers fail to recognize ungrammatical constructions with estar as
ungrammatical. The results are consistent with previous studies that capitalize on
interpretative differences between ser and estar.
 Manuela Pinto & Alexia Guerra Rivera

5,00

4,00
Mean score

3,00 Groups
L1
L2

2,00

1,00

0,00
Estar – GRAM Estar–UNGRAM Ser–GRAM Ser–UNGRAM
Grammaticality judgment task

Figure 1.  Mean scores of correct responses

4.2  Discussion of the GJT

The picture that emerges from this study confirms that integration of morpho-
syntactic information with semantics and pragmatics may be one of the major
factors affecting the acquisition of Spanish copulas, in line with Sorace’s Interface
Hypothesis. Whereas L2 speakers show target-like performance with copula selec-
tion which purely relies on the morpho-syntax (in NP and VP constructions), they
have difficulties when information from other cognitive components is involved.
However, beyond the predictive value of this hypothesis, the real challenge con-
sists of determining the specific factors that make certain constructions particu-
larly difficult to acquire by speakers with a different L1. To say it with Rothman
and Slabakova (2011), the real puzzle is understanding why certain grammatical
constructions are more difficult than others.

5.  Task 2: Fill-in-the-Gap Task

The next step is an analysis of the inherent semantics of Dutch copulas compared
to ser and estar. Two hypotheses are being tested: (1) Inherent aspectual proper-
ties of copulas may not be visible in the input that L2ers receive. If this is ­correct,
 Copula choice in adjectival constructions in Dutch L1 Spanish L2 

Dutch L1 Spanish L2ers may fail to see that worden differs from estar in just one
aspectual property. This would predict a superficial transfer of Dutch worden
for all cases of Spanish estar. The other hypothesis says instead that (2) inher-
ent aspectual properties of copulas may be available to L2ers. In this case Dutch
L1 Spanish L2ers would make a distinction between adjectival constructions that
focus on the process of change and adjectival constructions that focus instead on
the result of a change.

5.1  F
 ill-in-the-Gap Task
In order to test these predictions, Dutch L1 Spanish L2 learners completed a
Fill-in-the-Gap Task, in which two types of adjectives were used: scalar gradable
adjectives, denoting a property which is the result of a gradual change, like for
example old, long, fat; and non-scalar gradable adjectives, that remain agnostic
about the modality of the change and just denote the end result, like elegant,
romantic or naughty (Kennedy 1999; Winter 2005). If, according to Hypothesis
1, Dutch L1-Spanish L2 speakers do not have access to the inherent semantics of
copulas, we predict that they will show non-target performance with adjectives
that do not involve a process, as they may use ser, on analogy with the require-
ments of their L1 in which non-scalar gradable adjectives are combined with
copula zijn (and scalar gradable adjectives with copula worden). Conversely,
Hypothesis 2 assumes that Dutch L1 Spanish L2ers do have access to copulas’
internal properties. In this case the prediction is that worden will correctly be
used with scalar gradable adjectives, and ser with non-scalar gradable adjectives
counterpart.
The Fill-in-the-Gap Task consisted of 17 sentences, 7 were fillers and 10 were
test sentences, 5 called for estar and 5 called for ser. The test sentences were divided
into scalar gradable and non-scalar gradable adjectives, and for each there was a
construction that required ser and one that required estar. Each type appeared 5
times during the task. The adjectives used are listed in Table 3.

Table 3.  scalar gradable vs. non-scalar gradable adjectives

Scalar gradable Non-scalar gradable

Flaco (thin) Romántico (romantic)

Largo (long) Elegante (elegant)
Grande (tall) Pelirroja (red haired)
Viejo (old) Revoltoso (naughty)
Gordo (fat) Simpático (nice)
 Manuela Pinto & Alexia Guerra Rivera

Each sentence was introduced by a short story that provided contextual infor-
mation eliminating ambiguity:

El abuelo Ángel siempre ha sido delgado y eso no ha cambiado incluso cuándo

(8) 
se ha casado con Maria, quien es una gran cocinera.
‘Grandpa Ángel has always been thin and he hasn′t changed, not even when
he married Maria, who is a great cook.’

5.2  Results of the Fill-in-the-Gap Task

An independent t-test revealed no significant difference in the performance of the
two groups: t (34) = –0,886, p = .382. The L1 group performed slightly better than
the L2ers:

1,00

,80

,60
Mean score

Groups
L1
L2

,40

,20

,00
Ser Sc. Grad Ser Non Sc Grad Estar Sc. Grad Estar Non Sc. Grad
Fill in the gap task

Figure 2.  L1 and L2 mean-scores in terms of correct vs. incorrect answers

Figure 2 shows that L1 and L2ers performed similarly with the ser condition,
but differently with the estar condition. The difference between the two groups is
 Copula choice in adjectival constructions in Dutch L1 Spanish L2 

significant, these results are consistent with previous findings: also in this task the
L2ers show more trouble with estar than with ser.4
When looking at different adjectives, a t-test revealed a significant difference
between scalar gradable and non-scalar gradable ones: t (12) = 2,843, p = .015.
For the L2ers, the scalar gradable adjectives seem to be harder to acquire than the
­non-scalar gradable ones.

5.3  Discussion of Fill-in-the-Gap Task

The results of the Fill-in-the-Gap Task sheds light into the internal grammar that
Dutch L1 speakers build of their L2. These Spanish L2ers correctly choose estar
for change-of-state constructions with non-scalar gradable adjectives. Note that
despite the change-of-state interpretation, in Dutch these constructions require
zijn. These data suggest that Dutch L1-Spanish L2 do have access to the inherent
semantics of the L2 copulas, supporting Hypothesis (1). However, if this was the
case, participants should perform correctly also with scalar gradable adjectives.
This prediction was not confirmed, as most of the errors were found in construc-
tions with scalar gradable adjectives. Spanish L2ers seem to fail to see that scalar
gradable adjectives require estar.
In order to shed more light on this unexpected result, the test sentences of the
Fill-in-the-Gap Task were submitted to an item analysis. This revealed that scalar
gradable adjectives may be distinguished into two further classes: (1) reversible scalar
gradables, i.e. adjectives that denote a gradual change which can go in both directions
(get thin and then get fat again); and (2) irreversible scalar gradables, i.e. adjectives that
denote a gradual change that brings to an irreversible end result, like for instance old
(once you get old you cannot become young again). It is with this last category that
Dutch L1 encounter problems with the choice of the Spanish copula (See Table 4).
Thus it seems that although the participants chose the wrong copula, this was
not a random choice, but was based on the evaluation of the inherent semantics of
the adjective and of the copula.

6.  General discussion

The analysis of the inherent semantic properties of estar in comparison with

the properties of Dutch worden showed that although both copulas denote a

.  Figure 2 also shows that the L2 speakers are better than the L1 in the ‘ser + scalar gradable
condition’. An item analysis revealed that this result was caused by a Chilean/Spanish differ-
ence in the use of the adjective grande.
 Manuela Pinto & Alexia Guerra Rivera

Table 4.  Choice of the Copula in Dutch and in Spanish According

to the Adjectival Class
ADJ Dutch Spanish

Non-scalar gradable ZIJN ESTAR

Reversible scalar gradable WORDEN ESTAR
Irreversible scalar gradable WORDEN ESTAR

c­hange-of-state, they crucially differ in the type of information they provide

about the modality of this change. Worden focuses on the process subsumed by
the change, whereas estar emphasizes its end result. A Fill-in-the-Gap Task tested
whether this information is accessible to L2ers. The results showed that Dutch L1
Spanish L2ers make the correct copula choice with non-scalar gradable adjectives,
suggesting that they have access to the inherent semantic features that distinguish
estar from worden. However, the participants’ performance with scalar gradable
adjectives was unexpectedly target-deviant. A subsequent item analysis showed
that the difficulties with these adjectives are determined by a specific subclass of
scalar gradable adjectives, those we called irreversible, i.e. scalar gradable adjectives
denoting a property that cannot be reverted.
The data, particularly those summarized in Table 4, confirm that the prob-
lems with the use of estar are not of a superficial nature. Dutch L1-Spanish L2ers
correctly use estar with non-scalar gradable adjectives despite the fact that their
L1 with these adjectives prefers zijn. So Spanish L2ers do not establish a trivial
correspondence between worden and estar. The choice of the correct copula is
determined by an analysis of certain semantic properties of the construction
involved. Dutch L1-Spanish L2ers seem to know that estar is employed to express
some change-of-state of the predicate. This evidence suggests that the failure to
choose the right copula for scalar gradable irreversible adjectives is the result of
an analysis of the aspectual properties involved. Dutch L1-Spanish L2ers evalu-
ate the semantics of a construction with a scalar gradable irreversible adjective
and come to the conclusion that in this specific case estar is not adequate.
The explanation for this unexpected result is still speculative. The property
denoted by an adjective like old in Juan está vejo, being irreversible, may in the
end be interpreted as a sort of permanent property. L2ers may have difficulties to
analyze this construction as the result of a change-of-state. The choice of ser could
then be seen as an alternative default solution for an ambiguous case. Copula ser,
being aspectually neutral, may be a more accessible option to L2ers.
 Copula choice in adjectival constructions in Dutch L1 Spanish L2 

7.  C
 onclusions

This paper presented a study on the acquisition of the Spanish copulas in adjectival
constructions by Dutch L1-Spanish L2ers at an advanced level. The first experi-
ment showed that the difficulties with Spanish copulas are not across the board,
but can be localized in ambiguous constructions and seem to involve mainly
copula estar. In the second experiment we explored the inherent semantics of
copula constructions in both languages, in the attempt to determine what kind of
­information L2ers access about L2 grammar. This preliminary study shows that
L2ers’ choice between ser/estar in adjectival constructions is not trivial. The target
forms used by the participants but, more importantly, their non-target forms indi-
cate that copula choice is the result of a detailed analysis of the inherent semantic
properties of both adjectives and copulas.
Although the present analysis requires a more robust empirical basis, this
study shows that the L2ers’ problems with Spanish copulas can be narrowed down
to a mismatch of very subtle semantic properties at the lexical level (or rather
beneath it).

Acknowledgements

Thanks to the editors, to two anonymous reviewers and particularly to Francesco-

Alessio Ursini for the many detailed comments to previous versions of this paper.
Thanks to Raquel Fernández Fuertes for convincing her students to become our
control group, and to Mattis van den Bergh for helping us with statistics.

References

Brucart, J.M. (2012). Copular alternation in Spanish and Catalan attributive sentences. Lin-
guística: Revista de Estudos Linguísticos da Universidade do Porto, 7, 9–43.
Bruhn de Garavito, J., & Valenzuela, E. (2006). The status of ser and estar in late and early
bilingual L2 Spanish. In C.L. Klee & T. Face (Eds.), Selected Proceedings of the 8th Hispanic
Linguistic Symposium and 7th Conference on the Acquisition of Spanish and Portuguese as
First and Second Languages (pp. 100–109). Somerville, MA: Cascadilla Press.
Camacho, J. (2013). Ser and estar: Individual/stage level predicates or aspect? In J. Ignacio
Hualde & A. Olarrea (Eds.), The Blackwell Handbook of Hispanic Linguistics (pp. 100–120).
Hoboken, NJ: Wiley-Blackwell.
Delbecque, N., Masschelein, D., & Vanden Bulcke, P. (1995). Gramática española aplicada. El uso
de ser y estar. Leuven: Wolters.
 Manuela Pinto & Alexia Guerra Rivera

Diesing, M. (1990). The Syntactic Roots of Semantic Partition. Unpublished Ph.D. dissertation.
University of Massachusetts, Boston.
Gallego, A., & Uriagereka, J. (2009). Estar=ser+ P. Paper presented at the 19th Colloquium on
Generative Grammar, Vitoria. Available at: 〈http://filcat.uab.cat/clt/membres/professors/
gallego/pdf/GALUR_Vitoria2.0.pdf〉 (14 May 14 2014).
Geeslin, K. (2001). Changing norms moving targets and the SLA of copula choice. Spanish
Applied Linguistics, 5(1–2), 29–55.
Geeslin, K. (2002). The second language acquisition of copula choice and its relationship to
language change. Studies in Second Language Acquisition, 24, 419–451.
DOI: 10.1017/s0272263102003030
Geeslin, K. (2003a). A comparison of copula choice in advanced and native Spanish. Language
Learning, 53, 703–764. DOI: 10.1046/j.1467-9922.2003.00240.x
Geeslin, K. (2003b). The role of adjectival features in the second language acquisition of copula
choice. In P. Kempchinsky & C. Piñeros (Eds.), Theory, Practice and Acquisition: Papers
from the 6th Hispanic Linguistics Symposium and the 5th Conference on the Acquisition of
Spanish and Portuguese (pp. 332–351). Somerville, MA: Cascadilla Proceedings Project.
Geeslin, K., & Guijarro-Fuentes, P. (2005). The acquisition of copula choice in instructed
­Spanish: The role of individual characteristics. In D. Eddington (Ed.), Selected Proceedings
of the 6th Conference on the Acquisition of Spanish and Portuguese as First and Second Lan-
guages (pp. 66–77). Somerville, MA: Cascadilla Proceedings Project.
Geeslin, K., & Guijarro-Fuentes, P. (2006). The second language acquisition of variable struc-
tures in Spanish by Portuguese speakers. Language Learning, 56(1), 53–107.
DOI: 10.1111/j.0023-8333.2006.00342.x
Gómez-Soler, I. (2009). The Morphosyntax-Lexicon Interface Breakdown: An Aspectual Account
of the L2 Acquisition of Ser and Estar by L1 English Speakers. Unpublished Master’s thesis.
University of North Carolina, Chapel Hill.
Gunterman, G. (1992). An analysis of interlanguage development overtime, Part II: Ser and
estar. Hispania, 75, 1294–1303. DOI: 10.2307/344396
Hanegreefs, H. (2004). The Dutch change-of-state copula worden and its Spanish counterparts.
A matter of aspect and voice. Annual Review of Cognitive Linguistics, 2, 1–29.
DOI: 10.1075/arcl.2.01han
Kennedy, C. (1999). Projecting the Adjective: The Syntax and Semantics of Gradability and Com-
parison. New York, NY: Routledge. DOI: 10.4324/9780203055458
Kennedy, C. (2007). Vagueness and grammar: The semantics of relative and absolute gradable
adjectives. Linguistics and Philosophy, 30(1), 1–45. DOI: 10.1007/s10988-006-9008-0
Mesa Alonso, M., Domínguez Herrera, M., Padrón Sánchez, E., & Morales Aguilera, N. (1993).
Ser y estar: Consideraciones sobre su uso en español. Islas, 104, 150–156.
Montrul, S., & Bowles, M. (2009). Back to basics: Incomplete knowledge of differential object
marking in Spanish heritage speakers.  Bilingualism: Language and Cognition,  12(3),
363–383. DOI: 10.1017/s1366728909990071
Montrul, S., & Slabakova, R. (2003). Competence similarities between native and near-native
speakers. Studies in Second Language Acquisition, 25(3), 351–398.
DOI: 10.1017/s0272263103000159
Prévost, P., & White, L. (2000). Missing surface inflection or impairment in second language
acquisition? Evidence from tense and agreement. Second Language Research, 16(2),
103–133. DOI: 10.1191/026765800677556046
 Copula choice in adjectival constructions in Dutch L1 Spanish L2 

Rothman, J., & Slabakova, R. (2011). The mind-context divide: On acquisition at the linguistic
interfaces. Lingua, 121(4), 568–576. DOI: 10.1016/j.lingua.2011.01.003
Ryan, J.M., & Lafford, B.A. (1992). Acquisition of lexical meaning in a natural environment:
“Ser” and “estar” and the Granada Experience. Hispania, 75(3), 714–722.
DOI: 10.2307/344152
Schmitt, C. (1992). Ser and estar: A matter of aspect. Proceedings of the North East Linguistic
Society, 22, 411–425.
Schmitt, C. (2005). Semi-copulas. In P. Kempchinsky & R. Slabakova (Eds.), Aspectual Inquiries
(pp. 121–145). Dordrecht: Kluwer. DOI: 10.1007/1-4020-3033-9_6
Schmitt, C., & Miller, K. (2007). Making discourse-dependent decisions: The case of the copulas
ser and estar in Spanish. Lingua, 117, 1907–1929.
DOI: 10.1016/j.lingua.2006.11.007
Sorace, A. (2011). Pinning down the concept of interface in bilingualism. Linguistic Approaches
to Bilingualism, 1, 1–33. DOI: 10.1075/lab.1.1.01sor
Uriagereka, J. (2001). Adjectival clues, keynote speech at Acquisition of Spanish & Portugues/­
Hispanic Linguistics Symposium, University of Illinois at Urbana-Champaign (USA),
­October 11–14.
Ursini, F.A. (2011). On the syntax and semantics of ser and estar. Macquarie University Research
Online.
VanPatten, B. (1985). The acquisition of ser and estar in adult second language learners: A pre-
liminary investigation of transitional stages of competence. Hispania, 68, 399–406.
DOI: 10.2307/342218
VanPatten, B. (1987). The acquisition of ser and estar: Accounting for developmental patterns.
In B. VanPatten, T.R. Dvorak, & J.F. Lee (Eds.), Foreign Language Learning: A Research
Perspective (pp. 61–75). New York, NY: Newbury House.
Winter, Y. (2005). Cross-categorial restrictions on measure phrase modification. Linguistics and
Philosophy, 28(2), 233–267. DOI: 10.1007/s10988-004-1469-4
Zagona, K. (2012). Ser and estar: Phrase structure and aspect. In C. Nishida & C. Russi
(Eds.), Building a Bridge between Linguistic Communities of the Old and the New World.
Current Research in Tense, Aspect, Mood and Modality (pp. 303–327). Amsterdam: Rodopi.
 Typological proximity in L2 acquisition
The Spanish non-native grammar of French speakers

Juana M. Liceras & Anahí Alba de la Fuente

University of Ottawa / Université de Montréal

Typologically-close languages such as French and Spanish share many

typological universals and macroparametric options but display different
microparametric options as well as obvious and subtle morphological,
syntactic and processing differences. This counterposed situation that
we conceptualize as typological similarity versus typological proximity
constitutes our first tool to investigate the specific characteristics of the
Spanish interlanguage of L1 French (L2Sp-L1Fr) speakers. The other tool is
the Competing Grammars Hypothesis (CGH) which we use as a framework
to determine the optionality that results from the simultaneous availability
of target-like, transferred and idiosyncratic L2Sp-L1Fr parametric options or
feature combinations in the L2Sp-L1Fr interlanguage.

1.  Introduction

For more than three decades now, typological proximity understood as member-
ship to the same language family – or lack thereof – has played a central role in the
study of second language learning. In fact, we can trace what we call ‘the linguis-
tic approach’ to second language acquisition (L2A) to the so-called Contrastive
­Analysis Hypothesis (CAH), which was in perfect harmony with typology.1 How-
ever, the concept of typological proximity is rather elusive because close languages
may display striking differences. A case in point is the one we discuss in this chap-
ter: the acquisition of Spanish by L1 French speakers.
While French and Spanish can be taken as an obvious example of typological
proximity because they belong to the Romance family, it is nonetheless obvious
that these two languages display clear-cut differences. In fact, belonging to the

.  See Ellis (1994), Liceras (1986) or Larsen-Freeman & Long (1991), among many others.

doi 10.1075/ihll.3.13lic
© 2015 John Benjamins Publishing Company
 Juana M. Liceras & Anahí Alba de la Fuente

same family may ensure macroparametric proximity (being a synthetic versus an

isolating language or a polysynthetic versus a fusional or agglutinative language)
but it does not ensure microparametric proximity. This explains why when inves-
tigating the acquisition of Spanish by French speakers, we find, on the one hand,
systematic references of how this typological proximity shapes the L2 Spanish – L1
French (L2Sp-L1Fr) interlanguage (IL) and, on the other hand, detailed accounts
of how obvious and subtle differences between the two languages cause interfer-
ence and are even candidates for fossilization (Han 2013; Selinker 1972; Selinker &
Lamendella 1978). For Selinker (1972), fossilization, a mechanism of the Latent
Psychological Structure that is activated in L2 acquisition, constitutes a core com-
ponent of the Interlanguage Hypothesis (ILH).2 According to this hypothesis, L2
acquisition is different from L1 language acquisition because in L2 acquisition it is
the Latent Psychological Structure rather than the Latent Linguistic Structure (Uni-
versal Grammar) that is activated.
One of the advantages of the generative agenda when dealing with typology
is that it provides a universal framework for the comparison of languages – the
principles – and a mechanism that specifies how they differ – the parameters. The
many tools implemented to analyse natural languages within the Principles and
Parameters model (Chomsky 1981, 1986; Chomsky & Lasnik 1993) and the sub-
sequent model, the Minimalist Program (Chomsky 1995 and subsequent work),
provide acquisitionists with a wealth of resources to investigate the acquisition
of any pair of languages. Specifically, we can benefit from the macroparametric
and the microparametric classification of languages as well as from the feature
accounts that configure these classifications.
The first issue that we address is how to define the term ‘typology’ when
investigating the acquisition of an L2 by speakers of an L1 that belongs to the
same family.3 We take this issue as the motto of this paper because the field of L2

.  To date, fossilization has been related by many L2 researchers to the presence of L2 con-
structions that show permanent resistance to incorporate L2 input (change L1 structures or
idiosyncratic IL structures) and thus fail to progress towards the target (Han 2013). Some
researchers (Liceras 1986), taking a similar approach to the one portrayed by the Competing
Grammars Hypothesis (Kroch 1994, 2001), defined fossilization as the permanent co-­existence
of the two options of a given parameter.
.  The issue of typological proximity plays a central role in L3 acquisition. The Typological
Primacy Model (Rothman 2011, 2013, in press) states that at the initial stages of multilingual
language acquisition transfer is determined by the structural similarity between the L3 and
the previously acquired linguistic systems. For Rothman (2013), structural similarity refers to
linguistic properties that overlap cross-linguistically at the level of mental representation. In
other words, in a proposal that echoes Kellerman’s (1979) psychological proximity, what the
 Typological proximity in L2 acquisition 

acquisition cannot but deal, whether directly or indirectly, with the “two jarringly
counterposed problems of today’s linguistic research: why human languages are
as different from each other as they are, and why they are as similar to each other
as they are” (Newmeyer 1990, p. 203). While, in principle, the first problem would
be mainly taken up within the fields of sociolinguistics or anthropological lin-
guistics or the subfields that have traditionally focused on linguistic diversity,
the second problem has been addressed by the scholars who have dealt with lan-
guage universals. On the one hand, we find what has specifically been referred
to as ‘typological’ universals (i.e. Greenberg’s (1966) implicational universals or
Keenan & Comrie’s (1977) Accessibility Hierarchy). On the other hand, we have
what are considered ‘formal’ universals (i.e. the ones defined within Chomsky’s
(1981) Government and Binding tradition such as Chomksy & Lasnik’s (1977)
negative filters).
Returning to the languages under investigation in this chapter, we make use
of the terms typological similarity and typological proximity to refer to French
and Spanish because they belong to the same family. These two languages are
typologically similar with respect to many of Greenberg’s (1963) universals (i.e.
left-headedness: VO, N-Adj., Det-N) and also with respect to Baker’s (1996)
macroparameters.4 French and Spanish are also equal in terms of the Acces-
sibility Hierarchy of relativization in that both can relativize subjects, direct
objects, indirect objects, possessives and objects of preposition but not objects
of comparison. However, when it comes to formal universals, both Spanish and
French respect the empty complementizer filter but are different with respect

Typological Primacy Model states is that the lexical or grammatical level of underlying or true
grammatical similarity is assessed and determined subconsciously by the linguistic parser
very early in the L3 acquisition process.
.  Baker (1996) maintains that the diversity found in human language from a morpho-­
syntactic perspective falls within the boundaries of three types of languages and the potential
combination of the various categories, as depicted in Table 1.

Table 1.  Three-way typology underlying human language from a morphosyntactic

point of view (Baker, 1996, p. 5).

I II III

Morphological type Isolating Dependent marking Head marking

Word order type Head initial Head final Free
Exemplar English Japanese Mohawk
 Juana M. Liceras & Anahí Alba de la Fuente

to the that-t filter.5 The two languages are also different in that they have been
said to realize the opposing options of the null-subject parameter: Spanish is a
[+null subject] language like Italian, while French is a [–null subject] language
like English. This is not normally expected to be the case for two languages that
belong to the same family and, in fact, the status of French as a [–null subject]
language has been questioned by some syntacticians, as we will briefly discuss
in Section 2.2.
While ‘typological’ as such may not be conceptualized as an unambiguous
term, we think that it captures the characterization that we want to make about
pairs of languages such as Spanish and French versus English or Arabic. However,
the term ‘typological proximity’ seems to come short when trying to account for
the effects that arise when we deal with subtle structural differences that we find in
typologically close languages. For this reason, we feel that the notion of ‘typologi-
cal similarity’ must also be introduced in these contexts. That being said, before
we proceed with our analysis, it is important to note that typological similarity
and typological proximity cannot be defined as univocal concepts for at least four
reasons: (1) there is a continuum when it comes to language families (English
or Russian are typologically more distant from Spanish than French but less so
than they are from Arabic or Japanese); (2) typologically-close languages such
as ­Spanish and French may differ in terms of a number of formal universals or

.  According to Keenan and Comrie’s (1977) Accessibility Hierarchy, if a language can
relativize a subject position, there is no guarantee that it will relativize a genitive position.
However, if it can relativize the latter, it will relativize all the positions which are above in the
hierarchy, which they characterize as shown in (1).
(1) SU(bject) > D(irect)O(bject) > I(ndirect) O(bject) > O(bject) of P(reposition)
> G(enitive) > O(bject) of C(omparison).

Chomsky and Lasnik’s (1977) *[COMP Ø] filter states that the complementizer cannot be
phonetically null, which is the case in both French and Spanish as shown in (2).
(2) a. C’est le livre *Ø/que je viens d’acheter
b. Es el libro _*Ø/que_(yo) acabo de comprar
‘It is the book (that) I’ve just bought.’

However, French and Spanish are different with respect to the so-called *[that-t] filter since
French, but not Spanish, requires the presence of a subject marker in a subject restrictive relative
clause such as (3a) versus (3b).
(3) a. C’est le facteur *que/qui parle l’arabe
b. Es el cartero que/*quien habla árabe
‘It is the mailman that/who speaks Arabic.’
 Typological proximity in L2 acquisition 

occupy a different place in terms of Accessibility Hierarchies;6 (3) typologically

distant languages such as Spanish and Arabic may share typological or formal uni-
versals (i.e. have post-nominal adjectives or allow null subjects) but we maintain
that this does not make them typologically more similar to Spanish than English,
a language that does not share these two properties with Spanish; (4) the distinc-
tion between typological similarity and typological proximity may depend on the
linguistic analysis that we adopt, as is the case with the status of French as a [–null
subject] language that we will discuss in 2.2.
Based on the above, we state that typological proximity pertains to Spanish
and French more so than to Spanish and English, but we will differentiate between
typological similarity and typological proximity between Spanish and French as
follows: there is typological similarity when a typological or a formal universal is
equally realized in these two typologically-close languages; otherwise, we will talk
about typological proximity. Thus, there is typological proximity in that both share
the same option of macroparameters but there is typological similarity in that both
are equal with respect to Keenan and Comrie’s (1977) Accessibility Hierarchy of
relativization or with respect to Chomsky and Lasnik’s (1977) *[COMP Ø] filter.
In this chapter, we approach the characterization of the L2Sp-L1Fr IL in
terms of whether and how its idiosyncrasy is related to the fact that typologically-
close languages may display obvious morphosyntactic differences in the way they
realize formal universals.7 Specifically, we aim to: (1) identify and account for

.  Both Spanish and French have plain passives as in (1) and (2) and reflexive passives as
in (3) and (4) but only Spanish has impersonal passives as in (5). The difference between the
Spanish example and the French example in (5) is that French cannot use a se-morpheme
there but on which, unlike Spanish se, functions as the subject of the sentence. English only
has plain passives.
(1) Esta catedral fue construída en la Edad Media
(2) Cette cathédrale a eté construite dans la Moyen Age
‘This cathedral was built in the Middle Ages.’
(3) Esta catedral se construyó en la Edad Media
(4) Cette cathedral s’a construit dans la Moyen Age
‘This cathedral was built in the Middle Ages.’
(5) Aquí se respeta a los investigadores
Ici on respect les chercheurs
‘Here one respects researchers.’/‘Here researchers are respected.’
.  We use the terms interlanguage, non-native grammar and L2 grammar interchangeably in
spite of the fact that the term ‘interlanguage’, as coined by Selinker, was meant to imply that a
non-native or L2 grammar (it could also be L3 or Ln) is fundamentally different from a native
grammar. We use the term without linking it to the ‘interlanguage hypothesis’ or the ‘fun-
damental difference hypothesis’ (Bley-Vroman 1990; Slabakova 2009, among many others),
namely, as being synonymous with the terms non-native or L2 grammar.
 Juana M. Liceras & Anahí Alba de la Fuente

L2Sp-L1Fr structures that neither occur in the L1 French nor in the L2 Spanish
grammar (#demasiado niños versus demasiados niños or trop d’enfants, ‘too many
children’); and (2) identify and account for L2Sp-L1Fr structures that are not pos-
sible in Spanish but surface as an effect of negative transfer or interference from
the L1 (*hombres ranas as in hommes grenouilles versus hombres rana, ‘frogmen,
divers’).
When we approach the simultaneous availability of target-like, transferred
and idiosyncratic L2Sp-L1Fr parametric options or feature combinations, we will
attempt to test the Competing Grammars Hypothesis (CGH), as defined by Kroch
(1994) or Yang (2002), among others. According to this hypothesis, which was
originally formulated to account for optionality or variability in diachronic change,
a speaker has access to the two options of any given parameter. The options com-
pete during a period of time until one wins over the other, which is what normally
happens in diachronic change. Similarly, an L2 speaker will also have access to the
parametric options realized in the L1 and the L2 (Zobl & Liceras 2006). In the case
of L2 acquisition, unlike what happens in diachronic change, none of the options
may end up wining. If this is the case, the two options (or grammars) may compete
permanently, something that for L2 acquisition would qualify as an instance of
fossilization (Liceras 1986).
In what follows we discuss whether and how typology has played a role in the
analysis and description of L2Sp-L1Fr IL. We first discuss two constructions that
allow us to differentiate typological proximity from typological similarity. We
then outline how typological proximity interacts with a selection of morphosyn-
tactic phenomena in L2 acquisition. We conclude by taking a step beyond mor-
phosyntax by discussing typological proximity in relation to interface conditions
and to processing preferences in anaphora resolution by native and non-native
(L1 French) Spanish speakers.

2.  Th
 e Spanish grammar of L1 French speakers: Typological proximity
versus typological similarity

It has systematically been shown that, while non-native grammars are shaped by
the L1 and the L2 (Selinker 1972, among others), they also display what can be
labelled idiosyncratic constructions that they share with natural languages other
than the L1 or the L2 (Adjémian 1976; Adjémian & Liceras 1984; Liceras 1986;
Schwartz & Sprouse 1994, among others). It has also been shown that there is
competition between features, feature combinations, or parametric options from
these different systems. Competition, however, has not always been identified as
 Typological proximity in L2 acquisition 

such but rather as variability or optionality.8 With this in mind, we take as our
point of departure the assumption that optionality will be an intrinsic character-
istic of IL systems. Furthermore, we argue that the variability that characterizes
the L2Sp-L1Fr IL is, in fact, a reflection of the optionality that may stem from
competition between the L1 and the L2 but also between features of other natural
languages, available to the speaker through access to UG.9
We have already mentioned that typological proximity does not equate to
typological similarity because typologically-close languages may reveal striking
differences that may result in transfer and optionality. To illustrate these issues we
discuss two familiar characteristics of Spanish and French: clitic pronouns and
null and overt subjects.

2.1  Object clitics in L2Sp-L1Fr

From a general point of view, the two salient differences between French and
­Spanish object clitics – apart from the actual lexical differences – relate to word
order in the case of infinitival complements and to the position of direct and indi-
rect object clitics in clitic clusters.
The order of direct object/indirect object clitic clusters in French and ­Spanish
is the opposite. Namely, in Spanish third person clusters, the indirect object
(dative) always precedes the direct object (accusative), as shown in (1), while the
opposite is the case in French, as shown in (2):
(1) ¿Los libros? Se[DAT] los[ACC] enviaré[1SG] mañana
   The books? to him/her them [I] will send tomorrow
(2) Les livres? Je les[ACC] lui[DAT] enverrai demain
The books? I them to him/her will send tomorrow
‘The books? I will send them to him/her tomorrow.’

.  While in the L2A field optionality and variability may be used and have been used
i­ ndistinctly (Pérez-Leroux & Liceras 2002), within the generative grammar approach, the
term optionality has taken preference because variation has been defined at the level of com-
petence. The Competing Grammars Hypothesis is meant to address optionality resulting from
the speakers’ access to two different rules or parametric options (two competing grammars).
In our characterization of the L2Sp-L1Fr IL, we will try to account for relevant instances of
optionality resulting from the speakers’ access to more than one grammar and will use the
terms optionality and variability indistinctly.
.  Tarone (1982) uses the term vernacular to refer to IL structures that are neither L1 nor
L2-like and Goodluck (1986) discusses the fact that any child L1 as well as L2 structures
should exist in a given natural language, namely, that there are no “wild” grammars.
 Juana M. Liceras & Anahí Alba de la Fuente

It has also been shown that the two languages differ, as we will illustrate in the
forthcoming sections, in terms of the restrictions that apply to clitic cluster combi-
nations (Alba de la Fuente 2010, 2012; Bonet 1991, 1994, 2008; Rivero 2004, 2008;
among others), as well as in terms of the availability of clitic-doubling in Spanish
but not in French (Fontana 1994, 1997; Liceras 1985; Perales & Liceras 2010; Rivero
1986, 1997; Zobl & Liceras 2006; among others).
In relation to uninflected verbs, in Spanish, clitics can appear with a large
number of predicates in canonical position (after the verb which appears in the
infinitival form), as in (3a), or undergo clitic climbing and appear before the verbal
form that carries inflection, as in (4a). With this type of construction, as illustrated
in (5a), Spanish clitics can never occur before the infinitival, which happens to be
the obligatory position in French, as attested by the grammaticality of (5b) versus
the ungrammaticality of (3b) or (4b):
(3) a. Yo quiero[1SG] verla
b. *Je veux voir-la
  I want to see-her
‘I want to see her.’
(4) a. Yo la quiero ver
b. *Je la veux voir
  I her want to see
‘I want to see her.’
(5) a. *Yo quiero la ver
b. Je veux la voir
I want her to see
‘I want to see her.’

A superficial approach to transfer would predict that L1 French speakers would

place Spanish clitic pronouns before the infinitival form. Thus, in spite of typo-
logical proximity, this difference could shape L2Sp-L1Fr in a way that would not
be expected in the case of other Spanish ILs (e.g. L1 English IL). In turn, if learn-
ers have access to UG and therefore to constructions available in any natural lan-
guage, L2Sp-L1Fr could display idiosyncratic clitic structures. In other words,
it would display clitics that are different from those of French and Spanish but
were possible in Old Spanish, and are also possible in languages such as Czech,
as in (7) below. However, why would these IL speakers produce or accept those
constructions?
One possible answer to that question could come from linguistic theory, as
in Liceras (1985), where she investigated whether L2 learners of Spanish might
locate clitics in the lexicon (as affix-like elements) or post-lexically (as words in
the syntax) rather than giving them a unidimensional value. In other words, there
 Typological proximity in L2 acquisition 

could be competition between the location of a given process in the different com-
ponents of the grammar. Thus, assuming that L2 learners of Spanish might locate
clitics in the lexicon or post-lexically would imply that IL clitics might share the
properties assigned to Old Spanish clitic pronouns or to Modern Spanish clitic
pronouns, as in Table 1.

Table 1.  Old versus modern Spanish clitics

Old Spanish (XP) Modern Spanish (Xº)

-Interpolation -No interpolation

-Precede or follow the inflected verb -Only precede inflected verbs
-No clitic doubling -Clitic doubling

These properties of Old Spanish are exemplified in (6)–(8).

– Interpolation
(6) otro dia queles este buen mandado dixo Moysen
another day that them this good directions said Moysen
‘the day after Moses had given them directions’
(Fontana 1997, p. 229)
–  V-CL and CL-V orders with inflected verbs
(7) Rogaronle que les diesses la llave
(they) begged him that them (you) give the key
‘They asked him to give them the key.’
(Fontana 1997, p. 228)
–  Clitics and NPs in complementary distribution
(8) Ael llamaban otrossi amosis
to him (they) called also Amosis
‘They also called him Amosis.’
(Fontana 1994, p. 89)

The properties of Modern Spanish, which it shares with French, are depicted in (9)
to (11).
–  No Interpolation
(9) a. Se lo di ayer
him it (I) gave yesterday
‘I gave it to him/her/you yesterday.’
b. *Se lo ayer di
 him it yesterday (I) gave
‘I gave it to him/her/you yesterday.’
 Juana M. Liceras & Anahí Alba de la Fuente

–  Cl-V order before inflected verbs

(10) a. Lo trajo ayer
it (he/she) brought yesterday
‘He/she brought it yesterday.’
b. *trájolo ayer
  (he/she) brought it yesterday
–  Clitic doubling
(11) a. (Les) contaron un cuento (a los niños) (optional)
(them) (they)told a story (to the children)
‘They told a story to the children.’
b. A los niños les contaron un cuento (obligatory)
to the children them (they)told a story
‘They told a story to the children.’

Perales and Liceras (2010) analyzed oral and written data produced by 30
intermediate-­advanced L1 French speakers of Spanish. The only IL property that
was compatible with Old Spanish was the presence of instances of clitics in prever-
bal and post-verbal position with inflected verbs. Thus, the authors did not find
clear evidence for a phrasal (Old Spanish) account of IL clitics. However, since the
authors found instances of post-verbal clitics with tensed verbs, which are neither
possible in French nor in Spanish, they speculated that learners may generate clit-
ics in both argument and non-argument positions. In other words, the IL grammar
may have competing options, one of which is neither available in the L1 nor in the
L2.10 The authors also found evidence of competition between the L1 and the L2
in the written experimental data where participants produced some instances of
clitics before the infinitive as in (5). When learners were asked to substitute clitic
pronouns for DPs, they sometimes used their French L1 order in the Spanish IL.
In the above-mentioned analysis both French and Spanish clitics are affix-
like elements (typologically identical), which, under the Competing Grammars
Hypothesis and the assumption that the initial state of L2 acquisition is the L1,
would not predict the presence of any of the properties associated with Old
Spanish phrasal-like clitics but rather obvious positive transfer from French into
S­ panish.11 However, there is a recent feature account of object pronouns that makes

.  As we have seen, this was possible in Old Spanish and it is possible in languages such as
Czech and Serbo-Croatian.
.  It could be argued, under a ‘psychotypology’ approach as in Kellerman (1979) or the
­Typological Primacy Model (Rothman 2011, 2013), that French learners might perceive
Spanish object pronouns as not being typologically similar to French object pronouns. While
this is a logical possibility, we will not discuss ‘psychotypology’ in this paper.
 Typological proximity in L2 acquisition 

French and Spanish closer to each other than to languages, like English (Camacho
Taboada 2006), but not identical. Perales and Liceras (2010) and Liceras (2014)
draw from this account to provide a feature definition of object pronouns in terms
of the [±] combination of two features which relate to their phonological nature
[±phonological] and their phrasal [±XP] nature.

Table 2.  A feature account of object pronouns: From full pronouns to agreement
markers12
Full pronouns > Simple clitics > 2P clitics12 > Categorial clitics > Agreement markers
English English Czech, Old Sp Italian, French Spanish, Bulgarian
[–phon]             [+phon]            [+phon]
[+XP]            [+XP]              [–XP]

As Table 2 shows, French clitics are not classified as agreement markers but
as categorial clitics. This implies that typological proximity is a relative term and
cannot be automatically equated to typological similarity. The feature combina-
tion approach allows us to account, among other things, for the fact that only
languages whose clitics are agreement markers have clitic-doubling. This in itself
does not seem to be the only reason why clitic-doubling has systematically been
found to be problematic for all Spanish learners (Bruhn de Garavito 1999b; Liceras
1985; Perales & Liceras 2010; Zobl & Liceras 2006; among others), but it separates
French from Spanish when it comes to object pronouns. Thus, L2Sp-L1Fr may
display optionality due to the competition between the L1 French categorial clitics
and the L2 Spanish agreement markers.

2.2  N
 ull/overt subjects
There have been analyses of French (Authier 1992; Roberge 1986, 1990) that clas-
sify it as a [+null subject] language like Spanish or Italian because its subject clitics
are analyzed as person agreement morphology. However, the majority of syntacti-
cians consider French a [–null subject] language like English or German, and it
has in fact been used to illustrate diachronic change from [+null subject] to [–null
subject] (Adams 1987; Roberts 1993). Thus, while being a Romance language
makes French typologically close to Spanish, there is no typological similarity
when it comes to the null subject parameter. This was the reason why White (1985)
used a group of L1 French learners as control in a study intended to investigate

.  Second position (2P) clitics have been related to Verb Second phenomena. This was first
pointed out by Wackernagel (1892). See Anderson (1993) for a discussion of both phenomena.
 Juana M. Liceras & Anahí Alba de la Fuente

whether L1 Spanish learners transferred the three properties of the null subject
parameter depicted in (12)–(14) to their English IL.
–  Missing Subjects
(12) a. —telefonearé mañana
b. *__ téléphonerai demain (Je téléphonerai demain)
c. *__ will call tomorrow
‘I will call tomorrow.’
–  Free Subject-Verb Inversion
(13) a. Lea ha telefoneado/Ha telefoneado Lea
b. __ Lea a télephoné/*a téléphoné Lea
(‘Lea a téléphoné’)
c. __ Lea has phoned/*has phoned Lea
‘Lea has phoned.’
– That-trace effects
(14) a. ¿Quiéni dice Lea que ti acaba de llegar?
b. *Qui dit Lea que ti vient d’arriver?
(Qui dit Lea qui ti vient d’arriver?)
c. *Whoi does Lea say that ti has just arrived?
(Who does Lea say __ ti has just arrived?)
‘Who does Lea say has just arrived?’

These examples show that the equivalent of the Spanish sentences is ungrammati-
cal in both French and English, which in principle explains why L1 Spanish speak-
ers (mainly those with lower proficiency) accepted many of the ungrammatical
sentences listed in (13) and (14) as grammatical, while the French speakers did
not. However, a closer look at the construction in (14) shows that French and
English differ in the way in which they avoid the that-t sequence (that is not real-
ized in English whereas French requires qui to mark the subject position).13 This
implies that transfer may not work the same way in English and French when it
comes to their respective Spanish ILs. In other words, English and French adopt
different solutions to avoid the sequence que+t to comply with the *[that+t] filter
proposed by Chomsky and Lasnik (1977). This is another example of how typo-
logical proximity cannot be equated to typological similarity, either in terms of
parametric options or in terms of surface structure filters. However, we argue that

.  This qui has been analyzed as the result of the fusion of the complementizer que and
the subject pronoun il which took place in XIV century French (Kayne 1975, 1976), and it is
referred to as the que → qui (que + il) rule.
 Typological proximity in L2 acquisition 

in the case of French, we are dealing with a degree of typological proximity that
almost reaches typological similarity. This is not the case for English.
Furthermore, even if French and English select the [–null subject] option,
typological proximity between French and Spanish when it comes to verbal
morphology may give L1 French learners an advantage over L1 English learners
when learning Spanish. This is what was argued in Liceras (1989) when compar-
ing grammaticality judgments produced by L1 French and L1 English learners
of ­Spanish, since the rejection of overt subject pronouns that are redundant in
­Spanish increased regularly towards the target norm (Spanish) in the case of the L1
French group while the English-speaking group showed less consistency. S­ econd,
the L1 French group also performed better with subject-verb inversion. Verbal
morphology may have also helped the French group here, since they seemed to
have fewer problems interpreting the post-verbal subject as being the subject of
the sentence rather than assigning them objective case, as the English group did.
The results for some that-t Spanish ungrammatical sentences missing the comple-
mentizer, as in (15), were difficult to interpret even for the control group.
(15) *¿Quién dices ___ estudia español contigo?
who say (you) ___ study (3rd. p.) Spanish with you
‘Who do you say ___ is studying Spanish with you?’

Participants either inserted an overt subject (tú, ‘you’) or erased the verb decir. In
other words, whether or not the three properties are linked to a general parameter
which licenses null subjects only in Spanish, but not in English or French, there
are either microparameters (Holmberg 2005; Sheehan 2006) or specific morpho-­
phonological realizations of the properties that interfere with assigning French
and Spanish to either the same or the opposite option of the [±null subject]. None-
theless, the typological proximity, albeit not similarity, which stems from verbal
morphology (i.e. both languages are synthetic, and accusative) makes French
closer to Spanish and seems to play a role in shaping the L2Sp-L1Fr IL.
Having discussed our view of the potential role of typological proximity ver-
sus typological similarity in L2A, in the next section we will look at how typo-
logical proximity has been approached in recent analyses intended to account for
some specific characteristics of L2Sp-L1Fr.

3.  The Spanish grammar of L1 French speakers: Morphosyntactic issues

The available analyses of the Spanish grammar of L1Fr speakers that we will dis-
cuss are word formation (compounding), the plural feature in nouns, clitic cluster
constraints, quantifiers and passive constructions. We have chosen to discuss these
areas of grammar for two reasons. First, they allow us to show how typological
 Juana M. Liceras & Anahí Alba de la Fuente

proximity shapes the L2Sp-L1Fr IL at different structural levels and in relation to

different features. Second, they all illustrate how subtle differences between French
and Spanish may remain in competition even when learners achieve native-like
competence. This is important because it allows us to go beyond the macropara-
metric view that has modeled the CGH (Fontana 1994; Kroch 1994; Yang 2002;
Zobl & Liceras 2006) and explore feature accounts linked to microparametric
views (Martínez-Sanz 2011; Perales & Liceras 2010; Sheehan 2006). The latter are
particularly suitable to investigate IL variability or optionality that goes beyond
competition between the L1 and the L2 and incorporates feature realizations or
combinations available in other natural languages.

3.1  P
 roductivity, head directionality and distinct phono-morphological
features in French and Spanish compounds: Acceptance versus
production
Typological proximity between French and Spanish is obvious in relation to both
N–N compounds, as in (16), and deverbal (V+N) compounds, as in (17).
(16) homme-grenouille
hombre  rana
man frog ‘frogman’
(17) ouvre-boîte
abrelata-s
open can/s  ‘can open-er’

As the English equivalent shows, French and Spanish N–N compounds are left-
headed, while English N–N compounds are right-headed. This is also the case for
deverbal compounds, which are VO in French and Spanish but OV in English. The
deverbalization process is also similar in French and Spanish. Specifically, regard-
less of the interpretation of the third person verbal morpheme (the -e in ouvr-e or
in abr-e), neither French nor Spanish deverbal compounds depict an instrumental
or agentive morpheme like English -er.
The productivity that characterizes deverbal compounds in French and
­Spanish and the lack of productivity of N–N compounds in both languages are
also a reflection of their typological proximity.14 However, in spite of this high

.  In fact, many English N–N compounds or deverbals are realized as Noun Phrases or as
Nouns with derivative morphology in both French and Spanish, as shown in (1) and (2).
(1) doll house
casa de muñecas
maison de poupée
 Typological proximity in L2 acquisition 

degree of typological proximity, there are differences that may be difficult to detect
by non-native speakers precisely due to the proximity between the two languages.
Without taking into consideration the orthographic differences, the two languages
differ in two main respects.
First, with deverbal compounds, Spanish marks the noun complement with a
generic -s both when the compound is singular and when it is plural while French
only marks the plural as shown in (18).15
(18) a. un cascanuec-es → unos cascanuec-es
b. un casse-noisette → des casse-noisette-s

Second, in French, but not in Spanish, the two Nouns of an N–N compound get
the plural morpheme when the compound is plural and the two members consti-
tute an apposition, as in (19), but only the head takes the plural when the other
Noun is a determinative, as in (20).16
(19) a. un homme-grenouille → des hommes-grenouille-s
b. un hombre rana → unos hombre-s rana
(20) a. un timbre-poste → des timbre-s-poste
b. un sello/unos sello-s17

Pomerleau (2000) compared the status of these compounds in the Spanish L3 of

French speakers with English as an L2. In the production task, none of the French
speakers had problems with directionality or with the morphology of N–N com-
pounds, but they had problems with the morphology of deverbal compounds,
which seems to indicate that the L2, English, did not play a relevant role. In other

(2) clothes line

tendedero
étendoir
.  When the noun complement is a mass noun, it looks as if both options (with and without
plural marker) are possible:
(1) un chasse-neige → des chasse-neiges
a catch snow some catch snows
‘a snow catcher’ *a snows catcher
(2) un chasse-neige → des chasse-neige
.  An anonymous reviewer points out that in some Spanish compounds mass nouns occur
in the singular (i.e. tragaluz -swallow light- “skylight”). We do not think that this is relevant
here because we are discussing N–N compounds in order to show that only the head takes the
plural morpheme. Thus, the Spanish equivalent of light house would be casa luz and, in plural,
light houses would correspond to casas luz but not *casas luces. If maison lumière were pos-
sible in French, then maisons lumières would be (and in fact is) the expected plural.
.  There is no N–N equivalent in Spanish.
 Juana M. Liceras & Anahí Alba de la Fuente

words, the typological proximity of the L1 won over a potential recency effect
(influence from the second language acquired). However, in the ­grammaticality
judgment task all participants accepted some N–N right-headed compounds.
Pomerleau argues that when these subjects have to produce compounds they sys-
tematically produce left-headed ones, while they seem to need substantial expo-
sure to Spanish to be able to categorically reject right-headed compounds. Since
her French participants had been exposed to English, it is difficult to determine
whether this exposure to right-headed compounds can account for this acceptance.
In terms of the morphology of these structures, all participants had problems
in that they judged as grammatical compounds that were missing the generic -s
marker that characterizes Spanish deverbal compounds. This specific characteris-
tic of Spanish, which does not allow us to equate typological proximity with typo-
logical similarity when comparing it to French, seems to be a potential candidate
for fossilization and evidences that competition persists between the L1 and the L2
grammars with respect to this specific marking.18
The fact that the L1 French learners have problems with these types of com-
pounds shows that when lack of typological similarity between two typologically-
close languages is instantiated by very subtle differences (the plural agreement of
N–N compounds and generic plural of the deverbal complement), the difficulties
encountered by learners to detect these differences in the input may be compa-
rable to the difficulties they encounter when dealing with differences related to the
implementation of parametric properties or surface structure filters.

3.2  A prosody account of the realization of plural in L2Sp-L1Fr nouns

Unlike the case with the generic -s of deverbal compounds, French and Spanish
share not only the plural marker -s but also the noun-adjective word order. Bruhn
de Garavito (2008) assumes that this word order results from a number projection
with a [+strong] number feature that is available for DPs to which nouns raise,
resulting in the noun–adjective word order. The author investigates the acquisi-
tion of Spanish plural by French L1 speakers in order to determine the role of
transfer, focusing on the production of plural morphology in Spanish nouns. She

.  This is a potential candidate for fossilization because learners systematically alternate
between accepting and using the generic -s and rejecting/not producing it (Desrochers,
Liceras, Spradlin & Fernandez 2003; Liceras, Mongeon, Cuza, Senn & Spradlin 2004). An
anonymous reviewer wonders whether competition is the main cause of variability. We argue
that competition is not the cause of variability but rather that variability is the outcome of the
competition between the two grammars that are accessed by the speakers or for which they
have actual mental representations.
 Typological proximity in L2 acquisition 

observes that, if one compares English- and French-speaking learners of Spanish,

one would expect an advantage for the French speakers in regard to the noun–
adjective order, which French shares with Spanish, and that, taken at this superfi-
cial level, neither group should experience difficulty with plural morphology, as it
exists in both L1s – see (21) – and is frequent in the L2 input data.
(21) las luces rojas
les lumières rouges
*the lights reds
‘the red lights’

However, as the author points out, French differs from Spanish in several respects:
(i) French doesn’t have word markers linked to the formation of the plural; (ii) in
French a postvocalic consonant is only pronounced when it is a sonorant; (iii) a
final -s can only be realized if it attaches to a following syllable or if it is marked in
the lexicon and (iv) the syllabification of a final obstruent is not saved by a rule of
epenthesis.
In order to investigate whether these differences were a source of transfer for
L1 French speakers learning Spanish, Bruhn de Garavito (2008) elicited data from
42 high school students (30 beginners and 12 intermediate) learning Spanish in
Montreal via a production task administered individually. The author found that
these speakers, mainly the beginners, transferred syllable structure restrictions
from their L1, since they omitted the plural frequently, particularly with words
ending in consonants. Bruhn de Garavito argues that, unlike French speakers,
English-speaking learners of Spanish do not seem to have similar problems in
producing the plural. Thus, in this specific case, typological proximity does not
provide an advantage to the French speakers. Their relatively poorer performance
should be sought in the syllable structure of the two languages. The French speak-
ers’ main problem was acquisition of the long plural, since it requires learning the
rule of epenthesis, which does not apply in French. Once more, lack of systematic
realization of this rule in L2Sp-L1Fr points to an area of competition between the
two grammars that is not overcome by overall typological proximity.

3.3  N
 ow you see it, now you don’t: Plural and case marking
in L2Sp-L1Fr quantifiers
Another category where typological proximity between French and Spanish is
compromised is quantifiers. In contrast with plural morphology, as reported in
the previous section, it is not differences related to prosody but rather differ-
ences related to plural and case marking in the case of some quantifiers that
may trigger L1 transfer that will not result in target-like forms. This is what
Androutsopoulou, Español-Echevarría and Prévost (2010) found when they
 Juana M. Liceras & Anahí Alba de la Fuente

examined the L2 acquisition of Number specification on Spanish quantifiers by

French speakers.
In French and Spanish, quantifiers can combine with NPs by means of the
preposition de, as in (22), or directly, as in (23):
(22) algo de vino
some of wine
‘some wine’
assez de vin
enough of wine
‘enough wine’
(23) varios perros
plusieurs chiens
‘several dogs’

The need to value the case feature of the NP when Number agreement is not pos-
sible accounts for the use of de as a case marker. While the general pattern is the
one in (23) for Spanish, in French most quantifiers require de. In fact, only quanti-
fiers that can establish Number agreement with the Noun can modify the Noun
directly. Quantifiers such as bastante (enough), mucho (much/many), or demasi-
ado (too much/too many) establish Number agreement with the Noun in Spanish
but require de in French, as shown in (24).

(24) bastantes/muchos/demasiados perros

assez de/beaucoup de/trop de chiens
‘enough/many/too many dogs’

Androutsopoulou, Español-Echevarría and Prévost (2010) administered a gram-

maticality judgment task and a production task to 9 intermediate and 17 advanced
French-speaking learners of Spanish at the University of Laval (Quebec, Canada).
They found that these learners performed poorly on plural inflection with some
quantifiers, and rejected the use of de. This suggests that they had acquired Num-
ber specification on these quantifiers, which allows case marking on the following
noun, but did not produce the appropriate morphology. The authors argue that
this dissociation between the syntax and the morphology of quantifiers evidences
a mapping problem rather than a deficit due to the absence of the corresponding
abstract properties, the reason being that de was rejected when it was not pos-
sible in Spanish and was accepted and produced in obligatory contexts such as
those in (22). They also argue that neither input nor the L1 (the behavior of these
specific quantifiers in French) can account for these results, since neither French
nor ­Spanish quantifiers can occur without plural marking when they are directly
followed by nouns (demasiado perros). Thus, the IL of these speakers seems to
 Typological proximity in L2 acquisition 

include bare quantifiers (without a plural marker agreeing with the Noun) that are
neither possible in French nor in L1 Spanish but can be found in other Romance
languages such as Catalan. This provides us with an idiosyncratic L2Sp-L1Fr con-
struction that cannot be accounted for by L1 transfer from French or by overgener-
alization of a construction available in Spanish but that occurs in other languages.

3.4  R
 estrictions on clitic clusters: Typological proximity and
the representation/processing divide
The effects of typological similarity are explored in Alba de la Fuente’s (2012) study
of clitic cluster constraints, which provides data from two experiments admin-
istered to advanced L2 speakers of Spanish whose L1s are English, French and
Romanian.19
Bonet (1991), in her discussion of the so-called Person-Case constraint (PCC)
presents two types of languages: Strong PCC languages, like French, which reject
combinations of 1st and 2nd person clitics, and Weak PCC languages, like Spanish
and Romanian, which accept such clitic combinations.
(25) a. Te me presentaste en la fiesta [Spanish]
b. Mi te-ai prezentat la petrecere. [Romanian]
c. *Tu te m’es présenté à la fête [French]
‘You introduced yourself to me at the party.’

Given the different versions of clitic combinations that are available in natural lan-
guages, this phenomenon provides a scenario that allows us to explore the effects
of typological similarity in contrast with typological proximity.
Generally speaking, both French and Romanian may be considered to be
typologically-close to Spanish, as they are all Romance languages. However, as we
can see in (25), Spanish and Romanian pattern together with respect to clitic clus-
ters, since both languages accept 1st and 2nd person combinations under certain
conditions, whereas French categorically rejects such combinations.20
If language typology plays a role in the acquisition of clitic cluster restrictions,
we would expect L1 French and L1 Romanian speakers to outperform their L1
English counterparts. If, on the other hand, only typological similarity plays a role,

.  Since the second experiment does not provide clear-cut answers with respect to the
­typological similarity/proximity dichotomy, only the results of experiment 1 are reported here.
.  According to Alba de la Fuente’s (2010) Narrow Plural-Blocking Effect, which would
apply in both Spanish and Romanian, “in a combination of 1st and 2nd person clitics with a
dative, the non-dative cannot be plural” (p.213). As such, combinations of two singular clitics
are allowed and combinations of two plural clitics are rejected.
 Juana M. Liceras & Anahí Alba de la Fuente

L1 Romanian speakers would be at an advantage in this case, since the restrictions

in their L2 match those of their L1. In turn, L1 French speakers would have to
restructure the restrictions available to them in their L1 to accommodate to the
less restrictive set of constraints of the L2 in their IL. L1 English speakers would
need to acquire a new set of restrictions altogether, as these constraints do not
apply in English due to the different category of its pronouns.
The results of experiment 1, a scaled grammaticality judgment task, showed
that, despite the fact that the overall acceptance rates differed significantly in all
groups and that all non-native groups accepted ungrammatical combinations in
Spanish, only the L1 Romanian speakers patterned with the native control group
in that they assigned higher scores to grammatical combinations and lower scores
to ungrammatical ones. Even though all of the experimental combinations are
categorically unacceptable in their L1, the L1 French speakers presented a certain
degree of acceptance for all combinations, including ungrammatical ones, and dis-
played a pattern that matched that of the L1 English group. These results suggest
that typological proximity in itself does not necessarily provide an advantage to L2
speakers. Rather, it is typological similarity that serves learners by providing some
sort of positive reinforcement, as indicated by the high acceptance rates of gram-
matical combinations by the L1 Romanian speakers (although they still accepted
ungrammatical conditions at a much higher rate than the native control, prefer-
ence for grammatical conditions was significantly higher).

3.5  P
 assive constructions are not different when it comes
to differential object marking
Tremblay (2006) investigated the L2 acquisition of Spanish reflexive passives, as in
(26), and reflexive impersonals, as in (27), by French- and English-speaking adults
at an advanced level of proficiency.
(26) Esos pisos se[ACC] construyeron hace dos siglos
Ces appartements se[ACC] sont construits depuis deux siècles
These apartments were built after two centuries
‘These flats were built two centuries ago.’
(27) En esta ciudad se[NOM] puede entrar con mucha facilidad
Dans cette ville on[NOM] peut entrer très facilement
In this city one can enter very easily
‘In this city one can enter very easily.’/‘One enters this city very easily.’

A main difference between the two constructions stems from the fact that reflexive
passives can only be formed with transitive verbs (verbs that can assign accusa-
tive case), while impersonal passives can be formed with copulative, unergative or
unaccusative verbs. Another important difference is that the internal argument of
 Typological proximity in L2 acquisition 

impersonal passives checks accusative case and must be preceded by the so-called
personal a that precedes Spanish [+animate] [+specific] direct object comple-
ments, as in (28).
(28) Aquí se[nom.] respeta a los niños
Ici on respecte les enfants
‘Here one respects children.’/‘Here children are respected.’

In principle, as Tremblay herself hypothesized, typological proximity would pre-

dict that native French speakers would evidence convergence on these properties
because neither of the two constructions exist in English and even if imper-
sonal passive does not exist in French, this construction and the reflexive passive
construction are superficially similar (se V DP) to the French reflexive passive
construction.
Tremblay (2006) administered a grammaticality judgment task to 16 French,
13  English and 27 Spanish (control) speakers in order to determine how the
advanced non-native speakers fared when compared to native speakers and
whether typological proximity gave an advantage to the L1 French group. She
found that the two non-native groups differed significantly from the native group.
As for the two non-native groups, typological proximity played a role in that
the French group significantly outperformed the English group on grammatical
reflexive passives with a pre-verbal [-animate] DP that had to agree with the verb.
Grammatical and ungrammatical impersonal test items involving [+animate]
DPs – preceded or not by the object-marking preposition a – were particularly
problematic, as L2 learners judged them both as grammatical. This implies, once
more, that subtle differences between typologically proximate languages are as
problematic for the speakers of one of the two languages as for speakers from
languages where the same differences exist and are typologically more distant.21
In this specific case, the L2Sp-L1Fr IL may fossilize the omission of personal a
across the board, not only with impersonal passives but also with any transitive
sentences, as is the case with Spanish Heritage speakers and other Spanish ILs
(Guijarro-Fuentes 2012; Montrul & Bowles 2009).
The acquisition of the four constructions discussed in this section evidences
that when typological proximity between French and Spanish is not realized as
typological similarity, the L2Sp-L1Fr IL displays instances of competition between
the two grammars as well as constructions that are possible in other natural
languages.

.  Differential object marking is not particular to Spanish but occurs in natural languages
such as Hindi, Yiddish, Persian or Turkish (Montrul & Gürel, this volume), which are typo-
logically distant from both Spanish and French (Aissen 2003; Leonetti 2004; Torrego 1998).
 Juana M. Liceras & Anahí Alba de la Fuente

4.  The Spanish grammar of L1 French speakers: Beyond morphosyntax

There are two areas of language competence that have received special attention in
this century: one pertains to structures that have a special status at the interfaces
in general and at the syntax-pragmatic interface in particular, and the other per-
tains to processing preferences. In this section, we discuss anaphora resolution in
ambiguous contexts, an area of grammar often considered as a “crossroads” case.
In her study on the topic, Carminati (2002) used Italian experimental data to
test the Position of Antecedent Hypothesis (PAH), which states that the distinct
uses of null and overt pronouns is based on the preferences that speakers have
for anaphor antecedents differing along a ‘prominence’ scale. According to this
scale, which is based on syntactic notions, “antecedents in the highest specifier
projection (the Spec IP, i.e. the subject position) are considered to be more promi-
nent than antecedents in lower projections (e.g. the direct object and the indirect
object position)” (Carminati 2002, p. 6). Thus, when the parser encounters a null
pronoun, it searches for an antecedent in the subject position, as shown in (29a).
However, when the parser encounters an overt pronoun, it searches for an ante-
cedent in a non-subject position, as in (29b).
(29) a. Alexi vio a Juan mientras Øi montaba en bicicleta
Alexi saw dom Juan while (hei) was riding on bike
‘Alex saw Juan while he was riding a bike.’
b. Alex vio a Juani mientras éli montaba en bicicleta
Alex saw dom Juani while hei was riding on bike
‘Alex saw Juan while he was riding on bike.’

In other words, null pronouns prefer the antecedent in Spec IP (the subject posi-
tion) while overt pronouns prefer to be coindexed with the antecedent in object
position, regardless of them being in forward or backward anaphora construc-
tions, as we can see in (29) versus (30).
(30) a. Mientras Øi montaba en bicicleta Alexi vio a Juan
while (hei) was riding on bike Alexi saw Juan
b. Mientras éli montaba en bicicleta Alex vio a Juani
while hei was riding on bike Alex saw Juani
‘While he was riding on bike Alex saw Juan.’

The PAH has been tested using experimental data in Italian (Belleti, Bennati &
S­ orace 2007; Sorace & Filiaci 2006), Spanish (Alonso-Ovalle, Fernández-Solera,
­Frazier & Clifton 2002), and Arabic (Bel & García-Alcaraz, this volume), among
other languages.
These experimental data have shown that native speakers have rather cate-
gorical processing preferences in the case of null subjects as in (29a) and (30a), in
 Typological proximity in L2 acquisition 

contrast with the less clear-cut picture offered by overt subjects (29b) and (30b), in
which the overt pronoun can also take an outside referent (neither Alex nor Juan
but rather someone else). As for non-native speakers (L1 English speakers), the
processing preferences were native-like for the anaphora resolution of null sub-
jects but significantly different from the native preferences for the anaphora reso-
lution of overt subject pronouns.
These results have been interpreted as evidence that non-native speakers do
not master the pragmatic principles that determine the use of overt subjects in
Italian (Belleti et al. 2007; Sorace & Filiaci 2006) or Spanish (Valenzuela, Liceras &
López-Morelos 2011). These results also show that non-native speakers do not
master the processing preferences related to overt subjects in that they do not
associate overt subjects to a non-prominent (non-subject) antecedent as proposed
by Carminati (2002) but rather seem to treat them as null subjects.
Taking into account that the interface problem does not take into consider-
ation the fact that not all Spanish overt pronouns have a pragmatic value, Liceras
and Alba de la Fuente (Forthcoming), further investigate this issue in a study that
analyzes data from a grammaticality judgment task administered to L1 and L2
speakers of Spanish (L1 French). Assuming a microparametric approach to the
null subject parameter (Martínez-Sanz 2011; Sheehan 2006), the authors propose,
as in Liceras and Fernández-Fuertes (2013), that Spanish has two different types
of overt subjects: strong pronouns that have a focus or pragmatic value and overt
weak pronouns that are phonetic realizations of the null subject option (Holmberg
2005). In other words, Spanish overt weak pronouns, which are said to be the
marked option in relation to null pronouns for native Spanish speakers, could be
interpreted as French clitics by L1 French speakers. The authors further argue that,
rather than their pragmatic value, it is the marked status of these pronouns with
respect to their null counterparts that L1 French speakers (or L1 English speak-
ers) may not capture. This interpretation would make it difficult for the L1 French
speakers to grasp the processing differences associated with null and overt pro-
nouns which, in turn, could lead to a lack of specialization of the two different
types of pronouns. This scenario would imply that we are in fact confronting a
processing problem, rather than an interface problem, because the choice would
not have to be linked to the pragmatic value of Spanish strong pronouns but rather
to the successful mapping of the two different types of overt pronouns.
The results of the grammaticality judgment task revealed significant differ-
ences between the native and non-native groups. In the case of the L1 ­Spanish
group, the results were compatible with the authors’ proposal that Spanish overt
pronouns can have a pragmatic value but they can also be weak pronouns,
namely, the phonetic realization of null subjects (Holmberg 2005; Sheehan 2006).
­Specifically, and in contrast with Carminati’s PAH, the differences between subject
 Juana M. Liceras & Anahí Alba de la Fuente

and object referents were not straightforward, as the results revealed differences
between subject and object across different conditions, but no main effect for pres-
ence of pronoun was found. Indeed, whereas participants did prefer the object of
the main clause as the referent for an overt subject, they provided similar rates for
subject and object referents in the case of forward anaphora with a null subject–a
configuration, shown in (30a), where the PAH would predict a clear preference for
subject over object- and a certain degree of preference for both subject and object
referents was found in all conditions.
In the case of the L1 French group, speakers showed a preference for subject
referents in all conditions, although the differences between subject and object
referents were quite narrow in the forward anaphora conditions–as in  (29a)
and (29b).
In sum, the anaphora resolution preferences displayed by the native ­Spanish
speakers of this experiment do not follow Carminati’s PAH when it comes to null
subject forward anaphora–shown in (29a). However, the native group differs sig-
nificantly from the L1 French group in that only the former differentiates between
null subjects and overt subjects, which the authors interpret as evidence that it
is only for the native speakers that the marked status of Spanish weak overt pro-
nouns plays a role when processing anaphora resolution. In the case of the non-
native speakers, the marked status of the overt pronoun does not seem to be part
of their grammar.

5.  C
 onclusions

Taking as a point of departure the differences and similarities between French

and Spanish object pronouns (Section 2.1), we have argued that the feature
analysis that differentiates them provides a more refined framework for compar-
ing the two languages and analysing the L2Sp-L1Fr IL than a parametric analy-
sis that considers the two types of pronouns to be affix-like elements. We have
further argued that this provides a clear example of how typological proximity
differs from typological similarity, namely, French and Spanish evidence typo-
logical proximity but their clitic systems do not evidence typological similarity.
In Section 2.2, we have used the status of French and Spanish with respect to
the null subject parameter to argue that even though French has systematically
been considered a [–null ­subject] language like English, the fact that, unlike Eng-
lish, French has clitic subject pronouns and adopts a different solution to abide
by the so-called that-trace filter distances this language from English, a fact that
the traditional view of parameters cannot capture. These specific characteristics
of French, together with the fact that it has maintained some verbal agreement
 Typological proximity in L2 acquisition 

markers, may facilitate the­acquisition of Spanish by L1 French speakers. In this

respect, typological proximity may ­supersede the fact that these two languages
differ in terms of the microparameters (or properties) associated to the null
subject parameter (Sheehan 2006). In fact, it is only when sophisticated and up-
to-date linguistic analyses tease apart typological proximity and typological simi-
larity that a clearer picture emerges (as in Alba de la Fuente (2012) or in Perales
and Liceras (2010), where a feature account of object pronouns is used to provide
an analysis of L2Sp-L1Fr Spanish clitics).
We have shown throughout this chapter that typological proximity may
obscure some obvious and subtle differences that separate two closely related
languages such as French and Spanish, thus resulting in transfer or idiosyncratic
forms that co-exist with the L2 forms and may eventually fossilize. Such would
be the case of N-N and deverbal compounds, as shown in Pomerleau (2000), the
variability in the production of plural markers reported by Bruhn de Garavito
(2008), the idiosyncratic bare quantifier structures identified in Androutsopoulou,
Español-Echevarría and Prévost (2010), or the omission of the personal a marker in
reflexive passives and impersonals with a [+animate] DP.
To account for the simultaneous availability of target-like, transferred and
idiosyncratic L2Sp-L1Fr parametric options or feature combinations, namely to
account for the optionality that is so pervasive in IL systems, we have turned to
the CGH. Based on this hypothesis we have argued that the L2Sp-L1Fr speakers,
like L1 learners and the L1 speakers whose grammars are depicted to explain dia-
chronic change, may make use of two competing grammars (the two options of a
given parameter or the feature combinations displayed by two different languages)
when using their IL. This seems to be the case when language-specific morphol-
ogy such as the generic -s of Spanish deverbal compounds or the plural marking
of some Spanish quantifiers, constitute a problem for L1Fr speakers and, in fact,
seem to be candidates for permanent variability (fossilization). This implies that
it is not only macroparametric options between the L1 and the L2 that compete
but rather microparametric options or specific rules that may be available in other
languages, as is also the case with differential object marking (the personal a of
direct objects) or the specific realization of the that-t filter.
We have also shown that even though the differences between null and overt
pronouns in anaphora resolution have been accounted for in terms of difficulties
posed by the pragmatic value of overt pronouns (an interface condition), the fact
that Spanish overt pronouns can be weak (the marked option of the null real-
ization) has to be taken into consideration. We have specifically argued that this
influences the grammaticality judgements of both the native Spanish speakers and
the L2Sp-L1Fr speakers when it comes to anaphor ambiguity resolution and seems
to indicate that it is not necessarily the pragmatic value of strong pronouns that
 Juana M. Liceras & Anahí Alba de la Fuente

is problematic for the L2Sp-L1Fr speakers, but rather the processing of null and
overt pronouns, in the sense that they do not seem to capture the marked value
of Spanish overt pronouns. We are aware of the fact that more research is needed
to tease apart difficulties encountered by non-native speakers at the pragmatic-
interface level from processing differences between native and non-native speak-
ers. However, we conclude that empirical research aimed at providing an in-depth
analysis of these difficulties should take into consideration sophisticated linguistic
analyses that provide a suitable framework for investigating non-native compe-
tence and non-native processing.

References

Adams, M.P. (1987). Old French, Null Subjects and Verb-second Phenomena. Unpublished Ph.D.
dissertation. UCLA, Berkeley.
Adjémian, C. (1976). On the nature of interlanguage systems. Language Learning, 26, 297–320.
DOI: 10.1111/j.1467-1770.1976.tb00279.x
Adjémian, C., & Liceras, J.M. (1984). Universal grammar, the intake component and L1:
Accounting for adult acquisition of relative clauses. In F. Eckman, L. Bell, & D. Nelson
(Eds.), Universals of Second Language Acquisition (pp. 101–118). Rowley, MA: Newbury
House.
Aissen, J. (2003). Differential object marking: Iconicity vs. economy. Natural Language and
­Linguistic Theory, 21, 435–483. DOI: 10.1023/a:1024109008573
Alba de la Fuente, A. (2010). More on the clitic combination puzzle: Evidence from S­ panish,
Catalan and Romanian. In S. Colina, A. Olarrea, & A. M. Carvalho (Eds.), Romance
­Linguistics 2009. Selected papers from the 39th Linguistics Symposium on Romance Lan-
guages (LSRL), Tuscon, Arizona, March 2009. (pp. 203–215). Amsterdam: John Benjamins.
DOI: 10.1075/cilt.315.12alb
Alba de la Fuente, A. (2012). Clitic Combinations in Spanish: Syntax, Processing and Acquisition.
Unpublished Ph.D. dissertation. University of Ottawa, Ottawa.
Alonso-Ovalle, L., Fernández-Solera, S. Frazier, L., & Clifton. C. (2002). Null vs. overt pronouns
and the topic-focus articulation in Spanish. Journal of Italian Linguistics, 14,151–170.
Anderson, S.R. (1993). Wackernagel’s revenge: Clitics, morphology, and the syntax of second
position. Language, 69(1), 68–98. DOI: 10.2307/416416
Androutsopoulou, A., Español-Echevarría, M., & Prévost, P.H. (2010). The syntax/­morphology
interface in Spanish L2 acquisition: Focus on quantified DPs. The Canadian Journal of
­Linguistics, 55(2), 149–180. DOI: 10.1353/cjl.2010.0005
Authier, M. (1992). A parametric account of V-governed arbitrary null arguments. Natural Lan-
guages and Linguistic Theory, 10(3), 345–374. DOI: 10.1007/bf00133367
Baker, M.C. (1996). The Polysynthesis Parameter. Oxford: OUP.
DOI: 10.1017/s0022226798227113
Belleti, A., Bennati, E., & Sorace, A. (2007). Theoretical and developmental issues in the syntax
of subjects: Evidence from near-native Italian. Natural Language and Linguistic Theory, 25,
657–689. DOI: 10.1007/s11049-007-9026-9
 Typological proximity in L2 acquisition 

Bley-Vroman, R. (1990). The logical problem of foreign language learning. Linguistic Analysis,
20, 3–49. DOI: 10.1017/cbo9781139524544.005
Bonet, E. (1991). Morphology after Syntax: Pronominal Clitics in Romance Languages. Unpub-
lished Ph.D. dissertation. MIT.
Bonet, E. (1994). The person-case constraint: A morphological approach. MIT Working Papers
in Linguistics 22. The Morphology-Syntax Connection (pp. 33–52).
Bonet, E. (2008). The person-case constraint and repair strategies. In R. D’Alessandro,
S. Fischer, & G. Hrafnbjargarson (Eds.), Agreement Restrictions (pp. 103–128). Berlin: De
Gruyter. DOI: 10.1515/9783110207835.103
Bruhn de Garavito, J. (1999). The Syntax of Spanish Multifunctional Clitics and Near-native Com-
petence. Unpublished Ph.D. dissertation. McGill University, Montreal.
Bruhn de Garavito, J. (2008). Acquisition of the Spanish plural by French L1 speakers: The role
of transfer. In J. M. Liceras, H. Zobl, & H. Goodluck (Eds.), The Role of Formal Features in
Second Language Acquisition (pp. 270–298). Mahwah, NJ: Lawrence Erlbaum Associates.
Camacho Taboada, V. (2006). La arquitectura de la gramática. Los clíticos pronominales románi-
cos y eslavos. Sevilla: Publicaciones de la Universidad de Sevilla.
Carminati, M.N. (2002). The Processing of Italian Subject Pronouns. Unpublished Ph.D. disserta-
tion. University of Massachusetts, Amherst.
Chomsky, N. (1981). Lectures on government and binding. Dordrecht, Netherlands: Foris.
DOI: 10.2307/2273965
Chomsky, N. (1986). Knowledge of Language: Its Nature, Origin, and Use. New York, NY: Praeger.
DOI: 10.9793/elsj1984.6.213
Chomsky, N. (1995). The Minimalist Program. Cambridge MA: The MIT Press.
DOI: 10.1017/s0272263197241070
Chomsky, N. & Lasnik, H. (1977). Filters and control. Linguistic Inquiry, 8, 425–504.
DOI: 10.7551/mitpress/9780262062787.003.0002
Chomsky, N., & Lasnik, H. (1993). Principles and parameters theory. In J. Jacobs, A. von
­Stechow, W. Sternefeld, & T. Vennemann (Eds.), Syntax: An International Handbook of
Contemporary Research, Vol.1 (pp. 506–569). Berlin: Walter de Gruyter.
Desrochers, A., Liceras, J.M., Spradlin, K.T., & Fernández, R. (2003). Can an on-line response
latency task shed light on native and non-native competence in the deverbal compounds
of Spanish? In J. M. Liceras, H. Zobl, & H. Goodluck (Eds.), Proceedings of the 2002.
Generative Approaches to Second Language Acquisition (GASLA-6) Conference: L2 Links
(pp. 64–70). Somerville, MA: Cascadilla Press.
Ellis, R. (1994). The Study of second language acquisition. Oxford: OUP.
DOI: 10.1017/s0272263100014479
Fontana, J. (1994). A variationist account of the development of the Spanish clitic system. In
K.  Beals (Ed.), Papers from the 13th Regional Meeting of the Chicago Linguistic Society,
Vol. 2: The Parasession on Variation in Linguistic Theory (pp. 87–100). Chicago IL: Chicago
Linguistic Society.
Fontana, J. (1997). On the integration of second position phenomena. In A. Kemenade &
N.  Vincent (Eds.), Parameters of Morphosyntactic Change (pp. 207–249). Cambridge:
CUP.
Goodluck, H. (1986). Language acquisition and linguistic theory. In P. Fletcher & M. Garman
(Eds.), Language Acquisition (2nd ed.) (pp. 49–68). Cambridge: CUP.
DOI: 10.1017/cbo9780511620683.005
 Juana M. Liceras & Anahí Alba de la Fuente

Greenberg, J. (1963). Universals of Language. Cambridge, MA: The MIT Press.

DOI: 10.1017/s0022226700001109
Guijarro-Fuentes, P. (2012). The acquisition of interpretable features in L2 Spanish: Personal a*.
Bilingualism: Language and Cognition, 15(4), 701–720. DOI: 10.1017/s1366728912000144
Han, Z.-H. (2013). Forty years later: Updating the Fossilization Hypothesis. Language Teaching,
46, 133–171. DOI: 10.1017/s0261444812000511
Holmberg, A. (2005). Is there a little pro? Evidence from Finnish. Linguistic Inquiry, 36, 533–64.
DOI: 10.1162/002438905774464322
Kayne, R. (1975). French Syntax. Cambridge, MA: The MIT Press.
DOI: 10.1017/s0022226700005272
Kayne, R. (1976). French relative que. In M. Luján & F. Hensey (Eds.), Current Studies in
Romance Linguistics (pp. 255–299). Washington DC: Georgetown University Press.
Keenan, E., & Comrie, B. (1977). Noun phrase accessibility and universal grammar. Linguistic
Inquiry, 8, 63–99.
Kellerman, E. (1979). Transfer and non-transfer: Where we are now. Studies in Second Language
Acquisition, 2, 37–57. DOI: 10.1017/s0272263100000942
Kroch, A. (1994). Morphosyntactic variation. In K. Beals (Ed.), Papers from the 30th Regional
Meeting, Chicago Linguistic Society (pp. 180–201). Chicago IL: Chicago Linguistic Society.
Kroch, A. (2001). Syntactic change. In M. Baltin & C. Collins (Eds.), Handbook of Contemporary
Syntactic Theory (pp. 699–729). Oxford: Blackwell. DOI: 10.1002/9780470756416.ch22
Larsen-Freeman, D., & Long, M.H. (1991). An Introduction to Second Language Research.
­London: Longman. DOI: 10.1177/026765839200800205
Leonetti, M. (2004). Specificity and differential object marking in Spanish. Catalan Journal of
Linguistics, 3, 75–114.
Liceras, J.M. (1985). The value of clitics in non-native Spanish. Second Language Research, 1(2),
151–168.
Liceras, J.M. (1986). Linguistic Theory and Second Language Acquisition: The Spanish Nonnative
Grammar of English Speakers. Tubingen: Gunter Narr. DOI: 10.1017/s0272263100007403
Liceras, J.M. (1989). On some properties of the pro-drop parameter: Looking for missing sub-
jects in non-native Spanish. In S. Gass & J. Schachter (Eds.), Language Acquisition: A Lin-
guistic Approach (pp. 109–133). Cambridge: CUP. DOI: 10.1017/cbo9781139524544.009
Liceras, J.M. (2014). The Multiple Grammars Theory and the nature of L2 grammars. Commen-
tary to L. Amaral and T. Roeper’s Multiple Grammars and second language representation.
Second Language Research, 30(1), 47–54. DOI: 10.1177/0267658313517841
Liceras, J.M., & Alba de la Fuente, A. (Forthcoming). Native and non-native acceptability
­judgments on the anaphoric value of null and overt Spanish subject pronouns.
Liceras, J.M., & Fernández-Fuertes, R. (2013). Subject omission/production in child bilingual
English and child bilingual Spanish: The view from linguistic theory. Workshop on Cross-
linguistic influence in language contact. 19th International Congress of Linguistics (ICL
2013) – Geneva, July 2013.
Liceras, J.M., Mongeon, C., Cuza, A., Senn, C., & Spradlin, K.T. (2004). La adquisición en el aula
sin input formal: Los compuestos exocéntricos de las interlenguas del español. Red-ELE,
0, 1–28.
Martínez-Sanz, C. (2011). Null and Overt Subjects in a Variable System: The Case of Dominican
Spanish. Unpublished Ph.D. dissertation. University of Ottawa, Ottawa.
 Typological proximity in L2 acquisition 

Montrul, S., & Bowles, M. (2009). Back to basics: Incomplete knowledge of differential object
marking in Spanish heritage speakers. Bilingualism: Language and Cognition, 12, 363–383.
DOI: 10.1017/s1366728909990071
Newmeyer, F.J. (1990). Explaining language universals. Journal of Linguistics, 26, 203–222.
DOI: 10.1017/s002222670001450x
Perales, S., & Liceras, J.M. (2010). Looking for universals in the acquisition of L2 Spanish object
clitics. In P. Guijarro-Fuentes & L. Dominguez (Eds.), New Directions in Language Acqui-
sition: Romance Languages in the Generative Perspective (pp. 419–452). Newcastle upon
Tyne: Cambridge Scholars.
Pérez-Leroux, A.T., & Liceras, J.M. (2002). The Acquisition of Spanish Morphosyntax: The L1/L2
Connection. Dordrecht: Kluwer. DOI: 10.1007/978-94-010-0291-2
Pomerleau, J. (2000). La adquisición del español en el aula: Los compuestos nominales de los fran-
ceses de Canadá. Unpublished Master thesis. Universitat de Barcelona, Barcelona.
Rivero, M.L. (1986). Parameters in the typology of clitics in Romance and Old Spanish. Lan-
guage, 62(4), 774–807. DOI: 10.2307/415172
Rivero, M.L. (1997). On two locations for complement clitic pronouns: Serbo-Croatian,
­Bulgarian and Old Spanish. In A. Kemenade & N. Vincent (Eds.), Parameters of Morpho-
syntactic Change (pp. 170–206). Cambridge: CUP.
Rivero, M.L. (2004). Spanish quirky subjects, person restrictions and the person-case constraint.
Linguistic Inquiry, 35(3), 494–502. DOI: 10.1162/ling.2004.35.3.494
Rivero, M.L. (2008). Oblique subjects and person restrictions in Spanish: A morphological
approach. In R. D’Alessandro, S. Fischer, & G. Hrafnbjargarson (Eds.), Agreement Restric-
tions (pp. 215–250). Berlin: De Gruyter. DOI: 10.1515/9783110207835.215
Roberge, Y. (1986). Subject doubling, free inversion and null argument languages. Canadian
Journal of Linguistics, 31(1), 54–79.
Roberge, Y. (1990). The Syntactic Recoverability of Null Arguments. Montreal: McGill-Queen’s
University Press. DOI: 10.7202/602763ar
Roberts, I. (1993). Verbs and Diachronic Syntax: A Comparative History of English and French.
Dorcrecht: Kluwer. DOI: 10.1007/978-94-011-2910-7_2
Rothman, J. (2011). L3 syntactic transfer selectivity and typological determinacy: the typologi-
cal primacy model. Second Language Research, 27(1), 107–217.
DOI: 10.1177/0267658310386439
Rothman, J. (2013). Cognitive economy, non-redundancy and typological primacy in L3 acqui-
sition: evidence from initial stages of L3 Romance. In S. Baauw, F.A.C. Drijkoningen, &
M.  Pinto (Eds.), Romance Languages and Linguistic Theory (pp. 217–247). Amsterdam:
John Benjamins. DOI: 10.1075/rllt.5.11rot
Rothman, J. (In press). Linguistic and cognitive motivations for the typological primacy model
of third language (L3) transfer: considering the role of timing of acquisition and profi-
ciency in the previous languages. Bilingualism: Language and Cognition.
DOI: 10.1017/s136672891300059x
Schwartz, B., & Sprouse, R. (1994). L2 cognitive states and the full transfer/full access model.
Second Language Research, 12(1), 40–72. DOI: 10.1177/026765839601200103
Selinker, L. (1972). Interlanguage. IRAL, 10(2), 209–231. DOI: 10.1515/iral.1972.10.1-4.209
Selinker, L., & Lamendella, J. (1978). Two perspectives on fossilization in interlanguage learning.
Interlanguage Studies Bulletin, 3(2), 143–191. DOI: 10.1111/j.1467-1770.1979.tb01075.x
 Juana M. Liceras & Anahí Alba de la Fuente

Sheehan, M. (2006). The EPP and Null Subjects in Romance. Unpublished Ph.D. dissertation.
Newcastle University, UK.
Slabakova, R. (2009). L2 fundamentals. Studies in Second Language Acquisition, 31(2), 155–173.
DOI: 10.1017/s0272263109090263
Sorace, A., & Filiaci, F. (2006). Anaphora resolution in near-native speakers of Italian. Second
Language Research, 22(3), 339–368. DOI: 10.1191/0267658306sr271oa
Tarone, E. (1982). Systematicity and attention in interlanguage. Language Learning, 32(1), 69–84.
DOI: 10.1111/j.1467-1770.1982.tb00519.x
Torrego, E. (1998). Nominative subjects and pro-drop Infl. Syntax, 1(2), 206–219.
DOI: 10.1111/1467-9612.00008
Tremblay, A. (2006). On the second language acquisition of Spanish reflexive passives and
reflexive impersonals by French- and English-speaking adults. Second Language Research,
22(1), 30–63. DOI: 10.1191/0267658306sr260oa
Valenzuela, E., Liceras, J.M., & López-Morelos, L.P. (2011). Ambiguous anaphora in L2 English
and L2 Spanish. Paper presented at the BUCLD 36.
Wackernagel, J. (1892). Uber ein Gesetz der indogermanischen Wortstellung. Indogermanische
Forschungen, 1, 333–436.
White, L. (1985). The pro-drop parameter in adult second language acquisition. Language
Learning, 35(1), 47–62. DOI: 10.1111/j.1467-1770.1985.tb01014.x
Yang, C. (2002). Knowledge and Learning in Natural Language. Oxford: OUP.
Zobl, H., & Liceras, J.M. (2006). Competing grammars and parametric shifts in second lan-
guage acquisition and the history of English and Spanish. In D. Bamman, T. Magnitskaia, &
C. Zaller (Eds.), Proceedings of the 30th Boston University Conference on Language Develop-
ment (BUCLD) (pp. 713–724). Somerville, MA: Cascadilla Press.
 Index
A Aspectual
[+STATE]  312
Accessibility Hierarchy  delimitation  312
331–333
Accusative
case  140, 284–288, 291,
294, 349
see also Differential Object
Marking (DOM)
clitic  69, 110, 121–122, 261,
269, 271–272
object  294, 335
Activation  21–26, 32, 34–35,
37, 39, 42–43, 108, 127, 140,
177–178
Affricate  77, 79–81, 85–86
Afrikaans  235–236
Alveolar  76–77, 80–81, 86–88
see also Apico-alveolar,
Dento-alveolar
Anaphora resolution  203,
209–210, 224–227, 334,
350–353
Animacy  282–283, 285–290,
300, 302–303
Animate  114, 265, 281–282,
284–285, 288–293, 295,
297, 300, 302–304,
349, 353
Inanimate  114, 127, 129, 265,
284, 288–289, 291–293,
295–297, 300–302, 304
Apico-alveolar  77, 80, 86
see also Alveolar
Arabic  115, 171, 201, 203,
206–207, 209–210,
224–226, 235–236,
332–333, 350
Moroccan Arabic  115, 171,
201, 203, 206, 209–210,
224, 226
Modern Standard
Arabic  206, 235
Darija  206
Aspect  28, 30, 33, 35, 37, 52,
94, 234–236, 241, 244, 281,
284, 312, 315
content  312
difference  109, 315
feature  23–24, 34–35, 37–38,
42, 312, 316
information  312
interpretation  155, 312, 316
markers  312
morphology  235
property  139, 289, 309,
312–316, 320–321, 324
suffix  30, 33
Attributive adjective  135–136,
146, 157, 162
B 175–177, 194–195, 349
Bantu languages  282
Basque  56, 75–76, 78–101
Bilingual acquisition  76, 78,
113, 157, 208
Bilingual first language
acquisition  21
Brazilian Portuguese (BP)  173,
260, 265, 271–276
C 148–149, 157, 160–162,
Catalan  75–76, 81, 96, 99,
105–118, 120–129, 135–137,
140–142, 144, 146, 148–152,
155–159, 161, 163, 174,
214–215, 225, 290, 318, 347
Case  28–30, 32–34, 42, 78, 81,
121, 140, 149, 281, 284–288,
291, 294, 303–304, 345–349
Change of state  143, 313,
315–316, 323–324
CHILDES  119, 150, 289
Chinese  40, 115–116, 257, 260,
262, 268–277
Mandarin Chinese
(MC)  260
Clitic  27–28, 30–34, 37, 40,
42, 56, 66, 69, 105, 108,
110, 112, 117, 121–123, 127,
262–267, 269–273, 276,
335–339, 347, 351–353
cluster  335–336, 341, 347
doubling  31–32, 40, 336–339
non-argumental  21, 23, 31,
33, 42
see also Object clitic
Coda
complex  79–80
consonant  79
Contrastive Analysis
Hypothesis (CAH)  329
Contrastive Focus  169, 171,
173–174, 178–183, 185–186,
189–191, 194–195
Convergence  21, 23, 25–26,
31, 35, 37, 39, 42–43, 105,
107–109, 112, 129, 169–170,
Coronal  76, 81, 86
Copula  108–110, 112–116, 127,
135–146, 148–149, 151–157,
160–161, 163, 309–317,
319–321, 323–325
Copula + ADJ  310, 317
Cross-linguistic
influence  21–25, 30–31,
35, 37–42, 135, 145–146,
202, 209–210, 225–226,
228, 239, 240, 271
CV  83, 96
CVC  76, 78, 83–85, 89–91, 96
D
Dative  27, 33–34, 42, 108,
284–286, 291–295, 297,
302–304, 335, 347
marker  285, 302–303
Default  50, 52, 56, 58–59,
64–65, 68, 110, 112, 175,
194, 324
Definiteness  27–28, 105,
111–112, 116, 118, 122–125,
128–129, 261–263, 265,
267–268, 270–273, 276,
282, 284, 286, 288, 302
Definite  28, 30, 39–40,
105, 111–112, 119–120,
 Index
122–124, 126–129, 206,
261–269, 272–276, 282,
284, 286–288, 290–291,
295–296, 300
Definiteness Effect  105,
111–112, 116, 118, 125, 129,
286
Indefinite  81, 111–112,
119–120, 124–125, 128,
261–270, 272, 274–276,
284, 286–288, 295–296,
299–301
Dento-alveolar  80
Diploma de Español como
Lengua Extranjera
(DELE)  118, 177, 291, 318
Differential Object Marking
(DOM)  281–291, 293–296,
298–304, 348–349, 353
Direct Object  27–28, 31–32, 40,
56, 69, 281–282, 284–288,
290–293, 295, 297–303,
331, 335, 349–350
marking  34, 284
Doubling  see Clitic doubling
Dutch  208, 226, 235–236,
309–310, 315–321, 323–325
E 173–174, 177–183, 185–186,
Empty complementizer
filter  331
English  33, 53, 56–57, 75–79,
81, 96, 111, 113, 116, 128,
136, 145–146, 157, 160,
175–177, 180, 202–203,
207–210, 215, 226,
235–237, 240, 248–250,
257, 260, 263–265, 268,
271–277, 282–283,
289–291, 303–304,
309–311, 313–315, 317–318,
331–333, 336, 339–345,
347–349, 351–352
Endpoint encoding  234–236,
239, 241, 243–244,
246–250
Ergative  78, 81–82, 94
Estar  53, 57, 65, 105–110,
112–121, 123–127, 129,
137–142, 144–146, 149,
151–157, 160–162, 309–316,
319–325
Être  142–144, 161
European Portuguese  257,
260, 262–266, 268–169,
281–287
Event construal  233, 237, 241,
247, 250
Event type property  312
Eventive effect  105
Evidentiality  23–25, 28,
35–37, 109
F 348–349, 351
Farsi  169–174, 177–178, 186,
193–194, 282
Feature Reassembly
Hypothesis  21, 23, 25–26,
32, 35, 41–42, 108, 169, 257,
250–260, 272, 275–277,
282–283, 290, 303
Fill-in-the-Gap Task  309–310,
317, 320–324
French  57, 77–78, 114, 135–136,
142–144, 146, 148, 150–152,
158–161, 177, 206, 210, 215,
250, 282–283, 289, 314,
317–318, 329–353
Focus  21, 23–24, 26, 30, 41–43,
50, 56, 68, 106–107, 110,
136, 149–150, 153, 169, 171,
189–191, 193–195, 234, 243,
259, 270, 283, 285, 288,
290, 295, 311, 314, 321, 351
see also Contrastive Focus
Formal universal  331–333
Fossilization  289, 314, 330, 334,
344, 353
Full Transfer/Full Access
Hypothesis  258, 282–283,
290–291
Frequency  22, 40, 59, 75–77,
79–82, 91, 94, 96, 98–101,
107, 115, 117, 120, 122, 141,
152, 228, 236, 241, 244,
246, 248–250, 275, 289, 314
Fricative  77, 80–81, 86–87
Functional Convergence
Hypothesis (FCH)  37
Functional convergence  21, 23,
25, 31, 35, 37, 105,
108–109, 129
Functional
feature  21, 23–27, 37, 39,
42–43, 108
interference  21, 23,
25–26, 31, 33–35, 37–38, 42,
108–109, 127
G
Genitive  31–34, 286, 288, 332
Grammatical aspect  234–236
Grammaticality Judgment
Task  42, 289, 310, 314,
317–318, 320, 344, 346,
Scalar Grammaticality
Judgment Task  263, 267
Greek  175, 202
H
Haber  65, 105, 110, 115–116,
118–127
Hebrew  282, 302
Heritage speaker  289, 292,
309, 349
Hindi  282, 302, 349
I
Imperfective  35–37, 94,
235–236, 248
Implicational hierarchy  64
Indian Spanish  50, 69
Indirect object  27–28,
33–34, 285–286, 291–295,
297–299, 302–304, 331,
335, 350
Individual-Level  311–312
Predicate  146
see also Stage-Level
Predicate
Interface  21, 23–24, 26, 31,
35, 37, 39, 41–43, 148,
161, 169–170, 174–176,
195, 202–203, 225, 227,
309–310, 320, 334,
350–351, 353–354
Interface Hypothesis  169, 202,
309–310, 320
Interlanguage Hypothesis 
330, 333
Interpolation  337
Italian  112, 128, 160, 171,
174–176, 202, 204–208,
214, 223, 225, 332, 339,
350–351
Island  262–264, 266–268,
270–272, 274–275
 Index

J N
Japanese  75–76, 331–332
L 63, 67–68, 73
L2 acquisition  24, 59, 113, 170,
241, 257–259, 261, 277–278,
282–283, 289–290,
309–310, 313–315, 329–330,
334, 338, 346, 348
Language contact  22, 24, 37,
106, 108, 113–115, 283
Language identity  107
Language Specific Grammar
Hypothesis  76
Latent Linguistic
Structure  330
Latent Psychological
Structure  330
Lexicon  21–24, 26, 30, 37, 39,
41–43, 58, 68, 139, 281,
336–337, 345
Lexico-morphosyntactic
interface  23
Liquid  80–81, 85–88, 98
Locative  34, 81, 105, 107–118,
120–123, 126–127, 129, 141,
145, 152, 239–240, 243, 285,
313, 315
expression  34, 120, 141, 243
M Predicative adjective  135, 153,
Macroparameter  54, 331, 333
Manner  37, 117, 154, 233–235,
237–241, 243–247,
249–250, 277, 283
Mapping  21, 23–27, 30–31,
33–35, 37–43, 56, 58, 69,
108, 116, 128, 259–260, 316,
346, 351
Mapudungun  56
Microparameter  341, 353
Microvariation  201, 206,
210, 226
Minimal Word  75
Minimalist Program  54, 330
Missing Surface Inflection
Hypothesis  69, 116, 314
Modularity of language  22
Mood  52, 55, 171, 281
Morphological coda  78, 82,
94–95, 99
Motion event  233–237,
239–242, 247, 250
N-N compound  342–344
Nahuatl  49–52, 54–57, 59–61,
Nasal  77, 80–81, 85–89, 98,
100–101
Near-native  169–170, 174–178,
193, 195
Nominative  286, 303
Norwegian  76, 78, 235–236
Nucleus  75, 79, 285
Null
object  21, 39–40, 42, 264,
266, 268–270, 272, 274
pronoun  203–205, 209,
217–219, 221–224, 285, 350
subject  26–27, 53, 171–173,
175, 180–184, 189–191,
195, 202–203, 205–206,
208–209, 214–215, 220,
223–224, 226, 332–333,
339–341, 351–353
Number  26–28, 30, 52–55,
63–65, 67, 69, 83–86, 95–96,
135, 148, 150, 152, 158–159,
163, 171, 182–183, 185, 193,
206, 210, 213, 228, 234–235,
240, 246–247, 261, 264, 269,
271, 281, 290, 297, 310, 314,
319, 332, 336, 344, 346
O 160–161
Otomí  50, 56
Oblique  28, 32–34, 37, 42, 108
Object
agreement  54–56, 59
clitic  56, 66, 261–264,
270–274, 276, 335 see also
clitic
drop  54, 110, 257, 260,
263–264, 268, 270,
272–274, 276–277 see also
null object
Old Spanish  336–338
Ongoingness  236, 248–250
Onset  60–61, 79, 83, 98, 215,
227, 285
Optionality  50, 110, 202,
224, 226, 300, 309, 329,
334–335, 339, 342, 353
Overt pronoun  203–208, 210,
215, 217–221, 223–224,
350–352
P
Past tense  23, 28, 30, 35, 54–55,
64, 81
Path  110, 140, 148, 163, 234–235,
237–240, 246, 249, 277
Perfectivity  30
Permanent property  137–138,
141, 145–146, 152, 155, 157,
160–161, 312, 324
Person  26–28, 30, 33–34, 40,
52–56, 59, 62–66, 68,
81–82, 94, 99, 107, 110, 118,
150, 171–172, 174, 178–179,
201, 204, 206–207, 210,
228, 236, 242, 261, 264,
266, 271, 285, 311, 335, 339,
342, 347
Person-Case constraint
(PCC)  347
Phonological
process  75, 100–101
system  75–76, 100–1
Pluperfect past tense  23
Plural marking  77, 346, 353
Polysynthetic  54, 56, 330
Position of Antecedent
Hypothesis  204, 350
Praat  83
Pre-emption  257, 259–260,
276–277
Predorsal  77, 80–81
Preposition
a  28, 32, 110, 127, 281–282,
284, 294–295, 298, 349
de  32–33, 346
Principles and Parameters  53,
330
pro  52–54, 56, 171–173, 178,
203–205, 209, 211, 217,
266, 270, 272–274,
285–286, 304
pro-drop  52, 56, 209, 286,
304
Processing  22, 169–170, 174,
176–177, 180, 193–195, 202,
204, 224–225, 268, 281,
309, 329, 334, 347,
350–352, 354
load  309
Production of morphology  50,
68–69
 Index
Progressive  38, 53, 113, 140,
235–236, 243, 248, 313
aspect  235–236
Q 261–262, 270–271, 276,
Quantifier  284, 341, 345–347,
353
Quechua  21, 23–43, 56,
108–109, 117, 127
R 170, 206, 215, 219, 222, 225,
Resyllabification  79, 83
Rhyme  79, 81
Romanian  115–116, 282, 302,
347–348
Russian  116, 128, 175, 235, 282,
332
S Stop  77, 80–81, 85–88, 98, 107
Satellite-framed language  Stress  75–78, 80, 82, 90–94, 99,
237, 249
Scalar gradable adjective  Stressed  77–78, 80, 82, 85,
309–310, 317, 321, 323–324
Irreversible scalar
gradable  310, 317, 323–324
Non-scalar gradable  321,
323–324
Reversible scalar
gradable  323–324
Sein  57, 142–144, 161
Selective access  310
Non-selective access  24
Ser  57, 105–110, 112–121,
124–127, 129, 137–140, 142,
144–146, 149, 151–155,
157, 160–162, 179, 181, 191,
309–316, 318–325
Ser/estar + adjective  309
Sibilant  75, 77, 80, 85–86, 88,
98–101, 107
Spanish  21, 23–43, 49–64,
66–69, 73, 75–101,
105–115, 117–123, 125–129,
133, 135–142, 144–163,
169–178, 185–186, 193–194,
201–206, 208–210, 214–
216, 218, 220, 222–229,
233–235, 238–250, 255,
257, 260–277, 281–292,
295–296, 298, 300,
302–304, 309–321,
323–325, 329–354
Spec
CP  147
IP  147, 350
Specificity  27–28, 128,
282–288, 290, 300,
302–304
Non-specific  270, 284
Specific  24, 28, 42, 75–77,
81, 85, 87, 99, 100, 128, 137,
227, 235, 237, 247, 258, 261,
262, 269, 277, 281, 282,
284, 285, 287–291, 293,
300, 302, 304, 309, 315–317,
319, 320, 324, 329, 341,
344–346, 349, 352–353
Stage-Level Predicate  144, 146
101, 249
89–93, 315
Unstressed  77–78, 80, 82,
85, 90–93, 95, 99
Subject
agreement  26, 52, 55–56
overt  171–175, 180–184, 189,
205, 208, 341, 351–352
pronoun  54, 169, 170–171,
174–175, 179, 185, 201–203,
205, 208–210, 214, 228,
340–341, 351–352
Subject-verb inversion  Verb root  55, 238
340–341
Subset Principle  259, 265
Swedish  233–236, 238–239,
241–250, 282
Syntactic complexity  151, 160
Syntax-pragmatics
interface  23
Syntax-before-discourse
hypothesis  208
T 304, 331, 335, 344
TAM  52–53
Temporal property  137–138,
143–146, 151, 154–157,
160–161
Tense  23, 25, 28, 30, 35, 52–57, 59,
62–64, 67–68, 81, 110, 115,
138, 171, 206, 248–249, 281
That-trace effect  332, 340–341,
352–353
Topic  23–24, 30, 40, 54, 106,
169, 171–175, 177–185, 190,
192–195, 202–209, 213,
224, 227–228, 248, 269,
287, 312, 350
Topic Shift  9, 10, 169,
171–175, 177–181, 184, 185,
190, 193–195, 203, 205, 208,
209, 227
Topic Maintenance  9, 10,
169,171, 173, 174, 178–180,
182–185, 192, 194, 195
Turkish  116, 128, 176, 208,
226, 281–284, 286–288,
290–293, 296–298,
300–304, 349
Typological
proximity  329–336, 339–345,
347–349, 352–353
similarity  329, 331–335,
339–341, 344, 347–349,
352–353
U
Universals  329, 331–333
Universal Base Hypothesis  147
Universal Grammar  58,
258, 330
Hypothesis  76, 81
V
Verb-framed language  237
Vibrant  87–89, 98
Visibility  309
Voiceless  77, 82, 86–87, 98,
100, 139
W
Weak crossover effect  41, 42
Word order  27, 30, 41, 54, 56,
159, 175, 206, 286–288,
Word length  90, 92
Worden  309–310, 315–316, 321,
323–324
Z
Zijn  309–310, 315–316, 321,
323–324