Hispanic Contact Linguistics

Theoretical, methodological and empirical

perspectives

Luis A. Ortiz-López
Rosa E. Guzzardo Tamargo

Melvin González-Rivera

doi: 10.1075/ihll.22
ISBN: 978 90 272 6171 7 (ebook)
Cataloging-in-Publication Data available from Library of Congress:
LCCN 2019044649
© 2020 – 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 Company · [Link]

John Benjamins Publishing Company

Amsterdam/Philadelphia
Table of contents

Acknowledgments VII
Introduction 1
Luis A. Ortiz López, Rosa Guzzardo and Melvin González-Rivera
Theoretical and methodological approaches
Chapter 1. The New Spanishes in the context of contact linguistics: Toward a
unified approach 11
Donald Winford
Chapter 2. Chocó Spanish: An Afro-Hispanic language on the Spanish
frontier 43
Sandro Sessarego
Chapter 3. Methodological considerations in heritage language studies: A
comparison of sociolinguistic and acquisition-based tasks 61
Zoe McManmon
Phonetics, phonology, prosody
Chapter 4. Social change and /s/ variation in Concepción, Chile and Lima,
Peru: The role of dialect and sociolectal contact 85
Brandon M. A. Rogers and Carol A. Klee
Chapter 5. The acento pujado in Yucatan Spanish: Prosodic rhythm and the
search for the yucateco accent 115
Jim Michnowicz and Alex Hyler
Morphology
Chapter 6. First person singular subject expression in Caribbean heritage
speaker Spanish oral production 139
Ana de Prada Pérez
Chapter 7. Use of the present perfect indicative in New York Dominican
 Spanish 163
Cecily Corbett, Juanita Reyes and Lotfi Sayahi
Chapter 8. Transfer and convergence between Catalan and Spanish in a
bilingual setting 179
Amelia Jiménez-Gaspar, Acrisio Pires and Pedro Guijarro-Fuentes
Syntax
Chapter 9. The distribution and use of present and past progressive forms in
Spanish-English and Spanish-Brazilian Portuguese bilinguals 215
Julio César López Otero and Alejandro Cuza
Chapter 10. Portuguese-Spanish contacts in Misiones, Argentina: Probing
(for) code-switching constraints 235
John M. Lipski
Language variation, linguistic perceptions and attitudes
Chapter 11. Real perception or perceptive accommodation?: The Dominirican
ethnic-dialect continuum and sociolinguistic context 263
Luis A. Ortiz López and Cristina Martínez Pedraza
Chapter 12. Andean Spanish and Provinciano identity: Language attitudes
and linguistic ideologies towards migrants in Lima, Perú 283
Daniela Salcedo Arnaiz
Chapter 13. On the effects of Catalan contact in the variable expression of
Spanish future tense: A contrastive study of Alcalá de Henares, Madrid and
Palma, Majorca 315
Andrés Enrique Arias and Beatriz Méndez Guerrero
Index 335
Acknowledgments

This volume would not have been possible without the support and collaboration
of many colleagues who served as reviewers, including Héctor Aponte
(University of Puerto Rico, Río Piedras), Osmer Balam (College of Wooster),
José Luis Blas Arroyo (University Jaume 1), Anna Bartra (Universitat
Autonoma of Barcelona), Mary E. Beaton (Denison University), Rebeka
Campos-Astorkiza (Ohio State University), Ana María Carvalho (University of
Arizona), Alicia Cipria (University of Alabama), Gibran Delgado (University of
Puerto Rico, Mayagüez), Ana de Prada Pérez (National University of Ireland,
Maynooth), Jenny Dumont (Gettysburg College), Iraida Galarza (University of
Puerto Rico, Mayagüez), Timothy Gupton (University of Georgia), Chad Howe
(University of Georgia), Matthew Kanwit (University of Pittsburgh), Carol Klee
(University of Minnesota), Gillian Lord (University of Florida), Patrick Mather
(University of Puerto Rico, Río Piedras), Antje Muntendam (Florida State
University), Yolanda Rivera (University of Puerto Rico, Río Piedras), Aaron
Roggia (Oklahoma State University), Liliana Sánchez (Rutgers University),
Hiram Smith (University of Wisconsin, Madison), Eva-María Suárez
Büdenbender (Shepherd University), (Jorge Valdés Kroff (University of Florida),
Don Walicek (University of Puerto Rico, Río Piedras), and Eva Michelle
Wheeler (Oakwood University). To all of them we are extremely grateful, as
well as to the authors whose work appears in this volume.
We also owe the completion of this book to a number of people who
supported us in a variety of ways during its preparation: to the administrative
and academic staff at the University of Puerto, Río Piedras for providing us with
a great academic environment to research Spanish sociolinguistics from a variety
of disciplines and points of view, especially Carmen Pérez and Bárbara Santana;
to our colleagues from the University of Puerto Rico, particularly Rose Marie
Santiago Villafañe, Amárilis Torres, Héctor Aponte, and Alexandra Morales; to
the graduate students who helped us during this process, including Hernán
Rosario, Cristina Martínez, Jonathan Cruz, and Jessica Vélez. We are deeply
appreciative of Cristina Maymí for all her editorial assistance. Finally, we would
like to express our profound gratitude to the series editors, Kimberly L. Geeslin
and Jennifer Cabrelli Amaro, and to the editorial staff at John Benjamins
Publishing Company for their assistance and support.
Hispanic contact linguistics
Theoretical, methodological, and empirical
perspectives
Luis A. Ortiz López, Rosa E. Guzzardo Tamargo &
Melvin González-Rivera
Universidad de Puerto Rico-Río PiedrasUniversity of
Puerto Rico-Mayagüez

This volume brings together current research in the field of Hispanic Contact
Linguistics that contributes to proposed sociolinguistic and variationist theories
on language contact and change. Each of the chapters included provides an in-
depth linguistic analysis of the diverse phenomena presented. The chapters
comprise a wide scope of language situations in which Spanish is in contact with
other languages (e.g., Catalan, English, Brazilian Portuguese, Quechua,
Aymara), as well as varied participant groups, ranging from second-language
learners and heritage speakers to more balanced bilinguals and code-switchers
(Guzzardo Tamargo & Dusisas, 2013; Guzzardo Tamargo, Valdés Kroff &
Dussias, 2016). A large breadth of geographical areas is also considered in the
chapters, including the United States, Mexico, Puerto Rico, Colombia, Brazil,
Peru, Argentina, and Spain. Taken together, the chapters provide rich empirical
descriptions of data pertaining to different levels of language, diverse –
naturalistic and experimental – methodological approaches to data collection, as
well as theoretical implications of the findings. The interdisciplinary perspective
adopted by the authors proves to be fruitful to the analysis of linguistic data in
contact situations and provides important insights into theoretical linguistics in
general, as well as into theories of bilingualism and second language acquisition.
The thirteen papers contained in the volume are organized around five thematic
sections, as follows: (i) Theoretical and Methodological Approaches, which
includes three chapters; (ii) Phonetics, Phonology, and Prosody, comprising two
chapters; (iii) Morphology and (iv) Syntax, each consisting of three chapters; and
(v) Language variation, linguistic perceptions and attitudes, that contains two
chapters.
The first section begins with Donald Winford’s Chapter 1, in which the
author focuses on the set of contact languages that arose in European colonies in
Africa, Asia, the Caribbean, the Pacific and elsewhere due to contact between
European and various indigenous languages. Winford offers a unified account of
these contact languages by assuming a comprehensive model of language
contact that establishes links between linguistic, sociolinguistic, and
psycholinguistic approaches to contact-induced change. The author argues that
an integration of all three components is possible, and even necessary, to reach a
complete account of linguistic change triggered by contact. He offers a
preliminary account of how this unified approach can be applied to the
emergence of Iberian-lexified contact languages.
Chapter 2 comprises Sandro Sessarego’s inspection of the linguistic and
socio-historical evidence available for Chocó Spanish, an Afro-Hispanic
vernacular spoken in the Colombian Pacific lowlands. Findings indicate that the
long-assumed creolizing conditions for Chocó Spanish might not have been in
place in colonial Chocó and that the grammar of this language can be better
analyzed as the result of intermediate and advanced second language acquisition
processes, which do not necessarily imply a previous creole stage (cf. Ortiz
López, 1998, 2010). Sessarego also provides an analysis of the evolution of
Chocó Spanish in relation to the recently-proposed Legal Hypothesis of Creole
Genesis (Sessarego, 2015, in press). In so doing, the author’s work examines the
extent to which such a hypothesis makes valid predictions for a language like
Chocó Spanish, which developed in a region described by many as ‘remote’ and
‘on the frontier’ (cf., Whitten, 1974; Sharp, 1976), hence, far away from legal
courts, in areas where law was not likely to be properly enforced.
In Chapter 3, Zoe McManmon examines the differences found in linguistic
accuracy of a group of Mexican children who are heritage Spanish speakers. The
author measures accuracy across linguistic tasks involving sociolinguistic
interviews as well as an acquisition-based task (story retelling), by invoking the
work of Beebe (1987) on style-shifting in Sociolinguistic Variation and Second
Language Acquisition. She also refers to research on pronouns by Montrul and
Sánchez Walker (2013), which uses formal tasks, work by Shin and Van Buren
(2016) that employs sociolinguistic tasks, as well as other studies on gender
agreement in child bilinguals (Montrul & Potowski, 2007) in order to explore the
ways in which different linguistic methods may affect target language accuracy
and the ways in which style shifting influences heritage speakers’ performance.
In the next section, Chapter 4, by Brandon Rogers and Carol Klee, focuses
on -/s/ variation in Lima, a city that has experienced intense migration from the
Peruvian provinces over the past sixty years, and which presents a particularly
interesting case of dialect/language contact, given that many migrants to the city
are from the Andean region of the country, where indigenous languages
(Quechua and Aymara) are spoken by large sectors of the population and where
the regional dialect, Andean Spanish, has been influenced by Quechua and
Aymara. Using Praat (Boersma & Weenink, 2015), the authors examine the
word- and syllable-final sibilant productions of 26 first-, second-, and third-
generation speakers in the city of Los Olivos, the most affluent migrant area in
the Northern part of metropolitan Lima, and categorize them as [s], [h], or [0].
Their results indicate that the strongest predictor of -/s/ variation is immigrant
generation. While [s] is the overall preferred pronunciation for all speakers, the
probability that speakers will produce a weakened variant (i.e., [h] or [0])
increases significantly with each subsequent generation of speakers. These
findings indicate that speakers are gradually assimilating to the Limeño norm
and distancing themselves from their parents’ and grandparents’ way of
speaking, as has been the case in other immigrant communities (Labov, 2014).
In Chapter 5, Jim Michnowicz and Alex Hyler analyze prosodic features in
Yucatan Spanish, a distinctive regional variety of Spanish in contact with an
indigenous language, Yucatec Maya, in Mexico. By means of sociolinguistic
interviews, the authors collected production samples from ten Yucatan Spanish
monolinguals and ten Maya-Yucatan Spanish bilinguals, mostly balanced for sex
and age. Their findings demonstrate a changing pattern of rhythmic timing in
Yucatan Spanish that parallels those of previous studies on segmental features:
younger monolingual speakers of Yucatan Spanish use prosodic cues to
distinguish themselves linguistically from older generations and Maya-Yucatan
Spanish bilinguals, who speak a more ‘traditional,’ Maya influenced variety.
The third section starts off with Chapter 6, by Ana de Prada Pérez,
addresses the inconsistent results regarding the contact effects on subject
pronoun expression that have arisen in different regions of the United States (cf.
Aponte Alequín & Ortiz López, 2015; Ortiz López & Aponte Alequín 2018). To
contribute to the debate with additional data collected by means of
sociolinguistic interviews, the author examines the production of first-person
singular pronouns in the speech of Caribbean Spanish (Cuban, Puerto Rican,
Dominican, and Venezuelan) heritage speakers in Florida, belonging to different
generations and proficiency levels. In addition to examining the rates of overt
pronominal subjects, de Prada Pérez considers several linguistic variables:
subject form, switch reference, tense, aspect, mood continuity, verb ambiguity,
and semantic verb type. Although the results display Spanish proficiency effects,
they lack any contact effects. Therefore, instead of coinciding with findings from
New York City, which has a similar dialectal distribution as that present among
her participants, the results are comparable to those of previous studies with non-
Caribbean Spanish-speaking population conducted in the U.S. Southwest.
In Chapter 7, Cecily Corbett, Juanita Reyes, and Lotfi Sayahi, examine the
usage of the present perfect tense in the speech of 16 Dominican Spanish
speakers who were born in New York or who arrived in the area before the age
of 12 (NY Dominicans). The authors also include sociolinguistic interviews with
four older Dominican immigrant speakers and the same number of monolingual
speakers from El Cibao region in the Dominican Republic. Since the present
perfect in Spanish bears strong similarities to the same tense in English, the
paper aims to assess its maintenance in the speech of Spanish-English bilinguals,
and to test the hypothesis that perfect tenses tend to be reduced in cases of
language contact (Silva-Corvalán, 2014). This study is compared to those of
Howe and Schwenter (2003, 2008) regarding the extension in domains of use of
the present perfect in Latin American Spanish to test whether Dominican
Spanish, both in New York and in the Dominican Republic, is used in non-
sequential perfective discourse. The authors findings exhibit the normative use
of the present perfect by New York Dominican Spanish speakers.
Chapter 8, by Amelia Jiménez, Pedro Guijarro-Fuentes, and Acrisio Pires,
constitutes an analysis of the clitic system properties of Catalan and Spanish as
they are spoken in Majorca, to examine the patterns of clitic realization that
bilingual Catalan-Spanish speakers use in different language contact situations.
The authors collected spontaneous speech data from 45 participants by means of
a production task in both Catalan and Spanish. Their findings indicate that
Majorcan Spanish does not differ substantially from Peninsular Spanish;
however, Majorcan Catalan exhibits specific patterns which are clearly distinct
from Peninsular Catalan. The authors try to determine whether there is evidence
for transfer or convergence between Catalan and Spanish in the bilingual
Catalan-Spanish context of Majorca. Although they find cases of semantic
extension in Catalan, possibly influenced by Spanish, no substantial convergence
between the two languages arises.
In Chapter 9, Julio César López Otero and Alejandro Cuza examine the use
and distribution of present and past progressive forms in two groups of Spanish-
English and Spanish-Brazilian Portuguese simultaneous bilinguals: nine heritage
speakers of Spanish, born and raised in the United States (with English as their
dominant language), and 15 heritage speakers of Spanish, born and raised in
Brazil (with Brazilian Portuguese as their dominant language). The inspection of
these language pairs sparks interest because of the constraints of English,
Brazilian Portuguese, and Spanish regarding the use of the present progressive
and the simple present (cf. Aponte Alequín & Ortiz López, 2010; González
Rivera, 2014, 2015). In Spanish, for example, both the simple present and the
progressive form can select an ongoing meaning, but this is not the case in
Brazilian Portuguese nor English, where only the present progressive is used for
ongoing events. In addition, all three languages have a past progressive form, but
Spanish and Brazilian Portuguese, typologically similar languages with a very
close verbal morphology, also instantiate imperfect morphology, something
unallowed in English. These differences may create a learnability problem for
English-speaking heritage speakers of Spanish who must expand their grammar
and recognize the possibility of the imperfect form expressing a continuous
event in the past, and of the simple present tense selecting an ongoing meaning
in the present. The authors interpret the results in terms of crosslinguistic effects
from the speakers’ dominant language.
In the next section, Chapter 10, by John M. Lipski, describes the
Portuguese-Spanish code-switching practices of second-language learners and
balanced bilinguals living along the Brazilian border as well as those of fluent
bilinguals in Misiones, Argentina. The data comprise spontaneous speech
production and more controlled production provided through interactive tasks
(speeded translation, language classification, and memory-loaded repetition).
Although the results show that the presence of bilingual settings with structurally
and lexically congruent languages does not always result in code-switching,
when individuals in these environments do engage in this bilingual practice, they
follow constraints, as has been discussed in previous work with habitual code-
switchers in other bilingual communities (Guzzardo Tamargo, Valdés Kroff, &
Dussias, 2016; Valdés Kroff, Guzzardo Tamargo, & Dussias, 2018). In this
study, the use of code-switching constraints even arises among individuals who
do not frequently code-switch. Moreover, the use of intra-sentential code-
switching is not affected by the speakers’ level of bilingualism, but fluent
bilinguals obey the proposed constraints more consistently than second-language
learners.
The final section starts with Chapter 11, in which Luis [Link] López and
Cristina Martínez Pedraza use a verbal guise test to study the perceptions that
Puerto Ricans display towards samples of Dominican Spanish with different
degrees of contact with Puerto Rican Spanish, in terms of social class,
educational level, skin color, reliability, intelligence, affability, and correctness.
The results show that the perceptions, attitudes, and stereotypes are correlated
with the characteristics of the stimuli and of the participants, mainly with respect
to the amount of years that the Dominican Spanish speakers have been in contact
with Puerto Rican society and with the participants’ perception of nationality.
The authors’ findings provide evidence of individuals’ ability to distinguish
dialects as well as of the process of de-dialectalization of immigrant speech. The
verbal guise technique proves useful in the identification of the level of
acceptance or rejection, of positive and negative attitudes, and of the stereotypes
of immigrants and their descendants regarding their own dialect.
In Chapter 12, Daniela Salcedo Arnaiz studies the reactions that both
Limeños and Andean migrants have towards a selection of (stigmatized)
morphosyntactic features from Andean Spanish. By using the matched-guise
technique and a questionnaire, the author collects data in two complementary
fieldwork projects and shows that there is no dialect awareness towards Andean
Spanish. When participants are confronted with the selected morphosyntactic
features of that dialect, they do not associate them with a dialect per se, but with
‘incorrect’ Spanish spoken mostly by Andean migrants, who allegedly share
stereotypical characteristics, such as a low level of education, indigenous race,
and poverty. Furthermore, Salcedo finds that these stigmatized morphosyntactic
features are not found to index Andeanness when they are considered in
isolation, but only when they are taken together.
In the last chapter of the volume, Andrés Enrique-Arias and Beatriz
Méndez Guerrero, examine the effect of Catalan-Spanish contact on inhibiting
the spread of the Spanish periphrastic future. The authors aim to explore the
influence of Catalan bilingualism on the variable expression of futurity in
Spanish through a contrastive study of production data collected by means of
sociolinguistic interviews in a Spanish monolingual community (Alcalá, Madrid)
and a bilingual one (Palma, Majorca). This study centers around two internal
factor groups explored in previous studies of the expression of futurity (temporal
distance and sentence modality) as well as gender, age, and language dominance.
The results exhibit an increased use of morphological future among the educated
bilinguals, although the expression of futurity in both communities seems to be
similarly influenced by the linguistic factors examined. In the discussion of their
findings, the authors consider the possibility of a general change in progress in
which the morphological future is being replaced by the periphrastic future.
Some key points emerge from the joint consideration of the studies
presented in the chapters described above. First, several studies exhibit an
upsurge of innovative and distinctive speech among younger generations of
speakers, which can potentially lead to more general changes in language
practices. In addition, various studies discussed in this volume bring to the
picture new data from different language communities, such as the use of code-
switching types by bilinguals who are not normally associated with them, and, in
so doing, they challenge previously proposed hypotheses and contribute to
existing debates with conflicting evidence. While some of the studies display
expected crosslinguistic effects because of bilingualism and language contact,
others present a lack of contact effects and the absence of convergence.
Moreover, the chapters in this volume are inclusive, in that they not only discuss
language contact, but also focus on situations of dialect contact, in which
speakers’ present an ability to distinguish dialects and modify their own and use
different linguistic traits to characterize dialectal identities in complex and
intricate ways. Finally, these chapters reveal the importance of using
methodological techniques that are innovative in certain language communities,
of carefully examining one phenomenon using different methodologies in search
of converging evidence, and of bringing together different approaches in order to
propose wide-ranging models of language contact.
The chapters included in this volume constitute a selection of papers by
well-known scholars and promising researchers, presented at the 8th
International Workshop on Spanish Sociolinguistics (WSS8), which was held at
the University of Puerto Rico, Río Piedras campus on April 13th–16th, 2016, as
well as papers by distinguished academics in the field of Hispanic Linguistics
who kindly provided their recent work on the topic. We are very honored to have
had the opportunity to collaborate with all of them, as the collective product of
their contributions represents a state-of-the-art perspective of the field of
Hispanic Contact Linguistics.
References
Aponte Alequín, H. & Ortiz López, L. A.
(2010.) Una perspectiva pragmática del presente progresivo con valor de futuro en el español del Caribe
C. Borgonovo, M. Español-Echevarría, & P. Prévost (Eds.), Selected Proceedings of the Hispanic
 Linguistics Symposium (pp. 109–121). Somerville, MA: Cascadilla. ①
Aponte Alequín, H. & Ortiz López, L. A.
(2015.) Variación dialectal e interfaz sintáctica/semántica/pragmática: la anteposición de sujetos en
cláusulas subordinadas en infinitivo. Spanish in Context, 2(3): 396–418. ①
Beebe, L. M.
(1987.) Sociolinguistic variation and style shifting in second language acquisition. In G. Ioup & S. H.
Weinberger (Eds.), Interlanguage phonology: The acquisition of a second language sound system (pp.
378–388). New York, NY: Newbury House. ①
Boersma, P., & Weenink, D.
(2015.) Praat: Doing phonetics by computer [Computer program]. Retrieved from [Link]
González-Rivera, M.
(2015.) El gerundio en el español portocruceño. Muenchen: Lincom Europa. ①
González-Rivera, M.
(2014.) Puerto Rican/Porto-Crucian Spanish and the -ndo morpheme. International Journal of the
Linguistic Association of the Southwest, 33(1), 39–59. ①
Guzzardo Tamargo, R. E., & Dussias, P. E.
(2016.) Comprehension patterns of two groups of Spanish-English bilingual codeswitchers. In R. E.
Guzzardo Tamargo, C. M. Mazak, & M. C. Parafita Couto (Eds.), Spanish-English Codeswitching in
the Caribbean and the US (pp. 301–322). Amsterdam: John Benjamins. ①
Guzzardo Tamargo, R. E., & Dussias, P. E.
(2013.) Processing of Spanish-English code-switches by late bilinguals. In Proceedings of the Annual
Boston University Conference on Language Development. Boston University Conference on Language
Development, 37, 134–146. ①
Guzzardo Tamargo, R. E., Valdés Kroff, J. R., & Dussias, P. E.
(2016.) Examining the relationship between comprehension and production processes in code-switched
language. Journal of Memory and Language, 89, 138–161. ①
Howe, C., & Schwenter, S.
(2003.) Present perfect for preterit across Spanish dialects. University of Pennsylvania Working Papers in
Linguistics, 9(2), 61–75. ①
Howe, C., & Schwenter, S.
(2008.) Variable constraints on past reference in dialects of Spanish. In M. Westmoreland & J. A. Thomas
(Eds.), Selected proceedings of the Fourth Workshop on Spanish Sociolinguistics (pp. 100–108).
Somerville, MA: Cascadilla Proceedings Project. ①
Labov, W.
(2014.) The sociophonetic orientation of the language learner. In C. Celata & S. Calamai (Eds.), Advances
in sociophonetics (pp. 17–28). Amsterdam: John Benjamins. ①
Montrul, S., & Potowski, K.
(2007.) Command of gender agreement in school-age Spanish English bilingual children. International
Journal of Bilingualism, 11(3), 301–328. ①
Montrul, S., & Sánchez-Walker, N.
(2013.) Differential object marking in child and adult Spanish heritage speakers. Language Acquisition
20(2), 109–132. ①
Ortiz López, L. A.
(1998.) Huellas etno-sociolingüísticas bozales y afrocubanas. Madrid: Iberoamericana. ①②
Ortiz López, L. A.
(2010.) El español y el criollo haitiano: contacto lingüístico y adquisición de segunda lengua. Madrid:
Iberoamericana.
Ortiz López, L. A. & Aponte Alequín, H.
(2018.) Microvariación sintáctico-semántica-pragmática: El infinitivo frente al subjuntivo en el español
(caribeño y mexicano). Studies in Hispanic and Lusophone Studies, 11(1), 131–162. ①
Sessarego, S.
(2015.) Afro-Peruvian Spanish: Spanish slavery and the legacy of Spanish Creoles. Amsterdam: John
Benjamins. ①
Sessarego, S.
(in press). Language contact and the making of an Afro-Hispanic vernacular: Variation and change in the
Colombian Chocó. Cambridge: Cambridge University. ①
Sharp, F.
(1976.) Slavery on the Spanish frontier. Norman, OK: University of Oklahoma. ①
Shin, N. L., & Van Buren, J.
(2016.) Maintenance of Spanish subject pronoun expression patterns among bilingual children of
farmworkers in Washington/Montana. Spanish in Context, 13(2), 173–194. ①
Silva-Corvalán, C.
(2014.) Bilingual language acquisition: Spanish and English in the first six years. Cambridge: Cambridge
University. ①
Valdés Kroff, J. R., Guzzardo Tamargo, R. E. & Dussias, P. E.
(2018.) Experimental contributions of eye-tracking to the understanding of comprehension processes while
hearing and reading code-switches. In A. Munarriz-Ibarrola, M. C. Parafita Couto, & E. Wyngaerd
(Eds.), Methodologies for intra-sentential code-switching research (pp. 98–133). Amsterdam: John
Benjamins. ①
Whitten, N.
(1974.) Black frontiersmen. New York, NY: Wiley. ①
Theoretical and methodological approaches
CHAPTER 1

The New Spanishes in the context of

contact linguistics
Toward a unified approach
Donald Winford
The Ohio State University

Abstract
Weinreich (1953) argued that a comprehensive model of language
contact must integrate linguistic, sociolinguistic and psycholinguistic
approaches. This paper discusses how such a model can be applied to
the contact languages that arose in the Spanish colonies as a result of
contact between Spanish and various other languages. These contact
languages include close approximations to Peninsular Spanish, Afro-
Hispanic varieties, “indigenized” varieties and of course creoles. I
argue that all these contact varieties should be amenable to description
and analysis within the same framework. To this end, I discuss the
ways in which Weinreich’s model of an integrated approach can be
linked to a naturalistic second language acquisition (NSLA) account of
the formation of the contact varieties of Spanish.

Keywords
Spanish contact varieties; naturalistic SLA; social ecology;
language processing; imposition
1. Introduction
Since the pioneering work of Weinreich (1953), researchers over the last sixty
years or so have repeatedly pointed to the need for a unified framework for
language contact phenomena – one that would allow us to explain language
contact phenomena in terms of three types of factor:
social factors and motivations
structural factors and linguistic constraints
psycholinguistic factors involved in language processing

In other words, a unified framework must address the sociolinguistic, linguistic

and psycholinguistic aspects of contact-induced change and contact language
formation. The sociolinguistic component focuses on the socio-historical
contexts of language contact, including the social networks and broader social
structure within which changes originate, are propagated, and become
conventionalized as part of a new communal language. Linguistic approaches
focus primarily on the structural aspects of language contact and try to explain
“the forms of mutual interference of languages that are in contact” (Weinreich,
1953). The psycholinguistic component addresses how the mechanisms or
processes of change involved can be explained in terms of the ways in which
linguistic systems or inputs interact in the individual mind. A combined
examination of each of these aspects, and of their relationships to one another,
offers us the most comprehensive explanation of the processes of contact
language formation (See Winford, 2017 for further discussion). Such an
integrated framework has so far eluded scholars in Contact Linguistics as a
whole, and particularly in the study of contact languages such as indigenized
languages and creoles. Ultimately, a unified framework should not be confined
to explaining the origins of contact languages but should extend to all cases of
language acquisition and language change.
Weinreich’s framework was of course meant to cover a broad range of what
he called “interference phenomena” resulting from different kinds of transfer
across languages. He made a distinction between interference phenomena due to
“an outright transfer of elements” (1953, p. 7), and those due to transfer of a
“structural” type that involve some “rearrangement of patterns” (1953, p. 1).
Building on this pioneering work, a variety of competing frameworks for the
study of contact phenomena have been proposed, for instance by Johanson
(2002), Matras (2009), Thomason & Kaufman (1988), and van Coetsem (1988).
In addition, since Weinreich’s pioneering work, we have gained a fuller
understanding of the full range of language contact phenomena, the mechanisms
underlying them, and the diversity of contact languages (see for example, the
various anthologies devoted to language contact, including Bakker & Matras
(2013), Filpulla et al. (2017), Hickey (2010), Siemund & Kintana (2008), etc.).
A truly unified framework of the type that Weinreich had in mind, should of
course address the full range of outcomes of language contact, whatever the
mechanisms that created them. Such an ambitious goal is clearly beyond the
scope of the present paper. Instead, I am concerned with only one specific
category of contact languages – those that arose through processes of naturalistic
SLA. More specifically, I focus my attention here on the contact varieties of
Spanish that arose in colonial settings via such processes. These include not just
the varieties of Spanish that are close approximations to Peninsular Spanish, but
also Afro-Hispanic varieties (e.g. Bozal Spanish, Afro-Bolivian Spanish, etc.),
and “indigenized” varieties such as Andean Spanish, Amazonian Spanish.1 They
also include, of course, creoles such as Palenquero, Chabacano and Papiamentu.2
I argue that all these contact varieties, as well as those related to other European
languages, should be amenable to description and analysis within the same
framework.
 My objective is to discuss the ways in which Weinreich’s model of an
integrated approach including linguistic, sociolinguistic and psycholinguistic
components, can be linked to a naturalistic SLA account of the formation of the
contact varieties of Spanish. The idea that creoles and related outcomes of
language contact such as indigenized varieties originate from processes of
naturalistic SLA is, of course not new. As early as the nineteenth century,
Schuchardt (1882) and other pioneers of creole studies such as Coelho (1880–
1886) and Hesseling (1897) explored issues such as the relationship between
creole formation and first and second language acquisition. This relationship has
been explored in numerous studies since then, from Schumann (1978) and
Andersen (1980, 1983), to more recent works by Kouwenberg (2006), Mather
(2006), Siegel (2006), and others in Lefebvre et al. (2006), among many others.
Both Lipski (1994) and Schwegler (2010) emphasize the need to integrate
the study of Hispanic contact languages into the field of Hispanic Linguistics
more generally. Lipski’s earlier research (1994, 2001, 2005) was particularly
instrumental in drawing attention to these neglected offspring of Spanish. Yet
Schwegler (2010, p. 1) points out the startling fact that, “until recently,
dialectologists and general linguists with an interest in Hispanic or Lusophone
studies essentially ignored pidgin or creole languages”. This is unfortunate, since
these contact varieties of Spanish and Portuguese “may be key for an
understanding of the evolution of Spanish and Portuguese” (ibid.). In addition,
the genesis and evolution of these varieties have much to offer to our
understanding of contact-induced change and can play a more central role in
contributing to theoretical issues in the study of such change. The relative
neglect of contact varieties of Spanish in the past was in some ways a product of
profoundly different points of view concerning their evolution and relationship
to Peninsular Spanish – a fact reflected in similar disagreements among
Lusophone linguists about the relationship between Portuguese and its colonial
offspring. Schwegler (2010, p. 12) pointed out that “… prior to 2000,
dialectologists and general linguists with an interest in Latin America (both the
Spanish- and the Portuguese-speaking territories) had essentially ignored pidgin
or creole languages, and along with it the vibrant emerging field (creolistics) that
was studying them.” The same was applied to other parts of the world (Asia,
Africa) where Portuguese and Spanish had been introduced during colonial times
and later. This was a primary reason why the study of Iberian-Romance contact
languages has lagged behind that of their English- and French-lexicon
counterparts.
The neglect of the contact varieties has also led to a certain fragmentation
in the study of colonial offspring of Spanish and Portuguese. This is reflected in
the division of the field into fairly autonomous sub-fields such as “dialectology”
versus “creolistics”, or “native” versus “nativized” or “L2” varieties and the like,
not to mention the distinction between varieties that are due to “normal” versus
“abnormal” transmission. A similar division of interests is found in the study of
the colonial offspring of English and French. Such fragmentation ignores the fact
that all post-colonial contact language arose in similar kinds of social ecology,
involved natural second language acquisition, and share common processes of
restructuring. The idea of a unified treatment of Hispanic (and Lusophone)
contact varieties under a single theoretical framework is not new of course. It has
been pursued most notably by Clements (2009) within the Emergent Grammar
Model of language acquisition and change. Like the present approach, this
model treats contact varieties of Spanish and Portuguese as outcomes of
naturalistic SLA. Clements’ framework also emphasizes the role of the social
ecology in shaping the nature and degree of change in such contact situations.
The approach I employ here, goes beyond this, and is based on Weinreich’s
(1953, p. 4) proposal that “[p]urely linguistic studies of languages in contact
must be coordinated with extralinguistic studies on bilingualism and related
phenomena” in the fields of sociology, anthropology, and psychology. A truly
integrative model of language contact must establish links between linguistic,
sociolinguistic and psycholinguistic approaches to language change and
development. It should also explain why there is such diversity in the range of
contact languages, including those that are the outcomes of naturalistic second
language acquisition. This allows us to address some of the fundamental
questions that still challenge the field of contact linguistics more generally, and
creole linguistics in particular.
Within the latter field, as Schwegler (2010) points out, we still lack
consensus on such key issues as:
What defines creole languages and distinguishes them from other
languages, including the varieties of their lexifier languages that emerged in
European colonies (2010, p. 5ff)?
What theoretical framework best explains the origins of creoles as opposed
to other contact varieties? (2010, p. 8ff)

These two questions are in fact closely related, but they have been addressed in
very different ways by two competing frameworks – the so-called “Ecology of
Language” (EL) framework and the Naturalistic second language acquisition
framework. The EL framework ascribes all kinds of language acquisition,
including creole formation, to a process of “competition and selection” and
explicitly rejects the relevance of SLA models to the formation of creoles and
related contact varieties such as indigenized varieties. The NSLA framework
sees such languages as the result of a rather unusual kind of naturalistic SLA, in
which transfer from learners’ L1s goes to an extreme not found in most cases of
SLA. Concerning the relationship between creoles and other cases of naturalistic
SLA, the EL framework takes the position that the differences between the two
are more social than structural in nature. Mufwene (2008, p. 44) argues that
“external history suggests that there are actually no differences in regard to the
restructuring process involved” in the formation of creoles as opposed to
indigenized varieties.3 While I agree with Mufwene that there are in fact
similarities in the types of restructuring found across the outcomes of naturalistic
second language acquisition, I argue that there are in fact significant differences
that set creoles apart from other outcomes. The NSLA approach allows us to
account for both the similarities and the differences between creoles and other
contact varieties, and therefore address the issue of what defines creoles as a
distinct set of outcomes of contact.

2. The natural SLA framework

An NSLA framework must itself conform to the requirements of the more
general unified framework proposed by Weinreich. It must be based on social,
structural (linguistic), and psycholinguistic explanations of contact-induced
change and contact language formation. It must also explain why differences in
the contributions from social factors, linguistic inputs, and pycholinguistic
principles account for differences in the outcomes of contact-induced change.
Mufwene’s (2001, 2008) Ecology of Language approach also emphasizes the
role of social factors, but it offers a very different approach to the linguistic
aspects of contact induced change – one that is based on the notion of
competition and selection, and one which explicitly rejects an explanation of
creoles and similar contact varieties in terms of second language acquisition.
Contra the EL framework, I argue that the process of creole formation is
essentially one of “restructuring” similar to that found in cases of naturalistic
second language acquisition, of which creole formation is a somewhat unusual
type. This view of creole formation is now widely accepted and has been argued
for by a variety of scholars, including, more recently, Field (2004); Lefebvre
et al. (2006); Plag (2008a, 2008b); Siegel (2008) and others.
I use restructuring here in the sense in which it used by scholars in the
fields of first and second language acquisition. With regard to SLA, it refers to
the gradual and cumulative process of building and rebuilding the learner’s
developing grammar – the IL (Hulstijn, 1990, p. 32). In all cases of SLA, there
are three main contributions to this restructuring process:
Input, or intake, from the target language
Input from learners’ L1s
Internally motivated innovations in learners’ developing L2 systems

Each of these is associated with particular aspects of the language learning

process. Thus, the nature of the input largely determines the kind of intake that
learners internalize – in short, the extent of acquisition of the TL. Input from
learners’ L1s shapes the learner variety through processes of what SLA scholars
call “transfer”, or what I will refer to as “imposition”, following van Coetsem
(1988). Finally, all learners resort to creative strategies which have no
counterpart in either the TL or their L1s, thus creating new structures that are
purely the result of internal motivations. Various social factors relating both to
the macro-level of social structure and organization, and the micro-level of
individual backgrounds and relationships, constrain and regulate each of these
processes of learning.
Hence, differences in the social ecology, and consequently, differences in
the contributions of the three major components of the learning process can lead
to quite different outcomes. This explains why creoles and other contact varieties
that result from NSLA constitute a highly heterogeneous group of languages,
which arose in different socio-historical settings and with different inputs to, and
degrees of, restructuring. They include languages that closely approximate their
superstrate languages, as well as others that depart radically and to varying
degrees from them. Following from this, the second language varieties of
superstrate languages that emerged in colonial settings constitute a continuum, in
which no absolute demarcations can be established between a “dialect”, a
“partially-restructured variety”, a “semi-creole” and a “creole.” I illustrate this in
Figure 1 with examples from contact varieties of Spanish.

Figure 1. The continuum of Spanish-lexicon contact varieties

All of these outcomes represent cases of group second language acquisition, and
the consensus among most creolists today is that creoles share much in common
with other cases of “natural” second language acquisition in terms of the
processes of restructuring that are involved. Distinctions among the various
outcomes, then, have to do largely with differences in the linguistic inputs and in
the degree and type of restructuring that applied to the inputs. The specific
differences that distinguish the outcomes relate primarily to the roles played by
internally motivated creations by learners, and innovations due to transfer or
imposition from their L1s. As suggested earlier, social factors play a key role in
determining the extent to which such types of innovations arise and become
conventionalized as part of the grammar of the emerging contact language.

3. Social factors in the emergence of naturalistic second

language varieties
Thomason and Kaufman claim that “it is the sociolinguistic history of the
speakers, and not the structure of their language, that is the primary determinant
of the linguistic outcome of language contact” (1988, p. 35). We now have a
great deal of evidence that social factors do in fact play a significant role in
shaping the consequences of language contact (see Winford, 2013 for an
overview). Such factors operate at two broad levels – the macro-level of social
structure, and the micro-level of individual social networks and relationships.
The latter level is far more difficult to investigate and analyze, particularly for
past contact situations which we can no longer observe. Hence it is not
surprising that scholars of language contact have paid much more attention to the
macro-level of social structure. For instance, the EL model has identified the
following macro-level social factors that influence the products of language
contact (Mufwene & Vigouroux, 2013, p. 111ff):
Differences in economic systems (which influence population structure).
Population structures (which are relevant to who interacts with whom).
Demographic strengths of populations.

Mufwene (2001, 2008) has argued that different colonization types, and the
kinds of economic systems they engendered, led to differences in population
structures, which directly affected the linguistic inputs to colonial settings of
language contact, thus shaping the evolution of European languages in these
settings. Thus, colonies established mostly for trade purposes in places like
Africa and the Pacific produced pidgins based on non-standard dialects of the
lexifier language. “Exploitation” colonies like those in India and South East Asia
gave rise to “indigenized” varieties. By contrast, “settlement” colonies such as
those in the US, Australia, New Zealand, as well as those in Latin America,
produced outcomes that were closely related to their European sources. A subset
of these colonies – the plantation colonies in places like the Caribbean – relied
on agro-industry that required huge numbers of imported slave laborers. Such
colonies produced creoles that were structurally different from the varieties that
emerged in other settlement countries.
The general classification proposed by Mufwene is a promising first step
toward exploring the broad social contexts in which the contact varieties of
European languages emerged. It also sheds some light on the reasons for the
significant differences in the outcomes of contact in the European colonies. So
far, however, the EL framework has focused primarily on the macro-level of
social structure, saying little about the micro-level of social relationships and
interactions. The latter aspects of the social motivations for contact language
formation have in fact been seriously neglected in the literature as a whole. This
is one reason why explanations of the outcomes of contact primarily in terms of
economic structures, demographic ratios and the like, have fallen short, and
admit of too many exceptions. For instance, “settlement” colonies such as those
in Australia, produced creoles and indigenized varieties, in addition to outcomes
that were close approximations to the colonizers’ languages. Referring to the
forms of English that emerged in English colonies, Mufwene (2013, p. 214)
observes that “settlement colonies were not all of the same kind: variation in the
ways they were settled and in the ensuing population structures bore on how
English evolved.” The same applies to the Hispanic contact varieties that
emerged in Spanish colonies. Thus, we are still faced with the problem of
finding social explanations for the continuum of contact varieties related to
Spanish, from pidgins and creoles to bozal varieties, contemporary Afro-
Hispanic and Afro-Lusophone varieties, to the indigenized varieties like Andean
Spanish, the dialectal varieties and the local standard varieties.
To address this issue fully, we need a socio-historical approach to contact
between Spanish and other languages in the colonies, which goes beyond the
macro-level. Mintz’ (1971) framework offers a more comprehensive approach to
understanding how the social ecology shapes the outcomes of language contact
in colonial societies. It proposes three broad social conditions that shaped the
emergence of creole languages in the New World:
The proportions of Africans, Europeans and other groups, over time,
present in particular Antillean societies.
The codes of social interaction governing the relative statuses and the
relationships of these differing groups in particular societies;
The specific sorts of community settings, within which these groups
became further differentiated or intermixed.

The framework is an interesting blend of the macro-level of community structure

and the micro-level of individual statuses and relationships. At the former level,
it appeals to such factors as the social structure of the community, the
demographics or population ratios, the community settings, and legal
institutions. At the same time, it emphasizes such micro-level factors as the
statuses of the groups and their members, and the nature of their social
relationships, and the types and patterns of contact among them.

3.1 Social factors in the emergence of Hispanic colonial dialects

The macro-level factors go some way toward explaining difference in outcomes

of the contact between Spanish and other languages in the colonies. As already
suggested, the demographics of the original settlement of the colonies influence,
among other things, the primary linguistic inputs to the contact situation, and the
respective contributions they make to the emerging contact language. The
dialectal varieties of Spanish (and Portuguese) that emerged in the settlement
colonies arose as a result of relatively successful transmission of the European
languages. This was due to the fact that, first, their speakers traveled from the
Iberian Peninsula, and second, they were successfully acquired by a majority of
the indigenous colonial population. In these cases, the input and intake from the
target languages outweighed the effects of L1 influence and internal
developments. In other settings that involved plantation or other economies that
required large numbers of mostly imported laborers, learners had more restricted
access to the superstrate language, leading to sometimes significant L1 transfer
and other creative innovations. The effects of such creativity distinguish creoles
and other contact varieties from those forms of colonial languages which were
due to migration of their speakers, or relatively successful SLA. Unfortunately,
we still know far too little about the details of the social history of the contact
situations that involved Spanish on the one hand, and the languages of
subjugated populations on the other.
There has been increasing attention paid of late to the sociolinguistic
history of Latin American colonies. Much of the interest was sparked by the
debate over the “missing” Spanish creoles in the Americas, which is still a topic
of intense and contentious discussion (Lipski, 2005; McWhorter, 2000).
McWhorter proposed that creoles did not arise to the same extent in Latin
America primarily because Spanish-based pidgins which could form the basis
for creole formation did not arise either in the Spanish trade areas in Africa, or in
the colonies. He claims that this explains why, even in community settings where
there were high ratios of slaves to Spanish settlers, no creoles emerged. The
alternative view, held by most scholars, is that the paucity of Spanish-lexicon
creoles is due primarily to the social structure and demographics of Spanish
colonies. In fact, Mintz himself offered social explanations for the absence of
Spanish creoles in the Caribbean. He noted that, “Generally speaking, the
Hispano-Caribbean colonies were never dominated demographically by
inhabitants of African origin” (1971, p. 481). But he added to this the very
important observation that “moreover, in those colonies, movement from the
social category of ‘slaves’ to that of ‘freemen’ was almost always relatively rapid
and relatively continuous” (ibid. – italics in original). This meant that slaves had
more opportunities to interact with speakers of Spanish, and thus more access to
their language. Moreover, Mintz points out that the lack of large-scale economic
development in 17th century Hispanic colonies such as Cuba and Puerto Rico
contributed to frequent manumission of slaves. Thus, “when slavery became
important in Cuba and Puerto Rico (but not in Spanish Santo Domingo), and the
importation of slaves rose, there was already in these islands a large Spanish-
speaking population of mixed physical antecedents” (1971, p. 482). This
explains why a variety of Spanish that closely approximated the peninsular
variety came to prevail in those colonies.
Studies by Diaz-Campos & Clements (2005, 2008), Sessarego (2013, 2014,
2017) and others have argued along similar lines as Mintz. They show that the
demographic and other social conditions conducive to creole formation were not
present in places like the Chota Valley (Ecuador), Los Yungas (Bolivia), coastal
Venezuela, and so on, contra claims by McWhorter and others. However, the
contact situations in these places did allow for the emergence of second language
varieties characterized by features that can also be found in creoles. For instance,
Lipski (2008) lists such features of Afro-Bolivian Spanish as invariant verb
forms, lack of inversion in questions, lack of gender concord in NPs, and
possibly, preverbal TMA markers. Sessarego points out, however, that “many of
the key features mentioned as prototypical of Spanish and Portuguese creoles
can often be found in advanced second language varieties and non-standard
dialects of Spanish and Portuguese, for which a creole origin is not feasible”
(2013, p. 369). Such features include most of those listed by Lipski above, such
as use of invariant verb forms, non-inversion in questions, lack of gender
agreement, and others like the use of bare nouns to convey definiteness
(Sessarego, 2013, pp. 369–373).
 Most of these discussions of Latin American Spanish have focused on
macro-level factors such as demographic ratios and the like. But Mintz’s
explanation of the fate of Spanish in the colonies goes far beyond just
demographics. He shows how each of the main “background conditions” of his
framework – demographic proportions, the codes of social relations and the sorts
of community settings – played a role in shaping the linguistic outcomes. He
shows, for example, why Spanish in Puerto Rico did not undergo the kinds of
restructuring that led to creole formation in places like Jamaica, arguing that
“..the sociolinguistics of these two cases differed dramatically, since they
involved populations of different proportions, living by different social codes,
and with significantly different historical backgrounds” (1971, p. 485).
Moreover, Mintz’ approach provides a basis for investigating aspects of micro-
level social organization as well. Perhaps his most important contribution to our
understanding of what happened to Spanish comes from his detailed account of
the role played by “the codes of social relations governing the statuses and social
interactions of different groups in each society” (p. 487). He considers the degree
to which Spain controlled colonial political structures and decision making; the
liberal nature of Spanish slavery codes (laws); and the ideologies of the
dominant group, which led to the emergence of local loyalties and identities that
linked them more strongly to the colonies (p. 487). Mintz’ call for “a careful
evaluation of such factors in sociolinguistic terms” (p. 488) is still for the most
part unanswered. A welcome exception is Sessarego’s recent proposal of the
“Legal Hypothesis of creole genesis”, which holds that “the relative paucity of
creole languages in the Spanish Americas may be seen – in part – as the
byproduct of differences in the European legal tradition” (2017, p. 4). He argues
that slave law in the Spanish colonies were favorable to slaves’ “chances of
climbing the social ladder”, attaining freedom, owning property and having
families of their own – all of which increased their access and incentive to learn
the colonial language (ibid.). Sessarego’s account is well argued and supported
by a wealth of socio-historical evidence, and it provides at least a glimpse into
the conditions that affected the daily lives of slaves and their descendants in the
Spanish colonies. It is a first step toward the kind of micro-level explanation that
Mintz’ model calls for. But a full account of this level of sociolinguistic
explanation calls for more detailed exploration of the social networks, patterns of
interaction, social motivations, and ideological of the various groups that made
up colonial societies. This is in fact what the Ecology of Language model seems
to have in mind when it calls for a sociolinguistic explanation of how changes
propagate and become conventionalized as part of both individual community
language norms through processes of competition and selection. This aspect of
the model is still in the early stages of development (Mufwene, 2014; Mufwene
& Vigouroux, 2013). But that is true of most frameworks that address the role of
social factors in language change and language creation. Among the attempts
toward such a goal are Roberts’ (2004) fascinating account of the role played by
style and identity in the emergence of Hawai’i Creole English, and Arends’
(2001) account of the role of social networks in the development of Sranan.
Further investigations of how sociolinguistic factors impact the outcomes of
variation and change in contact settings can be found in Léglise & Chamoreau
(2013). More research of this type is desperately needed for contact varieties in
general, including those related to Spanish. On the whole, we are still far away
from understanding how macro- and micro-level social factors interact to
regulate change and resolve competition among linguistic variants, leading to
conventionalization of new contact languages.

4. Substrate inputs to the formation of Spanish-related

creoles
By contrast with the cases of relatively successful SLA that led to the emergence
of the colonial “dialects” of Spanish just discussed, the formation of Spanish-
related creoles involved quite different contact situations, which favored
different processes of restructuring, and hence outcomes that diverged in
different ways from the colonial varieties. The differences in the social ecologies
that produced these contact languages have been well documented in the
literature, so I will not provide a full account of them here (See Singler, 2008
and Winford, 2008 for overviews). Suffice it to say that differences in the
population ratios between colonizers and the enslaved, as well as differences in
social relationships and patterns of interaction between the groups and other
factors, produced conditions under which input from, and access to, the lexifier
language as a target was limited to varying degrees, depending on the particular
contact situation. Population ratios, particularly in the early stages of contact
language formation, are particularly relevant to the kinds of linguistic input
available in a given contact situation, as well as to the degree of influence they
exert on the emerging contact language. Fortunately, recent research has begun
to provide clearer pictures of the socio-historical contexts in which Ibero-
Romance creoles and indigenized varieties emerged, though much work still
needs to be done.
As a result, we now know much more about the substrate languages that
played a role in the emergence of Ibero-Romance contact languages, more so
perhaps in the case of Portuguese-lexicon creoles. For example, work by Lang
(2009, 2013), Biagui & Quint (2013) and others have shown that the substratal
inputs to the Upper Guinea Creoles (UGC) of the Cape Verde islands and
communities in Senegal and Guinea-Bissau, were primarily languages of the
Atlantic (Wolof) and Mande (Mandinka) families. By contrast, the primary
substratal inputs to the Gulf of Guinea creoles (GGC), spoken on islands off the
coast of Equatorial Guinea, consisted mainly of Edo and Kikongo, and to some
extent Yoruba (Hagemeijer, 2013; Maurer, 2009; Schang, 2000). Such
differences in inputs explain to a large extent the structural differences between
the Upper and Lower Guinea Lusophone creoles.
There are still gaps in our information about the sociohistorical and
sociolinguistic contexts in which creoles were created in Spanish colonies, but
recent research has begun to address this. For instance, studies by Fernández
(2011, 2012), Fernández & Sippola (2017), Grant (2011), Steinkrüger (2008)
have provided new information on the socio-historical contexts in which
varieties of Chabacano (Philippine Creole Spanish) arose, though most attention
has been paid to the Zamboangueño variety of Mindanao, and less to those that
arose in the Manila Bay Area, or in other parts of Mindanao. There still remains
uncertainty as to the precise periods in which each variety emerged, the
historical relationships among them (Fernández & Sippola, 2017; Grant, 2011;
Steinkrüger, 2008). Thus, there is also some uncertainty about the exact inputs to
their formation. Steinkrüger (2008, p. 152) suggests that the principal substrates
of Chabacano include Malayic, Visayan [Bisayan] languages (only for
Zamboangueño), and Hokkien. Grant (2011, p. 318) identifies the main substrate
languages of Zamboangueño as Yakan, Sinama and arguably Hiligaynon.
Adding to the problem of characterizing the precise inputs to the formation of
Chabacano is the fact that all of its varieties have been subject to continuing
contact with one or another Filipino language, including Tagalog and English at
various times in their history (Grant, 2011, p. 209); Steinkrüger, 2008, pp. 153–
154). Grant (ibid.) points out that:
The same language has often been both substrate and superstrate or substrate and
adstrate language, and the same language has served as both superstrate and
adstrate. This raises problems for any model of creole formation which assumes
that substrate, superstrate and adstrate are categories usually containing or
involving different languages.
Hence there is need for continued research on the social contexts of these contact
varieties at different periods of their development.
As far as the Caribbean is concerned, the precise linguistic inputs to the
formation of Papiamentu and Palenquero, particularly the former, remain a
matter of debate. Earlier scholars debated whether Papiamentu was originally
Spanish-based (Munteanu, 1996) or derived from an Afro-Portuguese ancestor
(Martinus, 1996) or a Brazilian Portuguese-lexicon creole brought to Curaçao by
Sephardic Jews (Goodman, 1987). At the center of this debate was the source of
the Portuguese elements in the creole. Recent work by Quint (2000a, 2000b) and
Jacobs (2008, 2009a, 2009b, 2010a, 2010b) argues that the Upper Guinea
Portuguese-lexicon creoles of Cape Verde and Guinea Bissau, transported to
Curaçao in the mid-seventeenth century, were a primary input to early
Papiamentu. Jacobs in particular provides compelling evidence of this, based on
both historical sources and a wide range of linguistic correspondences. The issue
remains a matter of debate since the evidence is far from conclusive or
straightforward.
Similar questions arise with regard to Palenquero, which is claimed to have
close affinities to Lower Gulf of Guinea Portuguese-lexicon creoles spoken on
São Tomé Principe, and Annobon (Lipski, 2009, p. 547) Indeed, Lipski
(2009, p. 548) claims:
It is almost certain that Palenquero arrived in Palenque de San Basilio at least
partially formed, modeled on the already emergent São Tomé Portuguese creole,
if not identical to the latter language.
Scholars like Schwegler (1996a, 1996b) and Megenny (1983) point to several
structural features that Palenquero shares with the Lower Gulf of Guinea creoles,
including plural subject pronouns, postposed possessives, and postposed
negation. Lipski (2009, p. 546) also points to the documented presence of early
São Tomé creole in nearby Cartagena as early as 1627.
 If the scenarios outlined above for Papiamentu and Palenquero are accurate,
they raise new questions about the substrate languages that may have exerted
most influence on these creoles. A Cape Verdean input or even origin for
Papiamentu implies that the substrate languages that shaped the former, i.e.
languages like Wolof and Mandinka, also contributed to the Spanish-related
creole. Similarly, languages like Edo and Kikongo, which were the primary
substratal inputs to the Gulf of Guinea creoles (GGC), must also have shaped the
grammar of Palenquero. Yet Schwegler (2010, p. 32) informs us that “Research
during the last 25 years suggests that Kikongo must have been the only African
language to significantly impact the formation of Palenquero.” The bulk of the
evidence for this comes from etymologies of recently discovered Palenquero
Africanisms, all of which are traceable to Kikongo. But is it possible that such
Kikongo influence exerted itself only later, on an already established
Palenquero? Moreover, does evidence of lexical influence from a particular
substrate language necessarily imply that the grammar of the creole in question
could not have been more profoundly influenced by other, earlier, substrates?
These questions remain unresolved for both Palenquero and Papiamentu. It may
explain why comparatively little attention has been paid to exploring the sources,
particularly the substrate sources, of many aspects of the grammar of these
creoles. Solid socio-historical evidence of the linguistic inputs to the relevant
contact situation is of course vital to the issue of such sources. Obviously, if we
lack a clear understanding of the primary superstratal and substratal inputs to the
formation of particular creoles, it is difficult to describe the processes of
innovative change that shaped the creole’s grammar. I turn my attention now to
these processes.

5. SLA processes at work in the formation of creoles

The NSLA approach to the formation of creoles which I adopt here departs
significantly from the approach taken by Mufwene, who claims that “[….] while
research on SLA can inform us about conditions that favor transfer during L2
“acquisition”, it cannot inform us on how substrate elements influence the
development of creoles or any other languages for that matter” (2008, p. 158).
Mufwene’s view is based on his assumption that: “substrate influence is a
population-level phenomenon that results from both the repeated occurrences of
xenolectal elements in some idiolects and their spread within the population of
speakers what other speakers simply copy them” (2008, p. 159). But this ignores
the clear parallels between the workings of L1 influence at the level of
individual learners’ ILs in both naturalistic SLA and creole development. Here
the focus is on the process of restructuring that yields individual idiolects or
interlanguage systems. It is during this process that innovations of various types
enter the individual IL systems and are thereby made available for the processes
of “selection” and propagation that yield conventionalized communal systems.
There has been increasing demonstration of the role of transfer in both classroom
SLA (Helms-Park, 2003) and naturalistic SLA (Siegel, 2006; Sanchez, 2006),
and this includes the robust literature on the development of indigenized
varieties of English, Spanish and other languages (Siegel, 2006, p. 36).
At the same time, however, there is general consensus that the development
of creoles and similar contact varieties differs from more “typical” cases of
naturalistic SLA because of the extent and persistence of L1 influence, and the
extent of internally motivated innovations, which characterize the former. As
argued earlier, varying degrees of L1 influence and internal change result in a
continuum of outcomes ranging from closer approximations to the lexifier target,
to radical departures from it. The formation of creoles and indigenized varieties
is a multi-dimensional process of creative language acquisition involving three
interrelated processes, which are shared to varying degrees with SLA:
 reduction and/or simplificatory restructuring of the input from the TL or
superstrate (intake)4
L1 influence via the mechanism of “transfer” or imposition
creative internal processes such as grammaticalization

Each of these lends itself to explanation in terms of the kinds of language

processing that are typical of second language acquisition. The innovations
produced by these processes have long been the primary focus of a theory of
creole formation. They relate directly to the question of the actuation of change
(Weinreich et al., 1968). Here we must distinguish between actuation at the level
of the individual, which is a psycholinguistic process related to individual
idiolects, and the propagation of change – a process of inter-idiolectal
competition and selection which is in fact a sociolinguistic process. Since the
locus of actuation (of a contact-induced innovation) is the individual, then the
mechanisms or processes of change have to be explained in terms of how
linguistic systems or inputs interact in the individual mind – i.e, in
psycholinguistic terms.

6. The psycholinguistic mechanisms of change in contact

language formation
There is growing evidence that the innovations in creoles and related contact
languages, which have been attributed to mechanisms like “simplification”,
hybridization, relexification, transfer, etc., can be understood in terms of
language processing by learners. There are two aspects of contact-induced
change that can be profitably explained in terms of processing constraints. One
has to do with the lack of inflectional morphology, the regularization of word
order in declarative, interrogative and negative sentences, and other features, all
of which are characteristic of the early stages of SLA, as well as creole
formation. The second aspect of contact-induced change that lends itself to
treatment in terms of language processing (particularly production), is the way
structural features are transferred from learners’ L1s into a developing IL. I will
deal with each of these in turn.

6.1 Processing constraints on the acquisition of TL structures

Plag (2008a, 2008b) has provided an excellent account of how a processing

model of SLA can explain such aspects of creole structure as the absence of
inflectional morphology and other TL structures. For instance, as Schumann
(1978), Andersen (1980), Plag (2008a, 2008b, 2011) and others have argued,
both creoles and early Interlanguage share such features as the following:
lack of bound inflectional morphology
the regularization of word order in declarative, interrogative and negative
sentences
other phenomena such as the preference for free forms and transparent
form-to-function relationships

Thus creoles generally lack both “inherent” inflection such as number, degree
and tense-aspect (though some of this is found in certain creoles), and
“contextual” inflection such as subject-verb agreement, case markers and the
like.
As Plag has argued, these similarities between creoles and early ILs are due
to the fact that both creole creators and early L2 learners devise similar strategies
in constructing their ILs in the early stages of acquisition. This is because they
are subject to similar processing constraints inherent in early SLA. This explains
the similarities we find across vernaculars that have resulted from naturalistic
SLA, which Chambers (2004) refers to as “vernacular universals.”
Contact varieties of Spanish provide robust evidence of such universals of
restructuring. For example, Lipski (2008) describes aspects of Traditional Afro-
Bolivian Spanish, which he ascribes to the incomplete acquisition of Spanish
morphology during the formative period of this variety. The principal
morphological features he lists include the following (Lipski, 2008, p. 184):
Non-differentiation of grammatical gender in nouns and adjectives, as in la
río [cf Span el río] “the river”; nuestro cultura antigo [Span nuestra cultura
antigua] “our traditional culture”, etc.
Invariant plurals, as in lu mujé [las mujeres] “the women”; lu patron [los
patrones] “the landowners”, etc.
Absence of definite articles in subject position (required in other Spanish
varieties), e.g., perro ta flojo [los perros están flojos] “dogs are worthless”,
etc.
Lack of formal vs familiar 2nd person distinction in the pronoun system,
e.g., oté <ustéd ‘you’ for both formal and familiar.
Lack of masculine-feminine distinction in the 3rd person pronouns, eg. ele
“he” or “she [Spanish él “he” vs ella “she”].
Use of the Spanish 3rd singular verb form with subjects of all persons and
numbers, e.g., nojotro tiene [nosotros tenemos] jrutita ‘we have fruit’

As far as syntax is concerned, Lipski identifies phenomena such as the

following, among others:
Non-inverted questions, e.g., ¿cuánto hijo pue oté tiene? [¿cuántos hijos
tiene usted?] “How many children do you have?”
The use of preverbal tense-aspect markers with unconjugated verbs, instead
of conjugated verbs as in Spanish, e.g., ¿quién ta comprá? [¿quién está
comprando?] ‘who is buying [coca]?’

Lipski points out that many of the features of ABS can be found in other contact
varieties of Spanish, including creoles like Papiamentu, Palenquero and
Chabacano, and concludes that Afro-Yungueño Spanish appears to be more
creole-like than the Helvécia dialect of Brazilian Portuguese, though less “deep”
than Palenquero or Chabacano (2008, p. 183). However, as noted above,
Sessarego (2013, p. 369) points out that many of these so-called “creole” features
are also found in advanced second language varieties and non-standard dialects
of Spanish and Portuguese. He concludes that “the linguistic elements found in
ABS appear to be the result of fossilized second language acquisition strategies”
(2013, p. 402). Such strategies are typical of both early second language
acquisition, and creole formation.

6.2 L1 influence and processing in creole formation and other cases of

naturalistic SLA
Let us now consider how processing accounts of SLA can contribute to our
understanding of the way structural features are transferred from learners’ L1s
into a developing IL, and by extension, in cases of naturalistic SLA. The
literature on second language acquisition has shown that learners’ versions of an
L2 are subject to significant structural transfer from their more dominant L1 in
both classroom situations (MacWhinney, 2005; Jarvis & Pavlenko, 2008) and
naturalistic learning settings (Odlin, 1989; Sanchez, 2006). Most of the research
on this has concentrated on the acquisition of L2 phonology and lexicon. There
is now a substantial body of evidence from both psycholinguistic and SLA
studies that “the acquisition of a native language has a fundamental and long-
lasting effect on how the sub-lexical and lexical units of a language are
perceived and acquired” (Schwartz & Kroll, 2006, p. 975). Similarly,
psycholinguistic studies of second language acquisition have provided
substantial empirical support for transfer from the L 1 to the L2 at the level of
sentence processing (MacWhinney, 1997, 2005, 2008). In cases of second
language acquisition, learners compensate for their lower proficiency in the
target of learning by appealing to the language production procedures of their
dominant L1s. Van Coetsem’s (1988, 2000) framework for the study of language
contact proposes that the mechanism underlying these kinds of transfer is
imposition, which we can define broadly as the strategy of employing the
language production procedures of a dominant language in producing a less
dominant language. Van Coetsem’s view of language dominance as a crucial
factor in determining the outcomes of language contact is quite in keeping with
psycholinguists’ views of the role of dominance in language production. As
Schwartz & Kroll (2006, p. 968) point out:
Bilinguals are rarely equally proficient or balanced in their use of the two
languages, rendering one of the languages the more dominant language.
Typically, the more dominant language will be the first or native language, but
for bilinguals who have lived in their L2 environment for many years, the L2
may be functionally more dominant, at least for certain language processing
tasks.
Psycholinguists have attempted to account for these kinds of cross-linguistic
influence by appealing to models of bilingual language production based on
Levelt’s (1989) model of monolingual language production. Space does not
permit a full account of language production models here, but a brief overview
of Levelt’s (1989) model will suffice for our purposes. This model proposes that
three types of mental processes are involved in language production:
conceptualization processes that specify which concepts are to be expressed
verbally; formulation processes that select appropriate lexical items and
construct the syntactic and phonological structure of the utterance; and
articulation processes that realize the latter as overt speech (Roelofs,
1993, p. 108). The basic outline of the model is shown in Figure 2.
Figure 2. A simplified model of language production
As Levelt (1993, p. 5) explains, conceptual structures are the preverbal messages
that form the input to the Formulator, “whose task it is to map the message onto
linguistic form.” The final output of the Formulator is “a phonetic plan that can
be executed by the articulatory motor system” (ibid.). The formulator therefore
involves two processes: grammatical and phonological encoding. Only the first
of these concerns us here. Levelt explains: “Grammatical encoding takes a
message as input, retrieves lexical items from the mental lexicon, and delivers a
surface structure as output” (ibid.). Bock & Levelt (1994) provide a detailed
outline of this model of language production. As they point out, (1994, p. 946),
grammatical encoding involves two subcomponents: functional processing and
positional processing. The primary subcomponents of functional processing are
lexical selection, which involves the identification of lexical concepts and
LEMMAS that are suitable for conveying the speaker’s meaning; and function
assignment, which involves the assignment of grammatical roles or syntactic
functions to the relevant lexical concepts. Positional processing, on the other
hand, involves constituent assembly (the creation of an ordered set of word slots)
and inflection (creation of morphological slots) (ibid.)
De Bot (2001 [1992]) adapted Levelt’s model of monolingual bilingual
production so as to address the additional requirements that such a model has to
meet in a bilingual context. As De Bot points out, the model must not only
account for crosslinguistic influences of various types, but also “deal with the
fact that the speaker does not master both language systems to the same extent”
(2001, p. 425). Hence, such a model must be able to account for the following
aspects of bilingual language production (ibid):
the two language systems involved can be used independently of each other,
or they may be mixed in various ways (as in code switching, for example)
the two systems may influence each other
the bilingual speaker may have different degrees of proficiency in each
system
interactions can take place between languages that are typologically similar
or dissimilar

The point most relevant to our present concern with L1 transfer is De Bot’s
observation that the majority of bilinguals will not have a complete command of
both languages, and that this can lead to interference from the more dominant
language. As he puts it:
It is clear that when the speaker has very little knowledge of the L2 he can still
make utterances in that L2 by making some (internal) extensions to the L1
system. In this way it is plausible to think that it is only the morpho-
phonological information for lexical items in the L2 which is L2 specific, while
syntactic information from the L1 translation equivalent is activated.
(2001, p. 441)
This kind of procedure seems to lie behind cases of lexical as well as
grammatical interference in the production of speakers for whom one of the
languages is dominant. In such cases, the speaker may treat a lexeme in the less
dominant language as an alternative phonological shape to that of an L1 lexeme
and associate the latter’s lemma with both. This would explain phenomena that
are common in the early stages of second language acquisition, where the learner
acquires L2 lexical items but not all aspects of their grammatical properties,
including often their grammatical category, their precise semantic content, and
their subcategorization requirements. Similarly, it explains why early learners
cannot produce L2 sentences with the appropriate constituent structures, since
that relies crucially on the syntactic subcategorization of lexical material, as
contained in lemmas (Pienemann, 1999, p. 51). Faced with the need to produce
utterances in an L2 with which they are not familiar, learners often resort to L1
knowledge and procedures. Interestingly, this is in fact essentially the same
explanation for L1 transfer that is assumed by adherents of the Relexification
Hypothesis and the Full Access/Full access model of SLA. Sprouse
(2006, p. 174) suggests that:
[…] the only way for the “abstract properties of the L1 to transfer and comprise
the initial state of L2 acquisition is via retention of the L1 lexicon (minus
phonetic labels). That is, Full Transfer’s “abstract properties” appear to
correspond in Minimalist terms to Relexification’s lexical “features”.(minus
phonetic features)
It would seem then, that both creole creators and L2 learners can transfer
abstract features of their L1 to the L2 they are attempting to produce. Following
van Coetsem, I propose that the mechanism by which they accomplish this is
imposition. The empirical evidence from studies of bilingualism and second
language acquisition demonstrates that there is potential for imposition at every
stage of the language production process, from the level of conceptualization, to
functional processing, to positional processing, all the way to the level of lexeme
selection and phonological encoding. For present purposes, I confine my
attention to imposition at the levels of positional and functional processing.
7. Mechanisms of imposition in cases of naturalistic SLA
In creole formation, just as in other cases of naturalistic second language
acquisition, learners tend to appeal to L1 knowledge to compensate for their lack
of proficiency in their L2. This is reflected in their L2 production, which
depends to varying degrees on the grammatical encoding procedures of their L1,
which is dominant. This is the essence of what the mechanism of imposition
involves. The insights of the psycholinguistic models discussed above suggest
that we can view imposition as a type of cross-linguistic influence in which the
production procedures of a dominant source language are transferred to the
production of a recipient language in which speakers are less proficient. This
may happen at any stage of the grammatical encoding process. For example, at
the level of constituent assembly, word order is particularly susceptible to
transfer. A good example of this comes from the well-known indigenized variety,
Andean Spanish, which is characterized by the transfer of basic word order
patterns from Quechua. Hence the SVO order of general Spanish is often
replaced by an SOV pattern, as in the following examples:

(1)
Y mi hermano aquí otro paloma hembra había chapado.
And my brother here other dove female had caught (Odlin,
1990: 103)
‘And my brother caught another female dove’

(2)
Mi santo de mí lo han celebrado.
My birthday of me it have.3PL celebrated (Lipski, 2004)
‘They celebrated my birthday’
Similarly, imposition can take place at the level of functional processing. Recall
that functional processing includes lexical selection, which involves the
identification of lexical concepts and LEMMAS that are suitable for conveying
the speaker’s meaning; and function assignment, which involves the assignment
of grammatical roles or syntactic functions to the relevant lexical concepts.
Another common strategy of imposition that learners employ is to impose the
lemmas of L1 lexemes onto their counterparts in the emerging L2 variety. We
can assume that, in the early stages of creole formation, just as in the early stages
of second language acquisition in general, learners have not yet acquired all of
the grammatical information associated with the L2 lexemes they can produce.
As Pienemann suggests:
If the L2 word is simply attached to the L1 lemma as an alternative
morphophonological form, then the complete L1 syntactic information would be
available upon accessing the lemma.(1999, p. 83)
This is especially true if access to the L2 is restricted, or if the L2 input consists
only of highly simplified structures or pidginized varieties, as in many cases of
creole formation. For imposition to take place in these cases, it is sufficient for
learners merely to have access to the semantics of an L2 lexical item, and its
syntactic category. Similarity between the L2 item and a corresponding L1 item
in these respects is all that is needed to trigger imposition of the L1 lemma on
the L2 item. This takes place at the level of function assignment, where the
argument structure of verbs is mapped onto appropriate syntactic functions of the
constituents of a sentential structure. Verbs are associated with specific thematic
roles (such as ‘agent’, ‘patient’, ‘theme’, etc.), which in turn are mapped onto
particular grammatical functions (e.g., ‘subject’, ‘object’ etc.) in language-
specific ways. There is ample evidence from the literature that learners often
apply the functional processing procedures of their L1 to their version of an L2,
resulting in L2 structures that replicate those of the L1. For example, Hagemeijer
and Ogie (2011) show that Santome has a wide range of serial verb constructions
modeled on those of Èdó. These include constructions in which the V2 conveys
directed motion, comparative, completive and other meanings. The following
illustrates a directed motion SVC (Hagemeijer & Ogie, 2011, p. 47)

(3)
1. Santome:
Inen kôlé lentla ke.
3PL run enter house
‘They ran into the house’
2. Èdó:
Íran rhùé̱ làá òwá.
3PL run enter house
‘They ran into the house’

We can explain such creole structures in terms of transfer of the lemmas

associated with substrate verbal lexemes to their superstrate counterparts. In the
process of producing the incipient creole, Èdó speakers imposed the lemmas of
verbs like rhùé̱ onto superstrate lexemes such as kôlé. In other words, the
argument structure and subcategorization properties of Èdó rhùé̱ are imposed on
Portuguese-derived kôlé. Like rhùé̱, this triggers generation of a structure that
has the same constituency as its Èdó counterpart. We can represent this pattern
of function assignment as follows:
Situations of attrition also provide examples of the transfer of argument structure
and subcategorization properties from an L2 that has become dominant over a
heritage first language. In L.A. Spanish, for instance, we find structures like the
following in the speech of English-dominant bilinguals (Silva-Corvalán,
1998, p. 233).

(4)
1. LA Spanish:
Yo gusto eso.
I like-1s that
“I like that.”
2. Gen Span.:
A mí me gusta eso.
To me pro please-3s that
“I like that.”

The LA Spanish example is a case where the syntactic properties of the English
verb like have been imposed on the Spanish verb gustar. In General Spanish,
gustar has a Theme that is mapped to a subject and an Experiencer that is
mapped to an indirect object. This mapping can be represented as follows:
In LA Spanish, by contrast, gustar adopts the subcategorization properties of
English like, and replicates the latter’s mapping of thematic roles onto syntactic
form, in which the Experiencer role is mapped onto the subject and the Theme
onto the object.
8. The need for future research on Spanish-related contact
languages
Currently, there are serious gaps in our understanding of the mechanisms of
change that led to the emergence of contact varieties of Spanish, both creoles
and indigenized varieties. As noted earlier, a key prerequisite for this is precise
identification of the substrates involved, and detailed comparisons of their
grammars with those of the respective contact varieties. Unfortunately,
comparatively few studies of this type have been done for Spanish-lexicon
contact varieties, by contrast with the substantial research on substrate influence
on contact varieties of other lexical affiliation. A welcome exception is recent
work on the history of varieties of Chabacano by Grant (2002, 2011), Lipski
(2001), Spitz (2001) and others. Grant provides a detailed overview of substrate
influence in Zamboangueño, which he refers to as Mindanao Chabacano (MC),
and compares structural features of the creole with those of Central Philippine
languages. He demonstrates a variety of similarities in both the phonological and
syntactic systems, as well as substantial lexical borrowing from the Philippine
languages into Chabacano. For example, MC employs tyene (< Sp tiene 3rd
person of tener ‘to have’) to mean ‘there is’ and its negative form nway (< Sp. no
hay “there isn’t”), in constructions like the following:

(5)
Tyene komida.
1. have food
‘There is (scil. some) food.’ (Frake, 1971, p. 235)
Nway komida.
2. [Link] food
‘There is no food.’ (Frake, 1971, p. 235)

According to Spitz (2001), these patterns correspond very closely with those
found in Hiligaynon where the form meaning ‘have’ is negated not by a simple
negator, but by a particle wala’ meaning ‘there is not’, as in the following
examples:

(6)
May kan’un.
1. [Link] food
‘There is food, I have food.’
Wala’ kan’un.
2. [Link] food
‘There is no food.’

Other features of MC that are modeled on those of Central Philippine languages

include VSO word order, and use of kon ‘with’ or konel, (literally ‘with the’) to
mark direct objects, as in the following example from Lipski & Santoro
(2007, p. 390). Note that si is a borrowing from Philippine languages, which has
the function of a “personal article” used only with personal names.

(7)
 ta-mira si Juan konel el kasa.
ASP -see ART John OBJ MARKER ART house
‘John sees the house’.

Grant (2011, p. 310) concludes:

Almost all the features examined for Zamboangueño could be found either in
Spanish or Hiligaynon and frequently in both. The considerable Philippine
lexicon in MC is only part of the story. Typologically and syntactically, MC is as
Philippine as it is Spanish.
The replication of Philippine grammatical patterns in Mindanao Chabacano is
strongly reminiscent of what we found in other cases of creole formation and
suggests that imposition played a key role in the emergence of Chabacano
grammar.

9. Conclusion
A unified model of language contact must explain the interaction between the
linguistic processes generated by language contact, on the one hand, and the
sociolinguistic and psycholinguistic factors that regulate them, on the other. The
approach I have attempted to sketch here is based on various models of contact-
induced change proposed in the field, some concerned with more linguistic
accounts of contact phenomena, others with the sociolinguistic and
psycholinguistic aspects of language contact. I have attempted to show how
these three approaches can be integrated into a unified account of the New
Spanishes, within the Naturalistic SLA framework. A cross-disciplinary
approach such as this is most likely to achieve the kind of unified theoretical
framework that we are seeking for contact languages in general. As Weinreich
long ago suggested, the linguistic component of such a framework must, in the
first place, deal with the structural aspects of contact-induced change – the
linguistic processes involved in the types of grammatical replication as well as
internal developments found in contact situations. Second, the framework must
describe the social contexts in which contact-induced change arises, and the
social processes by which change is propagated through social networks, leading
to conventionalization in community grammars. In addition, an integrated model
must account for the psycholinguistic processes involved in the actuation of the
innovations that emerge at the level of individual idiolects and eventually
become conventionalized as part of the new contact language. Finally, we have
to find a way of integrating all three of these components into the kind of inter-
disciplinary model that Weinreich (1953, p. 4) viewed as indispensable to the
field. Iberian-lexicon contact languages, in all their variety, offer an ideal testing
ground for developing such an integrated model of the origins and evolution of
contact languages in general.

Notes
1. I adopt Schwegler’s (2006, p. 71) description of Bozal Spanish as “the Spanish formerly spoken in Latin
America by newly arrived African slaves, [which] varied considerably, ranging from distorted pidginized
language to easily understandable foreigner talk.” ①

2. An anonymous reviewer comments that “Papiamento is not a variety of Spanish”, but “is Portuguese-
based.” I discuss recent views on the sources of Papiamentu later in the paper. ①

3. In Mufwene’s framework, the “external ecology” includes a wide range of both social and linguistic
factors, the latter including the languages that are in contact with the language undergoing formation or
development. So external history here refers to all aspects of the socio-historical context that have to do
with the external ecology. ①

4. The term “target language” (TL) is highly controversial in creole studies, primarily because it is unlikely
that the putative TL remained the same at different points in time. Presumably the TL during the early
stages of settlement of a colony would have been the varieties of the lexifier language spoken by native
speakers of that language. I assume that the kinds of restructuring due to constraints on acquisition, which I
describe in this section, would have applied at this point. However, over time, the TL that would have
provided input to newly arrived slaves would have been second language varieties of the lexifier that had
already undergone varying degrees of restructuring (Baker, 1990). ①
References
Andersen, R.
(1980.) Creolization as the acquisition of a second language as a first language. In A. Valdmann & A.
Highfield (Eds.), Theoretical orientations in Creole studies (pp. 273–295). San Diego, CA: Academic
Press. ① ②
Andersen, R. (Ed.) (1983.) Pidginization and creolization as language acquisition. Rowley, MA:
Newbury House. ①
Arends, J.
(2001.) Social stratification and network relations in the formation of Sranan. In N. Smith & T. Veenstra
(Eds.), Creolization and contact (pp. 291–308). Amsterdam: John Benjamins. ①
Baker, P.
(1990.) Off target? Journal of Pidgin and Creole Languages, 5(1), 107–119. ①
P. Bakker, & M. Yaron (Eds.) (2013.) Contact languages: A comprehensive guide. Berlin: De Gruyter.
①
Biagui, N.-B., & Nicolas, Q.
(2013.) Casamance Creole. In S. Michaelis, P. Maurer, M. Haspelmath, & M. Huber (Eds.), The survey of
Pidgin & Creole languages (2(3), pp. 40–49). Oxford: Oxford University Press. ①
Bock, K., & Levelt, W.
(1994.) Language production: Grammatical encoding. In M. A. Gernsbacher (Ed.), Handbook of
psycholinguistics (pp. 945–984). New York: Academic Press. ① ②
Chambers, J. K.
(2004.) Dynamic typology and vernacular universals. In B. Kortmannm (Ed.), Dialectology meets
typology: Dialect grammar from a cross-linguistic perspective (pp. 127–145). Berlin: Mouton de
Gruyter. ①
Clements, C.
(2009.) Linguistic legacy of Spanish and Portuguese: Colonial expansion and language change.
Cambridge: Cambridge University Press. ①
Coelho, F. A.
(1880–1886.) Os dialectos românicos ou neolatinos na África, Asia, e América. Bolletim da Sociedade de
Geografica de Lisboa. Republished in J. Morais-Barbosa (Ed.), Estudos linguísticos crioulos (1967)
Lisbon: Academia Internacional de Cultura Portuguesa. ①
De Bot, K.
(2001.) A bilingual production model: Levelt’s ‘speaking’ model adapted. In L. Wei (Ed.), The bilingualism
reader (pp. 420–441). London: Routledge. Reprint of De Bot (1992) Applied Linguistics, 13, 1–24. ①
②③
Diaz-Campos, M., & Clements, C.
(2005.) Mainland Spanish colonies and creole genesis: The Afro-Venezuelan area revisited. In S. Lotfi &
M. Westmoreland (Eds.), Proceedings of the Second Workshop on Spanish Sociolinguistics (pp. 41–
53). Someville, MA: Cascadilla. ①
Diaz-Campos, M., & Clements, C.
(2008.) A creole origin for Barlovento Spanish? A linguistic and sociohistorical inquiry. Language in
 Society, 37, 351–383. ①
Field, F.
(2004.) Second language acquisition in creole genesis: The role of processability. In G. Escure & A.
Schwegler (Eds.), Creoles, contact, and language change: Linguistic and social implications (pp. 127–
160). Amsterdam: John Benjamins. ①
Fernandez, M.
(2011.) Chabacano en Tayabas: Implicaciones para la historia de los criollos hispano-filipinos. Revista
Internacional de Linguistica Iberoamericana, 17, 189–218. ①
Fernandez, M.
(2012.) Leyenda e historia del chabacano de Ermita. UniverSOS: Revista de Lenguas Indigenas y
Universos Culturales, 9, 9–4/65–70. Retrieved from [Link]
.pdf ①
Fernández, M., & Sippola, E.
(2017.) A new window into the history of Chabacano: Two unknown mid-19th century texts. Journal of
Pidgin and Creole Languages, 32(2), 304–338. ①②
Filpulla, M., Klemola, J., & Sharma, D.
(2017.) The Oxford handbook of World Englishes. New York, NY: Oxford University Press. ①
Frake, C. O.
(1971.) Lexical origins and semantic structure in Philippine Creole Spanish. In D. Hymes (Ed.),
Pidginization and creolization of languages (pp. 223–242). Cambridge: Cambridge University Press.
①②
Goodman, M.
(1987.) The Portuguese element in the American creoles. In G. G. Gilbert (Ed.), Pidgin and Creole
languages: Essays in memory of John E. Reinecke (pp. 361–405). Honolulu, HI: University of Hawaii
Press. ①
Grant, A.
(2002.) El chabacano zamboangueño: Una lengua mezclada. Papia, 12, 7–40. ①
Grant, A.
(2011.) Substrate influences in Mindanao Chabacano. In C. Lefebvre (Ed.), Typological studies in
language: Creoles, their substrates, and language typology (pp. 303–324). Amsterdam: John
Benjamins. ①②③④⑤⑥
Hagemeijer, T.
(2013.) Santome. In S. Michaelis, P. Maurer, M. Haspelmath, & M. Huber (Eds.), The survey of Pidgin and
Creole languages (pp. 50–58). Oxford: Oxford University Press. ①
Hagemeijer, T., & Ogie, O.
(2011.) Edo influence on Santome: Evidence from verb serialization. In Lefebvre, C. (Ed.), Creoles, their
substrates, and language typology (pp. 37–60). Amsterdam: John Benjamins. ①②
Helms-Park, R.
(2003.) Transfer in SLA and creoles: The implications of causative serial verbs in the interlanguage of
Vietnamese ESL learners. Studies in Second Language Acquisition, 25, 211–244. ①
Hesseling, D. C.
 (1897.) Het Hollandsch in Zuid-Afrika. De Gids, 60(1), 138–162. Reprinted in English in D. Hesseling
the origin and formation of Creoles: A miscellany of articles (1979) Ann Arbor, MI: Karoma. ①

Hickey, R. (Ed) (2010.) The Handbook of Language Contact. Malden, MA: Wiley-Blackwell.
①
Hulstijn, J.
(1990.) A comparison between information processing and the analysis/control approaches to language
learning. Applied Linguistics, 11, 30–45. ①
Jacobs, B.
(2008.) Papiamentu: A diachronic analysis of its core morphology. Phrasis 2008(2), 59–82. ①
Jacobs, B.
(2009a.) The origins of Old Portuguese features in Papiamentu. In N. Faraclas, R. Severing, C. Weijer, &
L. Echteld (Eds.), Leeward voices: Fresh perspectives on Papiamentu and the literatures and cultures
of the ABC Islands (pp. 11–38). Curaçao: FPI/ UNA. ①
Jacobs, B.
(2009b.) The Upper Guinea origins of Papiamentu: Linguistic and historical evidence. Diachronica, 26,
319–379. ①
Jacobs, B.
(2010a.) Upper Guinea Creole: Evidence in favor of a Santiago birth. Journal of Pidgin and Creole
Languages, 26(2), 289–343. ①
Jacobs, B.
(2010b.) Los verbos estativos en el papiamento y el caboverdiano de Santiago. Revista Internacional de
Lingüística Iberoamericana VIII, 2(16), 129–155. ①
Jarvis, S., & Pavlenko, A.
(2008.) Crosslinguistic influence in language and cognition. New York, NY: Routledge. ①
Johanson, L.
(2002.) Contact-induced change in a code-copying framework. In M. C. Jones & E. Esch (Eds.), Language
change: The interplay of internal, external and extra-linguistic factors (pp. 285–313). Berlin: Mouton
de Gruyter. ①
Kouwenberg, S.
(2006.) L1 Transfer and the cut-off point for L2 acquisition processes in creole formation. In C. Lefebvre
L. White, & C. Jourdan (Eds.), L2 acquisition and Creole genesis (pp. 205–219). Amsterdam: John
Benjamins. ①
Lang, J.
(2009.) Les langues des autres dans la créolisation: Théorie et exemplification par le créole d’empreinte
wolof à l’île Santiago du Cap Vert. Tübingen: Gunter Narr. ①
Lang, J.
(2013.) Cape Verdean Creole of Santiago. In S. Michaelis, P. Maurer, M. Haspelmath, & M. Huber (Eds.),
The survey of Pidgin and Creole languages (p. 3–11). Oxford: Oxford University Press. ①
Lefevbre, C., White, L., & Jourdan, C. (Eds.) (2006.) L2 acquisition and Creole genesis. Dialogues.
Amsterdam: John Benjamins. ① ②
Léglise, I., & Chamoreau, C. (Eds.) (2013.) The interplay of variation and change in contact settings.
Amsterdam: John Benjamins. ①
Levelt, W.
(1989.) Speaking: From intention to articulation. Cambridge, MA: The MIT Press. ① ②
Levelt, W.
(1993.) Accessing words in speech production: Stages, processes and representations. In W. Levelt (Ed.),
Lexical access in speech production (pp. 1–22). Oxford: Blackwell. ①
Lipski, J.
(1994.) Latin American Spanish. New York, NY: Longman. ① ②
Lipski, J.
(2001.) The place of Chabacano in the Philippine linguistic profile. Estudios de Sociolingüística, 2(2), 119–
163. ① ②
Lipski, J.
(2004.) El español de América: Los contactos bilingües. In R. Cano (Ed.), Historia de la lengua Española
(pp. 1117–1138). Barcelona: Ariel. ①
Lipski, J.
(2005.) A history of Afro-Hispanic language: Five centuries, five continents. Cambridge: Cambridge
University Press. ①②
Lipski, J.
(2008a.) Afro-Bolivian Spanish. Frankfurt & Madrid: Vervuert & Iberoamericana. ①②③
Lipski, J.
(2008b.) Spanish-based creoles in the Caribbean. In S. Kouwenberg & J. V. Singler (Eds.), The handbook
of Pidgin and Creole studies (pp. 541–564). Malden, MA: Wiley-Blackwell.
Lipski, J.
(2009.) Spanish-based creoles in the Caribbean. In S. Kouwenberg & J. Singler (Eds.), The handbook of
Pidgin and Creole studies (pp. 543–564). Oxord: Wiley-Blackwell. ①②③
Lipski, J., & Santoro, M.
(2007.) Zamboangueño Creole Spanish. In J. A. Holm & P. L. Patrick (Eds.), Comparative Creole syntax
(pp. 373–398). London: Battlebridge. ①
Martinus, E. F.
(1996.) The kiss of a slave: Papiamentu’s West-African connections (Unpublished doctoral dissertation).
University of Amsterdam. ①
Mather, P.-A.
(2006.) Second language acquisition and creolization: Same (i-) processes, different (e-) results. Journal of
Pidgin and Creole Languages 21(2), 231–274. ①
Matras, Y.
(2009.) Language contact. Cambridge: Cambridge University Press. ①
Maurer, P.
(2009.) Principense: Grammar, text and vocabulary of the Afro-Portugese Creole of the Island of Principe,
Gulf of Guinea. London: Battlebridge. ①
McWhorter, J.
(2000.) The missing Spanish Creoles: Recovering the birth of plantation contact languages. Berkeley, CA:
University of California Press. ①
MacWhinney, B.
(1997.) Second language acquisition and the Competition Model. In J. Kroll & A. de Groot (Eds.),
Tutorials in bilingualism: Psycholinguistic perspectives (pp. 113–142). Mahwah, NJ: Lawrence
Erlbaum Associates. ①
MacWhinney, B.
(2005.) A unified model of language acquisition. In J. F. Kroll & A. M. B. de Groot (Eds.), Handbook of
bilingualism: Psycholinguistic approaches (pp. 49–67). New York, NY: Oxford University Press. ① ②
MacWhinney, B.
(2008.) A Unified Model. In P. Robinson & N. Ellis (Eds.), Handbook of cognitive linguistics and second
language acquisition (pp. 341–371). Mahwah, NJ: Lawrence Erlbaum Associates. ①
Megenney, W. W.
(1983.) La influencia del portugués en el palenquero colombiano. Thesaurus, 28, 548–563. ①
Mintz, S. W.
(1971.) The socio-historical background to pidginization and creolization. In D. Hymes (Ed.),
Pidginization and creolization of languages (pp. 481–96). Cambridge: Cambridge University Press. ①
②③④
Mufwene, S.
(2001.) The ecology of language evolution. Cambridge: Cambridge University Press. ①②
Mufwene, S.
(2008.) Language evolution: Contact, competition, and change. London: Continuum. ① ② ③ ④ ⑤
Mufwene, S.
(2013.) Driving forces in English contact linguistics. In D. Schreier & M. Hundt (Eds.), English as a
contact language (pp. 204–221). Cambridge: Cambridge University Press. ①
Mufwene, S.
(2014.) Language ecology, language evolution, and the actuation question. In T. Äfarli A. & B. Mælhum
(Eds.), The sociolinguistics of grammar (pp. 13–35). Amsterdam: John Benjamins. ①
Mufwene, S., & Vigouroux, C. B.
(2013.) Individuals, populations, and timespace: Perspectives on the ecology of language. In F. Gadet (Ed.),
Construction des connaises sociolinguistiques: Du terrain au positionnement théretique (pp. 111–137).
Brussels: EME & intercommunications. ① ②
Munteanu, D.
(1996.) El papiamento, lengua criolla hispánica. Madrid: Gredos. ①
Odlin, T.
(1989.) Language transfer: Crosslinguistic influence in language learning. Cambridge: Cambridge
University Press. ①
Odlin, T.
(1990.) Word-order transfer, metalinguistic awareness, and constraints on foreign language learning. In
Van Patten & J. F. Lee (Eds.), Second language acquisition/Foreign language learning (pp. 95–117).
Clevedon: Multilingual Matters. ①
Pienemann, M.
(1999.) Language processing and second language development: Processability Theory. Amsterdam: John
Benjamins. ①
Plag, I.
(2008a.) Creoles as interlanguages: Inflectional morphology. Journal of Pidgin and Creole Languages,
23(1), 114–135. ①②③
Plag, I.
(2008b.) Creoles as interlanguages: Syntactic structures. Journal of Pidgin and Creole Languages, 23(2),
307–328. ①②③
Plag, I.
(2011.) Creolization and admixture: Typology, feature pools, and second language acquisition. In P. Bhatt
& T. Veenstra (Eds.), Creoles and typology. Special Issue of Journal of Pidgin and Creole Languages,
26(1), 89–110. ①
Quint, N.
(2000a.) Grammaire de la langue cap-verdienne. Paris: l’Harmattan. ①
Quint, N.
(2000b.) Le cap verdien: Origines et devenir d’une langue métisse. Paris: l’Harmattan. ①
Roberts, S.
(2004.) The role of style and identity in the development of Hawai’ian Creole English. In G. Escure &
Schwegler (Eds.), Creoles, contact and language change: Linguistic and social implications (pp. 331–
350). Amsterdam: John Benjamins. ①
Roelofs, A.
(1993.) A spreading-activation theory of lemma retrieval in speaking. In W. J. M. Levelt (Ed.), Lexical
access in speech production (pp. 107–142). Cambridge MA – Oxford: Blackwell. ①
Sanchez, L.
(2006.) Bilingual grammars and creoles: Similarities between functional convergence and morphological
elaboration. In C. Lefebvre, L. White & C. Jourdan (Eds.), L2 acquisition and Creole genesis:
Dialogues (pp. 277–294). Amsterdam: John Benjamins. ①②
Schang, Emmanuel
(2000.) L’émergence des créoles portugais du Golfe de Guinée (Unpublished doctoral dissertation).
Université de Nancy II. ①
Schuchardt, H.
(1882.) Kreolische Studien I. Ueber das Negerportugiesische von S. Thomé (Westafrika). Sitzungsberichte
der kaiserlichen Akademie der Wissenschaften zu Wien, 101(2), 889–917. ①
Schumann, J. H.
(1978.) The pidginization process: A model for second language acquisition. Rowley, MA: Newbury
House. ① ②
Schwartz, A. L., & Kroll, J.
(2006.) Language processing in bilingual speakers. In M. J. Traxler & M. A. Gernsbacher (Eds.), Handbook
of psycholinguistics (2nd ed., pp. 967–999). Amsterdam: Elsevier. ①②
Schwegler, A.
 (1996a.) Chi ma nkongo: Lengua y rito ancestrales en El Palenque de San Basilio (Colombia). Frankfurt:
Vervuert. ①
Schwegler, A.
(1996b.) La doble negación dominica y la genesis del español caribeño. Hispanic Linguistics, 8, 246–315.
①
Schwegler, A.
(2006.) Bozal Spanish: Captivizing new evidence from a contemporary source (Afro-Cuban “Palo
Monte”). In L. L. Thornburg & J. M. Fuller (Eds.), Studies in Contact linguistics: Essays in honor of
Glenn. G. Gilbert (pp. 71–101). Bern: Peter Lang. ①
Schwegler, A.
(2010.) Pidgin and Creole studies: Their interface with Hispanic and Lusophone linguistics. Studies in
Hispanic and Lusophone Linguistics, 3(2), 1–51. ①②③④⑤⑥⑦
Sessarego, S.
(2013.) On the non-Creole bases for Afro-Bolivian Spanish. Journal of Pidgin and Creole Languages,
28(2), 363–407. ①②③④⑤
Sessarego, S.
(2014.) On Chota Valley Spanish origin: Linguistic and sociohistorical evidence. Journal of Pidgin and
Creole Languages, 29(1), 86–133. ①
Sessarego, S.
(2017.) The legal hypothesis of Creole genesis: Presence/absence of legal personality, a new element to the
Spanish creole debate. Journal of Pidgin and Creole Languages, 32(1), 1–47. ①
Siegel, J.
(2006.) Links between SLA and Creole studies: Past and present. In C. Lefevbre, L. White, & C. Jourdan
(Eds.), L2 acquisition and Creole genesis. Dialogues (pp. 15–46). Amsterdam: John Benjamins.
①②③
Siegel, J.
(2008.) The emergence of Pidgin and Creole languages. Oxford: Oxford University Press. ①
Siemund, P., & Kintana, N. (Eds.) (2008.) Language contact and contact languages. Amsterdam: John
Benjamins. ①
Silva-Corvalán, C.
(1998.) On borrowing as a mechanism of syntactic change. In A. Schwegler, B. Tranel, & M. Uribe-
Etxebarria (Eds.), Romance linguistics: Theoretical perspectives (pp. 225–246). Amsterdam: John
Benjamins. ①
Singler, J.
(2008.) The sociohistorical context of creole genesis. In S. Kouwenberg & J. V. Singler (Eds.), The
handbook of Pidgin and Creole studies (pp. 332–358). Malden, MA: Wiley-Blackwell. ①
Spitz, W.
(2001.) Hiligaynon/Ilongo. Munich: Lincom. ① ②
Sprouse, R. A.
(2006.) Full transfer and relexification: Second language acquisition and creole genesis. In C. Lefebvre
White, & C. Jourdan (Eds.), L2 acquisition and Creole genesis. Dialogues (pp. 169–183). Amsterdam:
 John Benjamins. ①
Steinkrüger, P.
(2008.) The puzzling case of Chabacano: Creolization, substrate, mixing and secondary contact. Studies in
Philippine Languages and Cultures, 9, 142–147. ① ② ③ ④
Thomason, S., & Kaufman, T.
(1988.) Language contact, creolization and genetic linguistics. Berkeley, CA: University of California
Press. ① ②
van Coetsem, F.
(1988.) Loan phonology and the two transfer types in language contact. Dordrecht: Foris. ①
③
van Coetsem, F.
(2000.) A general and unified theory of the transmission process in language contact. Heidelberg: Winter.
①
Weinreich, U.
(1953.) Languages in contact: Findings and problems. The Hague: Mouton. ① ② ③ ④ ⑤ ⑥ ⑦
Weinreich, U., Labov, W., & Herzog, M. I.
(1968.) Empirical foundations for a theory of change. In W. P. Lehman & Y. Malkiel, Directions for
historical linguistics: A symposium (pp. 95–188). Austin, TX: University of Texas Press. ①
Winford, D.
(2008.) Processes of creolization and related contact-induced change. Journal of Language Contact, Thema
2, 124–145. ①
Winford, D.
(2013.) On the unity of contact phenomena: The case for imposition. In de C. Féral (Ed.), In and out of
Africa: Languages in question. In honour of Robert Nicolaï, 1: Language contact and epistemological
issues (43–71). Louvain: Peeters. ①
Winford, D.
(2017.) The ecology of language and the new Englishes: Toward an integrative framework. In A.
Mauranen, J. Klemola, S. Vetchinnikova, & Filppula, M. (Eds.), Changing English: Global and local
perspectives (pp. 25–55). Berlin: De Gruyter. ①
CHAPTER 2

Chocó Spanish
An Afro-Hispanic language on the Spanish
frontier
Sandro Sessarego
University of Texas at Austin | Freiburg Institute for
Advances Studies | Helsinki Collegium for Advanced
Studies | Foro Latinoamericano de Antropología del
Derecho | Institut Universitari de Drets Humans de la
Universitat de València

Abstract
This paper presents sociohistorical and linguistic data to cast light on
the origin and nature of Chocó Spanish (CS), an Afro-Hispanic dialect
spoken in the Pacific lowlands of Colombia. This research suggests
that neither the Decreolization Hypothesis (Granda, 1977; Schwegler,
1991a, 1991b) nor the Afrogenesis Hypothesis (McWhorter, 2000) can
account for the phenomena encountered in contemporary CS.
Conversely, the present study indicates that the grammatical elements
found in this dialect may be better analyzed as the result of advanced
second language acquisition strategies, which were conventionalized at
the community level in a sociohistorical context in which black
captives had relatively good access to the colonial language.
 Keywords
Chocó Spanish; Decreolization Hypothesis;
Afrogenesis Hypothesis; Spanish Creole debate; black slavery

1. Introduction
The origins of the Afro-Hispanic contact varieties of the Americas have been a
hot topic in the field of contact linguistics for several decades now. In fact, the
contemporary lack of Spanish creoles in Latin American regions consists of an
unsolved puzzle for the pidgin and creole studies community. One of the most
important “pieces” in the “Afro-Hispanic creole puzzle” is certainly Chocó
Spanish (CS), an Afro-Hispanic dialect spoken in the Pacific lowlands of
Colombia, in the Department of Chocó, a region in which black people represent
95% of the total population (CIDCM, 2015). These people are the descendants
of the slaves who were taken to this region during the colonial period to work
the rich gold mines of the area.
Even if the morphosyntax and phonology of CS is somehow divergent from
standard Colombian Spanish, this dialect is perfectly intelligible by any speaker
of standard Spanish. Thus, this contact variety does not present the more radical
traces of grammatical restructuring that can be easily encountered in creole
languages (Holm & Patrick, 2007). At a first look, the lack of a Spanish creole in
this region may appear quite surprising. In fact, it has often been said that
colonial Chocó presented all the conditions that have generally been held
responsible for language creolization in other colonial territories: (a) thousands
of African-born slaves speaking a variety of African languages, (b) minimal
presence of white people speaking Spanish, (c) harsh working conditions in gold
mines, (d) a highly isolated region, far away from Spanish-speaking urban
centers (McWhorter, 2000, p. 9).
Two main hypotheses have been proposed in the literature to account for
this situation. Granda (1977) and Schwegler (1991a, 1991b) indicate that a
creole language may have existed in colonial Chocó, as well as in several other
Latin American Spanish colonies, and that it would have subsequently
decreolized due to more recent contact with standard varieties of Spanish
(Decreolization Hypothesis). They suggest that several of today’s Afro-Hispanic
contact dialects probably went through the same decreolizing process, and that
they would have originally derived from one Portuguese-based pidgin/creole,
originally formed in Africa and subsequently taken to the Americas. On the
other hand, McWhorter (2000, Chapter 2), in line with his Afrogenesis
Hypothesis of creole formation, claims that the sociohistorical conditions for a
creole to emerge were well in place in colonial Chocó, but a Spanish creole
never formed in this region. The reason behind this situation would be that Spain
did not trade directly in African slaves. As a result, a Spanish pidgin was not
introduced into the Americas from Western Africa and, consequently, the
linguistic bases for the development of a full-fledged creole language would
have been missing in the Colombian Pacific lowlands, and in the rest of the
Spanish colonies overseas.
This paper is meant to assess the origin and nature of CS and its
implications for creole studies. There is plenty of research that has been carried
out in other fields such as history and law, to which, so far, not much attention
has been paid by linguists. In recent decades, some attempts to combine
historical and linguistic data to cast light on the origin of certain Afro-Hispanic
Languages of the Americas (AHLAs) have been carried out for the Caribbean
and the Andean region (Mintz, 1971; Laurence, 1974; Díaz-Campos &
Clements, 2008; Clements, 2009; Sessarego, 2013a, 2014, 2015). Nevertheless,
for the Colombian Pacific lowlands almost no research of this type has ever been
done. This study explores the historical and linguistic evidence available for CS
and challenges the postures that would picture colonial Chocó as the perfect
place for a Spanish creole to develop (Afrogenesis Hypothesis) or as a location
in which a Spanish creole once existed and then gradually disappeared
(Decreolization Hypothesis).

Figure 1. Overhead view of the Department of Chocó, Colombia

This paper consists of five sections. Section 1 is the introduction to this work.
Section 2 offers an analysis of the so-called “Spanish Creole debate”. In so
doing, it situates Chocó Spanish (CS) within such a context and illustrates the
hypotheses that have been proposed in the literature to account for the
development of this Afro-Hispanic dialect. Section 3 provides an analysis of CS
“creole-like” features and shows that such grammatical phenomena can be
conceived as the traces of advanced second language acquisition strategies,
which do not necessarily imply a previous creole stage – contrary to what has
been traditionally indicated in the literature. Section 4 is a sociohistorical
analysis of slavery in the Chocó. It shows that this region did not present the
characteristics that have traditionally been held responsible for creole formation
in other American colonies. Finally, Section 5 summarizes the content of this
paper and provides the conclusions.
2. The place of Chocó Spanish in the Spanish creole
debate
For more than four decades now scholars interested in the origin and evolution
of Afro-European contact languages in the Americas have tried to figure out why
Spanish creoles are only spoken in two very circumscribed regions of Latin
America, in contrast to the much more widespread use of English- and French-
based creole varieties (cf. Lipski, 2005, Chapter 9). In fact, it is a well-known
fact that contemporary Latin American Spanish creoles can only be found in the
former maroon community of San Basilio de Palenque (Colombia) – where
Palenquero is spoken – and in the so-called ABC-triangle, the Caribbean islands
of Aruba, Bonaire and Curaçao (Netherland Antilles), where Papiamentu is
used.1
If we consider that Spain was the European country that ‘discovered’ the
‘New World’ and that became one of the most influential powers in the
colonization of the Americas, it may appear somehow surprising to observe that
Spanish creoles are so reduced in number and size. Moreover, according to some
scholars (cf. Goodman, 1987; Schwegler, 1993; McWhorter, 2000; Jacobs, 2012;
among others), Papiamentu and Palenquero should only be classified as Spanish
creoles in a strictly synchronic sense, since – in their view – these languages
most likely started as Portuguese creoles and only in a subsequent phase went
through a process of Spanish relexification.
The relative paucity of Spanish creoles in the Americas has captured the
attention of several researchers, who proposed a number of models to account
for this state of affairs. One early attempt to provide an analysis of the given
facts is the hypothesis initially postulated by Granda (1968, 1970), and
subsequently supported by a number of followers (e.g., Otheguy, 1973;
Megenney, 1984a, 1984b; Schwegler, 1991a, 1991b; etc.). According to these
authors, a once-widespread Afro-Iberian creole used to be spoken by the black
communities of colonial Spanish America. This language would have derived
from one single Portuguese-based creole, which originally developed around the
15th century from the early contacts between the Portuguese merchants and the
African groups found throughout the Western African coast. This hypothetical
Pan-American creole would have approximated to Spanish over time
(Decreolization Hypothesis), due to the pressure exerted by the standard norm
and the stigmatization attached to this black vernacular. For this reason,
according to the supporters of the Decreolization Hypothesis, besides
Papiamentu and Palenquero, which preserved most of the original Portuguese-
creole structure, all the remaining Afro-Hispanic varieties of the Americas
would have verged on Spanish to the point of presenting today only a few
creole-like features. According to these scholars, such grammatical elements,
which can also be found in colonial literary texts depicting black speech,
represent a key piece of evidence in support of the monogenetic model, since
their parallel existence in these black communities scattered across the Americas
would be very difficult to explain in terms of independent developments
(Schwegler, 1991b, p. 77).
This early attempt to account for the current paucity of Spanish creoles has
faced some criticism. Some researchers, in fact, have objected that due to a
convergence of historical factors, the Spanish Caribbean differed quite
significantly from the English and the French Antilles, so that the demographic
and socioeconomic conditions in colonial times would not have been ripe for
Spanish creolization to take place in Cuba, Puerto Rico and the Dominican
Republic (Mintz, 1971; Laurence, 1974; Lipski, 1986, 1994; Chaudenson, 2001;
Clements, 2009). For these reasons, the grammatical features attested in the
colonial Spanish Caribbean for the speech of bozales (black captives born in
Africa, who did not speak Spanish natively) should not be taken as evidence of a
Pan-American Afro-Iberian creole, but rather as the traces of non-target-like,
substrate-driven, learning strategies, “which arose spontaneously each time
Spanish and African languages came into contact” (Lipski, 1986, p. 171).
McWhorter (1997, 2000) acknowledges that the Spanish Antilles,
especially before the sugar boom of the 19th century, did not present the
conditions that have traditionally been considered responsible for creole
formation. However, he claims that in the mainland Spanish colonies, and
particularly, in Veracruz (Mexico), Chincha (Peru), Chota Valley (Ecuador),
coastal Venezuela, and the Chocó (Colombia), the conditions for creole
formation were perfect, but yet Spanish creoles did not develop (McWhorter,
2000, Chapter 2). In his view, the reason why this did not happen should not be
sought in the Americas but rather on the other side of the ocean, in Africa, from
where captives were shipped to the New World. McWhorter (2000) believes that
the creole languages spoken today in the Americas developed out of the pidgins
that formed on the Western African coasts from the contacts between European
traders and the Africans involved in the slave trade (Afrogenesis Hypothesis).
According to this model, the real cause behind the non-creolization of Spanish in
the Americas would be that Spain, unlike the other European powers involved in
the colonization of the New World, did not trade directly in African slaves, such
that a Spanish pidgin never developed in Africa and, as a result, a Spanish creole
could not possibly form in the Americas.
McWhorter (2000, pp. 7–10) presents the Chocó case as the most
emblematic example in which the conditions for creolization were perfect, but
yet a Spanish creole could not possibly form. He states:
Starting in the late seventeenth century, the Spanish began importing massive
numbers of West Africans who spoke a wide variety of languages into the
Pacific lowlands of northwestern Colombia to work their mines […]. In the
Chocó region […] there were no fewer than 5,828 black slaves by 1778, while
there were only about 175 whites – a mere 3 percent of the total population
(West, 1957, pp. 100, 108) […]. What is important is that creole theory predicts
that the Chocó context would have generated not a second-language dialect
diverging only slightly from the local standard, but a more radically reduced,
pidginized register, with much higher levels of structural interference from West
African languages. In short, the modern situation in the Chocó is a striking
counterexample to current creole genesis theory, all strains of which would
predict a Spanish creole in this region.2

3. Chocó Spanish “creole-like” features

Besides providing considerable ground for hypotheses on creole genesis and
evolution, CS and the rest of the AHLAs also have much to offer to linguistic
theory, since these varieties are rich in structures that would be considered either
ungrammatical or pragmatically infelicitous in standard Spanish and that may be
used as a powerful testing ground for linguistic hypotheses, which have usually
been built on standardized language data (Kayne, 1996; Sessarego, 2013b,
2014). In fact, some common features that have repeatedly been reported for the
vast majority of these Afro-Hispanic dialects, and that in some cases have been
identified as potential indicators of a previous creole stage, represent deviations
from standard Spanish that are extremely interesting from a microparametric
point of view (Sessarego, 2012, 2013b, 2014). Some of these recurring
grammatical phenomena are: (a) presence of bare nouns in subject position; (b)
variable number and gender agreement across the DP; (c) invariant verb forms
for person and number; (d) use of non-emphatic, non-contrastive overt subjects;
(e) lack of subject-verb inversion in questions.
CS presents all of the grammatical features indicated in (a-e), which are
here reproduced for this dialect in Table 1 and that have been taken by the
supporters of the Decreolization Hypothesis as indicators of a previous creole
stage for this and other Afro-Hispanic dialects (cf. Granda, 1976; Schwegler,
1991b; among others). The main argument adopted by these scholars is that,
given the simultaneous presence of these grammatical elements in several black
vernaculars, it would be difficult to accept that such linguistic patterns arose as
the result of independent parallel developments; rather, these authors claim that a
single proto-creole presenting those features must have existed in colonial times,
and that the Afro-Hispanic dialects showing such grammatical characteristics
today should be seen as the direct offspring of such a linguistic ancestor.

Table 1. Five commonly reported Chocó Spanish features traditionally ascribed to

a previous creole stage

Phenomenon Example

a. Presence of Cabeza mía no es pa’ eso no, maestro. ‘My head is not for this, sir’ (Ruiz García,
bare nouns in 2009: 45). Pollo vive allá. ‘Chickens live here’ (Sessarego 2019: Chapter 3).
subject
position
b. Variable Hacen unos caney. ‘They make some huts’ (Flórez, 1950: 423). Quieren cosa ligero ‘They
number and want something light’ (Ruiz García, 2009: 43).
gender
agreement
across the DP
c. Invariant Ellas dijo <dijeron> así, no sé si é veldá. ‘They said so, I do not know if that is the truth’
verb forms for (Sessarego 2019: Chapter 3). Esa vez vino <vinieron> catorce familias, catorce familias
person and vinieron. ‘Fourteen families came this time, fourteen families came’ (Dieck 1993: 21).
number
d. Use of non- Cuando él hace en la casa de él, me llama él. ‘When he makes it at his house, he calls
emphatic, non- me’ (Rodríguez Tocarruncho, 2010: 61). Mi papá, yo casi no pasé con él no. Yo casi no
contrastive me crié con él. ‘My father, I did not spend much time with him. I almost did not grow up
overt subjects with him’ (Ruiz García, 2009: 49).
e. Lack of ¿Cómo ella se llama? ‘What is her name?’ (Sessarego 2019: Chapter 3). ¿Cuántos usted
subject-verb tuvo, entre vivos y muertos, cuántos hijos usted llegó a tener? ‘How many did you have,
inversion in between alive and dead, how many sons / children did you have?’ (Ruiz García,
questions 2009: 49).

As the examples in Table 1 suggest, CS, in line with many other Afro-Hispanic
varieties, diverges quite significantly from standard Latin American Spanish.
However, the features encountered in this dialect should not necessarily be taken
as traces of a previous creole stage, since they are also well-attested patterns in
advanced second-language varieties and, in some cases, in certain native Spanish
dialects (cf. Sessarego, 2013b, 2014, Chapter 2).
During the past few years, I have proposed an analysis of the features
reported in Examples (a–e) that builds on recent theoretical models on the nature
of the language faculty, its modularity, and the challenges that certain
constructions may pose for the interaction between different language modules
(Sessarego, 2013b, 2015; Sessarego & Gutiérrez-Rexach, 2018; Romero &
Sessarego, 2018; Sessarego & Rao, 2016; Rao & Sessarego, 2016). In line with
recent theoretical proposals within the field of language development, I assume
that certain constructions involving high processing demands on the interface
between different linguistic modules may be more difficult to master in second
language acquisition (SLA) (Sorace, 2003; Sorace & Serratrice, 2009; Rothman
& Slabakova, 2011). Along these lines, I analyze the linguistic patterns found in
several AHLAs as the result of advanced SLA strategies. I claim, therefore, that
several grammatical aspects of these contact varieties can actually be seen as
advanced, conventionalized L2 features, which do not necessarily imply any
previous creole stage for CS and the other Afro-Hispanic dialects of the
Americas (Sessarego, 2013b).
A closer look at the so-called “creole-like” features clearly shows that they
can be systematically encountered in advanced L2 varieties of Spanish. The
existence of bare nouns in argument position in L2 Spanish speech is not a rare
feature to encounter, especially if the speakers’ L1 has a significantly divergent
determiner system (Sánchez & Giménez, 1998; Leonini, 2006; García Mayo &
Hawkins, 2009). CS, in line with the majority of the AHLAs I have analyzed (cf.
Sessarego, 2013a, 2014, 2015), presents a determiner system that closely
resembles that of standard Spanish; the main deviation has to do with the fact
that in CS bare nouns may appear in argument position and take on a variety of
semantic readings (i.e., plural/singular, specific/generic), if certain pragmatic
requirements are met (cf. Sessarego, 2019, Chapter 3; Gutiérrez-Rexach &
Sessarego, 2011).
Variable gender/number agreement in the DP is another well-known feature
detectable in L2 varieties of Spanish. Franceschina (2005) shows how gender
features are acquired late in SLA and sometimes their complete mastery is never
obtained. Moreover, it should be pointed out that variable number agreement is
not only common in L2 varieties; rather, there are several native dialects of
Spanish presenting similar cases of agreement variability (i.e., Caribbean
Spanish, Andalusian Spanish, etc.) (cf. Poplack, 1980; Lipski, 1994).
Given their lack of semantic import to the interpretation of phrases (cf.
Béjar, 2008), phi-features tend to be difficult to master in an L2. This is why phi-
agreement morphology has been labeled by Slabakova (2008) as the “bottleneck
of acquisition”. Such a situation not only accounts for the reduction of gender
and number agreement morphology across the AHLA DP, it also provides a
coherent explanation for the variable subject-verb agreement configurations
encountered in these contact vernaculars, in line with recent SLA analyses (cf.
McCarthy, 2008; Slabakova, 2009).
The use of non-emphatic, non-contrastive overt subjects in the AHLAs (as
well as in Caribbean Spanish) can be seen as a compensatory strategy related to
the weakening of the subject-verb agreement system (Sessarego, under review,
Chapter 3). Moreover, it has long-been shown by researchers working on SLA
that the proficient mastery of overt subject pronouns in Null Subject Languages
(cf. Rizzi, 1982) represents a big L2 challenge. In fact, it is highly constrained by
processes applying at the syntax/pragmatics interface (Sorace, 2000, 2003), since
both structural and discourse features are involved.
Finally, the lack of subject-verb inversion in questions not only appears to
be common in a number of non-creolized Spanish dialects (i.e., those from the
Caribbean), it also persists in the speech of advanced L2 and heritage speakers
cross-linguistically (Pienemann, 1998, 2005; Cuza, 2013; Guerra Rivera et al.,
2015). The evidence provided by the aforementioned studies clearly indicates
that phenomena (a-e) should not be necessarily taken as indicators of a previous
creole stage for CS and the rest of the AHLAs, since they are also systematically
found in advanced L2 speech.3

4. Sociohistorical factors with linguistic consequences:

Black slavery in the Chocó
The colonial and post-colonial history of the Chocó is strongly connected to the
socio-political development of its surrounding regions. In fact, the powerful
colonial miners residing in Antioquia, Cali and Popayán were those who pushed
the Spanish colonial enterprise toward this remote frontier until conquering the
region, one of the richest mineral areas of the Americas (Colmenares, 1979).
In particular, the principal actors in this colonial mission were the mining
families from Popayán, who, after several attempts to penetrate the region,
finally managed to defeat the native populations by the end of the 17th century
(Hansen, 1991). From that point until the abolition of slavery in 1851, several
white and mestizo entrepreneurs entered the region with their gangs of black
slaves (cuadrillas) to exploit the rich gold mines of the province. In 1851 slavery
was formally abolished in Colombia. This resulted in a shortage of laborers in
the Chocó mines. Consequently, the majority of the white and mestizo people
residing in the region abandoned the area leaving behind their former slaves,
whose descendants form today more than ninety percent of the local population
(Sharp, 1976).

4.1 Spanish conquest and mineral exploitation (1500–1851)

The northern part of the Chocó was one of the first mainland regions explored by
the Spaniards. Since the first decades of the 16th century, the conquerors knew
that this region was rich in mineral resources; nevertheless, for more than two
centuries this region remained out of their colonial control (Sharp, 1976, p. 15).
Indeed, the presence of bellicose native populations, tropical diseases and a hot
climate discouraged Spanish establishment in the area. Even if they were not
attracted by the option of living and residing in the Chocó, the Spaniards tried to
conquer the area on several occasions, but, as it is well known, this did not turn
out to be an easy task (cf. Sharp, 1976, Chapter 2) and according to Hansen
(1991, p. i), it took them “nearly 300 years to bring the Chocó under the Crown’s
control”.
The invasion of the Chocó became more systematic by the mid-1660s and
proceeded from the surrounding regions of Popayán and Antioquia, which were
already under Spanish control by that time. During this phase, both payaneses
and antioqueños penetrated the Chocó. The former entered from the southern
part of the region, establishing a foothold among the Noanamá Indians and
subsequently among the Tatamá and the Chocó, while the latter advanced from
the northern side, thus fighting against the Citará tribe. Soon after the first
successful conquering attempts, the gobernación of Popayán and the
gobernación of Antioquia began disputing the jurisdictional control over the
areas recently colonized. The conflict over the Chocó between these two
gobernaciones lasted for several years and was eventually resolved by the end of
the 1680s, when Popayán took effective administrative control over the region
(Hansen, 1991, p. 97).
Several Indian uprisings took place during those years. Some of them were
so brutal that they resulted in the killing of many Spanish miners, their slaves
and some missionaries. In 1669 four Franciscan missionaries were killed by
natives living in the Atrato region. Another big Indian rebellion happened in
1684 in the Quibdó region. The natives joined forces with some black slaves and
managed to take control over the area. The Spanish army led by Governor Juan
Bueso de Valdés put an end to the Indian uprising in 1686. This was the last big
conflict that saw the Indians as protagonists of a successful rebellion. From that
point on, the Spaniards managed to systematically increase their control over the
Chocó and more and more entrepreneurs proceeding from Popayán entered the
region with their slave gangs to carry out mining activities (Hansen,
1991, p. 367).
Colmenares (1972, 1979) has researched the role played by Popayán in the
colonization of the Chocó and the use of black workers to exploit the mineral
resources of the region. He has shown that many miners from Popayán
transferred their slaves to the more recently conquered Chocó mines between the
end of the 17th and the beginning of the 18th century. He also indicates that the
majority of these captives were not directly proceeding from Africa, but rather
from other parts of Colombia or from already settled colonies, such as the
Caribbean; thus, they were not African-born bozales, but rather American-born
criollos. (cf. also West, 1953). This claim is in line with Cantor’s (2000) analysis
of early colonial Chocó. In fact, after studying the composition of several
cuadrillas, he concludes that “los mulatos (junto con los criollos) participaron
en el proceso de construcción cultural desde los primeros años de la ocupación
española [the mulattoes (as well as the criollos) took part in the process of
cultural construction from the very beginning of the Spanish occupation]”.
Remarks of this kind are not exceptional (cf. Colmenares, 1972; Marzahl,
1978) and are not only limited to the first phase of the colonization of the Chocó
(cf. West, 1957; Sharp, 1976; Cantor, 2000). An analysis of 18th century slave
gangs’ composition highlights how some 60% of the people working in the
Chocó at that time were either criollos or mulatos (cf. Cantor, 2000, pp. 49–50).
The author further points out how the majority of the captives consisted of
locally born slaves, who could speak Spanish natively. He also indicates that the
rest of the slaves proceeded from different parts of Africa, so that they were
highly heterogeneous, from both a linguistic and a cultural perspective. This
situation, in his view, eventually favored the spread of the Spanish language and
the Christian faith among the Chocoan cuadrillas (2000, p. 56):
Con base en la documentación histórica disponible se piensa que la agregación
de gentes caracterizadas por la extraordinaria diversidad de procedencias
determinó una situación de relativa fragmentación lingüística y cultural dentro
de cada cuadrilla, contexto dentro del cual se desarrolló el contacto entre los
diversos negros y de éstos con los blancos. Tal fragmentación implica la usencia
de una cultura, un idioma y un sistema de creencias comunes y compartidas por
todos los miembros de la cuadrilla […]. El concepto de fragmentación […] se
plantea como un hecho relativo por cuanto existía un alto número de mulatos y
criollos, quienes compartían como idioma común el castellano y eventualmente
conocían el cristianismo.
[Based on the available historical documentation, we may think that the
aggregation of people proceeding from highly divergent backgrounds implied a
linguistic and cultural fragmentation within the slave gang, a context in which
the contact among blacks and whites developed. Such a fragmentation implies
the absence of shared culture, language and beliefs […]. The concept of
fragmentation was, nevertheless, relative since there was a big number of
mulattoes and criollos, who shared the Castilian language and probably knew the
Christian faith].
These arguments can also be encountered in Sharp’s (1976) analysis of colonial
Chocó, in which he states that a convergence of sociohistorical factors
differentiated blacks’ living conditions in this region from the conditions of
slaves in other colonial settings across the Americas (i.e., French, Dutch, English
plantations): slaves worked, on average, 260 days a year, since during the
remaining time they were off to provide for themselves and their families
(1976, p. 134); they could accumulate goods, gold and other properties to pay for
their manumission (1976, p. 135), abuse of slaves was remarkably rare
(1976, p. 136), they were instructed in the precepts of the Catholic religion daily
(1976, p. 139), marriage was encouraged and family units preserved
(1976, p. 140). These social dynamics appear to have favored the acquisition of
Spanish by the bozal community and thus decreased the chances of creole
formation and/or preservation in the region. The information provided by the
aforementioned historians, therefore, sharply clashes with the view that would
depict colonial Chocó as the perfect breeding ground for a Spanish creole to
develop (cf. McWhorter, 2000).

4.2 End of slavery and underdevelopment in present-day Chocó

With the end of slavery in 1851, the already-few white residents that remained in
the Chocó up to that point left the region, since they could no longer exploit the
rich mineral resources of the Department by relying on slaves (Sharp,
1976, p. 16). The departure of the ruling class resulted in an almost complete
lack of interest in developing this area on the part of the Government in the years
to follow.
 During the last century, private foreign companies have become interested
in the exploitation of the mineral resources of the region. Oftentimes powerful
transnational corporations, in agreement with the Colombian governments, have
been granted permission to carry out mining operations on a large scale in order
to extract mineral resources, in particular gold, silver and platinum. This has
caused considerable tension among the local farmers, fishermen and traditional
miners, who have often been deprived of their lands and forced to accept
“infamous compensations” in exchange (Escalante, 1971, p. 105). The use of big
excavators and dredges, combined with the dumping of chemical products in the
local rivers (especially mercury, related to the extraction of gold), has caused
serious damage to the local environment and its inhabitants (1971, pp. 113–126).
To worsen the already complex social context of the Chocó, starting in the
1960s, guerrilla warfare has spread throughout the region, so that the
Revolutionary Armed Forces of Colombia (FARC), and a number of other local
militias, have been carrying out a variety of illegal activities, often related to
extortions, kidnappings, and drug trafficking.
After almost two centuries since the end of slavery in Colombia, the Chocó
remains one of the poorest regions in the country. With a large comparative
deficit in terms of public infrastructures, schools, media communication, and
basic services, the Department of Chocó is a region that has remained – to a
significant extent – isolated from the rest of the country. Given these conditions,
the chances of a massive decreolization process pushed by the pressure of the
standard norm, which would have wiped away a pre-existent creole language
(i.e. the Decreolization Hypothesis), appear to be quite slim.

5. Conclusions
Of all the Afro-Hispanic languages of the Americas (AHLAs), Chocó Spanish
(CS) represents perhaps one of the most enigmatic varieties. In fact, CS is
spoken in a region that has always been considered as an ideal place for creole
formation and/or preservation but, from a linguistic point of view, this contact
vernacular classifies more as a “Spanish dialect” than a “Spanish creole”.
Two different hypotheses have been provided in the literature to account for
this situation. On the one hand, the Decreolization Hypothesis (Granda, 1968, et
seq; Schwegler, 1991a, 1991b; etc.), has proposed that CS, along with the rest of
the AHLAs, used to be a creole that decreolized due to more recent contact with
standard Spanish, so that certain grammatical features found in this variety (i.e.,
variable gender agreement across the DP, lack of subject-verb inversion in
questions, etc.) would be seen as the remaining traces of a previous creole stage.
On the other hand, the Afrogenesis Hypothesis (McWhorter, 2000) has claimed
that, even though the conditions were perfect for creolization in the Chocó,
Spanish never creolized in the Americas because a Spanish pidgin never formed
in Africa and – as a consequence – a Spanish creole could not possibly develop
in the New World.
The present paper has challenged both hypotheses and has suggested that
CS may be better analyzed as the byproduct of advanced SLA strategies, which
do not imply any previous (de)creolization phase for this variety (Sessarego,
2013a). Moreover, a closer analysis to the sociohistorical facts has indicated that
colonial and post-colonial Chocó did not present the conditions for a
(de)creolization process to take place. Consequently, the patterns reported in
Table 1 have been analyzed in this paper as advanced, conventionalized L2
features, which crystallized at the community level in an environment in which
black slaves had a relatively good access to the colonial language (cf. also
Sessarego, 2015).
An investigation on the genesis and evolution of any AHLAs, I believe,
should never divorce the linguistic analysis from the social one. For this reason,
I have not only tried to document and explain the nature of CS grammar; rather, I
have also explored the historical facts that lead to its formation. Thus, this study
has presented an analysis of CS that is rooted in both linguistic and
sociohistorical research. I think similar multidisciplinary approaches may help us
better understand the dynamics of contact-driven language evolution, in the
Americas and beyond.

Notes
1. An anonymous reviewer points out the existence of at least one other Spanish-based creole, Chabacano,
from the Philippines (cf. Sippola, 2011). This creole is certainly important to understanding the evolution of
Spanish-contact varieties in colonial settings. Nevertheless, Chabacano is not spoken in the Americas (but
in Asia) and its substrate languages did not proceed from Africa. For this reason, the genesis and evolution
of Chabacano is not central to the formation of the AHLAs. ①

2. With the general expression “creole theory”, McWhorter refers to what he has labeled “the limited
access model” or the misleading assumption (in his view) that creole formed because Africans did not have
access to the superstrate language. ①

3. The interested reader may want to consult Montrul (2004) and Geeslin (2013) and references therein for
additional information on the development of these grammatical features in L2 Spanish varieties. ①

References
Béjar, S.
(2008.) Conditions on Phi-Agree. In D. Harbour, D. Adger, & S. Béjar (Eds.), Phi-theory (pp. 130–154).
Oxford: OUP. ①
Cantor, E.
(2000.) Ni aniquilados, ni vencidos. Bogotá: Instituto Colombiano de Antropología e Historia. ① ② ③
Chaudenson, R.
(2001.) Creolization of language and culture. London: Routledge. ①
CIDCM
(2015.) Center for International Development and Conflict Management. Retrieved from [Link]
.[Link]/mar/[Link]?groupId=10001 ①
Clements, J. C.
(2009.) The linguistic legacy of Spanish and Portuguese. Cambridge: Cambridge University Press.
①②
Colmenares, G.
(1972.) Historia económica y social de Colombia 1.537–1.719. Bogotá: Ediciones Culturales. ① ②
Colmenares, G.
(1979.) Historia económica y social de Colombia: Popayán una sociedad esclavista 1680–1800. Bogotá:
La Carreta. ① ②
Cuza, A.
(2013.) Crosslinguistic influence at the syntax proper: Interrogative subject-verb inversion in heritage
Spanish. International Journal of Bilingualism, 17, 71–96. ①
Dieck, M.
(1993.) Notas breves sobre el español de Bahía Solano (choco). Colombia país plurilingüe, 21–3. ①
Díaz-Campos, M., & Clements, J. C.
(2008.) A Creole origin for Barlovento Spanish? A linguistic and socio-historical inquiry. Language in
Society, 37, 351–383. ①
Escalante, A.
(1971.) La minería del hambre: Condoto y la Chocó Pacífico. Medellín: Ediciones Universidades. ① ②
Franceschina, F.
(2005.) Fossilized second language grammars (Language Acquisition and Language Disorders 38).
Amsterdam: John Benjamins. ①
Flórez, L.
(1950.) El habla del Chocó. Thesaurus, 6(1), 110–116. ①
García Mayo, M., & Hawkins, R.
(2009.) The second language acquisition of articles (Language Acquisition and Language Disorders 49
Amsterdam: John Benjamins. ①
Geeslin, K.
(2013.) The handbook of Spanish second language acquisition. Hoboken, NJ: Wiley-Blackwell.
Goodman, M.
(1987.) The Portuguese element in the American Creoles. In G. Gilbert (Ed.), Pidgin and Creole
languages: Essays in memory of John E. Reinecke (pp. 361–405). Honolulu, HI: University Press of
Hawaii. ①
Granda, G. de
(1968.) La tipología criolla de dos hablas del área lingüística hispánica. Thesaurus, 23, 193–205. ① ②
Granda, G. de
(1970.) Un temprano testimonio sobre las hablas ‘criollas’ en África y América. Thesaurus, 25(1), 1–11.
Granda, G. de
(1976.) Algunos rasgos morfosintácticos de posible origen criollo en el habla de áreas hispanoamericanas
de población negra. Anuarios de Letras, 14, 5–22. ①
Granda, G. de
(1977.) Estudios sobre un área dialectal hispanoamericana de población negra: Las tierras bajas
occidentales de Colombia. Bogotá: Publicaciones del Instituto Caro y Cuervo. ① ②
Gutiérrez-Rexach, J., & Sessarego, S.
(2011.) On the nature of bare nouns in Afro- Bolivian Spanish. In J. Herschensohn (Ed.), Romance
linguistics 2010 (Current Issues in Linguistic Theory 318) (pp. 191–204). Amsterdam: John
Benjamins. ①
Guerra Rivera, A., Coopmans, P., & Baauw, S.
(2015.) On the L2 acquisition of Spanish subject-verb inversion. Procedia – Social and Behavioral
Sciences, 173, 37–42. ①
Hansen, A.
(1991.) Conquest and colonization in the Colombian Chocó, 1510–1740 (Unpublished doctoral
dissertation). University of Warwick. ① ② ③ ④
Holm, J. A., & Patrick, P. L. (Eds.) (2007.) Comparative Creole syntax. Parallel outlines of 18 Creole
grammars. London: Battlebridge. ①
Jacobs, B.
(2012.) Origins of a Creole: The history of Papiamentu and its African ties. Berlin: De Gruyter.
Kayne, R.
(1996.) Microparametric syntax: Some introductory remarks. In J. Black & V. Motapanyane (Eds.),
Microparametric syntax and dialect variation (pp. ix–xxviii). Amsterdam: John Benjamins.
①
Laurence, K.
(1974.) Is Caribbean Spanish a case of decreolization? Orbis, 23, 484–99. ① ②
Leonini, C.
(2006.) The acquisition of object clitics and definite articles: Evidence from Italian as L2 and L1
(Unpublished doctoral dissertation.) Università degli Studi di Firenze. ①
Lipski, J.
(1986.) Convergence and divergence in bozal Spanish. Journal of Pidgin and Creole Languages, 1, 171–
203. ①②
Lipski, J.
(1994.) Latin American Spanish. New York, NY: Longman. ① ②
Lipski, J.
(2005.) A history of Afro-Hispanic language. Cambridge: Cambridge University Press. ①
Marzahl, P.
(1978.) Town in the empire. Austin, TX: University of Texas Press. ①
McCarthy, C.
(2008.) Morphological variability in the comprehension of agreement: an argument for representation over
computation. Second Language Research, 24(4), 459–486. ①
McWhorter, J.
(1997.) Towards a new model of Creole genesis. Bern: Peter Lang. ①
McWhorter, J.
(2000.) The missing Spanish Creoles. Berkeley, CA: University of California Press. ① ② ③ ④ ⑤ ⑥ ⑦
⑨⑩
Megenney, W.
(1984a.) Traces of Portuguese in three Caribbean creoles: Evidence in support of the monogenetic theory
Hispanic Linguistics, 1(2), 177–189. ①
Megenney, W.
(1984b.) El habla bozal cubana ¿lenguaje criollo o adquisición imperfecta? La Torre, 33, 109–139. ①
Montrul, S.
(2004.) The acquisition of Spanish. Amsterdam: John Benjamins. ①
Mintz, S.
(1971.) The socio-historical background to pidginization and creolization. In D. Hymes (Ed.),
Pidginization and creolization of languages (pp. 481–498). Cambridge: Cambridge University Press.
①②
Otheguy, R.
(1973.) The Spanish Caribbean: A creole perspective. In C. Bailey & R. Shuy (Eds.), New ways of
analyzing variation in English (pp. 323–339). Washington, DC: Georgetown University Press. ①
Pienemann, M.
(1998.) Language processing and second language development: Processability Theory. Amsterdam: John
Benjamins. ①
Pienemann, M. (Ed.) (2005.) Cross-linguistic aspects of Processability Theory. Amsterdam: John
Benjamins. ①
Poplack, S.
(1980.) Deletion and disambiguation in Puerto Rican Spanish. Language, 56(2), 371–385. ①
Rao, R., & Sessarego, S.
(2016.) On the intonation of Afro-Bolivian Spanish declaratives: Implications for a theory of Afro-
Hispanic Creole genesis. Lingua, 174, 45–64. ①
Rizzi, L.
(1982.) Issues in Italian syntax. Dordrecht: Foris. ①
Rodríguez-Tocarruncho, L.
(2010.) La marca de plural y otros aspectos morfológicos y sintácticos en el español del Pacífico de
Colombia (Unpublished MA thesis). Consejo Superior de Investigación Científicas. ①
Romero, R., & Sessarego, S.
(2018.) Hard come, easy go: Linguistic interfaces in Istambul Judeo-Spanish and Afro-Ecuadorian Spanish
In J. King & S. Sessarego (Eds.), Language variation and contact-induced change: The Spanish
language across space and time (pp. 63–82). Amsterdam: John Benjamins. ①
Rothman, J., & Slabakova, R. (Eds.) (2011.) Acquisition at the linguistic interfaces. Special issue of
Lingua, 121(4), 567–687. ①
Ruiz García, M.
(2009.) El español popular del Chocó, Colombia. Saarbrücken: VDM Verlag Dr. Müller. ① ② ③ ④
Sánchez, L., & Giménez, M. J.
(1998.) The L2 acquisition of definite determiners: From null to overt. In A. Greenhill, M. Hughes, H.
Littlefield, & H. Walsh (Eds), Proceedings of the 22nd annual Boston University Conference on
Language Development (pp. 640–650). Somerville, MA: Cascadilla Press. ①
Schwegler, A.
(1991a.) El habla cotidiana del Chocó (Colombia). América Negra, 2, 85–119. ① ② ③ ④
Schwegler, A.
(1991b.) La doble negación dominicana y la génesis del español caribeño. Lingüística, 3, 31–88. ① ②
 ④⑤⑥
Schwegler, A.
(1993.) Rasgos (afro-)portugueses en el criollo del Palenque de San Basilio (Colombia). In C. Díaz Alayón
(Ed.), Homenaje a José Pérez Vidal (pp. 667–696). La Laguna, Tenerife: Litografia A. Romero S. A.
①
Sessarego, S.
(2012.) The contribution of Afro-Hispanic contact varieties to the study of syntactic microvariation. In
González-Rivera & S. Sessarego (Eds.). Current formal aspects of Spanish syntax and semantics (pp.
229–250). Newcastle upon Tyne: Cambridge Scholars. ①
Sessarego, S.
(2013a.) Chota Valley Spanish. Madrid & Frankfurt: Iberoamericana & Vervuert. ①②③
Sessarego, S.
(2013b.) Afro-Hispanic contact varieties as advanced second languages. IBERIA, 5(1), 96–122. ① ② ③
⑤
Sessarego, S.
(2014.) The Afro-Bolivian Spanish determiner phrase: A microparametric account. Columbus, OH: The
Ohio State University Press. ① ② ③ ④ ⑤
Sessarego, S.
(2015.) Afro-Peruvian Spanish: Spanish slavery and the legacy of Spanish Creoles. Amsterdam: John
Benjamins. ①②③④
Sessarego, S.
(2019.) Language contact and the making of an Afro-Hispanic vernacular: Variation and change in the
Colombian Chocó. Cambridge: Cambridge University Press. ①②③④
Sessarego, S.
(under review). Roots of some languages. Ms. University of Texas at Austin. ①
Sessarego, S., & Gutiérrez-Rexach, J.
(2018.) Afro-Hispanic contact varieties at the interface: A closer look at pro-drop phenomena in Chinchano
Spanish. In J. King & S. Sessarego (Eds.), Language variation and contact-induced change: The
Spanish language across space and time (pp. 83–110). Amsterdam: John Benjamins. ①
Sessarego, S., & Rao, R.
(2016.) On the simplification of a prosodic inventory: The Afro-Bolivian Spanish case. In A. Cuza, L.
Czerwionka, & D. Olson (Eds.), Inquiries in Hispanic linguistics (pp. 171–190). Amsterdam: John
Benjamins. ①
Sippola, E.
(2011.) Una gramática descriptiva del chabacano de Ternate (Unpublished doctoral dissertation). Helsinki:
University of Helsinki. ①
Slabakova, R.
(2008.) Meaning in the second language. Berlin: Mouton de Gruyter. ①
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). Somerville, MA: Cascadilla Proceedings
Project. ①
Sorace, A.
(2000.) Syntactic optionality in non-native grammars. Second Language Research, 16, 93–102.
Sorace, A.
(2003.) Near-nativeness. In C. Doughty & M. Long (Eds.), The handbook of second language acquisition
(pp. 130–153). Oxford: Blackwell. ①②
Sorace, A., & Serratrice, L.
(2009.) Internal and external interfaces in bilingual language development: Beyond structural overlap.
International Journal of Bilingualism 13(2), 195–210. ①
Sharp, F.
(1976.) Slavery on the Spanish frontier. Norman, OK: University of Oklahoma Press. ① ② ③ ④ ⑤ ⑥
⑧⑨⑩⑪
West, R.
(1953.) Colonial Placer Mining in Colombia. Baton Rouge: Louisiana State University Press. ①
West, R.
(1957.) The Pacific lowlands of Colombia. Baton Rouge, LA: Louisiana State University Press. ① ②
CHAPTER 3

Methodological considerations in heritage

language studies
A comparison of sociolinguistic and acquisition-
based tasks
Zoe McManmon
Morton College

Abstract
This study examines accuracy of heritage speakers with respect to
gender agreement of noun phrase (NP) constituents. How does
methodology affect participants’ accuracy with gender agreement
(GA)? Methods employed include an acquisition-based task, and the
sociolinguistic interview used in four groups: children who access
heritage language (HL) in school and in social network (SN); children
who access HL only in SN; children who access Spanish only at
school; and finally, children who access Spanish neither in school nor
in SN. Three groups showed comparable accuracy in sociolinguistic
interviews and acquisition task. For the group that did not access
Spanish at school/home, the accuracy was higher on a sociolinguistic
task (98.61%) compared to acquisition task (73%). Different
methodologies yielded different rates of accuracy. Methods influence
the participants’ task accuracy.
 Keywords
heritage language; language contact; child bilingualism;
methodology; gender agreement

1. Introduction
The purpose of the study is to analyze gender agreement in Mexican-American
child bilinguals who are heritage speakers of Spanish, aged six through twelve. I
analyze gender agreement, which I used here to refer to agreement of all
constituents in the noun phrase, by using two data samples: a story retelling task
and a sociolinguistic interview. Both data samples are collected from the same
participants, most often on the same day. This analysis will show the impact of
different data collection techniques on accuracy of gender agreement.
This study analyzes the gender agreement of bilinguals in a contact setting.
In comparison to assignment of gender to a noun’s determiner, gender agreement
shown in the following examples requires agreement of noun and adjective
across the phrase and is thought to be more difficult to acquire (Franceshina,
2001; Franceshina, 2005; Hawkins & Franceschina, 2004). Consider the
following examples of correct (1) and erroneous (2) gender agreement:

(1)
El bosque estaba muy oscuro

(2)
El lobo estaba muy llena*

A linguistic object is a target linguistic item I address, in this case gender

agreement. Here I look at gender agreement of noun with other constituents in
the noun phrase. I examine the accuracy in aggregate and an explanation of the
rationale will follow. In the literature review we will see that other authors refer
to other objects, such as case marking or pronoun expression, that I do not
address in this text.
When I refer to the participants’ accuracy with respect to their gender
agreement usage, I refer to the correct tokens the participant produces of gender
as a percentage of the total contexts of gender on constituents referring to a
known noun in the phrase. Next, I will explain what I mean by a “proficient
speaker” and the various participant groups I employed in the data collection.
A proficient heritage speaker child is one whose use of the heritage
language approaches that of their parents or other bilinguals raised in a
monolingual setting. Because some children used errors in their interviews and
tasks, we can refer to the speakers on a gradient of skills, where some speakers
are more proficient, and others less so. Since this text focuses on gender
agreement, as it pertains to agreement of all constituents in the noun phrase, we
can also refer to children who are more proficient in gender agreement, or less
so.
I employ four participant groups when I look at the two data samples:
sociolinguistic interviews and a story retelling task. The four participant groups
are based on exposure to input in Spanish, which was found to be a key factor in
Alarcón (Alarcón, 2010; Alarcón, 2011). The four participant groups include:
children who access heritage language in an academic class at school as well as
in a social network, children who access heritage language only in the dense
social network (not in class), children who access Spanish only in an academic
class at school (not in a network setting), and finally children who do not access
Spanish either academically in their class at school or in a social network.
Children exposed to the most input were enrolled in Spanish instruction at a
local school and lived in a dense migrant network setting with many immigrant
parents from Puebla, Mexico. Children exposed to the least input in Spanish
were enrolled in an English classroom at school and lived in traditional housing
where they had daily interactions with only one or two adult speakers of
Spanish. For this context, it was problematic to ask the parents to label the
children’s Spanish as “sequential” or “simultaneous”. In these cases, even if the
parents stated they spoke to the child only in Spanish before attending school,
this did not rule out English exposure from siblings or other community
members. Due to the significant use of both languages in the community setting,
most participants are simultaneous bilinguals.
In this text I also refer to the contact setting. When I use the term contact
setting, I refer to an area of contact between Spanish and English. I also use this
term to refer to the specific area near Chicago where most of the adult
population speaks Spanish and most of the child population is bilingual in both
Spanish and English (simultaneous bilinguals) and uses both languages on a
routine basis.
In the literature review, I will lay out the previous approaches to study of
bilinguals, specifically child bilinguals, in contact settings. Then I will explain
how this cross-sectional project encompasses two methodologies: sociolinguistic
and story retelling tasks, to provide a novel understanding of language use by
child bilinguals in language contact. Following the results, I will discuss
methodological issues and advantages and shortcomings of the two project
methods: oral narratives and sociolinguistic interviews. The concept of
‘incomplete acquisition’ Montrul (2008) will be challenged, supporting Otheguy
(2013). This paper will show that some speakers show “low proficiency” only in
one data sample, and not across various methods of data samplings. My
conclusions call for continued work investigating bilingual populations.
Differences in data from heritage speakers with low levels of input can be found
using different methods. This project draws attention to the need for multiple
method studies to further address these discrepancies in some groups of low
input users of heritage language Spanish.

2. Review of literature
In this section on literature review, I discuss the topic of diverse methods in
language research. Then we turn to recent studies of child bilinguals living in
contact settings where English and Spanish are used side by side in the United
States. Next, I discuss background about grammatical gender and agreement,
elaborating on the definition of gender agreement provided in the introduction.
Publications have drawn attention to the need for a variety of methods in
language research, as related to the field of second language acquisition (Beebe,
1987; Geeslin, 2010). Geeslin, in her article, “Beyond Naturalistic: On the Role
of Task Characteristics and the importance of Multiple Elicitation Methods”,
notes that “each elicitation task provides unique and essential information about
learner language” (Geeslin, 2010). The data in this study will show that
Geeslin’s, 2010 hypothesis applies not only to second language learners but to
heritage speakers as well. In her paper, Geeslin discusses the relevance of
classroom language and ‘naturalistic’ language and suggests that classroom
language may in fact be the most naturalistic language for the second language
learner. The reader can consider how this concept of ‘naturalistic’ and other
language tasks applies to the heritage speaker, who may have less exposure to
language in the classroom. Both Geeslin and Beebe (Beebe, 1987; Geeslin,
2010) find that second language learners have task preferences, with second
language learners preferring formal language, over vernacular tasks. The
question of how these concepts of task preference relate to heritage language
learning and research is an area that merits further investigation. Task differences
in adult heritage language learners have been elaborated in Montrul, Foote and
Perpignan (2008). In the adult population, Montrul, Foote, and Perpignan (2008)
focus on differences between written and verbal tasks in both second language
learners and heritage speakers. However, task differences in child heritage
language learners are yet to be explored.
While data on developing child bilinguals, heritage speakers who are
residents of the United States, is not overwhelming in presence, neither is it
novel. Anderson in an early conceptual account of language attrition and
incomplete development notes about bilingual children that, “the immigrant
language may be restricted to communication with older monolingual speakers
of that language and to home, church, and other family- and community-related
uses”. In later accounts of a case study of two Puerto Rican sisters, the author
found that gender agreement was susceptible to language loss after analyzing
videotaped recordings of the two girls (Anderson, 1999). The reader may be
familiar with the works of Montrul and Potowski (2007), Montrul and Sánchez
Walker (2013), and also Shin and Van Buren (2016), each of these focusing on
bilingual children from diverse perspectives and distinct linguistic phenomena.
Due to the limited data on child bilingualism I review these studies here. The
distinct linguistic phenomena may raise the question if these studies can be
compared. Wei cautions “it is only natural that findings from these studies are
not always compatible, as the research questions and methods of data collection
and analysis are so radically different” (Wei, 2000). Additionally, we might
expect different results when different populations are involved in the research
study. More detailed consideration of previous methodological approaches in
contact settings is justified because these approaches have resulted in very
different findings. Some participants were found to be accurate and proficient,
while others were not. This topic bears further consideration because of residual
questions about if the different results might be due to distinct participant
groups, distinct linguistic phenomena or other factors.
This section presents a review of previous work on child bilinguals in
contact settings (Anderson, 1999; Montrul & Potowski, 2007; Montrul &
Sánchez Walker, 2013). Much of the early work on bilinguals in contact
followed Anderson (1999) who studied a pair of siblings to immigrate to the
United States from Puerto Rico. Following up on declines in linguistic accuracy
in a contact setting, Montrul and Potowski (2007) collect school-based data on
gender agreement, including an oral narrative task, from participants at a
Chicago dual language school. Though the publication was ground breaking, it
should be noted that it focused on oral narrative and puzzle tasks and did not
reflect language as used in a peer communication or social setting. A description
of the oral narrative task follows:
For the elicited production task, all participants were shown pictures of the
children’s story Little Red Riding Hood and were asked to retell the story in the
past, with as much detail as possible. All speakers were audio recorded, and
speech samples were later transcribed by several research assistants, coded and
analyzed. (Montrul, 2009)
The picture naming task, in contrast, was an animal naming game, where the
child selected three cards and created a sentence “Veo (determiner) (noun)
(color)” (Montrul & Potowski, 2007). This task was completed with both
masculine and feminine gendered animals. The authors in this text note some
complications with use of grammatical gender in child bilinguals in the target
location in Chicago, Illinois, including their finding that few adjectives were
produced during the story retelling task. Additionally, children who were
simultaneous bilinguals performed like second language learners in terms of
their gender agreement. Moreover, input played a role in the findings but was not
a study variable.
In a later piece, Montrul and Sánchez Walker (2013) focus on similar tasks
in a non-school setting of child bilinguals, the authors find acquisition of the
direct object marker to be problematic, calling the acquisition “non-native like”.
The participants sample was not found in a school setting, but individuals known
to the author(s). Both Montrul and Potowski (2007) and Montrul and Sanchez
Walker (2013) focus their work on acquisition-based tasks, and one of these is a
story retelling task. The retelling task is an asset in terms of eliciting linguistic
features. However, a downside is the task focuses on a language style that is
different from casual conversation.
In contrast to Montrul and Potowski (2007), Montrul and Sánchez Walker
(2013), and Shin and Van Buren (2016) focus on social language in a study of
sociolinguistic interviews in a migrant community of Washington and Montana.
In contrast to the previous two studies, Shin and Van Buren’s work is distinct
because it focuses on documenting language as it is used for a communicative
purpose in migrant networks, through the sociolinguistic interview. Aside from
the unique cultural context of a migrant community, Shin and Van Buren (2016)
show that the use of pronouns by these farmworkers and families was indeed
native like, indicating “little to no change in pronoun expression”.
Alarcón (2011), in a study of adult heritage speakers, did not find sizeable
differences in highly proficient Spanish speaking adults, though other
researchers have found differences (Montrul et al., 2008; Montrul & Potowski,
2007). Alarcón (2011) attributes the differences found to pooling subjects who
are ‘low proficiency speakers’. The differing results in the literature between low
and high proficiency speakers, and absence of a study, with children, to account
for groups of both high and low input exposure, demonstrates a research need.
As a result, I propose to account for two educational groups: one exposed to
Spanish in school and the other exposed to English in the classroom. It is
possible that students who participated in the school’s bilingual program have a
language experience that is distinct from those who did not.
The reader is forced to consider a strand of linguistic research, commonly
focusing on bilingualism in contact settings (Chicago, Washington, Montana),
that uses diverse methods (acquisition based oral narratives, sociolinguistic
methods), focusing on diverse linguistic objects (grammatical gender,
differential object marker, pronoun expression), with quite differing results. Wei
has noted that differing results should be expected when distinct methods are
used (Wei, 2000). However, some of the results clearly eschew the children’s
acquisition of native like forms (Montrul & Potowski, 2007; Montrul & Sánchez
Walker, 2013), while others find that language is maintained in a native-like way
(Shin & Van Buren, 2016). The diverse results found from heritage speakers in
contact have caused some critique of key terminology in the acquisition field
such as ‘incomplete acquisition’ (Otheguy, 2013). Critics may suggest that the
different findings may be due to the different linguistic objects employed
(subject pronoun expression, direct object marker, as in Montrul & Bowles
(2009), or gender agreement), or due to distinct settings, or different participants.
In this project, I attempt to control for some factors by using the same participant
group and a focus on gender agreement with a variety of project methods.
Additionally, because the participants used in this type of research may not be a
homogenous group, I subdivide my participant group into four typological
groups explained in the introduction. Because the topic of acquisition is one that
has been challenged, a project considering children of varying proficiencies, and
using several methodologies is justified. I will address this topic here as it
pertains to gender agreement of a variety of constituents within the noun phrase.
For readers who point out the data may differ with other linguistic objects, I
encourage the critical reader to replicate a study of multiple methods with other
linguistic objects.
I turn to the question, then, to what is this difference in findings owed
where some speakers are considered ‘native like’ (Alarcon, 2011; Shin & Van
Buren, 2016) and others are considered to have acquired their parents’ language
in an incomplete way (Anderson, 1999)? Different studies have arrived at
different findings. Are these differences due project methods? In this study I will
try to solve this inquiry. I replicate the use of oral narrative task based on Little
Red Riding Hood, as in Montrul and Potowski (2007) and Montrul and Sánchez
Walker (2013). Additionally, I add a sociolinguistic task like as in Shin and Van
Buren (2016) in which I collect data from a same-age peer group. This task is
based on a peer-group interview where children answer questions about their
background information and discuss topics in a small group of peers as in Labov
(1969). I will show that in this study participants in the study show different
results. The methods replicate the previous studies. However, this is not strictly a
replication study because the population of participants is unique. The next
sections will focus on methods and results of this project.

3. Grammatical gender
To reiterate the definition I provided in the introduction, gender agreement is a
linguistic object prominently featured in bilingualism literature of acquisition.
Here we will address gender agreement, concord, and gender assignment in
detail. The distinction between gender assignment and concord agreement across
the whole phrase is a crucial one noted in second language acquisition research,
in particular because the agreement across the phrase is one that has been found
to be more difficult to acquire in studies of second language and heritage
language acquisition. Consider the following examples:

(1)
El bosque estaba muy oscuro

(2)
El lobo estaba muy llena*

Researchers hypothesize that agreement in phrases (1) and (2) would be more
difficult to acquire than simple assignments of determiners to a noun. The noun
is assigned gender via its corresponding determiner, which ideally agrees with
the noun. Gender agreement is a phenomenon that links the gender of a noun to
its agreeing elements, usually a determiner and often an adjective (Corbett,
1991). Adjective agreement, which is referred to as ‘concord’ is more difficult to
acquire than selecting a determiner. This distinction has been documented in the
SLA research (Franceshina, 2001, 2005; Hawkins & Franceschina, 2004). A key
reason to consider gender is that it shows up in a variety of places both in subject
or object position, including agreement with adjectives and determiners and also
agreement with objects throughout discourse. Gender agreement is an area of
interest because the phenomena of agreement by gender and number do not exist
in English, so these are features that have been shown to be difficult for child
bilinguals to acquire. Finally, Alarcón notes that “grammatical gender in an
interesting category for analysis because it provides a window on both lexical
access and syntactic processing” (Alarcón, 2011, p. 352). The topic of gender
agreement was found to be intriguing in French as well (DeWaele & Véronique,
2001). This study analyzes the accuracy rates of agreement by bilinguals in a
contact setting. I focus here on the accuracy of the constituents including
determiners and adjectives. I analyze the constituents in the aggregate because of
the high accuracy rate and minimal number of errors in this data set. The
accuracy rate will be analyzed across two methods: sociolinguistic interviews
and oral narrative tasks, both of these will be explained in the session on
methodology. Before turning to the study methods, I provide an overview of the
research questions:
1. How do data collection and methodology affect the participants’ accuracy
with use of gender agreement?
2. What are the advantages and disadvantages of the two project methods:
 interviews and oral narrative?

4. Methodology
The data presented in this text is part of a larger, multi methods project
encompassing four study methods in a bilingual community: a story retelling
task rooted in the acquisition literature, a gender concord task, sociolinguistic
interviews, and qualitative ethnographic methods. In this text I focus on results
of the story retelling task and sociolinguistic components, to draw contrasts
between these two methods as they were used to study Mexican background
bilingual children at a Chicago area public school. When I refer to the story task
as an “acquisition-based task” I suggest that it uses academic language. This
academic focus creates a task that differs from the language a heritage speaker
uses with one’s family. This concept relates to Geeslin’s (2010) proposal that
classroom tasks are relevant tasks for second language learners. The story
retelling task is an activity much like one a child would participate in during
school, a “performance” individually enacted for the researcher. When I refer to
the sociolinguistic task, I refer to the activity of participation in a conversation
with a same age peer group. This more closely approximates the type of
conversations a child would have with friends, classmates, or siblings. Certainly,
the story retelling task involved some “interaction”, however I present data from
two distinct tasks, where one involved the peer group (sociolinguistic) and the
other did not (retelling). I will present the findings from both data collections
and we will see that there are some differences in these two methods.
It was previously established in the literature review that both methods,
story retelling and sociolinguistic interviews, are commonly used in bilingualism
research. Yet, it is unusual for research to concurrently procure both data sets
from the same group of individuals. In this section I focus on the present project
to explain how its methods encompass both acquisition-based and sociolinguistic
tasks.
Children were recruited by a researcher at a public school in a Chicago
suburb. An anonymous reviewer inquired about the percentage of heritage
speakers in the school. This is an interesting question but unfortunately, we do
not have this data because children are only identified in their records as
participating, or not participating, in the school’s bilingual program. Thus, there
certainly are heritage speakers who do not participate in the school’s bilingual
program, due to parent choice or initial assessment data on their English level. In
fact, some of these participate in the study here. I cannot say with certainty the
percentage of heritage speakers in the school community. Based on census data
the percentage of Spanish speakers in the neighborhood is 71% (MLA Language
Map).
Recruited individuals were, with the permission of their parents, the school
administration at the area school district, and the Institutional Research Board at
the University of Illinois at Chicago, registered for an after-school session with
the researcher. At the after-school session, the recruited child bilinguals
participated in three tasks. First, they individually participated in an oral
narrative task, where they viewed illustrations from a children’s fairy tale, Little
Red Riding Hood (Marshall, 1987). As the child viewed the photo images on the
classroom smartboard, she told the story of the tale. Meanwhile, the child was
recorded using a digital voice recorder for sociolinguistics. Following the story
retelling, the child was issued a ‘Concord Task’, where she was asked the
following questions “¿Cómo era el lobo?” “¿Cómo era la abuela?”, and other
like questions related to other text characters. The purpose of this task was to
determine if the child was able to use an accurate gendered adjective in their
description of the character, when the gender was provided via a gendered
article. Finally, students interacted with their peers in a small group interview
session (sociolinguistic interview). In this present study I will focus mainly on
the oral narrative task and the sociolinguistic task, rather than the concord task.
Details of the sociolinguistic interview follow here. The third task for the
child attending the after-school session was the sociolinguistic interview. The
interview was distinct from the acquisition based oral narrative task, because the
sociolinguistic method employed a group of child participants as in previous
field work with children (Cameron, 2010; Labov, 1969). Furthermore, the
interview format of elicitation is considered unique, because, it is a speech event
“which has its own rules and produces a characteristic style (Wolfson, 1976).
Labov (1969) noted that children behave differently when they were interviewed
by a sole adult researcher, in contrast to their participation in a conversational
group with same age peers. A sample of the ‘conversational group’ will follow.
For the sociolinguistic interview, the children often selected their peer group
with whom they reported to the research session in the classroom. This was
generally a peer group of two to four individuals, comprised of siblings and/or
same age class peers. The sociolinguistic interview resulted in data that was
qualitatively different than the acquisition-based task. Though not the topic of
the present writing, code switching is one area that differed between the two
methods: oral retelling and sociolinguistic interview, and this would be an area
for future work. In contrast to the discourse level switching in the sociolinguistic
interviews, code switching was rare in the story retelling, where bilingual
children held to a greater use of Spanish, and when code switching occurred it
generally occurred only in one-word switches.
Here is a sample of a “conversational group” of second to fourth grade
female bilinguals who use English and Mexican dialect Spanish. The
researcher/Interviewer is labeled “Z” (Zoe) and the children are not referred to
by real names. Children were told that they would be discussing their opinions
about a variety of topics while they met with the group, and the instructions were
read in both Spanish and English.

(3)

Z:Okay let’s go back to La Llorona. ¿Creen en la Llorona?

C:Yes.

A:Yes,

C:Because we’ve heard a lot of stories of her.

Z:So do you think La Llorona only happens in Mexico or it happens here too?

A:It happens here too.

C:They think that the bloody Mary is real.

A:(inserts names of four male classmates) dicen que han encontrado bloody
Mary con las luces apagadas. Y velas. Y están tres veces: Bloody
Mary, bloody Mary, bloody Mary.

C:Están mintiendo.

Z:You don’t believe in Bloody Mary. But you do believe in La Llorona.

A:In the night.

C:In the night, and in the pueblo too, there are like clowns, they come for you
at three o’ clock in the morning,

A:You want to be careful in Mexico, porque a five o’clock these bad guys
 come. Because my mom went, like a week ago.

C:yeah, she went, and she was in my house.

A:You want to be careful porque allí salen muchachos muy malos, porque
cuando mi mama fue a comprar cosas en la tienda, les
preguntaron, “¿De quién es esa troca allí afuera?” Y dicen “son
de nosotros porque vinimos de Chicago”.

C:Why did she? Because they’re going to think if you came from Chicago
you’re going to have a lot of money.

A:yeah y dijeron, y luego dijeron “oh okay”. Y allí pasan muchos malitos.

The reader can note that one speaker, Ana, routinely uses Spanish while here
peer Chrissy uses more English.1 Crissy’s Spanish is limited to single word
usage, like “pueblo”, while Ana switches her Spanish and English use across the
phrase (“you want to be careful/porque allí salen muchachos muy malos’). This
speech act illustrates the use of discourse level code switching in the
sociolinguistic interview, and this was distinct from the recordings of the oral
narrative (acquisition-based task). In the oral narrative, the child performed
alone, and the speech was more formal and academic, with different grammatical
results to follow.

5. Participants
There are four participant categories used in this research study and these
categories will be used to compare participant groups. The categories will be
based on the following social characteristics. The sample pool was comprised of
participants who were six to twelve-year-old child bilinguals residing outside of
Chicago with Mexican-born parents. The speakers were second-generation child
bilinguals. Otheguy (2013) has pointed out that the concept of ‘heritage speaker’
in and of itself, lacks sufficient definition to define the field of study of these
individuals. Early work on incomplete acquisition was critiqued for not
including groups of heritage speakers of varying proficiencies or for not
accounting for proficiency (Alarcón, 2010, 2011; Otheguy, 2013). In response,
here I incorporate groups benefiting from diverse input attending Spanish
classes, English classes, and accessing social networks as well as groups that do
not.
Thus, by modeling a broad group of heritage speakers of differing
proficiencies, we can aim to better understand this linguistic group. The four
categories of speakers included children who sat for English class at the local
public school and were not living in a dense Spanish network setting (n = 6),
children who sat in an English class at the school but did live in a dense Spanish
network setting (n = 4), children who had been enrolled in a Spanish language
class at the local public school and also lived in the dense Spanish context
(n = 10), and finally children who were enrolled in the class where the teacher
used the heritage language at school, but this was the main setting in which they
were exposed to the language (n = 14). Participants were recruited and
interviewed, and their data was analyzed based on social factors including social
network (dense, not dense) and language of instruction at school (Spanish,
English). An explanation of how I operationalized measures of ‘dense’ social
network will follow in the next section on “context”.
Children’s ages were taken during the sociolinguistic segment along with
the child’s self-report of their language dominance, and usage of language with
adults and siblings. However, due to the large number of participants (40) and
variety of reports made, this data will not be elaborated on here.
 To elaborate upon the context, this setting should be noted as distinct from
previous data collected from bilingual school age children (Montrul & Potowski,
2007) because many participants in this study live in the migrant labor setting,
where they had daily interactions with some of four hundred residents from areas
nearby to Puebla, Mexico. As Wei notes, “bilingualism research can be
conducted in various physical settings, ranging from laboratory conditions to
community-based participant observations… the same speakers may display
very different behaviors in different settings” (Wei, 2000, p. 502). Thus, we can
consider not only the speakers and their context, but also the effect that the study
methodology may have on groups of contact speakers. To do this, an analysis of
the child participants’ accuracies on gender agreement in the two study
methodologies will follow.

6. Context
As mentioned previously, children were recruited and tested in an outlier
community of Chicago, one mentioned in Rancheros en Chicagocán by Marcia
Farr (2006). Though in an urban environment, the setting was unique because
approximately half of the study participants resided in a polideportivo for
migrant laborers (the parents), who worked in a cash based local industry and
were provided group housing. Considering the discussion of migrant networks
on language maintenance and change (Shin & Van Buren, 2016) the reader can
consider that the urban data may reflect some trends found in migrant labor
groups more typically found in a rural context. In order to operationalize this
social context into a category described in the previous section, children were
asked questions about who they lived with, how many people in their residence
spoke Spanish, and their network connections with friends and family. Here is an
account of the operationalization of the social network rating scale I used with
child participants.
1. I live in a territorially based cluster.
2. In my class at school there are at least three other students who live in the
same residence where I live.
3. I have at least three other playmates who live in the same residence where I
live.
4. During the sociolinguistic interview, I report that another family lives in the
residence with me (perhaps I pass my hand-me downs to them or stated
they help my parents when they need help).

Children were awarded one point for each statement that was true for them.
Points were totaled and each child that had two or more points was considered
part of a dense social network for purposes of this study. This line of questioning
took place as part of the sociolinguistic interview battery of questions, and
children were placed into two groups as categorized into dense or non-dense
social network, taking into account their responses to the open-ended questions
about their network outside of school.
The qualitative research component was collected during subsequent
recruitment interest and scheduling for a community-based follow up session,
and will not be reported on here. Qualitative findings are documented in greater
depth in the dissertation by McManmon (2016). While researchers have
generally focused on either acquisition based or sociolinguistics-based methods,
the present study will show that each method has its merit. The story retelling
permitted greater elicitation of grammatical forms that might be uncommonly
produced in immigrant groups, such as the direct object marker in Spanish
(Montrul & Sánchez Walker, 2013). The access of less commonly used forms
comes with a trade-off, where the researcher accesses mainly academic language
used to discuss print materials, rather than the conversational type of Spanish
used in bilingual Mexican communities, reflected in the peer based
sociolinguistic interview. Though there are merits to each method, the two
methods tell different stories about the accuracies and proficiencies of the four
social groups of these contact speakers of Mexican Spanish, with a distinct
benefit to multiple task projects as in Geeslin (2010). I now turn to a discussion
of the findings, as they relate to the contrasts in the results of the oral narrative
and sociolinguistic tasks from children raised in this contact setting.

7. Findings

7.1 Oral narrative/story retelling

A transcription and analysis of 1,752 tokens from the oral narrative task showed
that participants who did not access a dense Spanish network and did not access
Spanish at school had statistically significantly lower accuracy on grammatical
gender agreement (88.94 ± 21.95) compared to peers who accessed a dense
network and/or Spanish language instruction at the school (97.92 ± 3.30),
t(36) = 2.188, p = 0.035. Recall that in the introduction, accuracy was defined as
the number of correct tokens the participant produces of gender as a percentage
of the total contexts of gender on constituents referring to a known noun in the
phrase. Gender agreement accuracy was high and near ceiling, approximately
98% accuracy, for the following groups of participants: children who lived in the
migrant setting and had Spanish classroom instruction at school, children who
lived in the migrant labor setting but had English instruction at school, children
who lived in traditional housing but had Spanish classroom instruction at the
school. Only the group receiving the least input in Spanish demonstrated lower
accuracy on gender agreement.
Table 1. Tokens and accuracy by social group

Group Target tokens (felicitous) Non- felicitous Tokens % Accuracy

+BIL +SN (n = 12) 582 10 592 98%

+BIL −SN (n = 14) 543 9 552 98%
−BIL +SN (n = 5) 494 7 501 98.6%
−BIL −SN (n = 9) 77 28 105 73%

Individual analysis showed that twenty-three participants scored one hundred

percent accuracy on gender agreement in the oral narrative task. Ten individuals
made only two or three errors which were often in non-canonical forms. I
mention the non-canonical forms of gender agreement here because it explains
some of the small percentage of errors the children made in the above Table 1.
The remaining participants to score an accuracy rate lower than 95% (n = 7)
were more likely to be found in the group of children who were not in the dense
social network labor setting and were not exposed to Spanish instruction at the
school. In contrast to the sociolinguistic interview, the oral narrative has a “more
formal, less regular style” (Labov, 1977). This text turns to the statistical analysis
of retelling data.

7.2 Statistical analysis of gender assignment tokens (story retelling)

I submitted the data to a repeated measure ANOVA analysis with two factors
where the between subjects’ factor was the four social groups with treatments of
+/− for social network and bilingual education programs. The within subjects’
factors include grammatical gender feminine and masculine, and the dependent
variable was accuracy on gender. There was no main effect of grammatical
gender either masculine or feminine (F (1, 35) = 1.596, p = .215). There was no
main effect of group (F (3, 35) = 1.782, p = .169). There was no interaction (F
(3, 35) = .567, p = .641).
With respect to token count, there was no main effect of gender (F
(1, 35) = .792, p = .379). There was no main effect of group (F (3, 35) = .562,
p = .667). There was no interaction (F (3, 35) = .598, p = .620). There was no
significant difference in number of articles, though the group +SN, +BIL
produced more article tokens on average. Participants who were in the group
−BIL, −SN produced less feminine articles (M = 18.89, SD = 6.33) and
masculine articles (M = 18.44, SD = 12.70) than participants who were exposed
to a dense social network with bilingual education (M = 22.42, SD = 6.037) for
feminine tokens as well as masculine article tokens (M = 20.67, SD = 8.80).
With respect to the homogeneity, children in bilingual classrooms were
more homogenous than those who weren’t, and children placed in an English
class who lacked social network were the least homogenous of any group, with a
factor of 92%. This reflects that it was those children exposed to the least input
(English class and lacking social network) who were the least consistent in their
group responses. The indicators for homogeneity suggest that later qualitative
results of working with one participant from the group (−Bilingual, −Social
Network) should be interpreted with some caution because the group shows
diversity within itself. This is underscored by the high standard deviations found
in the group (−BIL, −Social Network), as high as 25 for feminine article
assignment and 8 for masculine article assignment. By contrast, the group +BIL,
+SN had a standard deviation of 1 for feminine article assignment. The
homogeneity values and standard deviations for the group (−BIL, −Social
Network) suggest this group as an area of further study within child
bilingualism.
Because the ANOVA did not indicate differences, I further investigated the
topic of grammatical gender. I compared the group −BIL, −SN with the other
three groups combined using an independent samples t-test. This study found
that participants who did not access a dense Spanish network and did not access
Spanish at school had statistically significantly lower accuracy on grammatical
gender agreement (88.94 ± 21.95) compared to peers who accessed a dense
network and/or Spanish language instruction at the school (97.92 ± 3.30),
t(36) = 2.188, p = 0.035. The results were statistically significant with the ENG
−SN group preforming differently than a combined set of children in the other
groups using an independent t-test. Some interesting points are made by these
results. First, for the group of students who access a dense social network, there
appears not to be differences in those groups that access bilingual education and
those who do not. In fact, I checked these groups and there was no statistically
significant difference between them, as we would expect from the table above.
For students in a dense network, there is no statistical effect for heritage
language instruction. A second point to be made is that, for students who lack a
dense social network in Spanish, children in a bilingual education setting
perform at 98% much like their dense network peers, while those who also lack
heritage language at school perform at a lower accuracy level. Thus, there appear
to be effects for instruction in the heritage language in cases where children do
not access a dense network in Spanish. However, while there is a percentage
difference between speakers in English and Spanish classes who did not access
the social network it was not statistically significant.

7.3 Sociolinguistic task (group conversation)

The sociolinguistic task recorded 520 tokens of gendered articles and 108 tokens
of gender on adjectives. For three of the four social groups considered in the
previous analysis, the accuracy during the sociolinguistic interviews was
comparable to the accuracy found during the oral narratives. As shown in
Figure 1, there is a different result for the participant group receiving the least
input in Spanish through their contact setting. For the group that did not access
Spanish instruction at school and did not access a dense social network, the
accuracy was dramatically higher on a sociolinguistic task (98.61%) compared
to an oral narrative task (73%).

Table 2. Accuracy on grammatical gender agreement by group.*

(sociolinguistic interview compared to story retelling)

Group Accuracy/Socio Accuracy/Retelling

+Bilingual Ed +Social Network n = 10 100% 98%

−Bilingual Ed +Social Network n = 4 100% 98.6%
+Bilingual Ed −Social Network n = 14 99.38% 98%
−Bilingual Ed −Social Network n = 6 98.61% 73%

* Note. “Table 2” includes data from 34 speakers because 6 speakers spoke in Spanish during the oral
narrative (retelling) and spoke only English during the sociolinguistic task. Thus, these speakers were
excluded from the comparison data. However, it is crucial to note that these were not the low-preforming
individuals noted in the oral narrative task. Rather, use of English appeared to be more stylistic than based
on a result of speaker proficiency. ①

7.4 Statistical analysis of sociolinguistic task

These findings show that heritage language speakers show a higher accuracy on
measures of gender assignment in a sociolinguistic interview compared to a
retelling. The group with the least input in Spanish, lacking instruction in
Spanish at school and lacking a social network outside, was still the lowest
performing group. However, when using the sociolinguistic task, the differences
between the −SN, −BIL group and other groups was cut to less than two
percentage points. On the sociolinguistic task, two groups scored means at 100%
accuracy, and these were the two groups accessing social network. Furthermore,
all four social groups were between 98%–100% accuracy, approaching or
meeting accuracies found by native speakers. As a result, we can expect that
there were not any significant differences between these four subject groups, and
in fact a repeated measures ANOVA showed this was the case. There was no
main effect of gender (F (1, 25) = .217, p = .607). There was no main effect of
group (F (3, 25) = .870, p = .470). There was no interaction (F (3, 25) = 1.685,
p = .201). An anonymous reviewer notes the fewer number of tokens of gender
agreement in the group conversation/sociolinguistic interview task. This is
indeed the case, and this finding was the result of the task context (group
conversation) where children felt free to use either language, Spanish or English,
in their conversation with their peers. In contrast the retelling task was an
individual performance and in this task, children used predominantly Spanish
(see also footnote 2 from Table 2).

Figure 1. Comparing accuracies of sociolinguistic and story retelling tasks

With respect to the quantity of data obtained, the sociolinguistic interview was
superior in eliciting pronouns (n = 875), while the oral narrative would be
preferred for eliciting objects of interest for future work such as differential
object marking in Spanish (n = 142). These topics of pronouns and differential
object marking are not addressed in this study but could be a possible topic for
further investigation.

8. Discussion
For the speakers from the English exposure groups (English schooling and not
from a dense Spanish network) there were differences in the data obtained from
the sociolinguistic and story retelling tasks. These low input speakers presented
as more proficient or more skilled when tasked with a sociolinguistic activity
based on peer conversational language. We can consider this finding in light of
Geeslin’s (2010) proposal that classroom tasks are very ‘naturalistic’ for second
language learners. The flip side of this coin is the question of an appropriate task
for a heritage speaker who may spend less time using the heritage language in
the classroom. For this group, the sociolinguistic task may be of key importance.
That finding, though, does not rule out the utility of other tasks that elicit unique
data forms difficult to obtain from sociolinguistic data. There is a value to be
obtained from both tasks, though the results should be interpreted with reference
to other study methods and not used merely to state that the speakers’ language
is “incomplete”. These results certainly call for further work with multiple task
studies, as suggested in Geeslin (2010).
A note about applications of the study with respect to school-based
language programs for children. Those students who had no Spanish instruction
at school nor at home (referred to as the ‘low input’ group) had only a 73%
accuracy rate on gender agreement in Spanish. These findings underscore the
role of Spanish language instruction for children who are heritage speakers and
do not access a dense network community. Keeping in mind that in the United
States most minors do not live in migrant labor contexts, instruction in the
heritage language is a fundamental component for accuracy. This statement is
underscored with a consideration of problematic features, such as gender
agreement. These findings are also pivotal for school programming decisions.
Comparing the Spanish instruction/no social network group to the English
instruction/no social network group, Spanish language instruction shows an
advantage for children like these second-generation speakers who live in a
contact setting. Children living in the migrant network comprise one school in
this district, in only a small area of the Chicago land area. We can only imagine
possible effects on language maintenance for children who have no access to a
migrant network and no heritage language instruction at their school.

9. Conclusion
Returning to the initial question posited, “Do child bilinguals in this contact
setting show greater morphosyntactic accuracy in sociolinguistic interviews or
oral narratives?” We can conclude that indeed there is a difference in accuracy
on sociolinguistic interviews and oral narratives, specifically with children
exposed to low levels of input of the heritage language in a contact setting. All
four participant groups scored between 98–100% on gender agreement in the
sociolinguistic interview task. However, the accuracies ranged from 73% to
98.6% in the oral narrative task. These results demonstrate that heritage speaker
children perform at higher levels with a sociolinguistic task, in particular notable
differences are found by those who get less input in heritage language in contact
settings. When the target language is a heritage language, higher accuracies for
sociolinguistic tasks result in the group of low proficiency heritage speakers.
Some speakers, in particular individuals exposed to less input, made more errors
in a retelling task than in a sociolinguistic task. However, for individuals with
more input (for example those enrolled in the Spanish bilingual class at school)
discrepancies were mitigated. These results are particularly relevant for any
consideration of appropriate school programming for these child bilinguals in
contact settings, where readers may consider dual language programs a relevant
addition to the children’s language repertoire.
This study has reported on a group of school-age bilingual children, G2
speakers who learn their Mexican parents’ language at home, in their
communities, and in some cases, in their schools in the contact setting in which
they live. We found that nearly all contact speakers had acquired grammatical
gender of Spanish. In an oral narrative task, however, some participants showed
more difficulty with gender assignment errors, suggesting that the children may
be subject to task-specific production errors. We found the low accuracy in
gender assignment in the narrative task to be most extreme in these participants
who had experienced the least input in Spanish, those who were in an English
class at school and lacked a dense network at home. These results should be
interpreted in light of the different findings between the two tasks. Though
contact bilinguals have acquired the language, as demonstrated by the
sociolinguistic task, the group of low input speakers may be less familiar with
academic language. The goal of this study is to draw the researchers’ attention to
the variety of data that child bilinguals can produce when examined with a
variety of different methods. Use of various study methods: either
task/acquisition based, or social interview based, provide the researcher different
data about the accuracies and competencies of the bilingual speakers. This study
shows that the task and methodology-based disparity is greater in participant
groups that have less input in the contact language. That being said, this study in
no way claims that the children have full acquisition of all linguistic objects in
Spanish, only that their accuracy scores were quite high, and near ceiling, with
gender agreement, in particular when provided the sociolinguistic task. These
findings differed from the oral narrative task. The researcher’s future work will
focus on other linguistic objects of interest in the Spanish of bilingual children,
and she hopes to call future attention to issues of methodology for others
researching bilingual children in contact setting, in particular heritage speakers.

Note
1. Not individuals’ real names. ①

References
Alarcón, I. V.
(2010.) Gender assignment and agreement in L2 Spanish: The effects of morphological marking, animacy,
and gender. Studies in Hispanic and Lusophone Linguistics, 3(2), 267–300. ①②
Alarcón, I. V.
(2011.) Spanish gender agreement under complete and incomplete acquisition: Early and late bilinguals’
linguistic behavior within the noun phrase. Bilingualism: Language and Cognition, 14(3), 332–350.
①②③④⑤
Anderson, R. W.
(1999.) Loss of gender agreement in L1 attrition. Bilingual Research Journal, 23, 319–338..
②③④
Beebe, Leslie
(1987.) Sociolinguistic variation and style shifting in second language acquisition. In G. Ioup & S.
Weinberger (Eds.), Interlanguage phonology: The acquisition of a second language sound system (pp.
378–388). Rowley, MA: Newbury House. ① ②
Cameron, R.
(2010.) Growing up and apart: Gender divergences in a Chicagoland elementary school. Language
Variation and Change, 22(2), 279–319. ①
Corbett, G. G.
(1991.) Gender. Cambridge: Cambridge University Press. ①
Dewaele, J. M., & Véronique, D.
(2001.) Gender assignment and gender agreement in advanced French interlanguage: A cross-sectional
study. Bilingualism: Language and Cognition, 4(3), 275–297. ①
Farr, M.
(2006.) Rancheros in Chicagoacán: Language and identity in a transnational community. Austin, TX:
University of Texas Press. ①
Franceschina, F.
(2005.) Fossilized second language grammars: The acquisition of grammatical gender. Amsterdam: John
Benjamins. ①②
Franceschina, F.
(2001.) Morphological or syntactic deficits in near-native speakers? An assessment of some current
proposals. Second Language Research, 17(3), 213–247. ①②
Geeslin, K.
 (2010.) Beyond “naturalistic”: On the role of task characteristics and the importance of multiple elicitation
methods. Studies in Hispanic and Lusophone Linguistics, 3(2), 501–520. ①②③④⑤⑥
⑦⑧
Hawkins, R., & Franceschina, F.
(2004.) Explaining the acquisition and non-acquisition of determiner-noun gender concord in French and
Spanish. In P. Prévost & J. Paradis (Eds.), The acquisition of French in different contexts. Focus on
functional categories (pp. 175–206). Amsterdam: John Benjamins. ①②
Labov, W.
(1969.) The study of non-standard English. Champaign, IL: National Council of Teachers of English. ①
③
Marshall, J., & Donnelly, D.
(1987.) Red riding hood. London: Dial Books. ①
McManmon, Z. C.
(2016.) Grammatical gender in child bilinguals: Language acquisition across sociolinguistic and
pedagogical spheres (Unpublished doctoral dissertation). ①
Montrul, S.
(2009.) Knowledge of tense-aspect and mood in Spanish heritage speakers. International Journal of
Bilingualism, 13(2), 239–269. ①②
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.
Montrul, S.
(2008.) Incomplete acquisition in bilingualism: Re-examining the age factor. Amsterdam: John Benjamins.
①
Montrul, S., Foote, R., & Perpiñán, S.
(2008.) Gender agreement in adult second language learners and Spanish heritage speakers: The effects of
age and context of acquisition. Language Learning, 58(3), 503–553. ①②③
Montrul, S., & Potowski, K.
(2007.) Command of gender agreement in school-age Spanish-English bilingual children. International
Journal of Bilingualism, 11(3), 301–328. ①②③④⑤⑥⑦⑧⑨⑩
Montrul, S., & Sánchez-Walker, N.
(2013.) Differential Object Marking in child and adult Spanish heritage speakers. Language Acquisition
20(2), 109–132. ①②③④⑤⑥⑦⑧
Otheguy, R.
(2013.) The linguistic competence of second-generation bilinguals: A critique of “incomplete acquisition”
In C. Tortora, M. den Dikken, I. L. Montoya, & T. O’Neill (eds.), Romance linguistics 2013: Selected
papers from the 43rd Linguistic Symposium on Romance Languages (LSRL) (pp. 301–320).
Amsterdam: John Bejamins. ① ② ③ ④
Shin, N. L., & Van Buren, J.
(2016.) Maintenance of Spanish subject pronoun expression patterns among bilingual children of
 farmworkers in Washington/Montana. Spanish in Context, 13(2), 173–194. ①②③④⑤
⑥⑦
Wei, L.
(2000.) Methodological questions in the study of bilingualism. In L. Wei (Ed.), The bilingualism reader
(pp. 475–486). London: Routledge. ①②③
Wolfson, N.
(1976.) Speech events and natural speech: Some implications for sociolinguistic methodology. Language in
Society, 5(2), 189–209. ①

Appendix 1. Data table of quantitative statistics

Accuracy +B+SN +B−SN −B+SN −B−SN

(N = 12) (N = 13) (N = 5) (N = 10)

# Accuracy # Accuracy # Accuracy # Accuracy

Tokens M Tokens M Tokens M Tokens M
M (SD) M (SD) M (SD) M (SD)

Total Total Total Total

*
Articles N = 248 N = 269 N = 258 N = 314 N = 85 N = 110 N = 211 N = 170
masc 20.66 98.43 18.54 99.3 17 98.67 18.44 95.56
(2.89) (2.52) (2.98) (8.81)
fem 22.42 99.33 17.15 96.97 22 95.17 18.89 88.74
(1.56) (6.27) (9.01) (25.23)
Total 43.08 99 37.23 98.39 39 96.45 37.33 89.37
**
Adjectives
masc 2.17 100, 1.77 100, 1.2 100, 3.11 100,
Total SD = 0 23 SD = 0 6 SD = 0 28 SD = 0
26
fem 2.83 100, 2.69 100, 2.0 100, 2.33 91.67
Total SD = 0 Total SD = 0 Total SD = 0 Total Pop
34 35 10 21 31.62
Total 5 100 4.46 100 3.2 100 5.44 92
Total Total Total Total
60 58 16 49
 * Refers to data from the story retelling task. ①

** Refers to data collected from the concord elicitation task, “¿cómo era el lobo?” ①
Phonetics, phonology, prosody
CHAPTER 4

Social change and /s/ variation in

Concepción, Chile and Lima, Peru
The role of dialect and sociolectal contact
Brandon M. A. Rogers & Carol A. Klee
Ball State University | University of Minnesota

Abstract
This study compares /s/ variation in the Spanish of Concepción, Chile,
and Lima, Peru in the context of sociolectal and dialect contact.
Results for the Chilean data stand in stark contrast to previous studies
in Chile, revealing an overwhelming tendency for elision in all social
groups and providing support for the hypothesis that in Chile
sociolectal leveling is occurring. In contrast, in Lima, where Andean
migrants encounter overt discrimination and where virtually exclusive
use of [s] indexes Andean speech, there is an increase in /s/ weakening
across the second and third generations of migrants, whose production
approximates coastal norms. Thus, in contrast to Chile, /s/ reduction in
Lima continues to serve as a marker of dialectally and sociolectally
salient differences.

Keywords
Chilean Spanish; Lima Spanish; sibilant variation;
sociolectal leveling; dialect contact
1. Introduction
During the twentieth century, many regions in Latin America underwent
significant social transformation. As a result of industrialization, rapid
urbanization and increased access to health services and education, the standard
of living improved and major gains were made in life expectancy and adult
literacy (Astorga, Berges, & Fitzgerald, 2005). Chile, in particular, underwent
rapid urbanization; by 1930 its urban population had surpassed its rural
population, while in Latin America as a whole this landmark was not met until
1960. In contrast, in 1960 in Peru, over half the population still lived in rural
areas, although by 2016 the percentage of rural dwellers had dropped to 21%,
while by that year in Chile only 10% of the population lived in rural zones
(World Bank).1
Rapid urbanization, and the social changes that accompanied it, had major
ramifications for the sociolects of Spanish spoken in urban centers, bringing
different regional varieties and different sociolects of Spanish into contact. This
study compares variation in the pronunciation of /s/ in two contrasting varieties
of Spanish: the Spanish of Concepción, Chile, where the weakened variants are
the most frequent and the Spanish spoken in Lima, Peru, where the prevailing
variant is the sibilant. We will explore the ways in which /s/ variation seems to
be evolving in these two varieties in the context of sociolect leveling in
Concepción and dialect contact in Lima. While /s/ variation is a widely studied
phenomenon in the Spanish-speaking world, the myriad of factors that have been
shown to motivate it have numerous and differing impacts and implications in
different varieties of Spanish. Thus, while sibilant variation is a well-
documented diachronic process in Spanish, it is not enough to simply study it
from within the context of a single variety. Through the study of /s/ variation in
Lima and Concepción, we show that the linguistic outcomes of mobility are
context dependent and are shaped by local historical and social factors.
This chapter begins with a synopsis of the previous literature on /s/
variation in both contexts. Following a description of the methodology of the
studies, we describe the results for each site, focusing on the social variables
relevant to /s/ variation and their interpretation. We conclude with a discussion
of the impact of social changes and dialect contact on /s/ variation in these two
contexts and recommend directions for future research on this topic.

2. Review of the literature

Penny (2000) has observed that the geographical division between regions where
coda /s/ tends to be retained as a sibilant and those in which it is weakened has a
sociopolitical and economic basis dating from the early years of Spanish
colonization. Thus, according to Penny (2000, pp. 148–149) “those areas which,
because of their political and economic importance in the Empire, attracted
prestigious speakers of central Castilian varieties are the ones that retain /-s/
most frequently (most of Mexico, much of Colombia, Ecuador, Peru and
Bolivia).” In contrast, Penny notes that coda /s/ undergoes intense weakening or
loss in the Caribbean, the Pacific coast, and in the southern Cone.

2.1 /s/ variation in Chile

During the colonial area, Chile’s isolation, both geographically and culturally,
allowed its variety of Spanish to develop independently, outside the influence of
the more prestigious and conservative variety spoken in Lima, the seat of the
Viceroyalty. This isolation facilitated the development of several characteristics
of vernacular Chilean Spanish that differentiate it from Limeño Spanish,
including verbal voseo, unique to the voseo observed in other Latin American
Spanish varieties, as well as the frequent weakening of coda /s/. It should be
noted that there is little regional variation in Chilean Spanish in spite of its
remarkable length, comprising 4270 kilometers. Sadowsky and Aninao (2019)
attribute this to the fact that Chile is characterized by hyper-centralization in
which “all forms of power, from political and to economic to social and
intellectual, are concentrated in Santiago,” resulting in an outsized influence of
Santiago, both politically and culturally. This influence extends to Chilean
Spanish, which shows “considerable uniformity of pronunciation” (Lipski,
1994, p. 199), including in the pronunciation of /s/. It must be noted that despite
what Lipski asserts, a number of studies (e.g. Soto-Barba, 2011; Sadowsky,
2012, 2015; Figueroa, Salamanca, & Ñanculeo, 2013; Rogers, 2016; Rogers &
Mirisis, 2016; among others) have documented extensive sociolinguistic
variation in Chilean Spanish. Thus, while geography does not seem to drive
variation and change within Chilean Spanish, social factors have been shown to
strongly pattern with rich and diverse allophonic variation in the region.
Weakening of coda /s/ in Chilean Spanish has been documented in colonial
writings beginning in the sixteenth century and was likely generalized by the
seventeenth century (Contreras, 2007). Previous studies of /s/ variation in Chile
have found high rates of reduction, even in word-final prevocalic position before
a stressed vowel (Cepeda, 1990a), which represents an advanced stage of /s/
weakening diachronically. Eight studies of /s/ variation in spontaneous speech in
Chile (Tassara & Duque, 1990; Cepeda, 1991; Valdivieso & Magaña, 1991;
Valencia, 1993; Cid-Hazard, 2003; Pérez, 2007; Soto-Barba, 2011; and Cerda-
Oñate, Fuentes, Soto, & Hamdam, 2015) are summarized in chronological order
in Table 7. It should be noted that only Pérez (2007) uses instrumental analysis
to identify the variants of /s/. Data come from cities in northern Chile (Iquique,
La Serena), central Chile (Valparaíso, Santiago), southern Chile (Concepción,
Temuco), and the austral zone (Coyhaique, Punta Arenas, Valdivia).
Table 1. Distribution of the three main variants of /s/ in Chilean Spanish in
spontaneous speech

City # of [s] % [h] % Ø % Total

speakers N

Valparaíso (Tassara & 16 – 14% – 69.5% – 16% 8644

Duque, 1990)
Concepción 16 1079 10.2% 8308 72.3% 2253 17.5% 11,640
(Valdivieso &
Magaña, 1991)
Valdivia (Cepeda, 36 21,839 49.8% 14,039 32.0% 7,935 18.1% 43,813
1991)
*
Santiago (Valencia, 18 – 1.3% – 82.5% – 16.2% 6,271
1993)
Santiago (Cid- 12 462 7.6% 3537 58.6% 2041 33.8% 6040
Hazard, 2003)
Chilean news 136 – 13.0% – 61.9% – 25.1% –
broadcasters (Pérez,
2007)
Ñuble Province 4 urban – 20% – 77% – 3% 400
(Soto-Barba, 2011) high
4 urban – 16% – 79% – 35% 400
low
4 rural – 3% – 84% – 13% 400
low
Iquique 4 – 5.8% – 52.5% – 41.7% 120
La Serena 4 – 10.0% – 65.0% – 25.0% 120
Valparaíso 4 – 2.5% – 86.7% – 10.8% 120
Santiago 4 – 7.5% – 72.5% – 20.0% 120
Concepción 4 – 10.8% – 67.5% – 21.7% 120
Temuco 4 – 10.8% – 75.8% – 13.3% 120
Coyhaique 4 – 3.3% – 92.5% – 4.2% 120
Punta Arenas Cerda- 4 – 10.8% – 80.0% – 9.2% 120

**
 **
Oñate et al. (2015)

* We included the .4% glottal stops that Valencia (1993) reports separate from the cases of elision with the
elision data in this table. ①

** A total of 217 interviews of professional with university degrees were conducted, but only the four
individuals from each city whose speech received the highest evaluation by 50 university-educated raters
were included in the analysis. ①

These previous studies have overwhelmingly concurred that the most common
variant of /s/ in Chilean Spanish is aspiration. Tassara and Duque (1990) affirm
that in the central city of Valparaíso, in spontaneous speech, aspiration is most
common and most prestigious production among both male and female speakers.
Valdivieso and Magaña (1991) in a study that included spontaneous speech in
Concepción also concluded that aspiration was the clear prestigious variant
followed by elision and then [s]. Cepeda (1990a), reported that both the sibilant
and aspiration shared the same level of social prestige among speakers in the
southern city of Valdivia and states that elision was stigmatized due to its
association with the lower socioeconomic stratum. In an expanded study of
Valdivian Spanish, Cepeda (1991) reported notably higher levels of sibilant
production than in any of the other studies in Table 1. Despite this, she showed
that female speakers preferred aspiration and that younger speakers showed
higher levels of elision. Based on this, she partially reaffirmed her previous
findings and concluded that /s/ was on a set trajectory of weakening, which was
still somewhat restricted by the elevated levels of remaining prestige that /s/ held
among speakers from more affluent socioeconomic strata.
In Santiago, Valencia (1993) slightly contradicts some of the earlier results
documented in Valparaíso and Concepción and reports higher levels of elision
than [s]. However, analogous to these previous studies, [h] was the clear
preference of speakers and the most prestigious of the three variants studied.
Similarly, Cid-Hazard (2003) indicated that aspiration was the most prestigious
and favored variant for male and female speakers in Santiago. Comparable
results are reported in Chilean media by Pérez (2007), who, like Valencia,
documents higher levels of elision than [s], but at much lower frequencies than
[h]. Soto-Barba (2011), also indicates that aspiration is the most common and
socially prestigious variant in the province of Ñuble.
Most recently, Cerda-Oñate, Fuentes, Soto, and Hamdam (2015) examined
the geographical distribution of [s], [h]. They gathered data in the cities of
Iquique, La Serena, Valparaíso, Santiago, Concepción, Temuco, Coyhaique, and
Punta Arenas. In a more formal reading task, there was an overall preference for
[s] in Iquique, La Serena, Concepción, Temuco, and Punta Arenas. Aspiration
was preferred in this context in Valparaíso, Santiago, and Coyhaique. Elision
was reported in the lowest socioeconomic stratum in all eight cities. However, in
spontaneous interviews, aspiration was the most common variant in all eight
cities, while the frequencies for [s] and elision varied at much lower levels from
city to city. Notably, the authors assert that Chilean Spanish is not a variety of
Spanish that exhibits advanced elision of /s/, as reported in other varieties such
as Dominican Spanish and the Spanish of Panama City.
With the exception of Soto-Barba’s study, which includes lower class
speakers from both an urban and rural context, the other studies summarized in
Table 1 focus on upper-middle class professionals, and the study by Cerda-Oñate
et al. (2015) analyzes only the four professionals in each city whose speech was
evaluated the highest or most correct by 50 college-educated raters. Thus, care
must be taken in interpreting the results of the studies as they tend to represent
the more or less formal speech of professionals in each of the cities and the
sample size is quite small. Even so, with the exception of Cepeda (1991) who
reports 49.8% sibilant production, percentages of sibilant [s] production in
spontaneous speech are relatively low, ranging from 2.5% to 20%, as shown in
Table 1. The most prevalent variant in spontaneous speech, even in the lower
class urban and rural speakers in Soto-Barba’s (2011) study in Ñuble, is
aspiration, which ranges from 52.5% to 86.7%. Reported elision rates range
from 4.2% to 41.7%. Overall, previous studies show that weakened variants of
/s/ are overwhelmingly preferred over the sibilant, and aspiration is preferred
over elision by professionals.
At the same time, recent accounts of Chilean Spanish sociophonetics
(Sadowsky & Salamanca, 2011; Sadowsky, 2015; Sadowsky & Aninao, 2019)
have concluded that the Spanish of the young generation of speakers is
undergoing a process of sociolectal leveling in which “the phonetic and
phonological differences between the consonant allophones of the various
socioeconomic groups are effectively being reduced” (Sadowsky & Aninao,
2019, p. 283). We will examine this claim in relation to coda /s/ weakening in the
discussion of our results.

2.2 /s/ variation in Peru

As noted above, Lima’s role as seat of the richest viceroyalty during the colonial
era, and as the center of all cultural and economic activity in the viceroyalty,
resulted in sustained contact with Spain during that period. Lipski (1994, p. 316)
describes the unbalanced socioeconomic and linguistic environment that
resulted: “large areas of Peru were linguistically marginalized right from the
outset, making the (upper-class) speech of Lima a modern enclave in the midst
of an archaizing and isolated dialect zone.” As Lipski (1994) points out, both
historically and linguistically, Lima shares ‘lowland’ and ‘highland’ antecedents,
which has given rise to a variety in which some /s/ weakening is found, but
where maintenance of the sibilant is favored much more than in other coastal
varieties of Spanish. Penny (2000) observes that /s/ weakening may be a
relatively recent change from below given its patterns of distribution in Lima.
 Previous studies of Lima speech have shown a clear preference for the
sibilant [s] by both professionals and working-class speakers. Carvedo’s (1983,
1989, 1990) sociolinguistic studies of Lima, part of the Proyecto de la norma
culta hispánica, did not reveal many differences between upper-middle and
working-class speakers in the production of the sibilant (i.e. 74.1% and 72.7%
respectively), as can be seen in Table 2. In regard to aspiration and elision,
upper-middle class speakers had higher proportions of aspiration than working-
class speakers (15.3% vs. 6.0%) and lower levels of elision (4.4% vs. 11.3%).
Additionally, while upper-middle class speakers aspirate /s/ word-internally, they
overwhelmingly prefer the sibilant before a pause (92.8%), while working class
speakers also prefer the sibilant in prepausal position (57.6%), but also use other
variants, including elision (23.1%), in that context.

Table 2. Distribution of /s/ variants by social class in Lima (adapted from

Caravedo, 1990: 136)

Variant Upper-middle class Working class

Tokens Percent Tokens Percent

[s] 10222 74.1% 5,817 72.7%

[h] 2105 15.3% 485 6.0%
Ø 611 4.4% 908 11.3%
[z] 137 1.0% – –
[θ] 727 5.3% 784 10.0%

Since the mid twentieth century, Lima has undergone significant social change,
primarily as a result of migration from the provinces. Migrants from rural
regions arrived in Lima seeking employment and better educational
opportunities for their children. In the 1980s migration increased as a result of
political violence in the southern Andes that primarily affected indigenous
communities. Following many decades of migration, 87% of the current
population of Lima comprises migrants from the provinces (36%), their children
(43.5%) and their grandchildren (8%), while only 13% of Lima’s population, i.e.
the so-called “classic Limeños,” has parents and grandparents that were born in
Lima (Arellano & Burgos, 2004). Many of the migrants are from the Andean
region of Peru, where Quechua and Aymara are spoken, and where the regional
variety of Spanish has a number of features that differentiate it from the Limeño
variety. Phonologically, one of the primary differences is in the pronunciation of
coda /s/, which in the Andean region is primarily pronounced as a sibilant. In a
study by Hundley (1983) of nine speakers from three socioeconomic classes in
Cuzco, the sibilant was produced 96.9% of the time, while aspiration occurred in
only 1.2% of cases and elision occurred 1.9% of the time.
In spite of the numerical predominance of migrants and their descendants,
the higher prestige of Limeño Spanish, combined with negative attitudes toward
the Andean variety (Caravedo, 2014; Mick & Palacios, 2013; Salcedo, 2013),
has resulted in accommodation by migrants, and especially by their children and
grandchildren, to the Limeño variety. For example, Klee and Caravedo (2006),
compared coda /s/ variation in a group of first- and second-generation migrants
living in a long-established shantytown with the /s/ production of classic
Limeños from the lower middle class. They found that Andean migrants and
their adult children produced a higher rate of elision (76% and 74% respectively
when measuring elision vs. aspiration) than classic Limeños (56%). As in
Caravedo’s (1990), study of working-class speakers, elision occurred more
frequently than aspiration in all three groups, although surprisingly the rate of
elision of the Andean migrants and their adult children surpassed that of the
classic Limeños. A subsequent study by Klee, Rogers, Caravedo, & Dietz
(2018), will be discussed in more detail below and the results will be compared
with new data on /s/ variation in Chile.

3. Methodology

3.1 Participant selection

The participants for the current chapter were taken from two different linguistic
corpora consisting of sociolinguistic interviews in the Bio-Bio Region of Chile
and in Lima, Peru.

Chile:
As part of an ongoing larger corpus of spontaneous Chilean Spanish, the
sociolinguistic interviews of 22 speakers (11 female and 11 male) were
analyzed. The speakers were from the Bio-Bio Region of Chile in the
southern-central city of Concepción, and the surrounding neighborhoods,
called poblaciones, of Michaihue, Boca Sur, Candelaria, Lomas Coloradas,
and Villa San Pedro. Participants were recruited through the social networks
of one of the authors and the social networks of their friends and family. The
interviews lasted between 15 and 35 minutes and were all conducted by the
same author in the participants’ homes, local recreation centers, office
buildings, and local churches. The speakers were divided into 3 age groupings
based on Rogers (2016): 18–24, 25–44, 45–49. Education was divided into six
different levels, and five different socioeconomic strata were used based on
Sadowsky’s (2012) modified version of Esomar (Adimark, 2000).2 The
socioeconomic strata used were: low, high-low, low-mid, mid-mid, high-mid,
and low-high. The education groupings were the following:
1. Without elementary education or elementary education incomplete
2. Elementary education complete
 3. High-school level technical/professional education finished or
incomplete or high school education incomplete
4. High school education complete and/or superior technical/professional
education incomplete
5. Superior technical/professional education complete or four-year
university education incomplete
6. Four-year university education complete.

See Table 1 in the Appendix for a summary of the Chilean speakers and the
social factors used in the analyses of the Chilean data.

Lima:
The Lima corpus of sociolinguistic interviews was gathered in 2012–2013 by
a fieldworker from the Limeño district of Los Olivos, which is considered to
be the most prosperous migrant district in the city. The fieldworker had
extensive social networks in Los Olivos through which he selected the
participants. Speakers were second- and third-generation migrants between 18
and 39 years old. In all, 45 speakers were interviewed in various parts of Los
Olivos, of which a subset of 26 speakers was selected for the current study.
This was due to the fact that some of the interviews had too much background
noise to accurately categorize the acoustic signal of the sibilant /s/. Speakers
were categorized based on migrant generation, heritage (Andean, mixed, non-
Andean, based on whether the speakers, their parents or their grandparents
were from Andean or non-Andean regions), gender, education, occupation,
and social networks, as seen in Table 2 in the Appendix. Fifteen participants
were second generation migrants with at least one parent born in the provinces
who subsequently migrated to Lima; ten participants were third generation
migrants with at least one second-generation parent. One speaker was
categorized as fourth generation, but was included with the third-generation
 speakers for the analysis as she was born and raised in Los Olivos. Education
was divided into five categories ranging from some higher technical
education, completed technical education, some higher education, completed
university to completed postgraduate studies (masters and/or doctorate). In
regard to occupation, participants were categorized as unskilled laborers,
skilled laborers, or students. With respect to social networks, speakers were
placed into three categories: in the first were speakers whose social networks
were primarily within Los Olivos and who had positive attitudes toward Los
Olivos and planned to continue living there over the next ten years; in the
second group were speakers whose social networks comprised family and/or
friends both inside and outside of Los Olivos; and, in the third group were
those whose social networks were primarily outside of Los Olivos and who
planned to move to a different district of Lima within the next ten years.

3.2 Data measurement and instruments

In the majority of cases, tokens in both the Peruvian and Chilean interviews were
taken after the 10-minute mark of each interview when speakers were more
acclimated to the interview and more comfortable with those conducting the
interviews. A total of 201 tokens of /s/ at syllable-internal (['[Link]]) and
word-final (['[Link]]) junctures were analyzed per participant. In the case of
four of the Chilean speakers, a sample size of 201 sibilants after the 10-minute
mark was not possible due to the shorter length of these interviews. In these
cases, the author moved back from the 10-minute mark at 30 second intervals
until the 201 tokens-per-speaker threshold was met.3 Tokens were categorized as
either sibilants ([s]), aspirated ([h]), or elided ([Ø]) as has been done in
numerous previous studies regarding /s/ variation in Spanish, primarily due to
the greater level of perceptual saliency between categories (e.g. Cepeda, 1990b,
1995; Caravedo, 1990; Hundley, 1983; Soto-Barba, 2011; Cerda-Oñate et al.,
2015 among many others). All tokens from the Chilean data and the Los Olivos
data were acoustically verified using Praat (Boersma & Weenink, 2015) using a
number of methods to determine their classifications. First, a sibilant was
determined to be present when aperiodicity was observed in the waveform along
with turbulence in the spectrogram. Aspirated tokens were labeled based on the
presence of glottalized turbulence in the spectrogram. The turbulence observed
for [h] and [s] differed due to the different manners of articulation. When a
speaker produces [s], the blade of the tongue either partially approaches or
makes complete contact with the alveolar ridge, resulting in a greater amount of
turbulence than when [h] is produced. The stridency and energy output that
results when [s] is articulated can be observed particularly in the upper limits of
the visible spectrogram. The voiceless aspirated fricative [h] results in a notably
smaller amount of oral constriction, which also reduces the amount and the
intensity of the stridency of the visible spectrographic turbulence. Additionally,
glottal turbulence is more evenly distributed throughout the visible spectrogram.
Finally, tokens were marked as elided if the spectrogram and the waveform did
not present any visible evidence of glottalization or turbulence, as well as
aperiodicity. Figures 1–3 illustrate examples of how [s], [h] and [Ø] were
determined. In Figure 1, the arrows point out the strident spectrographic
turbulence as well as the aperiodic waves present when producing [s]. Figure 2
demonstrates the spectral turbulence and aperiodicity of the aspirated [h]. Of
note is the how the distribution of the turbulence in the spectrogram of Figure 2
is weaker and more evenly distributed throughout the visible spectral space than
in Figure 1. Finally, Figure 3 is an example of elision at the end of the word
gratis, and lacks any acoustic evidence of turbulence, stridency, or aperiodicity.
Figure 1. Example of [s] according to the criteria used in the current study
Figure 2. Example of [h] according to the criteria used in the current study
Figure 3. Example of elided /s/ from the current study
Likewise, auditory cues were used simultaneously as an additional means to
confirm the data that was verified acoustically. Impressionistically, [s] is
different from [h] and Ø in that it sounds more strident and more constricted.
The aspirated variant is breathier than the sibilant, less strident, and sounds less
constricted because of its different manner and place of articulation. Cases of
elision lack any of the above-mentioned features associated with [s] and [h], and
at times were observed as brief glottal stops followed by non-sibilant segments.
Even though tokens were grouped into three distinct categories, there was
considerable phonetic variation in each category, especially in the Chilean data.
Cepeda (1990a, 1990b, 1991), Valencia (1993), and Sadowsky and Salamanca
(2011) have documented up to 13 different allophonic realizations of /s/ in
Chilean Spanish, as seen in Table 3.

Table 3. Allophonic realizations of /s/ in Chilean Spanish in previous studies

Allophonic symbol Description

[s] Voiceless dorso-alveolar fricative

[h] Voiceless glottal fricative
Ø Elided segment
[t͡s] Voiceless dorso-alveolar affricate
[s̞] Voiceless dorso-alveolar approximant
[z] Voiced dorso-alveolar fricative
[θ] Voiceless postdental fricative
[θ̞] Voiceless postdental approximant
[θ̟] Voiceless interdental fricative
[θ̝̝] Voiceless interdental approximant
[s̟] Voiceless dorso-alveolar sibilated fricative
[ɕ] Voiceless alveo-prepalatal fricative
[ʔ] Glottal stop

For the present study, of the variants listed in Table 3, in the Chilean data [s], [h],
Ø, [z], [θ], and [ʔ] were observed in the specific phonetic contexts analyzed.
Additionally, [ɦ] (voiced glottal fricative i.e. ['miɦ.mo]), [x] (voiceless velar
fricative i.e. ['[Link]]), and [ʃ] (voiceless dorso-palatal affricate) preceding [t͡ɹ̥̌]
['maʃ.t͡ɹ̥̌a.'β[Link]], were also observed as allophones for /s/. In the Los Olivos data
only [s], [h], [ɦ], [x], Ø, [z], and [ʔ] were observed.
Tokens were considered sibilants if they were realized as any of the
following allophones: [s], [θ], [ʃ], [z], [ˢ] (weakened voiceless alveolar fricative),
and [z] (weakened voiced alveolar fricative). Tokens categorized as [h] were any
of the following realizations: [h], [ɦ], [h] (weakened voiceless glottal fricative),
and [ɦ] (weakened voiced glottal fricative). It must be noted that in instances of
[z] and [ɦ], due to the fact that these segments are generally fully voiced, the
aperiodic oscillations present in voiceless productions were largely absent in the
waveform, so the spectrographic turbulence patterns were used to confirm their
presence. Tokens that were phonetically elided were those that left no acoustic
evidence of a sibilant or a glottal fricative. Additionally, in cases where words
ended in a glottal stop and were not followed by any acoustic evidence of
aspiration or assibilation, /s/ was considered to be elided (e.g. ['maʔ], ['[Link]ʔ],
['[Link]ʔ]). Consecutive orthographic representations of /s/ (e.g. {los sordos}) and
instances of [h] followed by [x] (e.g.[loh.'xu̯e.ɣ̞os]) were excluded.
Finally, a phenomenon observed most frequently in the Chilean data was
word final vowel devoicing. At times, in cases where speakers produced words
that ended in /s/, a devoiced or a voiceless vowel made it difficult to tell if the
speaker was aspirating or simply devoicing a vocalic segment and eliding /s/. In
cases where consistent and relatively stable formants were observed throughout
the entire segment with no following evidence of aspiration or assibilation, /s/
was marked as elided. In cases where the formants were less regular and stable
and accompanied and/or followed by evidence of glottalization or sibilant
stridency, /s/ was labeled appropriately as either [h] or [s]. Cases where it was
impossible to determine if the formants belonged to a vowel or a weakly voiced
glottal fricative were excluded. Figure 4 illustrates a case of a devoiced vowel,
[a], in the sequence “hartas cosas” where the following /s/ was considered to be
elided.

Figure 4. Example of a devoiced vowel and elided /s/ from the current study
3.3 Statistical analysis

As previously mentioned, the variables were coded in a categorical fashion even

though there was considerable intracategorical variation. Because of this
variation, previous studies, such as Erker (2012), have chosen to analyze /s/ as a
continuous variable using center of gravity measurements and linear regression
analyses. However, when dealing with categorical variables and repeated
measures, such as those of the current study, linear regressions are not
appropriate statistical frameworks from which to analyze the data. Despite this,
it is important to recognize the /s/ does exist along an acoustic continuum. The
true underlying acoustic value, represented as ψ, can only be captured by way of
observable ordinal categories. In other words, the sibilant undergoes weakening
within its category before it is considered [h]. The aspirated variant also has to
weaken before it is elided. As a result, an analysis that makes the overarching
assumption that [s], [h], and Ø are unordered categories, does not take into
account the natural ordering of the response. A plausible option for this type of
situation is a proportional-odds mixed effects model. First, because repeated
measures were coded for each participant, the assumption of independence can
no longer be made, and as a result mixed effects analysis should be used to
account for the dependence in the data. Second, within the proportional-odds
framework, the researcher can model an ordered dependent variable while also
restricting the overall complexity of the model by assuming that the categories
respond similarly to changes in independent variables.
To model the proportional odds, the logit function, also referred to as the
log-odds, models the cumulative distribution function (CDF) which is
represented as P[Y ≤ y]. Thus, with a matrix of covariates (X) for each
observation recorded, the logit function models the CDF as a linear combination
of a fixed effects vector (β) and a random effects vector (U):
Z is a random effects matrix that recognizes the correct random effect element of
U for each of the individuals that make up the given sample. The possible values
of j are determined by the number of response categories; in the current study j is
either 1 or 2. The model operates on the assumption that β is constant over the
various levels of the response. For the current study, ψ exists on an ordered
continuum observable only at Ø, [h] and [s]. This situation requires a model with
2 break points, α1 and α2. Y = y1 = Ø if ψ < α1, Y = y2 = [h] if α1 < ψ < α2, and
Y = y3 = [s] if ψ > α2.
Given that the goal of most studies, including the current investigation, is to
generalize to the extent possible for a larger population, random effects are
studied to better understand the variability of the specific populations analyzed.
By doing so, the effect of individual variation on the data can be observed.
Likewise, it is useful to be able to determine how different certain individuals
may be when compared to the rest of the population. Thus, random intercepts by
speaker were included in the model. Models were fit with an iterative algorithm
in R using clmm{ordinal} (Christensen, 2015).
In summary, the mixed effects proportional odds model used in the current
study is advantageous for several reasons. First, it considers individual variation
and the fact that multiple measurements were taken from each speaker. Second,
while the dependent variables were coded categorically due to the salient
differences between each category, the proportional odds portion recognizes that
there is variation within and across categories. In other words, it considers that
the responses are ordered. Thus, inherent in the analysis is the fact that the
responses exist on a continuum, which more accurately reflects the reality of the
production data. Finally, by setting thresholds, the model calculates different
likelihood ratios for various groups and independent variables. These ratios
allow for the analysis of different behaviors by using probabilities, and to an
extent, permit researchers to predict the trajectories of these behaviors. This is
also advantageous when comparing cross-dialectal differences, such as those
analyzed in the current study.

4. Results

4.1 Concepción, Chile

Similar to previous studies on /s/ in Chile summarized in Table 1, our results
indicate that speakers produce the sibilant 10.4% of the time. However, in
contrast to the results of previous studies, speakers elide the sibilant at notably
higher rates than those at which they aspirate, as summarized in Table 4. While
elision occurred in 71.2% of all observations, the aspirated variant was produced
at the second highest frequency, or 18.4% of the time.

Table 4. Descriptive statistics of /s/ production in Concepción

Production Cases Proportion Cumulative proportions

Ø 3150 0.712 1.000

[h] 814 0.184 0.29
[s] 458 0.104 0.10
Total 4422 1.000

Similarly, in Valdivieso and Magaña’s (1991), study of Concepción, the overall

frequency of the sibilant is virtually the same (10.4% vs. 10.2% respectively).
However, the frequencies of elision and aspiration have essentially flip-flopped;
in 1991 aspiration was the most frequent variant (72.3%) and elision was the
second most common (17.5%). The 2015 study by Cerda-Oñate et al., which was
only based on four upper-middle class speakers whose Spanish was rated as
“most valued,” also found a similar frequency of the sibilant (10.8%) and a
slightly lower frequency of aspiration at 67.5% and a slightly higher frequency
of elision at 21.7% than in Valdivieso and Magaña’s (1991), study. We will
examine the possible reasons for the discrepancy between our results and those
of previous studies of Concepción in the Discussion section of this chapter.
While the overwhelming tendency in our data was for speakers to elide, the
patterns of weakening played out in different ways according to different social
variables. All estimates that follow were produced on the logit (log-odds) scale,
however, the model can be used in other useful ways to help with the
interpretation of the results. For this reason, threshold estimates were converted
to a cumulative probability scale for the social variables. These can provide a
comparison with the observed cumulative proportions as shown in Table 4.
Typically, these should be reasonably close, but can vary depending on how well
the model fits to the data. In order to identify significant pairwise differences
within level of the factors being analyzed, the odds scale was used, which
exponentiates the logit scale difference estimates.

4.2 Chilean social model with three variants: [s], [h], Ø

Because education and stratus overlap in Esomar (Adimark, 2000), two different
social models were run for each combination of allophonic realizations of /s/. In
the proportional-odds mixed effects model used in the current study, the social
model is specified in the following way:

Social Model Specification 1- For the jth response given by ith individual,
logit(P(Yij ≤ Ø)) = α1 − [β1* education/stratus ij + β2 * age groupij + β3 *
genderij + ui]
logit(P(Yij ≤ Ø)) = α2 − [β1* education/stratus ij + β2 * age groupij + β3 *
genderij + ui]

The first social model was run with all three allophones. When stratus was
included with age and gender, only one significant pairing resulted (those in the
lowest stratum were 0.0964 times more likely to reduce /s/ than speakers in the
mid-mid stratum (p = 0.0185)). However, when stratus was replaced with
education, there were notably more significant results. The results of a likelihood
ratio test (LRT) for each of the social factors in the second social model in
Table 5, indicate that there are significant differences in the predicted cumulative
probability of the pronunciation of /s/ based on education (p < 0.001), and gender
(p = 0.0039). Age group was not significant at the 0.05 threshold (p = 0.08).

Table 5. Results of likelihood ratio test for each of the social factors for
Concepción

Variable LRT χ2 P-value

Education 20.554 <0.001

Gender 8.346 0.0039
Age group 5.051 0.08

Pairwise Tukey tests were run to determine which pairings within each of the
three social variables were statistically significant. The results are shown in
Table 6. Significant pairings for education showed that those speakers who had
not finished their primary education (i.e. Level 1) were slightly more likely (i.e.,
0.20 times more likely) to reduce /s/ than those in who had completed their
primary education (Level 2), 0.22 times more likely to reduce /s/ that those who
had completed high school (Level 4), and 0.16 times more likely to reduce /s/
that speakers who had completed a four-year university education (Level 6).
Those speakers who had not finished high school or high school level technical
education were 0.43 times more likely to weaken /s/ than those speakers who
had completed high school (Level 4) and were 0.44 times more likely to weaken
/s/ than those who had completed technical/professional education (Level 5) and
were 0.32 times more likely to weaken /s/ than those who had completed a four-
year university education (Level 6). In line with previous studies, males were
1.77 times more likely to reduce or elide /s/ than females.

Table 6. Pairwise results of Tukey comparisons (significant at the 0.05 level) for
Concepción

Pairing Logit scale difference estimate P-value Odds estimate (95% CI)
(S.E.)

Education: Level 1-Level −1.6053 (0.5494) 0.0407 0.2008 (0.0669,

2 0.6027)
Education: Level 1-Level −1.5272 (0.4542) 0.01 0.2171 (0.875, 0.539)
4
Education: Level 1-Level −1.8233 (0.5496) 0.0117 0.1615 (0.054, 0.485)
6
Education: Level 3-Level −0.8473 (0.2824) 0.0322 0.4286 (0.244, 0.754)
4
Education: Level 3-Level −0.8254 (0.2002) <0.001 0.438 (0.294, 0.654)
5
Education: Level 3-Level −1.1434 (0.3795) 0.0311 0.3187 (0.149, 0.681)
6
Gender: Female-Male 0.5689 (0.1788) 0.0015 1.7662 (1.235, 2.525)

Additional social models were run with two-variant combinations of [h]/[s], [h]/
Ø, and [s]/ Ø as well. When comparing the two weakened variants ([h]/Ø), there
were no conclusive patterns for stratus or education and there were no
statistically significant differences between age groups. Likewise, there were no
significant differences between genders, most likely due to the very high levels
of elision documented for both groups. When [s] and [h] were compared there
were no statistically significant results. However, when [s] and Ø were analyzed
with education, age group, and gender, the model produced several significant
results. While results did indicate that speakers who had not finished high school
were more likely to elide than produce the sibilant than speakers from the three
highest education levels, no other general conclusive patterns were observed.
With regards to age and gender, the 18–24-year-old speakers were 2.17 times
more likely to produce the sibilant than the 25–44-year-old speakers (p < 0.001)
and female speakers were 2.2 times more likely to produce the sibilant than elide
when compared to males (p < 0.0001).
In summary, overall the processes of /s/ retention and reduction in
Concepción are mostly driven by education and gender. Higher levels of
education resulted in slightly greater probabilities of sibilant production and less
/s/ weakening. The overall results for gender indicate that males are more likely
to weaken /s/ than their female counterparts, as has been found in previous
studies in other parts of the Spanish-speaking world. The fact that age is not a
significant variable when comparing all three variants to one another, would
seem to indicate that /s/ variation may be a stable sociolinguistic variable within
this community. Nonetheless, the rates of elision found in our study are much
higher than those recorded previously in Chile.

4.3 Lima, Peru

As mentioned earlier, the Lima results come from a previous study (Klee et al.,
2018) conducted by the authors. Contrary to the results from Concepción, in
Lima the sibilant remains the preferred variant, as indicated in Table 7. Overall,
the frequency of the sibilant in Los Olivos was 69.0%, while aspiration occurred
18.4% of the time and elision, which had a frequency of 12.7%, was the least
common of the three variants. While the rate of aspiration is identical to that
found in Concepción, the frequency of the sibilant and elision is the reverse: in
Concepción the sibilant occurred with a frequency of 10.4% and elision occurred
71.2% of the time.

Table 7. Descriptive statistics of /s/ production in Lima (Klee et al. 2018: 44)

Production Cases Proportion Cumulative proportions

[s] 3603 0.690 1.000

[h] 959 0.184 0.31
Ø 664 0.127 0.12
Total 5226 1.000

To determine which social variables explained the variation in /s/ production, we

used a proportional-odds mixed effects model in which the social model was
specified in the following way:

Social Model Specification 1- For the jth response given by ith individual,
logit(P(Yij ≤ Ø)) = α1 − [β1* genderij + β2 * generationij + β3 * originij + β4 *
educationij + β5* occupationij + β5* social-networkij + ui]
logit(P(Yij ≤ [h])) = α2 − [β1* genderij + β2 * generationij + β3 * originij + β4
* educationij + β5* occupationij + β5* social-networkij + ui]

The social factors that were significant at the .05 level were migrant generation
and education, as shown in Table 8. Generation and social network approached
the .05 level of significance and will be discussed as they contribute to our
understanding of /s/ variation in Los Olivos. Neither family origin nor
occupation were significant.4

Table 8. Results of likelihood ratio test for each of the social factors in Lima (Klee
et al. 2018: 45)

Variable LRT χ2 P-value

Migrant Generation 4.40 0.0359

Education 9.61 0.0475
Gender 3.53 0.0603
Social Network 5.10 0.0778
Origin 3.58 0.1533
Occupation 0.35 0.8386
As with the Chilean data, pairwise Tukey comparisons were performed to
determine which pairings within each factor were significant and the results are
displayed in Table 9. In regards to migrant generation, there were significant
differences between the second and third generations, with second generation
speakers 1.69 times more likely to produce [s] than third generation speakers.
Overall, second-generation speakers not only produce more [s] (72.0% vs.
65.3%) than third generation speakers, they also use less aspiration (16.9% vs.
19.9%) and elision (11.1% vs. 14.7%).

Table 9. Pairwise results of Tukey comparisons (significant at the 0.10 level) in

Lima (Klee et al. 2018: 47)

Pairing Logit scale difference estimate P- Odds estimate

(S.E.) value (95% CI)

Migrant: 2nd–3rd Generation 0.522 (0.237) 0.0280 1.68 (1.05, 2.71)

Education: Postgrad Studies – 1.218 (0.452) 0.0549 3.38 (1.37, 8.35)
Completed Tech
Gender: Female – Male 0.551 (0.289) 0.0644 1.74 (0.97, 3.09)
Social Network: 1–3 0.679 (0.296) 0.0653 1.97 (1.08, 3.61)

With respect to education, the only pairing that was significant was between
speakers who had completed postgraduate studies and those who had finished
technical education. Those with postgraduate studies were 3.38 times more
likely to produce the sibilant than speakers who had finished technical education.
There were no other significant differences between educational groups, but it
should be noted that, in contrast to the Chilean data, there were no participants
who had not completed high school and all speakers had some university-level
or technical education.
Regarding gender, female speakers were 1.74 times more likely than males
to produce [s], which mirrors the results from Concepción, as well as those in
other areas of the Spanish-speaking world. However, when examining the
relationship between generation and gender, it appears likely that there is an
interaction, as can be seen in Table 10. While in the second generation there are
large differences between males in females in their production of the sibilant
(61% vs. 77%) and elision (21% vs. 6%), by the third generation the differences
have diminished, and males use slightly more of the sibilant than females (67%
vs. 64%) and only slightly more elision (16% vs. 14%), while females tend to
produce more aspiration (17% vs. 22%). Previous studies of Limeño Spanish
(Hundley, 1983; Caravedo, 1987) also found no significant differences between
males and females in the production of /s/.

Table 10. Crosstabs of generation by gender (Klee et al. 2018: 51)

2nd generation 3rd generation

# % # %

Males [s] 617 61% 537 67%

Males [h] 178 18% 137 17%
Males Ø 210 21% 130 16%
Females [s] 1547 77% 902 64%
Females [h] 338 17% 306 22%
Females Ø 125 6% 199 14%

One final factor approached .05 significance in the Peruvian data and that was
whether speakers had social networks primarily inside vs. outside of Los Olivos.
Speakers with strong networks inside the community tended to use less
aspiration (13.7% vs. 20.2%) and more elision (16.1% vs. 10.5%) than speakers
with strong social networks outside Los Olivos. This is important because
previous research (Caravedo, 1990) showed that the middle class in Lima tends
to use more aspiration, while the working class produces more elision. Residents
of Los Olivos who planned on leaving the district for more affluent areas within
ten years appear to accommodate their /s/ weakening to the norms of middle-
class, rather than to working-class Limeños. Overall, the factor that correlates
most strongly with /s/ variation in Los Olivos is migrant generation, and other
factors that correlate to a lesser degree include education, gender (but only in the
second generation), and social networks.

5. Discussion
With regards to Concepción, the results of the current study stand in stark
contrast to what has previously been stated about /s/ variation in Chile. While
higher levels of overall weakening when compared to sibilant production were
reported in the current data set, as has been the case in most previous studies, the
assertion that aspiration is the overwhelming and most stable weakened variant
of /s/ in Chilean Spanish did not hold up. Overall, in just under 75% of all
observed cases, elision was clearly the most common variation of /s/ in
Concepción. The highest levels of aspiration, which were documented among
female speakers, only reached 22%. The use of proportional odds indicated that
certain groups were more likely to produce sibilants than others, but despite this,
among all groups, the overwhelming tendency was for elision. The main driving
factors behind the variation observed in the current data were gender and
education. As has been found in other studies, speakers with less education and
males were more likely to reduce than females and individuals with more
education. There were no significant differences between age groups or
socioeconomic strata. This clear preference for elision across different social
variables and dynamics is indicative that /s/ weakening, especially elision, is
much more advanced in Concepción, and possibly more generally in Chile, than
has been previously reported.
One possible explanation for the leveling of /s/ variation across different
social variables is rooted in what Sadowsky (2015), calls a sociolectal leveling in
Chile wherein previous sociolinguistic variation has been reduced due to
significant social changes and progress in Chilean society. These changes have
increased access to higher education and reduced the national poverty rate.
According to Sadowsky, this has resulted in greater social mobility, a generally
more educated populace, and perhaps most relevant to the current study, contact
between different socioeconomic rungs of Chilean society has increased, which
favors sociolinguistic leveling. Sadowsky indicates that this reorganization of
sociolectal markers that is occurring in Chile will not permanently reduce the
sociolinguistic variation in Chilean Spanish. Rather, it appears that speakers are
at an intermediate stage in a gradual progression toward the development and
distribution of new sociolectal markers. Sadowsky states that because significant
social gaps continue to exist and be reinforced in Chile, eventually new
allophones will develop and be distributed throughout the different social strata
in Chile, thus diversifying this variety once more.
Based on the assertions of Sadowsky, and the results reported in the current
study, it is possible that /s/ variation is part of this allophonic and sociophonetic
reorganization. The social markers assigned to [s], [h] and Ø are no longer as
notable and telling of social differences as they once were. If, in fact, these
sociolectal markers were still as relevant, it would be expected that the data
would reflect previously reported differences between different social groups
based on the factors analyzed. Sociophonetic studies on other segments in
Chilean Spanish have shown clear social differences and motivations. For
example, Sadowsky (2012), demonstrates vocalic differences among younger
speakers in Concepción and indicates that these differences are the result of
changes being primarily driven by young females from lower socioeconomic
strata. Rogers (2016), and Rogers and Mirisis (2016), showed that lenition and
deletion of stop consonants in Chilean Spanish are strongly driven by younger
speakers and males. Figueroa, Salamanca, and Ñanculeo (2013), demonstrate
that more fricativized productions of /t̪ɾ/, /r/, / /t͡ʃ̠ / and /ʝ̞/ are associated with the
speech of lower socioeconomic strata. Yet, these clear differences are not
observed in the data reported in the current investigation with regards to /s/
weakening. It appears that despite observed variation between different social
groups and factors, the increasing norm across gender, age, education, and
socioeconomic stratification, at this stage, in Concepción, is elision. These
results are curious, because previous studies on Chilean sibilant variation have
reported notable social differences and negative stereotypes associated with
elision. Thus, the data are suggestive of a type of sociolectal leveling similar to
that reported by Sadowsky (2015).
In contrast, in Lima /s/ variation continues to serve as a marker of
dialectally and sociolectally salient differences. The exclusive use of the sibilant
seems to be perceived in Lima as a marker of Andean speech, as reflected in the
following comment of a second-generation Limeña, whose mother had migrated
from the Andean region: “Los de la sierra tienden a … según los cusqueños
como que hablan mucho con la S al final también pronuncian mucho la S.”
[People from the mountains tend to … according to Cuzqueños they speak a lot
with S at the end they also pronounce the S a lot.”] Frequent use of the sibilant in
final position appears to index Andeanness in Lima. At the same time, aspiration
continues to be associated with middle class speakers and elision with the
working class. In Los Olivos there were significant differences between the
second and third generations in their production of /s/: the third generation
produced the sibilant less than the second generation and used a rate of
aspiration that slightly surpassed that of the upper-middle class in Caravedo’s
study (1990). In this respect, they seem to be accommodating to the middle-class
norms of /s/ weakening. However, at the same time their rates of elision were
over three times higher than those of the upper-middle class. This may reflect the
fact that third-generation migrants are establishing new norms in Lima, possibly
accelerating the diachronic trend of /s/ weakening. Further studies are needed to
determine whether classic Limeños still follow the same norms as Caravedo’s
(1990), participants or whether /s/ weakening has continued to evolve in this
group.

6. Conclusion
Sibilant weakening and reduction is one of the most extensively documented and
analyzed linguistic phenomenon in Spanish linguistics. Lipski (2011), states that
over half of the worldwide Spanish-speaking population speaks a variety that
exhibits a certain level of /s/ reduction and asserts that the phenomenon is
“perhaps the most robust phonetic differentiator of regional and social dialects”
(p. 73). Despite it being so common across a variety of dialects, it does not
pattern phonetically or socially in a uniform manner interdialectally. From a
purely sociolinguistic perspective, distinct social factors play differing roles in
the way that it is manifested and perceived by speakers of different varieties of
Spanish. While weakening (Caravedo, 1990) or extreme weakening (e.g.
Cepeda, 1995) can be stigmatized in some dialects, more extreme levels of /s/
conservation and maintenance can be stigmatized in other varieties based on
very different social values and perceptions (e.g. Mack, 2011; Klee et al., 2018).
The current chapter shows that increased /s/ reduction in Los Olivos is primarily
the result of contact between different dialects and sociolects. Because higher
levels of sibilant maintenance are associated with more rural and Andean speech,
subsequent Andean migrant generations are abandoning the norms of their
predecessors’ sociolects in favor of the more prestigious sociolect of Lima.
In regard to Chile, /s/ variation appears to be leveling off and normalizing
across different social variables and groups due, in part, to increased contact
between different sociolectal groups. If this is indeed the case it may no longer
be enough to simply study /s/ in terms of [s], [h] and elision. As previously
stated, within each of the three categories used in this and the majority of the
previous studies on the topic there is considerable phonetic variation. Many of
these intracategorical variants possess varying levels of social capital in Chile.
For example, all the interdental variants are stigmatized to differing degrees. The
whistled sibilant is a fairly recent phenomenon, and also has negative social
value. Yet within the tripartite framework that /s/ has been treated, there is social
variation within each category that is largely ignored, and any potential
sociophonetic reorganization that is occurring goes mostly ignored. Furthermore,
it merits mentioning that in the current data, apart from the allophones that were
documented in the methodology section, an additional previously unattested
allophone was observed repeatedly. In most cases it occurred in word-internal
position before /t/, best represented at this juncture as [hs]. It starts as an
aspirated production [h], but ends as a sibilant [s], as seen in Figure 5.

Figure 5. An example of a previously unattested allophone of /s/

Sadowsky (2015) indicates that several of the variants of /s/ are more recent
developments with varied social reception. The fact that so much variation exists
within the traditional tripartite system of /s/ categorization in Chilean Spanish,
and more recent variants are being reported, suggests that speakers of Chilean
Spanish may be in the process of reorganizing and redistributing new allophones
associated with /s/. According to Sadowsky’s predictions, this reorganization, if
allowed to run its natural course, would eventually result in at least a partial
restoration of the sociophonetic variation of /s/ that has possibly been lost to
sociolectal leveling, with new allophones being assigned differing levels of
social value. As a result, with Sadowsky’s assertions and what the current data
indicate, more precise approaches that do not group so many allophones into the
same categories may have to be taken in the future if the current sociolinguistic
status of /s/ in Chilean Spanish, especially in Concepción, is to be understood in
a more complete and thorough fashion.
Despite there being less intracategorical variation reported for Limeño /s/,
there was also a notable number of different variants that were coded as either
[s] or [h] in the Los Olivos data. Likewise, while not observed in the Lima data
used for the current chapter, Caravedo (1990), reported instances of [θ] for /s/ in
Lima. It is possible that some of the variants that were coded as [s] or [h] have
social capital and/or covert prestige that do not surface using more traditional
methods of analyzing /s/ variation. Thus, there may be levels of intracategorical
sibilant variation in Lima that have yet to be observed or documented and
require more detailed approaches than simply studying three categorical
variables.5
Finally, contact linguistics has traditionally examined how different
languages influence one another and has analyzed the resulting changes in the
linguistic subsystems of each language. However, as demonstrated by the Los
Olivos and the Chilean data in the current chapter, as well as previous studies of
dialect contact elsewhere (e.g. Trudgill, 1986 and many others), contact-induced
change can also occur within different sociolects of the same language. The
study of language-internal contact-induced change can demonstrate how
communities assign social capital to different linguistic phenomena over time
and how different groups react and adapt linguistically when social
transformations bring regional dialects and/or different sociolects into more
intimate contact. The phenomenon of /s/ weakening in Spanish, in particular,
continues to provide abundant opportunities to analyze language change
resulting from dialect and sociolectal contact.

Notes
1. <[Link] ①

2. Esomar is a system designed specifically to determine the socioeconomic strata of individuals in Chilean
society based on a number of social and economic factors. Sadowsky’s version modifies the methodology
of Esomar to more effectively create socioeconomic strata for sociolinguistic research carried out in Chile
using education and occupation as the primary determining factors. ①

3. The furthest back we had to go for one speaker was the 148 second (2.47 minutes) mark. ①

4. While it is common practice to set 0.05 as the threshold for p values, recently the American Statistical
Association (ASA) published a statement specifically addressing p-values, see <[Link]
.com/doi/full/10.1080/00031305.2016.1154108>. At one point they assert: “Researchers should bring many
contextual factors into play to derive scientific inferences, including the design of a study, the quality of the
measurements, the external evidence for the phenomenon under study, and the validity of assumptions that
underlie the data analysis…The widespread use of “statistical significance” (generally interpreted as
“p ≤ 0.05”) as a license for making a claim of a scientific finding (or implied truth) leads to considerable
distortion of the scientific process.” ①

5. For example, Schmidt & Willis (2011), and Chappell (2016), have indicated that [z] is socially
conditioned to varying degrees in Mexican and Costa Rican Spanish respectively. ①

References
Adimark, S. A.
(2000.) El nivel socio económico esomar. Manual de aplicación. Retrieved from [Link]
/idm/documentos/[Link] ① ②
Arellano Cueva, R., & Burgos Abugattas, D.
 (2004.) Ciudad de los Reyes, de los Chávez, los Quispe. Lima: Epensa/Arellano Investigacion de
Marketing. ①
Astorga, P., Berges, A. R., & Fitzgerald, V.
(2005.) The standard of living in Latin America during the twentieth century. The Economic History
Review, 58, 765–796. ①
Boersma, P., & Weenink, D.
(2015.) Praat: Doing phonetics by computer [Computer program]. [Link] ①
Caravedo, R.
(1983.) Estudios sobre el español de Lima I: Variación contextual de la sibilante. Lima: Pontificia
Universidad Católica del Perú. ①
Caravedo, R.
(1987.) Constricciones contextuales del español hablado en Lima: El caso de /s/. In M. T. Vaquero de
Ramírez & H. López Morales (Eds.), Actas del I Congreso Internacional sobre el español de América
(665–674). San Juan: Academia Puertorriqueña de la Lengua Española. ①
Caravedo, R.
(1989.) El español de Lima: Materiales para el estudio del habla culta. Lima: Pontificia Universidad
Católica del Perú. ①
Caravedo, R.
(1990.) Sociolingüística del español de Lima. Lima: Pontificia Universidad Católica del Perú. ① ② ③
⑤⑥⑦⑧⑨
Caravedo, R.
(2014.) Percepción y variación lingüística: enfoque sociocognitivo. Madrid & Franfurt: Iberoamericana &
Vervuert. ①
Cepeda, G.
(1990a.) La alofonía de /s/ en Valdivia (Chile). Estudios Filológicos, 25, 5–16. ① ② ③
Cepeda, G.
(1990b.) La variación de /s/ en Valdivia (Chile): Sexo y edad. Hispania, 73, 232–237. ①②
Cepeda, G.
(1991.) Las consonantes de Valdivia. Valdivia, Chile: América ltda. ① ② ③ ④ ⑤
Cepeda, G.
(1995.) Retention and deletion of word-final /s/ in Valdivian Spanish (Chile). Hispanic Linguistics, 6/7,
329–353. ① ②
Cerda-Oñate, K., Fuentes, D., Soto, J., & Hamdam, N.
(2015.) Variación diatópica, discursiva y fonética de /s/ en el habla de profesionales chilenos. Onomázein
32, 254–274. ①②③④⑤
Cid-Hazard, S. M.
(2003.) Proceso de debilitamiento consonántico en el español de Santiago de Chile: Variables fonológicas y
estilísticas (Unpublished doctoral dissertation). University of Southern California, Los Angeles. ① ②
③
Chappell, W.
(2016.) On the social perception of /s/ voicing in Costa Rican Spanish. Language Variation and Change
 28(3), 357–378. ①
Christensen, R. H. B.
(2015.) Regression models for ordinal data. R package version 2015. 6–28. [Link]
/package=ordinal/ ①
Contreras, M.
(2007.) Un momento en la historia del español de Chile. Estudios filológicos, 42, 59–77. < >
Erker, Daniel G.
(2010.) A subsegmental approach to coda /s/ weakening in Dominican Spanish. International Journal of
the Sociology of Language, 203, 9–26. ①
Figueroa, M., Salamanca, G., & Ñanculeo, M.
(2013.) El eje oclusión–fricción en el sistema sociofónico del castellano chileno. Estudios de Fonetica
Experimental, XXII, 233–273. ① ②
Hundley, J.
(1983.) Linguistic variation in Peruvian Spanish: Unstressed vowel and /s/ (Unpublished doctoral
dissertation). University of Minnesota, Minneapolis. ① ② ③
Klee, C. A., & Caravedo, R.
(2006.) Andean Spanish and the Spanish of Lima: Linguistic variation and change in a contact situation
C. Mar-Molinero & M. Stewart (Eds.), Globalisation and the Spanish-speaking world (pp. 94–113).
Houndmills: Palgrave Macmillan. ①
Klee, C., Rogers, B. M. A., Caravedo, R., & Dietz, L.
(2018.) Measuring /s/ variation among younger generations in a migrant settlement in Lima, Peru. Studies
in Hispanic and Lusophone Linguistics 11(1), 1–29. < >①②③④⑤⑥⑦
Lipski, J. M.
(1994.) Latin American Spanish. New York, NY: Longman. ① ② ③
Lipski, J. M.
(2011.) Socio-phonological variation in Latin American Spanish. In M. Díaz-Campos (Ed.), The handbook
of Spanish sociolinguistics (pp. 72–97). Malden, MA: Wiley-Blackwell. ①
Mack, S.
(2011.) A sociophonetic analysis of /s/ variation in Puerto Rican Spanish. In L. A. Ortiz-López (Ed.),
Selected proceedings of the 13th Hispanic Linguistics Symposium (pp. 81–93). Somerville, MA:
Cascadilla Proceedings Project. Retrieved from [Link], document #2477> ①
Mick, C., & Palacios, A.
(2013.) Mantenimiento o sustitución de rasgos lingüísticos indexados socialmente: Migrantes de zonas
andinas en Lima. Lexis 37, 341–380. ①
Penny, R.
(2000.) Variation and change in Spanish. Cambridge: Cambridge University Press. ①②③
Pérez, H. E.
(2007.) Estudio de la variación estilística del fonema /s/ en posición implosiva en el habla de los noticieros
de la televisión chilena. Revista de Lingüística Teórica y Aplicada, 45(1), 101–115. ① ② ③ ④
Rogers, B. M. A.
(2016.) The influence of linguistic and social variables in the spirantization of intervocalic /b,d,g/ in
 Concepción, Chile. Studies in Hispanic and Lusophone Linguistics, 9(1), 207–237. ①②③
Rogers, B. M. A., & Mirisis, C.
(2016.) Voiceless stop lenition and reduction as social phenomena in Concepción, Chile. Paper presented at
the 8th Workshop on Spanish Sociolinguistics, Universidad de Puerto Rico, 15 April. ① ②
Sadowsky, S.
(2012.) Naturaleza fonética y estratificación sociolingüística de los alófonos vocálicos del castellano del
Concepción (Chile) (Unpublished doctoral dissertation). La Universidad de Concepción, Concepción,
Chile. ① ② ③
Sadowsky, S.
(2015.) Variación sociofonética de las consonantes del castellano chileno. Sociolinguistic Studies, 9(1), 71–
92. < >①②③④⑤
Sadowsky, S., & Salamanca, G.
(2011.) El inventario fonético del español de Chile: Principios orientadores, inventario provisorio de
consonantes y sistema de representación (AFI-CL). Onomázein, 24, 61–84. ① ②
Sadowsky, S., & Aninao, M. J.
(2019.) Spanish in Santiago Chile: Globalization, internal migration and ethnicity. In A. Lynch (Ed.),
Spanish in the global city (pp. 277–329). New York, NY: Routledge. ① ② ③
Salcedo Arnaiz, D. C.
(2013.) Defining Andeanness away from the Andes: Language attitudes and linguistic ideologies in Lima,
Peru (Unpublished doctoral dissertation). The Ohio State University, Columbus. ①
Schmidt, L. B., & Willis, E.
(2011.) Systematic investigation of voicing assimilation of Spanish /s/ in Mexico City. In S. Alvord (Ed.),
Selected proceedings of the 5th Conference on Laboratory Approaches to Romance Phonology (pp. 1–
20). Somerville, MA: Cascadilla Proceedings Project. Retrieved from [Link], document
#2631> ①
Soto-Barba, J.
(2011.) Variación consonántica en el habla urbana y rural de la provincia de Ñuble. Revista de Lingüística
Teórica y Aplicada 49(2), 111–127. ①②③④⑤⑥
Tassara, G., & Duque, C.
(1990.) Variación lingüistica: /s/ implosiva en el habla semiformal culta de Valparaíso. Signos, 24(29),
131–139. ① ② ③
Trudgill, P.
(1986.) Dialects in contact. Oxford: Basil Blackwell. ①
Valdivieso, H., & Magaña, J.
(1991.) Variación fonética de /s/ en el habla espontánea. Revista de Lingüística Teórica y Aplicada, 29, 97–
114. ① ② ③ ④ ⑤
Valencia, A.
(1993.) Realizaciones e /s/, /ĉ/ y /ĵ/ en el habla adolescente. Nueva Revista del Pacífico, 38–39, 159–180.
①②③④⑤

Appendix 1. Summary of the speakers and social factors

used in the analyses of the Chilean data

Speakers Gender Stratus Age group Education grouping

A Female Mid-mid 25–44 5

B Female Mid-mid 25–44 5
C Female Mid-Mid 45–49 5
D Male High-low 25–44 3
E Male High-low 25–44 4
F Female Low-mid 25–44 5
G Male Low 25–44 1
H Female Mid-mid 25–44 5
I Male Mid-mid 18–24 3
J Male Low-mid 18–24 3
K Male Low-mid 18–24 5
L Female High-mid 25–44 5
M Male High-low 25–44 4
N Male Mid-mid 25–44 5
O Female High-mid 25–44 6
P Female Low-mid 25–44 4
Q Male High-low 18–24 3
R Male Mid-mid 18–24 5
S Female Low-mid 25–44 2
T Female Low-mid 18–24 3
U Male High-low 25–44 5
V Female High-low 25–44 3

Appendix 2. Summary of the speakers and social factors

used in the analyses of the Lima data
Speakers Migrant Heritage Gender Education Occupation Social
generation networks

A 2nd Mixed Male Finished university Skilled 1

B 2nd Mixed Female Some higher ed Student 3
C 2nd Non- Male Completed tech Unskilled 1
Andean
D 3rd Mixed Female Some higher ed Student 2
E 2nd Mixed Male Finished university Skilled 2
F 3rd Mixed Male Completed tech Skilled 1
G 2nd Mixed Female Postgrad studies Skilled 3
H 2nd Andean Male Some higher ed- Student 1
technical
I 2nd Non- Female Finished university Student 3
Andean
J 2nd Andean Female Some higher ed Student 3
K 2nd Mixed Male Finished university Student 3
L 2nd Andean Female Some higher ed Student 3
M 3rd Andean Male Some higher ed Student 1
N 3rd Non- Female Some higher ed Student 1
Andean
O 2nd Mixed Female Some higher ed Student 3
P 3rd Andean Male Postgrad studies Skilled 1
Q 2nd Mixed Female Postgrad studies Skilled 2
R 2nd Andean Female Some higher ed Student 1
S 3rd Mixed Male Some higher ed Student 3
T 2nd Mixed Female Finished university Skilled 3
U 3rd Non- Female Finished university Skilled 2
Andean
V 2nd Andean Female Some higher ed- Skilled 2
technical
W 3rd Mixed Female Some higher ed- Skilled 2
technical
X 4th Non- Female Postgrad studies Skilled 3
Andean
Y 3rd Andean Female Finished university Skilled 3

Z 3rd Andean Female Completed tech Skilled 1

CHAPTER 5

The acento pujado in Yucatan Spanish

Prosodic rhythm and the search for the
yucateco accent
Jim Michnowicz & Alex Hyler
North Carolina State University

Abstract
Previous research has indicated that the accent of Yucatan Spanish
(YS) differs from that of other dialects, in particular standard Mexican
Spanish from central Mexico. These studies attribute the YS accent,
described as halting/staccato (pujado), to contact with Yucatec Maya,
suggesting that yucatecos speak Spanish with a Mayan accent. This
claim has never been addressed explicitly in the literature, however.
The present study applies several metrics of prosodic rhythm (%V,
Cdev & PVI) to spontaneous speech samples from bilingual and
monolingual yucatecos. Results show a changing rhythm, with
younger speakers moving away from more traditional, possibly Maya-
influenced patterns in apparent time. Connections between rhythm and
segmental features of YS are discussed, along with possible social
implications.

Keywords
Yucatan Spanish; prosodic rhythm; rhytmic timing;
language contact; language change
1. Introduction
Due to a confluence of socio-historical factors, the Spanish of Yucatan, Mexico
shows marked differences from other surrounding varieties of Spanish, including
standard Mexican Spanish as spoken in and around Mexico City. These factors
include relative isolation during the colonial period and beyond (Mosely, 1980)
and the presence of an indigenous contact language, Yucatec Maya (hereafter
Maya), (see Michnowicz, 2015 for an overview). The presence of a long-term,
stable contact situation has led to Yucatan being one of three regions in Latin
America most likely to show both direct and indirect influence from an
indigenous language (along with the Andes and Paraguay – Klee, 2009). Many
of the linguistic differences examined in previous research center on the
phonetics/phonology of Yucatan Spanish (YS), which have been shown to differ
across a range of segmental features. These features include: the labialization of
absolute final nasals – [ˈpam] for pan ‘bread’ (Lope Blanch, 1987; Michnowicz,
2006, 2007, 2008; Yager, 1982, 1989); occlusive [bdg] where standard Spanish
would require an approximant [βðɣ] – [ˈ[Link]] for cada ‘each’ (Alvar, 1969;
Barrera Vásquez, 1937; García Fajardo, 1984; Lope Blanch, 1987; Michnowicz,
2009, 2011, 2012; Yager, 1982); aspirated /ptk/ – [ˈ[Link]] for kasa ‘house’
(Alvar, 1969; Barrera Vásquez, 1937; Coupal & Plante, 1977; García Fajardo,
1984; Lope Blanch, 1987; Michnowicz, 2012; Michnowicz & Carpenter, 2013;
Nykl, 1938; Suárez, 1979; Yager, 1982); and glottal stop insertion, especially
before vowel initial words – [ˈkwaɾ.to.ˈʔa.ɲo] for cuarto año ‘fourth year’
(Barrera Vásquez, 1937; García Fajardo, 1984; Lope Blanch, 1987; Michnowicz,
2012; Michnowicz & Kagan, 2016; Nykl, 1938; Suárez, 1979; Yager, 1982).
In addition to these segmental features, however, one of the most
commented traits of YS (both by previous researchers and by speakers
themselves) is the distinctive accent, which yucatecos describe as pujado –
strained or halting. Early studies on YS mention accent as an important feature,
but offer little in the way of analysis, instead claiming that the acento pujado
stems from direct influence from Maya (Barrera Vásquez, 1945; Mediz Bolio,
1951; Suárez, 1979). More recently, scholars have examined YS accent through
the lens of intonation (Michnowicz & Barnes, 2013; Butragueño, Mendoza, &
Orozco, 2015, 2016), finding an increased rate of early peak alignment that
contributes to the overall perceived accent in YS. Importantly, Michnowicz and
Barnes (2013), while noting the role of F0 peak alignment, also suggest that
other prosodic features, such as vowel lengthening (see Lope Blanch, 1987) and
the increased consonantismo of YS (such as the realizations of /ptk/, /bdg/, /ʔ/,
etc. outlined above, see Michnowicz, 2015) may have an important impact on
the YS accent. Differences in vowel length as well as more consonantismo can
lead to a marked difference in overall accent, which can be interpreted as
differences in rhythm (Hualde, 2005, p. 273).
The present study seeks to examine the possibility that differences in
vocalic and consonantal intervals in YS contribute to the acento pujado through
a quantitative analysis of prosodic rhythm. Results demonstrate that older
speakers and Maya-Spanish bilinguals often pattern together, in having higher
values for consonantal rhythm and lower values for measures of vocalic rhythm,
while some younger speakers have an innovative rhythm that distinguishes them
both from more traditional, possibly Maya-influenced YS as well as standard
varieties of Spanish. The rest of the paper is as follows: Section 2 provides
background on the YS accent, as well as reviews previous studies on prosodic
rhythm in Spanish; Section 3 provides the research questions and methodology;
Section 4 presents the results of the quantitative analysis of two paired metrics of
acoustic rhythm that measure both consonantal and vocalic intervals (%V/ΔC
and PVI – see Arvaniti, 2009), while Sections 5 and 6 provide discussion and
final conclusions.
2. Accent and rhythm in (Yucatan) Spanish
The earliest studies on YS single out the accent as one of the most distinctive
features of the dialect. The YS accent is consistently described as a result of
direct influence of Maya, as exemplified by the following quotes:
[L]o que primero llama la atención al extraño que por primera vez oye el español
yucateco, es su acento. Es allí donde está presente la influencia maya en su
forma más crónica… Barrera Vásquez (1945, p. 341)
[E]l acento yucateco, tan profundamente señalado, no es sino una consecuencia
del acento maya [Link] Bolio (1951, p. 19)
[P]or más que un yucateco no sepa pronunciar una sola palabra del idioma
nativo [Maya], la entonación de su castellano, por castizo que sea, es de una
inflexión y un estilo absolutamente mayas. Mediz Bolio (1951, p. 19)
…la entonación fraseal lenta y pausada, fenómeno[s] que no son sino reflejos de
la fonética nativa. Suárez (1979, p. 77)
In these quotes from foundational studies on YS, we see that researchers clearly
note the differences between YS and other varieties and connect the regional
accent with direct influence from Maya. Importantly, speakers themselves
frequently make reference to the regional accent as a distinctive trait. This
contrasts with many (but not all) of the segmental phonological variables that
define YS, which seem to lie below the level of consciousness for many speakers
(see Michnowicz, 2015). The following quote from an older female participant is
echoed frequently by speakers of YS:
[T]enemos la fama los yucatecos que al hablar tenemos así un pujadito…que
nadie tiene en otra parte de la República…[y] nos hacen mucha burla.(Female,
72, Spanish monolingual)
Here the speaker refers to how frequently the YS accent is imitated and made
fun of, both on TV and elsewhere, throughout Mexico. Much like a Southern
accent in US English, a YS accent is immediately recognizable throughout the
country. The participant also directly mentions the “pujadito” that characterizes
the accent, and hablamos pujado was a common answer by informants to the
question “What is YS like?”.1
So while it is clear that the accent of YS is somehow “different”, what is
less obvious is what aspects of YS phonology give this impression. In spite of
the clear differences noted above, prosodic features of YS have not received as
much attention in the literature as other segmental features, or as prosody in
other dialects of Spanish (Pfeiler, 1995; Michnowicz & Barnes, 2013;
Butragueño, Mendoza, & Orozco, 2015, 2016, and Uth, 2016 are the few
exceptions). Accent as a concept within linguistics can be conceived of as the
combination of segmental and suprasegmental features, such as intonation,
accentuation and rhythm, that distinguish varieties of a language (what Cristia,
Seidl, Vaughn, Schmale, Bradlow, & Floccia (2012, p. 479) call “within-
language accents”). Descriptions of YS suggest a role for at least two of these
features, as the halting or staccato nature of the accent could be attributable to
intonation, rhythm, or a combination of the two.

2.1 YS Intonation

Michnowicz and Barnes (2013) represents an initial attempt to quantify the YS

accent through a phonetic analysis of prenuclear F0 peak alignment in
spontaneous speech. Most varieties of Spanish align the F0 peak in non-
contrastive declaratives with the post-tonic syllable in pre-nuclear position.
Alignment with the tonic syllable, on the other hand, is generally reserved for
nuclear position, or when it occurs in prenuclear syllables, it signals emphasis or
contrast (Face, 2003). Peak alignment with respect to the tonic syllable has been
found to vary across dialects, with many contact varieties displaying an
increased rate of F0 peaks aligned within the tonic syllable (early peaks),
perhaps due to direct influence of the contact language, or as a general bilingual
simplification strategy (Colantoni, 2011). Since early peak alignment in
declaratives can give the utterance a staccato or insistent sound (Face, 2003),
corresponding to the impression given by the acento pujado, Michnowicz and
Barnes (2013) hypothesized that at least some of the YS accent may be due to
differences in peak alignment. Their results indicate a higher rate of early peaks
in spontaneous YS than in many other varieties (53% in YS; compared to 25%
early peaks in spontaneous Castilian Spanish – Face, 2003), which lends initial
support to the idea that F0 peak alignment plays a role in distinguishing the YS
accent. Butragueño, Mendoza, and Orozco (2015, 2016) examined YS peak
alignment from a phonological point of view, classifying pre-nuclear peaks using
the Spanish Tones and Break Indices (Sp-ToBI) framework, and included both
production and perception data. Their results confirm the increased rate of
prenuclear peaks in both read and spontaneous YS (with between 17% and 55%
more early peaks in spontaneous data according to different social groups
(Butragueño et al., 2016)). Results also indicated that early peak alignment was
correlated with “se oye yucateco” (Butragueño et al., 2015), reinforcing the
connection between intonational peak alignment and a perceived YS accent.
While the results of these studies indicate an important role for peak
alignment in the production of an “acento pujado”, Michnowicz and Barnes
(2013) suggest that other features reported in the literature on YS, such as tonic
vowel lengthening (Lope Blanch, 1987) and increased consonant complexity
(consonantismo – Michnowicz, 2015) may also play a role in distinguishing YS
from other varieties. Differences in vowel length and consonant complexity are
key components of what is perceived as prosodic rhythm (Hualde, 2005), which
is the focus of the present study.
2.2 Prosodic rhythm

Prosodic timing is a rhythmic classification of language that traditionally has

been defined in terms of a binary distinction between stress-timed and syllable-
timed languages (Pike, 1945; Abercrombie, 1967). In prototypically stress-timed
languages, the same distance is maintained between tonic syllables; to achieve
this, tonic vowels are lengthened, while atonic vowels undergo some sort of
reduction. Consonants are also likely to be more complex and variable.
Germanic languages, such as English, are considered to be prototypical stress-
timed languages. On the other end of the spectrum are syllable-timed languages,
in which each syllable lasts approximately the same duration, irrespective of
stress. Likewise, consonant clusters are simpler, and consonants show less
variation in their duration. Many Romance languages, such as Spanish, are
classified as having syllable-timing (see Thomas & Carter, 2006; Grabe & Low,
2002, among many others). Importantly, these differences are not absolute. More
recent research indicates that rhythmic differences are gradient and exist on a
continuum, and therefore may be quantified using a variety of quantitative
methodologies (Ramus, Nespor, & Mehler, 1999; White & Mattys, 2007; Low &
Grabe, 1995).
In recent decades, a number of rhythm metrics have been developed to
study differences in prosodic timing across and within languages. The first set of
metrics are interval measures, based on the duration of consonantal and vocalic
intervals. Ramus et al. (1999) examined several prototypical languages (Spanish,
Italian and French as syllable-timed; English and Dutch as stress-timed), along
with moraic Japanese and additional languages described as having mixed
rhythmic systems (Polish and Catalan). Metrics employed by Ramus et al.
(1999) include %V, the percentage of speech that consists of vocalic segments,
along with ΔC, the standard deviation of the consonantal intervals. Using these
two metrics, languages that are traditionally considered syllable-timed are
characterized by high %V and low ΔC values.
Low and Grabe (1995) take a different approach in developing their metric,
the Pairwise Variability Index (PVI). PVI consists of comparing the duration of
adjacent pairs of vowels or consonants, which produces an index score. Because
vocalic durations are more susceptible to the influence of speech rate (Grabe &
Low, 2002), the vocalic measure is Vocalic Normalized PVI(Vnpvi). Consonant
intervals, on the other hand, are measured as raw, non-normalized values, via the
Consonantal Raw PVI (Crpvi). For Vnpvi, higher values indicate more variation
among adjacent vowel duration, producing a more prototypically “stress-timed”
rhythm. Lower values indicate a more “syllable-timed” pattern. For Crpvi,
higher values indicate the presence of more complex consonant clusters and/or
more variation among consonant interval duration, suggestive of a more stress-
timed pattern. The formula for nPVI is found in Figure 1. The formula for Crpvi
is the same, except that the difference in interval length is not divided by the
mean (e.g. raw values).

Figure 1. Formula for nPVI

Because each rhythm metric examines a slightly different (but related) aspect of
prosodic timing, previous research has found that using multiple metrics within
the same study provides a more accurate picture of the rhythmic patterns in a
given language or dialect (White & Mattys, 2007; O’Rourke, 2008). And while
the validity of rhythm metrics for classifying languages as either syllable-timed
or stress-timed has been questioned (Arvaniti, 2009, 2012), the studies outlined
above (among others) have found that these measures can successfully be
applied to identify differences between languages, as well as within dialects of
the same language. In light of Arvaniti’s (2009, 2012) arguments, in this study
we will not interpret results as indicating a rhythmic classification for (different
varieties of) YS. Instead, we will treat the rhythm scores as a truly gradient
variable, speaking only of higher vs. lower, or more consonantal vs. more
vocalic values. The two sets of metrics employed (%V/ΔC; PVI) here are the
most widely used in studies of rhythmic timing (Arvaniti, 2009).
Research has shown that prosodic rhythm is susceptible to cross-linguistic
influence and convergence (Low, Grabe, & Nolan, 2000). Several studies on
Spanish in contact with other languages show converged patterns, as speakers
produce intermediate values between the two languages in contact. Carter (2005)
found that Latino English presented PVI values between those of Spanish and
Anglo English, indicating that the influence of Spanish rhythm carries over into
the bilingual speakers’ English. Shousterman (2014) finds similar results for
Puerto Rican speakers in Harlem, as their English shows Spanish-like rhythm.
Likewise, Carter & Wolford (2016) further demonstrate how rhythm can
converge across time, as their results for a community in south Texas show that
the Spanish rhythm of bilingual speakers moves towards the values of their
Latino English across apparent time. The speakers’ English PVI values are lower
than expected (between 40–50), indicating that Spanish has already had an effect
on English in the community. Among the youngest speakers, differences in PVI
values between English and Spanish disappear, resulting in a converged,
intermediate system in both languages. Gabriel and Kireva (2014) also find
evidence of prosodic convergence between Italian and Buenos Aires Spanish,
both of which differ from Castilian Spanish values (see also Benet, Gabriel,
Kireva, & Pešková, 2012). Shousterman (2014, p. 164) notes that these studies
support the idea that non-native speakers can play a critical role in the
development of contact varieties, and that this role can persist across
generations. This pattern may also be important in YS, as previous work has
suggested a similar path for segmental and morpho-syntactic features in that
dialect. Michnowicz and Carpenter (2013) and Michnowicz and Kagan (2016)
summarize how bilingual forms can filter through the community, entering into
the speech of even the monolingual majority (see also Lipski, 2004).
Unfortunately, to our knowledge there are no studies on Maya rhythm using
the methods summarized above. Vowel length is phonemic in Maya, with vowels
on average varying between 113 ms for short vowels and 216 ms for glottalized
vowels (Frazier, 2009). Likewise, Maya overall shows more variation in
consonant durations than Spanish. One clear example of this is the two distinct
types of voiceless stops. Maya possesses both aspirated and ejective /ptk/, with
average varying VOT values between 36 ms for aspirated stops and 53 ms for
ejectives (Burnett-Deas, 2009). The distinction between ejectives and plain stops
also exists for alveolar and post-alveolar affricates, and Maya likewise shows
variation between /ʔ/ and creaky voicing, etc. (Frazier, 2009). The variable
nature of both vowels and consonants in Maya would be predicted to translate
into lower %V and higher Vnpvi, ΔC and Crpvi. As outlined above, Maya and
Spanish are predicted to differ with respect to rhythm values, and previous
research suggests that the outcome of cross-linguistic influence/convergence for
prosodic rhythm is a possibility in YS.
Based on the studies above, the research questions for the present
investigation are as follows:
 1. How can the prosodic rhythm of YS be characterized according to the two
sets of rhythm metrics employed in this study?
2. Are there differences in prosodic rhythm based on speaker age, sex and
language background (Spanish monolingual vs. Maya-Spanish bilingual)?
3. Can the acento pujado be due, in part, to differences in rhythmic timing? If
so, what role do consonantal and vocalic measures play in these
differences?

3. Methodology
Data for this study come from sociolinguistic interviews with 20 native speakers
of YS, conducted in and around the capital city of Merida in 2004–2005.
Participants are balanced for language background (10 Spanish monolinguals; 10
Maya-Spanish bilinguals), and mostly balanced for sex and age group. Because
previous studies have found that being age 30 and older at the time of the
interview (i.e. born before 1974 or 1975) is the primary, significant distinction
for age groups in Merida, with speakers older than 30 producing more regional
forms (Michnowicz, 2015), participants in the present study were also divided
into age groups, defined as ‘younger’ (ages 18–29; mean age 22 years) and
‘older’ (ages 30+; mean age 62 years). Participant demographics are found in
Table 1.

Table 1. Participant demographic data

Speaker code Language Sex Age group Birth year

202 Span Only F Younger 1985

203 Span Only F Younger 1985
205 Span Only M Younger 1984
230 Span Only M Younger 1980
225 Span Only F Older 1933
229 Span Only F Older 1916
233 Span Only F Older 1968
235 Span Only F Older 1936
224 Span Only M Older 1929
232 Span Only M Older 1938
215 Maya-Span F Younger 1981
221 Maya-Span F Younger 1983
231 Maya-Span F Younger 1981
208 Maya-Span M Younger 1984
234 Maya-Span M Younger 1986
213 Maya-Span F Older 1940
222 Maya-Span F Older 1948
210 Maya-Span M Older 1951
211 Maya-Span M Older 1955
226 Maya-Span M Older 1961

Data were transcribed in Praat (Boersma & Weenink, 2015), and vocalic and
consonantal intervals were segmented by hand. Boundaries were placed at the
zero-crossing, and were all checked for consistency by the first author.
Approximately 200 vocalic intervals were measured per speaker. A sample
coding is seen in Figure 2.

Figure 2. Sample CV coding

Once coded, data were extracted from Praat using a script, and rhythm metrics
were computed using Correlatore (Mairano & Romano, 2010), a software
package designed for the analysis of prosodic rhythm. In order to account for
final lengthening effects, all prepausal intervals were excluded from analysis. A
series of linear regression analyses were carried out with Rbrul, including main
effects and interactions (Johnson, 2009). The continuous dependent variables
were the rhythm metrics (%V, ΔC, Vnpvi, Crpvi). The independent variables
were speaker sex, age group, and language background (Maya-Spanish bilingual
vs. Spanish monolingual).

4. Results
In this section, we present the results of the linear regressions for each dependent
variable (%V, ΔC, Vnpvi, Crpvi), along with plots of individual patterns by
group and crosstabulations/interactions for all independent variables (speaker
sex, age and language group).

4.1 %V and ΔC

Figure 3. Values for %V (x-axis) and ΔC (y-axis) by speaker

The linear regression found a significant effect of age (p. = 0.037) and sex
(p. = 0.0184) on %V, with younger speakers and women producing significantly
higher %V values than older speakers and men. Of the five speakers with %V
values below 45, all are older speakers (4 monolinguals and 1 bilingual).
Likewise, of the nine speakers with %V values above 47, seven are women, and
six are younger. Language group was not a significant predictor of %V
(p. = 0.752), and there were no significant interactions.
For ΔC, the overall trend was reversed, with older speakers and men
producing higher ΔC values. In the regression, there were no significant main
effects (age p. = 0.357; language group p. = 0.325; sex p. = 0.546) for ΔC, but
there were significant interactions of sex:age (p. = 0.00167) and sex:language
(p. = 0.00758).
Crosstabulation/interaction plots reveal further trends in the data.
Significant interactions in the regression analyses are indicated for each plot.

Figure 4. %V by sex and age. p. = 0.207

Figure 5. ΔC by sex and age. p. = 0.00167
The interaction of sex:age was not significant for %V; younger speakers of both
sexes had higher %V values. There was a significant interaction of sex:age for
ΔC, as seen in Figure 5. Younger women had overall lower ΔC values, while the
opposite is true for younger men.

Figure 6. %V by sex and language. p. = 0.721

Figure 7. ΔC by sex and language. p. = 0.00758
Figure 6 shows very few overall differences across sex:language groups for %V.
For the significant interaction for ΔC in Figure 7, on the other hand, monolingual
women have the lowest ΔC, followed by bilingual men, monolingual men and
finally bilingual women.

Figure 8. %V by language and age. p. = 0.155

Figure 9. ΔC by language and age. p. = 0.531
Although not a significant interaction for either variable, the crosstab values for
%V and ΔC by language group and age show interesting trends (Figures 8 and
9). While the differences are small, results show that younger speakers of both
language groups used a more vocalic variety, as indicated by higher %V. For ΔC,
bilinguals produced nearly identical values regardless of age (older = 54.94;
younger = 54.985). Bilinguals’ ΔC values overall were also virtually identical to
those of older monolinguals (54.054), suggestive of convergence between Maya-
Spanish bilinguals and older Spanish monolinguals. Young Spanish
monolinguals, on the other hand, showed less variation in their consonantal
production, as indicated by lower ΔC values (49.793), suggesting that this group
is moving away from the more variable consonantal rhythm produced by older
monolinguals and bilinguals of both age groups.

4.2 PVI

Figure 10 shows a comparison of Vnpvi on the X-axis, and Crpvi on the Y-axis.

Figure 10. Mean Vnpvi and Crpvi by age and language groups. Speaker codes
include sex (M/F)
The linear regression found a significant effect of age for Vnpvi, with younger
speakers producing significantly higher values (p. = 0.0155), although some of
this trend is due to the behavior of two young monolinguals, 205M and 221F,
with values above 55. Still, of the 14 speakers with Vnpvi averages above 50,
eight are younger. Language group (p. = 0.431) and sex (p. = 0.823) were not
significant predictors of Vnpvi. There was a significant interaction of
age:language (p. = 0.00582).
No significant main effects were found in the analysis of Crpvi (age
p. = 0.19; language group p. = 0.357; sex p. = 0.444), although for age, some
trends are evident. Of the 11 speakers with Crpvi values above 55, 7 are older,
and 6 are bilinguals. There was a significant interaction of sex:language
(p. = 0.0339). Crosstabulation/interaction plots for Vnpvi/Crpvi are found in
Figures 13–18.

Figure 11. Vnpvi by sex and age. p. = 0.92

Figure 12. Crpvi by sex and age p. = 0.103
Figure 11 shows that younger speakers overall produced higher Vnpvi values,
regardless of sex. For Crpvi in Figure 12, younger women produced lower Crpvi
values than the other groups. Older women showed the largest values, while
there were few differences among men regardless of age.
Figure 13. Vnpvi by sex and language. p. = 0.682
Figure 14. Crpvi by sex and language. p. = 0.0339
In Figure 13, we see that Spanish monolinguals had slightly lower Vnpvi values
across both sexes, although the difference is not significant. Figure 14, however,
shows the significant interaction of sex and language for Crpvi; the lowest Crpvi
values were produced by monolingual women (51.933). Bilingual women
(58.004) and monolingual men (57.631) had similar Crpvi values, with bilingual
men producing intermediate values (55.351).

Figure 15. Vnpvi by language and age p. = 0.00582

Figure 16. Crpvi by language and age p. = 0.775
The overall trends by age and language group can be seen in Figures 15 and 16.
For the significant interaction of age:language for Vnpvi, bilinguals produced
almost identical values, regardless of age group (older 52.179; younger 51.904).
Monolinguals, on the other hand, show greater differences across age groups,
with younger monolinguals producing higher Vnpvi values (older 47.64;
younger 54.932). For Crpvi, on the other hand, younger speakers of both
language groups produced lower mean values than older speakers. Regarding
language, younger bilinguals and older monolinguals produced very similar
Crpvi values (55.499 and 55.835 respectively), while the lowest Crpvi values
were found among younger monolinguals (51.809). As seen with ΔC, this
suggests that younger monolinguals are distancing themselves from the
consonantal rhythm of older speakers and Maya-Spanish bilinguals.
In sum, significant effects of age were found for both vocalic measures
(%V and Vnpvi). Younger speakers overall are employing a less consonantal,
more vocalic rhythm (%V), while at the same time showing more variability in
the pairwise (Vnpvi) durations of vocalic intervals. The analysis of
crosstabulations and significant interactions indicates that it is younger,
monolingual women in particular that are leading the change away from the
traditional consonantismo-influenced rhythm of ‘traditional’ YS, as seen in the
results for %V, ΔC and Crpvi.

5. Discussion
The results of this study demonstrate that rhythmic timing in YS is complex and
multifaceted and is defined by a large degree of individual variation as well as
cross-group patterns. Specifically, a significant effect of age for both vocalic
metrics, along with patterns for consonantal measures seen in crosstabulations
and significant interactions, suggest that the rapid changes occurring in YS at the
segmental level are also underway for rhythmic timing. Previous studies have
shown that younger yucatecos are quickly abandoning regional forms and
moving in the direction of standard Central Mexican Spanish for segmental
features such as aspirated /ptk/, occlusive /bdg, and /ʔ/ insertion (see
Michnowicz, 2015 for summaries), and the present results suggest that younger
speakers are also altering their rhythmic timing in the same direction. Assuming
that older speakers reflect the earlier stage of YS that may have shown more
direct Maya influence, as has been argued for segmental variables, ‘traditional’
YS can be characterized as having a lower %V and Vnpvi, along with higher ΔC
and Crpvi. In other words, ‘traditional’ YS is less vocalic overall, but with more
consistency across vocalic intervals, while at the same time showing a greater
variation in the length of consonantal intervals. Younger YS speakers, however,
show the opposite trends – higher values for vocalic measures, and lower values
for consonantal metrics. As Spanish has been described as having a high %V and
low ΔC and Crpvi, the differences seen across age groups can be interpreted as a
process of standardization among younger speakers.
Spanish has also been described as having low Vnpvi values, as a
prototypical syllable-timed language that shows consistent durations across
adjacent vocalic intervals. Therefore, a move toward ‘standard’ Spanish norms
might be expected to entail lower Vnpvi values. Younger speakers, however,
produced significantly higher Vnpvi values, suggesting that they are innovating
at the same time they are standardizing their speech in terms of consonantal
metrics and %V. The prediction that higher Vnpvi values might be indicative of
Maya-influenced traditional speech was not borne out in the data, at least when
comparing Vnpvi values across age and language groups. As seen in Figures 12
and 13, the trend toward higher Vnpvi in the current data is led primarily by
monolingual women. These young women, while moving away from
‘traditional’, possibly Maya-influenced consonantal patterns, are continuing to
distinguish themselves in their vocalic productions. One hypothesis is that this
new pattern is a way of marking a Yucatan identity without the stereotypical YS
rhythm typical of older speakers and bilinguals. Previous work has documented
that women in Yucatan often maintain distinctive forms more than men, as long
as those forms are not overtly stigmatized. For example, women use more final -
m (Yager, 1989; Pfeiler, 1992; Michnowicz, 2007, 2008), but lower rates of /ʔ/
(Michnowicz & Kagan, 2016). Importantly, in studies on intonation in YS,
women also produced higher rates of early peak alignment than men (L+H* in
Butragueño, Mendoza, & Orozco, 2016), further contributing to accent
differences for women. By producing a vocalic rhythm that possibly moves
away from other standard varieties (Vnpvi) but avoids the stigmatized speech of
the (Maya-influenced) past (consonantal metrics), these speakers are able to
maintain their identities as yucatecas in a way that balances the pressures of
regional vs. standard speech. Future real-time studies are needed to further
examine this possibility.
These results are also in agreement with the predictions made by
Michnowicz and Barnes (2013), namely that the segmental features of YS
reported in previous studies, such as stop /bdg/, aspirated /ptk/ and /ʔ/ insertion,
as well as differences in vowel length, would contribute to perceived differences
in YS accent. This increased consonantismo in ‘traditional’ YS (spoken by older
speakers and bilinguals) is evident in the higher values the consonantal intervals
(ΔC, Crpvi) and lower %V, a possible contact feature of the dialect. As
mentioned in Section 2.2, Maya on the whole displays a more complex
consonantal inventory and distribution than Spanish, which would be
hypothesized to result in higher ΔC and Crpvi values, while at the same time
leading to lower %V. As such, the rhythmic patterns of older speakers and
bilinguals could be interpreted as possibly reflecting the patterns of Maya, as
suggested by early studies on YS (Barrera Vásquez, 1945; Mediz Bolio, 1951;
Suárez, 1979). Given that these same speakers – older monolinguals and Maya-
Spanish bilinguals – are frequently associated with the most ‘traditional’ YS, it
seems likely that rhythmic differences contribute to the impression of the acento
pujado in YS. For example, Butragueño et al. (2016) found increased early peak
alignment among older participants and bilinguals – the same groups that were
determined to sound more yucateco in their 2015 study. Since the same groups
that show increased early peak alignment and were perceived as having a more
regional accent in Butragueño et al. (2015, 2016) also produced lower %V and
higher ΔC and Crpvi values in the present study, it is reasonable to propose that
different patterns in both intonation and rhythm contribute to the perceived
acento pujado. Future perception studies are required to confirm or refute this
hypothesis.
At the same time, the fact that significant main effects were not found
between language groups for any of the rhythm metrics provides further
evidence that what may have begun as Maya accented rhythm has been passed to
some of the monolingual population as well, in particular older speakers. The
same results were found for some segmental features, such as aspirated /ptk/,
where older Spanish monolinguals behaved more like Maya speakers than did
younger monolinguals (Michnowicz & Carpenter, 2013).
As in other studies of YS, while group behavior plays an important role in
determining the outcome of language variation, there is a great deal of individual
variation in the data, with some speakers showing patterns that do not
correspond to the age, sex, or language groups to which they belong. Intense
contact between Spanish and Maya has only taken place over the last 100 or so
years, and so it is very common to find a middle-class, monolingual Spanish-
speakers whose grandparents were dominant in Maya. Merida is a city that is
quickly changing and becoming more cosmopolitan, as increased immigration
from other areas of Mexico leaves its mark on the culture and language of the
city (Michnowicz, 2015). Butragueño, Mendoza, & Orozco (2015) note the
importance of social networks in determining the prosody used by particular
speakers. More traditional, regional patterns dominate in dense, locally-oriented
social networks, while more pan-Hispanic or Mexican patterns found among
members of more open social networks. The findings of the present study, which
indicate an overall pattern of rhythmic change across time, but with certain
individuals diverging from the behavior of their peers, suggest that the important
within-group variation might best be examined through the lens of social
network analysis, as we suspect that what appear to be individual differences in
rhythm are most likely correlated with different social networks within Merida.
Lope Blanch (1987) refers to the variable nature of YS as el polimorfismo
yucateco, and increased contact between Maya bilinguals and Spanish
monolinguals, as well as between native yucatecos and immigrants from other
regions results in highly variable patterns, as speakers of all backgrounds slowly
converge on a new koiné variety (Kerswill, 2002). This variation among younger
speakers is clearly visible in the present data.

6. Conclusions
The present study provides evidence for rhythmic differences in YS that
contribute to the acento pujado widely reported by scholars and speakers alike.
Many younger speakers (monolinguals in particular) are employing a less
consonantal, possibly Maya-influenced rhythm, while older monolinguals and
Maya-Spanish bilinguals show similar rhythmic patterns suggestive of
convergence. At the same time, some younger speakers (primarily female) are
showing innovative patterns for Vnpvi, producing a much more stress-timed
pattern than older speakers. It is hypothesized that this innovation is a way for
some young women to distinguish themselves as yucatecas while at the same
time distancing themselves from more ‘traditional’, stigmatized YS rhythm.
Finally, the individual variation evident in the data is likely due to the multiple
and complex social networks that exist in Merida, as noted by Butragueño et al.
(2015) for YS intonation. Future studies should examine prosodic patterns
among a new generation of younger speakers (the present data are from 2004–
2005), as well as undertake a detailed social network analysis of speakers, in
order to confirm or refute this possibility as a source of language variation in
YS. Finally, patterns outside of the capital city of Merida are likely to be quite
different, and a study of the Spanish prosody in Maya-dominant regions of the
peninsula would lend important insights into the origins of the acento pujado.

Note
1. The dictionary of the Royal Spanish Academy ([Link]) defines “pujar” as “Tener dificultad en explicarse,
no acabar de romper a hablar para decir algo; Vacilar y detenerse en la ejecución de algo”; or “Hacer
gestos o ademanes para prorrumpir en llanto, o quedar haciéndolos después de haber llorado”. Hablar
pujado refers to the halting, staccato nature of the YS accent. ①

References
Abercrombie, D.
(1967.) Elements of general phonetics. Edinburgh: University Edinburgh Press. ①
Alvar, M.
(1969.) Nuevas notas sobre el español de Yucatán. Iberoromania, 1, 159–89. ① ②
Arvaniti, A.
(2012.) The usefulness of metrics in the quantification of speech rhythm. Journal of Phonetics, 40, 351–73.
①②
Arvaniti, A.
(2009.) Rhythm, timing and the timing of rhythm. Phonetica, 66, 46–63. ①②③④
Barrera Vásquez, A.
(1937.) Mayismos y voces mayas en el español de Yucatán. Investigaciones Lingüísticas, 4, 9–35. ① ②
Barrera Vásquez, Alfredo
1945. El idioma español en Yucatán. In Enciclopedia Yucatanense. Mexico City: Gobierno de Yucatán.
Benet, A., Gabriel, C., Kireva, E., & Pešková, A.
(2012.) Prosodic transfer from Italian to Spanish: Rhythmic properties of L2 speech and Argentinean
Porteño. In Q. Ma, H. Ding, & D. Hirst (Eds.), Proceedings of the 6th International Conference on
Speech Prosody (pp. 438–441). Shanghai, China: Tongji University Press. ①
Boersma, P., & Weenink, D.
(2015.) Praat: Doing phonetics by computer [Computer program] Version 5.4.22. [Link]
Burnett-Deas, A.
(2009.) VOT of ejectives, implosives, and plain stops in Yukateko, Mopan, and Itzaj Maya (Unpublished
MA thesis). Southern Illinois University-Carbondale. ①
Carter, P. M.
(2005.) Prosodic variation in SLA: Rhythm in an urban North Carolina Hispanic community. University of
Pennsylvania Working Papers in Linguistics 11(4). ①
Carter, P. M., & Wolford, T.
(2016.) Cross-generational prosodic convergence in South Texas Spanish. Spanish in Context, 13, 29–52.
①
Colantoni, L.
(2011.) Broad-focus declaratives in Argentine Spanish contact and non-contact varieties. In C. Gabrie &
Lleó (Eds.), Intonational phrasing at the interfaces: Cross-linguistic and bilingual studies in Romance
and Germanic (pp. 183–212). Amsterdam: John Benjamins. ①
Coupal, L., & Plante, C.
(1977.) Las oclusivas sordas yucatecas: /ptk/: ¿Fuertes, aspiradas, glotalizadas? Langues et Linguistique
129–76. ①
Cristia, A., Seidl, A., Vaughn, C., Schmale, R., Bradlow, A. & Floccia, C.
(2012.) Linguistic processing of accented speech across the lifespan. Frontiers in Psychology, 3, 479.
 ①
Face, T. L.
(2003.) Intonation in Spanish declaratives: Differences between lab speech and spontaneous speech.
Catalan Journal of Linguistics, 2, 115–31. ①②③
Frazier, M.
(2009.) Tonal dialects and consonant-pitch interaction in Yucatec Maya. New Perspectives in Mayan
Linguistics, 59, 59–82. ① ②
Gabriel, C., & Kireva, E.
(2014.) Prosodic transfer in learner and contacts varieties. Studies in Second Language Acquisition, 36,
257–81. ①
García Fajardo, J.
(1984.) Fonética del español de Valladolid, Yucatán. Mexico City: Universidad Nacional Autónoma de
México. ① ② ③
Grabe, E., & Low, E. L.
(2002.) Durational variability in speech and the rhythm class hypothesis. In C. Gussenhoven & N. Warner
(Eds.), Papers in laboratory phonology, 7 (pp. 515–546). Berlin: Mouton de Gruyter. ①②
Hualde, J. I.
(2005.) The sounds of Spanish. Cambridge: Cambridge University Press. ① ②
Johnson, D. E.
(2009.) Getting off the GoldVarb standard: Introducing Rbrul for mixed effects variable rule analysis.
Language and Linguistics Compass, 3(1), 359–383. ①
Kerswill, P.
(2002.) Koineization and Accommodation. In J. K. Chambers, P. Trudgill, & N. Schilling-Estes (Eds.),
handbook of language variation and change (pp. 669–702). Malden, MA: Blackwell. ①
Klee, C. A.
(2009.) Migrations and globalization: Their effects on contact varieties of Latin American Spanish. In M.
LaCorte & J. Leeman (Eds.), Español en Estados Unidos y otros contextos de contacto:
Sociolingüística, ideología y pedagogía (pp. 39–66). Frankfurt & Madrid: Vervuert & Iberoamericana.
①
Lipski, J. M.
(2004.) El español de América. Cátedra: Madrid. ①
Lope Blanch, J. M.
(1987.) Estudios sobre el español de Yucatán. Mexico City: Universidad Nacional Autónoma de México.
①②③④⑤⑥⑦
Low, E. L., Grabe, E., & Nolan, F.
(2000.) Quantitative characterizations of speech rhythm: Syllable-timing in Singapore English. Language
and Speech, 43, 377–401. ①
Low, E. L., & Grabe, E.
(1995.) Prosodic patterns in Singapore English. In ICPhS 95(3), 636–639. ① ②
Mairano, P. & Romano, A.
 (2010.) Un confronto tra diverse metriche ritmiche usando Correlatore. In S. Schmid, M. Schwarzenbach
& D. Studer (Eds.), La dimensione temporale del parlato, (Proc. of the V National AISV Congress,
University of Zurich, Collegiengebaude, 4–6 February 2009) (pp. 79–100). Torriana (RN): EDK.
[Link] ①
Martín Butragueño, P., Mendoza, E. & Orozco, L.
(2015.) Aproximación a la prosodia del español de Mérida: Datos del COEM (Corpus oral del español
México, 2010–2015). Presented at AMLA, XIII Congreso Nacional de Lingüística, Tuxtla Gutiérrez,
7–10 October. ① ② ③ ④ ⑤ ⑥ ⑦
Martín Butragueño, P., Mendoza, E., & Orozco, L.
(2016.) Nueva visita al prealineamiento del español yucateco de Mérida. Presented at the 8th International
Workshop on Spanish Sociolinguistics, San Juan, PR, 13–16 April. ① ② ③ ④ ⑤ ⑥ ⑦
Mediz Bolio, A.
(1951.) Interinfluencia del maya con el español de Yucatán. Mérida, Mexico: Editorial Zamná. ① ② ③
Michnowicz, J.
(2006.) Final – m in Yucatan Spanish: A rapid and anonymous survey. In J. P. Montreuil (Ed.), New
perspectives on Romance linguistics, Vol. 2: Phonetics, phonology, and dialectology: Selected papers
from the 35th Linguistic Symposium on Romance Languages (LSRL) (pp. 155–166). Amsterdam: John
Benjamins. ①
Michnowicz, J.
(2007.) El habla de Yucatám: Final -m in a dialect in contact. In J. Holmquist (Ed.), Selected proceedings of
The Third Workshop on Spanish Sociolinguistics (WSS3) (pp. 38–43). Somerville, MA: Cascadilla
Proceedings Project. ① ②
Michnowicz, J.
(2008.) Final nasal variation in Merida, Yucatan. Spanish in Context, 5, 278–303. ①②
Michnowicz, J.
(2009.) Intervocalic voiced stops in Yucatan Spanish: A case of contact induced language change? In M.
LaCorte & J. Leeman (Eds.), Español en Estados Unidos y en otros contextos de contacto (pp. 67–84).
Frankfurt & Madrid: Vervuert & Iberoamericana. ①
Michnowicz, J.
(2011.) Dialect standardization in Merida, Yucatan: The case of (b d g). Revista Internacional de
Lingüística Iberoamericana, 18, 191–212. ①
Michnowicz, J.
(2012.) The standardization of Yucatan Spanish: Family case studies in Izamal and Mérida. In K. Geeslin
& M. Díaz-Campos (Eds.), Selected proceedings of the Hispanic Linguistics Symposium 2010 (pp.
102–115). Somerville, MA: Cascadilla Proceedings Project. ① ② ③
Michnowicz, J.
(2015.) Maya-Spanish contact in Yucatan, Mexico: Context & sociolinguistic implications. In S. Sessarego
& M. González Rivera (Eds.), New Perspectives on Hispanic contact linguistics in the Americas (pp.
21–42). Madrid & Frankfurt: Iberoamericana & Vervuert. ①②③④⑤⑥⑦
Michnowicz, J., & Barnes, H.
(2013.) Pre-nuclear peak alignment in Yucatan Spanish. In C. Howe, S. E. Blackwell, & M. Lubbers
Quesada (Eds.), Selected proceedings of the Hispanic Linguistics Symposium 2011 (pp. 221–235).
 Somerville, MA: Cascadilla Proceedings Project. ① ② ③ ④ ⑤ ⑥ ⑦
Michnowicz, J., & Carpenter, L.
(2013.) Voiceless stop aspiration in Yucatan Spanish: A sociolinguistic analysis. Spanish in Context, 10,
410–437. ①②③
Michnowicz, J., & Kagan, L.
(2016.) On glottal stops in Yucatan Spanish. In S. Sessarego & F. Tejedo-Herrero (Eds.), Spanish language
and sociolinguistic analysis, 8 (pp. 219–239). Amsterdam: John Benjamins. ①②③
Mosely, E. H.
(1980.) From conquest to independence: Yucatan under Spanish rule, 1521–1821. In E. H. Mosely & E. D.
Terry (Eds.), Yucatán: A world apart (pp. 83–121). Tuscaloosa, AL: University of Alabama Press. ①
Nykl, A. R.
(1938.) Notas sobre el español de Yucatán Veracruz y Tlaxcala. In P. Henríquez Ureña (Ed.), El español en
Méjico, Los Estados Unidos, y la América Central, (pp. 207–225). Buenos Aires: La Universidad de
Buenos Aires. ① ②
O’Rourke, E.
(2008.) Speech rhythm variation in dialects of Spanish: Applying the pairwise variability index and
variation coefficients to Peruvian Spanish. In Proc. Fourth Conf. on Speech Prosody, May 6– 9, 2008,
Campinas, Brazil. ①
Pfeiler, B.
(1992.) Así som, los de Yucatán: El proceso fonológico Vn>m/___(#, C) en dos lenguas en contacto.
Memorias del Primer Congreso Internacional de Mayistas (pp. 110–122). Mexico: San Cristóbal de
las Casas. ①
Pfeiler, B.
(1995.) Zur Prosodie des Spanischen in Yucatán, Mexiko. In K. Zimmerman (Ed.), Lenguas en contacto en
Hispanoamérica: Nuevos enfoques (pp. 119–134). Frankfurt: Vervuert. ①
Pike, K. L.
(1945.) The intonation of American English. Ann Arbor, MI: The University of Michigan Press. ①
Ramus, F., Nespor, M. & Mehler, J.
(1999.) Correlates of linguistic rhythm in the speech signal. Cognition, 73, 265–92. ①②③
Shousterman, C.
(2014.) Speaking English in Spanish Harlem: The role of rhythm. University of Pennsylvania Working
Papers in Linguistics, 20, 18. ① ②
Suárez, V. M.
(1979.) El español que se habla en Yucatán: Apuntamientos filológicos. Mérida, Mexico: La Universidad
de Yucatán. ① ② ③ ④ ⑤
Thomas, E. R., & Carter, P. M.
(2006.) Prosodic rhythm and African American English. English World-Wide, 27, 331–55.
Uth, M.
(2016.) Prosodic realization of contrastive focus in Yucatecan Spanish. Romanistisches Jahrbuch, 67, 255–
283. ①
White, L. & Mattys, S. L.
 (2007.) Calibrating rhythm: First language and second language studies. Journal of Phonetics, 35, 501–22.
①②
Yager, K.
(1982.) Estudio del cuadro consonántico del español de Mérida, Yucatán con consideraciones de posible
influencia maya (Unpublished MA thesis). University of California-Santa Barbara. ① ② ③ ④
Yager, K.
(1989.) La -m bilabial en posición final absoluta en el español hablado en Mérida, Yucatán (México).
Nueva Revista de Filología Hispánica XXXVII, 83–94. ①②
Morphology
CHAPTER 6

First person singular subject expression in

Caribbean heritage speaker Spanish oral
production
Ana de Prada Pérez
Maynooth University

Abstract
Research on contact effects on subject pronoun expression in the U.S.
has largely focused on Spanish in the U.S. Southwest and New York
City (NYC). The conflicting results reported in this literature could be
due to the differences in the distribution of varieties in these U.S.
regions. To test this hypothesis, this paper offers a variationist analysis
of first person singular (1SG ) data from eleven Caribbean heritage
speakers (HS) divided into two proficiency groups and who were
raised in Florida, where the distribution of varieties is similar to that in
NYC. Results are consistent with a lack of contact effects in the higher
proficiency group, while contact effects are attested in the lower
proficiency group.

Keywords
heritage speaker bilingualism; subject pronoun expression;
variation; Caribbean Spanish

1. Introduction
Variationist and generativist examination of Spanish heritage speaker (HS)
subject pronoun expression (SPE) reports a higher rate of overt third person
pronominal subjects, largely attributed to contact with English, irrespective of
data collection method (e.g., Otheguy & Zentella, 2012; Montrul, 2004). Results
for first person singular (1SG ), however, are less conclusive (Torres-Cacoullos &
Travis, 2010; Travis, 2007). It is possible that some of the differences have to do
with the region in the U.S. where the data is collected: New York City (NYC) vs.
the U.S. Southwest. In this paper, we discuss 1SG interview data from Caribbean
Spanish HSs from Florida to contribute novel data to the debate of whether
English has an effect on the production of overt pronominal subjects in U.S.
Spanish.
In addition to person, the variationist literature indicates that Spanish SPE is
restricted by a variety of linguistic factors (speech connectivity, verb form
ambiguity, clause type, and semantic verb type, etc.), as well as extralinguistic
factors, such as region and language contact (e.g. Carvalho, Orozco, & Lapidus
Shin, 2015; Otheguy & Zentella, 2012; Shin, 2012; Silva-Corvalán, 1982).
Contact effects have variably been reported both in terms of SPE rates,
significance of variables, and direction of effect. In this paper, we further
examine the effect of proficiency, a measure that has not been examined directly
in previous research on oral production. All our speakers have been raised in the
U.S. and consider themselves dominant in English. Additionally, they are all of
Caribbean heritage. Thus, with these data we can focus on the effect of
proficiency.
The aim of this paper is twofold. It examines subject pronoun expression in
Florida to further understand the effect that the demographic/dialectal
distribution of the communities may have on the variable expression of subject
pronouns in bilingual Spanish. Additionally, it examines the role that proficiency
in the heritage language has on subject pronoun expression in Spanish. With this
purpose in mind, the paper is organized as follows. In Section 2, we summarize
previous relevant work on subject pronoun expression. In Section 3, we offer the
motivation for the study, research questions and hypotheses. In Section 4, we
describe the participants and research design and present the results. In
Section 5, we discuss the results. Lastly, in Section 6, we offer some
conclusions.

2. Literature review
Previous literature on Spanish SPE present results both on rates of overt
pronominal subjects and patterns of use. With respect to rates of overt
pronominal subjects in Caribbean Spanish, there is some variation attested
across studies. Table 1 presents some of the percentages offered in the previous
literature for 1SG subjects.

Table 1. Overt 1SG pronoun rates across varieties of Caribbean Spanish

Place (Author) Percentage overt pronominal subject

Castañer, Puerto Rico (Holmquist, 2012) 34%

San Juan, Puerto Rico (Cameron, 1992) 50%
San Juan, Puerto Rico (Ávila-Jiménez, 1996) 51%
Puerto Ricans in NYC (Flores Ferrán, 2002) 52%
Dominican Republic (Martínez Sanz, 2011) 59.8%
Dominican Republic (Alfaraz, 2015) 51.6%
Coastal Colombia (Orozco, 2015) 44.5%
Cuba (Ortiz López, Dauphinais & Aponte Alequín, 2017) 24%

The percentages in Table 1 indicate that in 1SG , Caribbean speakers use overt
pronominal subjects as low as 24% of the time and as high as 59.8% of the time.
 Variationist analyses of Spanish subject expression indicate that the
distribution of null and overt pronominal subjects in Spanish is best explained by
a combination of variables, except in those cases traditionally excluded from
variable rule analyses: predicates that require an expletive subject, predicates
accompanying impersonal uses of the second person singular and third person
plural, reverse psychological predicates, predicates in subject relative clauses,
subjects with inanimate referents, and predicates in set phrases. In each case, the
null pronominal subject fails to alternate with an overt counterpart.
In variable contexts, on the other hand, the distribution has been best
accounted for by a combination of variables (Carvalho et al., 2015). In general
terms, null subjects tend to indicate continuity. Thus, CO-REFERENTIALITY (also
referred to as switch reference), or whether the subject in the preceding sentence
is the same or not, favors the use of a null subject.

(1)
CO-REFERENCE OR SWITCH REFERENCE

Y yo los bañaba, y los vestía, les daba de comer, los ponía a dormir.
‘And I would bathe them, dress them, feed them, put them to sleep.’ (Travis,
2007: 107)

In (1) the pronominal subject is expressed in the first instance and omitted
afterwards where it is co-referential.
Similarly, TENSE, ASPECT, MOOD (TAM) CONTINUITY favors null subjects, as
exemplified in (2).

(2)
TAM CONTINUITY

Mañana voy. Yo dejé diez paquetes allá.

‘I will go tomorrow. I left ten packets there.’ (Travis, 2007: 107)
The subject in (2) is null in the first clause and expressed in the second, where
there is a change in TAM.
Some of the verb forms in Spanish, namely the first- and third-person
singular forms, take the same form in several tenses (imperfect, conditional,
present subjunctive and related compound forms). These ambiguous forms have
sometimes been attested with more overt pronominal subjects, with
disambiguating purposes. Consider the following example:

(3)
VERB FORM AMBIGUITY

En la noche ella iba a mi lado y yo estaba temblando

‘At night she used to go by my side and I was shaking.’ (Silva-Corvalán,
1994, p. 149)

In (3), both verbs could refer to a first or a third person singular. In this case, to
disambiguate, both appear with an overt pronominal subject. While overt
subjects are favored with verbal forms that are ambiguous and not with
unambiguous ones (Silva-Corvalán, 1994) this result is not always attested
(Casanova, 1999; Morales, 1997; Ranson, 1991). Unlike in Example (3), often
times there is enough information in the context for the referent not to be
ambiguous even if the verb form is so, as in (2) above.
The distribution of overt subjects is also relevant to the establishment of the
speaker’s position on an idea. As a result, SEMANTIC VERB TYPE affects the
distribution. For instance, verbs that express opinion or estimative verbs favor
overt subjects (Enríquez, 1984; Morales, 1997; Otheguy et al., 2007; Silva-
Corvalán, 1982, 1994; Travis, 2007). For example, Morales (1997) shows that
the subjects of verbs like pensar ‘to think’ may be produced even in topic
continuation contexts, as in (4).
(4)
VERB TYPE

Porque se enfrascan tanto en que tienen que crear una obra que se olvidan, y
aunque yo sé que ya no se usa este sistema en música, yo les digo….
‘Because they immerse themselves so much in the fact that they have to create
a work of art that they forget and, although I know this system is no longer
used in music, I tell them…’ (Morales, 1997, p. 157)1

More recently, the classification of verb type has been reduced to external
actions (i.e., verbs that denote external activity), mental processes, and stative
verbs (see Orozco, 2015, for a fuller description). In general, mental and stative
predicates favor overt subjects, while external actions favor null pronominal
subjects.
These factors (cf. Carvalho et al., 2015, for a more complete list of factors)
do not all affect the distribution equally. Switch reference, for instance, is
significant and highly ranked (i.e., larger magnitude of effect) across studies.
Verb form ambiguity, however, tends to be ranked lower or not significant at all.
Subject pronoun rates, on the other hand, vary significantly from region to
region. Research on monolingual varieties in the Caribbean report significant
variation from study to study (Alfaraz, 2015; Bentivoglio, 1987; Cameron, 1995;
Cameron & Flores-Ferrán, 2004; Enríquez, 1984; Erker & Guy, 2012; Flores-
Ferrán, 2002, 2004; Hochberg, 1986; Holmquist, 2012; Martínez-Sanz, 2011;
Morales, 1997; Orozco, 2015; Orozco & Guy, 2008; Ortiz López, 2009; Otheguy
& Zentella, 2012; Shin & Otheguy, 2013). The rates reported for monolingual
Caribbean Spanish in 1SG subjects span from 34% in Castañer, Puerto Rico
(Holmquist, 2012) to 59.8% in the Dominican Republic (Martínez Sanz, 2011).
Importantly, although dialectal variation is largely attested in subject pronoun
rates and restricted to lower ranked variables (those with a smaller effect size),
these differences do not affect most linguistic variables. Thus, this combination
of variables is rather stable across varieties of Spanish.
With respect to subject pronoun expression, a comparison that has
generated great interest in the literature has been between monolingual and
bilingual varieties of Spanish. Researchers of Spanish in contact with English
have reported conflicting results: while some do not report differences (Bayley
& Pease-Alvarez, 1996, 1997; Flores & Toro, 2000; Flores-Ferrán, 2004; Silva-
Corvalán, 1994; Torres-Cacoullos & Travis, 2010; and Travis, 2007), others
report a higher rate of overt pronominal subjects and changes in the distribution
reflected in variables that are significant, their ranking, and/or their constraint
ranking (Erker & Guy, 2012; Erker & Otheguy, 2016; Lipski, 1994, 1996;
Otheguy & Zentella, 2012; Shin, 2012; Shin & Otheguy, 2013; Toribio, 2004,
among others). It is possible that these differences are due to the grammatical
person examined in these studies (cf. Prada Pérez & Gómez Soler, 2019) or to
differences in the communities. The data in studies that do not report a contact
effect are primarily from Mexican varieties in the U.S. Southwest (with the
exception of Flores & Toro, 2000; Flores-Ferrán, 2004). In contrast, the data in
studies that report a contact effect are from several varieties of Spanish or
Caribbean Spanish in NYC (with the exception of Toribio, 2004). The
demographics of the U.S. Southwest are rather different than those in NYC.
According to the 2010 U.S. Census, in New Mexico 46.3% of the population is
Latino, while in NYC 17.6% of the population is Latino. Within the Latino
population, only 1.44% of the Latino population in New Mexico is of Caribbean
origin (including Cuba, Dominican Republic, Puerto Rico, and Venezuela),
while it is 57.8% in NYC. In this paper we examine data from Florida, whose
Latino population is 22.5% of the total population. Within the Latino population,
55.3% is of Caribbean origin. Thus, in this paper we examine a novel
community (Caribbean speakers in Florida), which is similar to NYC in the
distribution of varieties to determine if the demographics, and the resulting
dialect contact, can shed some light on the conflicting previous results on SPE in
U.S. Spanish.

3. Motivation and research questions

The goal of this study is to explore subject expression in 1SG (yo ‘I’/Ø) through
spontaneous speech samples from Spanish HSs of Caribbean heritage living in
Florida in order to provide novel data to better understand differences in
varieties of U.S. Spanish. Similar to the contact situation in NYC reported by
Otheguy and Zentella (2012), in Florida there is a slightly higher population of
Caribbean Spanish speakers than non-Caribbean Spanish speakers. In New
Mexico, in contrast, the Caribbean population constitutes only 1.44% of the
Latino population. While Otheguy and Zentella (2012) report contact effects in
terms of rates of overt pronominal subjects and the strength of the variables
regulating the distribution, Torres-Cacoullos and Travis (2010) do not. Torres-
Cacoullos and Travis (2010) included only 1SG subjects, whereas Otheguy and
Zentella (2012) included all persons. In this paper, we follow Torres-Cacoullos
and Travis (2010) and only include 1SG subjects. Person is a highly ranked
variable in all previous studies and it tends to be collinear with other variables,
especially verb form ambiguity. Since the analyses presented in this paper
include the variable verb form ambiguity, only those cases where the participant
used a 1SG subject were included. Additionally, this analysis facilitates
comparisons with Torres-Cacoullos and Travis’ (2010) data.
Our research questions inquire about the two measures usually examined in
previous research. First, we examine rates of overt pronominal subjects, to
establish a possible increase due to contact with English, a language that rarely
allows the absence of the subject pronoun. Additionally, we examine patterns of
overt pronominal subjects, instantiated in the variables selected as significant,
their magnitude of effect, and the direction of the effect. These two measures are
examined through comparisons of data from our bilingual participants with data
from previous studies where the participants were monolingual, as well as a
comparison within our data between higher proficiency speakers and lower
proficiency speakers. We divided our participants, who were all raised in the
U.S. (and who were either born in the U.S. or migrated to the U.S. by age 11)
and dominant in English, according to their proficiency in Spanish, as measured
by the version of the Diploma de español como lengua extranjera (DELE) exam
commonly used in the field of Second Language Acquisition (Montrul &
Slabakova, 2003; Cuza, 2013). Thus, comparisons are established based on
language contact vs. no language contact, as well as different degrees of
language contact.
Our first research question examines rates of overt pronominal expression
and the effect that language contact has on it. In particular, this paper aims to
address the following question: Are there differences between Caribbean
Spanish HSs from Florida and their monolingual counterparts in the previous
literature or between higher and lower proficiency Caribbean Spanish HSs in
rates of overt pronominal subjects? This question will be tackled through the
examination of Caribbean Spanish HSs from Florida, through the two
comparisons: rates of overt pronominal subjects in monolingual Caribbean
speakers reported on previous research vs. rates in our participants; and rates of
participants with higher proficiency in Spanish vs. participants with lower
proficiency in Spanish. Previous research examines differences in rates between
monolingual and bilingual speakers as well as different groups of bilinguals,
based on their degree of contact with English. Irrespective of region of origin,
Otheguy and Zentella (2012) report a higher rate of overt pronouns in speakers
with more contact with English than those with less contact with English.
Degree of contact was determined based on the linguistic experience of the
bilinguals (those born and raised in NYC vs. those born in a Spanish-speaking
country who migrated to the U.S. in adulthood and those that were born in a
Spanish-speaking country and who were more established in the U.S.). In
contrast, Torres-Cacoullos and Travis (2010) do not find this effect in their
comparison of bilinguals in New Mexico with different degrees of English
proficiency (and presence of code-switching in their speech). If the differences
between the results in previous literature are attributed to the linguistic
composition of the communities, we anticipate to report contact effects for our
participants, since the linguistic composition of Florida is more similar to that in
NYC. All speakers are expected to exhibit higher rates of overt pronominal
subjects than those reported in the literature on Caribbean monolingual varieties
but more so in the case of lower proficiency speakers. Thus, in line with
Otheguy and Zentella (2012), we anticipate HSs in Florida to exhibit higher rates
of overt pronominal subjects than their monolingual counterparts and for these
contact effects to increase with intensity of contact, as in Spanish lower
proficiency speakers than higher proficiency speakers.
Our second research question inquires about the patterns of subject
expression across our participants and comparable monolingual data reported in
previous literature and between the two groups of participants: the higher
proficiency and the lower proficiency HSs. Torres-Cacoullos and Travis (2010)
report minor differences between those speakers with higher English proficiency
and those with lower English proficiency in the corpus they examined. Similarly,
Otheguy and Zentella (2012) report similarities in the variables that are
significant, with minor differences. Specifically, those groups of speakers that
are more established in the U.S. exhibit a weakening of the effect of some
variables (notably, the variable switch reference) and some differences in the
direction of effect in the variable Person. For example, third person singular was
the highest ranked constraint in the Mainland (or Non-Caribbean) Newcomers
and not significant in the Caribbean Newcomers. In the New York born and
raised (NYR), for the Mainland speakers, there was no difference with their
Newcomer counterparts. For the Caribbean speakers, in contrast, third person
singular was significant. Since we only examine 1SG subjects in this paper, we
anticipate little difference in the variables that regulate subject expression and
the direction in both groups of speakers. Differences, however, may exist in the
magnitude of effect of these variables, as in Otheguy and Zentella (2012).
Regarding the variables included in this study, then, the anticipated direction of
effect (if there is an effect) is for (i) contexts of different reference to favor overt
pronominal subjects more than those of same reference; (ii) continuity of TAM
to favor pronominal omission; (iii) mental and stative predicates to favor
pronominal omission; and (iv) ambiguous verb forms to favor overt pronominal
subjects. Thus, comparisons with previous monolingual research as well as
comparisons across the two proficiency groups are expected to show differences
in the magnitude of effect of the variables, which may result in lower ranked
variables (those with a smaller effect) in the monolingual literature not to be
significant in bilingual speakers, especially those of lower proficiency. We do
not anticipate changes in the direction of effect of variables.
It is important to point out that, in this paper, both groups of speakers are in
contact with English. Thus, we use proficiency as a measure of intensity of
language contact. If the data is consistent with Otheguy and Zentella’s (2012)
conclusion, both groups will be different from monolingual Caribbean speakers,
but more so in the case of the lower proficiency group. On the other hand, if the
data is consistent with Torres-Cacoullos and Travis (2010), at least the higher
proficiency group will not differ from monolingual Caribbean speaker data
discussed in previous research.
In order to test these hypotheses, we analyzed data from sociolinguistic
interviews carried out in Spanish with speakers of Caribbean heritage in Florida.

4. The current study

4.1 Participants

Eleven participants of Caribbean heritage volunteered for this study. All

participants were instructed HSs enrolled in university Spanish courses at the
time of the study. There were 10 females and 1 male. There were five
participants of Cuban heritage, three of Puerto Rican heritage, two of
Venezuelan heritage and one of Dominican heritage. As an independent measure
of proficiency, we utilized a customized version of the DELE exam commonly
used in the field of second language acquisition and heritage speaker
bilingualism (Montrul & Slabakova, 2003; Cuza, 2013) and presented it online
through Qualtrics. The test consists of 50 questions divided in two different
sections: a multiple-choice vocabulary section and a cloze passage that tests
grammatical knowledge. The mean score was 27 (range: 14–39). We calculated
the median (29/50) and split the participants into two groups: the higher
proficiency group, who scored above 29 (n = 5) and the lower proficiency, who
scored 29 or below (n = 6). Although none of the participants scored within the
range commonly referred to as advanced in the literature (40–50), most of the
participants in the higher proficiency group were perceived as native-like by the
author.
They belonged to different generations: two were born in Puerto Rico and
moved to the Mainland U.S. before age 11 (Generation 1.5), three were second-
generation HSs, two were born in the U.S. with one parent born in a Spanish-
speaking country and one parent who was a second-generation HS (Generation
2.5), and four third-generation HSs. Table 2 summarizes this information.
Table 2. Participant information

Generation 1.5 2 2.5

Proficiency Higher Lower Higher Lower Higher Lower

proficiency proficiency proficiency proficiency proficiency proficiency

Number of 2 0 1 2 0 2
participants

Given the diversity attested in the participant pool, in addition to group results
we will report individual results.

4.2 Methodology

We conducted 11 semi-structured sociolinguistic interviews. The interviews

lasted between 30 to 60 minutes and were always recorded in a quiet office. All
interviews were digitally recorded using a Marantz PDM 661 and a Shure SM
10-A head-mounted microphone. The interviewers used a Power Point with
questions to guide the interview and to assure that all participants spoke about
similar topics and in similar tenses. Participants were told that they should
consider the interview as an informal conversation and that they would simply
be talking about some events in their past, present, and future. The interviewer
initiated the conversation asking participants to talk about their birth city and
family. Then, participants were asked to talk about their childhood friends and
teachers, and to describe interesting events from their childhood. Next,
participants were asked to talk about their current experience at the university
(classes, friends, teachers) as well as about their current jobs. Finally, the
interview focused on future events by asking participants about their plans for
the next holiday (e.g., Thanksgiving/Spring break), their plans for their future
career, and life in the year three thousand.
The interview data was transcribed and coded for several linguistic
variables: SUBJECT FORM, SWITCH REFERENCE, TAM CONTINUITY, VERB
AMBIGUITY, and VERB TYPE . We also coded two extra-linguistic variables:
PROFICIENCY and GENERATION . Only tokens in 1SG , topic continuation, main
clauses, and with animate referents were included (with a total of 2,054 tokens).
Table 3 describes the variable and constraints considered for analysis.

Table 3. Linguistic and extralinguistic variables and associated constraints

Variable type Variable Constraint

Linguistic variables SUBJECT FORM (dependent variable) Null

Overt Pronominal
SWITCH REFERENCE Same referent
Different referent
TAM CONTINUITY Same TAM
Different TAM
VERB TYPE State
Mental External
Extralinguistic variables PROFICIENCY Higher proficiency
Lower proficiency
GENERATION 1.5
Second
2.5
Third

Next, we illustrate the above constraints from examples from the interviews.
Example (5) shows the dependent variable:

(5)
SUBJECT FORM
 1. Overt Pronominal
Yo nací en Plantation en la Florida, en el sur de la Florida. Entonces por
ahí todos hablan español.
‘I was born in Plantation, in Florida, in South Florida. So, everyone
speaks Spanish over there.’
2. Null
Ø Entonces tengo a dos hermanos que también hablan español y
aprendimos español de mi mamá.
‘So I have two brothers who also speak Spanish and we learned Spanish
from my mom.’

In (5a), the speaker produces the 1SG pronoun yo, while in (5b) she does not.
This distribution tends to be regulated by switch reference.
While in (5a) the participant uses the overt pronoun yo ‘I’, in (5b) they omit
the subject pronoun. This distribution is regulated by several independent
variables. The first independent variable that regulates subject expression is
switch reference, as in (6):

(6)
SWITCH REFERENCE

1. Same referent
Tenía clase en el college o me quedaba en la escuela.
‘I had class at the college or I stayed in school.’
2. Different referent
Mi quinto periodo era history creo que era y después mi séptimo periodo.
‘My fifth period was history, I believe that it was, and then my seventh
period.’

As shown in (6), the speaker uses a null subject with me quedaba in (a) whose
subject has the same referent as the previous verb (tenía), and with creo in (b),
even though the previous subject has a different referent.
Another variable that expresses speech connectivity is TAM continuity:

(7)
TAM CONTINUITY

1. Same TAM
No dormí mucho esos cuatro años pero aprendí cómo escribir ensayos.
‘I didn’t sleep much those four years but I learned how to write essays.’
2. Different TAM
Como yo hice dual enrollment no tenía clase
‘Since I did dual enrollment I didn’t have class.’

As can be seen in (7a), the speaker omits the subjects with aprendí, which is in
the same TAM as the previous verb, dormí. Some authors find that speakers are
more likely to use more overt subjects when the TAM is different, as in (7b). In
this example, however, the speaker omits the subject pronoun.
Some previous literature reports the use of overt pronominal subjects to
disambiguate. This trend is not categorical, as shown in (8).

(8)
VERB AMBIGUITY

1. Non-ambiguous
Me identifico mucho como cubano cuando voy a Miami.
‘I identify a lot as a Cuban when I go to Miami.’
2. Ambiguous
Les dije que era puertorriqueño.
‘I told them I was Puerto Rican.’

In (8), the speaker uses a null subject both with the morphologically
unambiguous verb form voy and with the ambiguous verb form era.
Lastly, speakers tend to use more overt pronominal subjects with mental or
stative predicates than with external action predicates, as shown in (9).

(9)
VERB TYPE

1. External
So fui para casa siempre y con el bus.
‘So I always went home and on the bus.’
2. Stative
Y leí, vi televisión, leí más. No me gusta jugar deportes y nunca hací eso.
‘And I read, watched tv, read some more. I don’t like playing sports and I
never did that.’
3. Mental
Ella, mi mamá, no quería que yo guiaba todos los días porque era muy
temprano y estaba muy cansado y ella se preocupó.
“She, my mom, didn’t want me to drive every day because it was very
early and I was very tired and she got worried.’

In (9), the speaker uses null subjects with external actions (fui ‘I went’) and with
stative predicates (leí ‘I read’). In (9c), however, they use an overt pronominal
subject with a mental (quería ‘I wanted’) predicate. All 1SG tokens with animate
referents in main clauses were coded and included in the analysis, carried out
using Goldvarb Yosemite software.

4.3 Results

The purpose of this study is to examine the distribution of overt pronominal

subjects in the Spanish of Caribbean HSs in Florida. The group of speakers
belong to different generations and proficiency levels. There was collinearity
between the variables generation and proficiency, such that all speakers in
generation 1.5 were of higher proficiency and none of the speakers in generation
2.5 were of higher proficiency. Given that regressions assume independence of
variables, separate analyses were carried out to better understand the effect of
these variables. This is the only interaction that was found in our analyses. All
the tables summarizing the results follow the same conventions. For the
variables not selected as significant the factor weights are provided in square
brackets. The selected application value was overt pronominal subjects. All
factor weights above .50 (bolded in Tables 4–8) indicate that the factor favors
the use of an overt pronominal subject. Thus, the direction of effects is discussed
based on the factor weight. The variables or factor groups are ranked according
to their magnitude of effect, from the highest range at the top of the table to the
lowest range at the bottom of the table.
A separate analysis for the variable generation revealed that it is a
significant variable (Table 4).

Table 4. 1SG subject expression in Caribbean Spanish-English bilinguals in

Florida: Generation
Multivariate regression analysis of the contribution of the factor generation to
the probability of producing a 1SG overt pronominal subject vs. a null subject;
non-significant factor groups in square brackets.

Factor weight % N % of data

Generation
2.5 0.78 51.2 196 18.6
Second 0.69 39.8 274 33.5
Third 0.29 10.6 51 23.5
1.5 0.23 8 40 24.3
Range 55
Total N 2,054
Corrected Mean .227
Log likelihood −1030.585
Significance .000

The different behavior for generation 1.5 and the second generation as well as
between the generation 2.5 and the third generation prevented the recoding into
two generations to facilitate interpretation. Generation 2.5 and the second
generation favored the overt form. In contrast, generation 1.5 and third
generation disfavored overt pronominal subjects. Given the low number of
participants in each generation, the results for generation have to be taken with
caution.
The results for proficiency, on the other hand, are more informative (see
Table 5). The participants with lower proficiency produced significantly more
overt pronominal subjects than those in the higher proficiency group.

Table 5. 1SG subject expression in Caribbean Spanish-English bilinguals in

Florida: Proficiency
Multivariate regression analysis of the contribution of external factor proficiency
to the probability of producing a 1SG overt pronominal subject vs. a null subject;
non-significant factor groups in square brackets.

Factor weight % N % of data

Proficiency
Lower 0.62 37 345 45.4
Higher 0.40 19.3 216 54.6
Range 55
Total N 2,054
Corrected Mean .264
Log likelihood −1164.089
Significance .000

For the lower proficiency level, in addition to producing an average rate of overt
pronominal subjects higher than that of the higher proficiency group, there was
more variation within the group. As shown in Table 6, within the higher
proficiency group, the range of overt pronominal subjects is 6% to 30.3%, while
it is 3% to 92.9% in the lower proficiency group.

Table 6. Percentage of use of overt pronominal subjects by participant in each

proficiency group

% N Country of origin

Higher proficiency
Participant 2 16.9 10 Puerto Rico
Participant 3 18.8 15 Puerto Rico
Participant 4 30.3 143 Venezuela
Participant 5 25.3 23 Cuba
Participant 10 6 25 Puerto Rico
Lower proficiency
Participant 1 3 4 Puerto Rico
Participant 6 7 14 Cuba
Participant 7 61.7 137 Dominican Republic
Participant 8 92.9 105 Cuba
Participant 9 25 26 Venezuela
Participant 11 36.6 59 Cuba

When compared with previous studies, the average number of overt pronominal
subjects in the higher proficiency group (19.3%) is lower than in all monolingual
Caribbean varieties (cf. Table 1 above), where the lowest percentage, 24% overt
pronominal subjects, is reported by Ortiz López et al. (2017). The lower
proficiency group uses overt pronominal subjects 37% of the time, which is
lower than the majority of monolingual Caribbean varieties (with the exception
of Ortiz López et al., 2017, and Holmquist, 2012). Although we cannot explore
this hypothesis further here, it is possible that it is the result of convergence with
other varieties of Spanish, given Florida’s linguistic landscape. It is also possible
that differences do exist between different varieties within the Caribbean region,
an area that merits further research. In this paper, they are all discussed as one
linguistic group. Some of the data, however, could be consistent with dialectal
differences, within the Caribbean, in rates of overt pronominal subjects.
Participant 7, for instance, was the only participant from the Dominican
Republic and one of the participants with the highest rates of overt pronominal
subjects. It is also important to point out that there were wide proficiency
differences, particularly within the lower proficiency group. For example, the
participant with the highest rate of overt pronominal subjects, Participant 8, was
also the one who obtained the lowest proficiency score (14/50 in the DELE) and
had difficulty maintaining a fluid conversation during the interview. Given the
diversity attested in our participant pool and the low number of participants per
cell, it is possible that the attested differences in overt pronominal subject rates
between the two proficiency groups are due to a variety of factors, including
dialectal variation within the Caribbean region and differences in proficiency
within each of the two proficiency groups, an issue that invites future research
with a higher number of participants per cell.
In previous literature, differences in rates across regions or groups of
speakers did not necessarily imply differences in their grammars as instantiated
in the variables regulating the distribution and the direction of effects (Carvalho
et al., 2015). In order to examine possible differences across the two proficiency
groups, we performed separate multivariate regressions for the higher and the
lower proficiency groups.
The grammars of the higher proficiency group were regulated by two
factors: switch reference and verb type (Table 7). TAM continuity and verb form
ambiguity were not selected as significant.

Table 7. Subject pronoun expression in the higher proficiency group

Multivariate regression analysis of the contribution of internal factors to the
probability of producing a 1SG overt pronominal subject vs. a null subject in the
higher proficiency group; non-significant factor groups in square brackets.
background

Factor weight % N % of data

Switch reference
Different referent 0.70 32.7 108 29.8
Same referent 0.41 13.7 108 70.2
Range 29
Verb type
Stative 0.60 25.2 113 40.1
External 0.46 15.8 65 36.7
Mental 0.39 14.6 38 23.3
Range 21
Tam continuity
Different TAM [0.54] 22 97 39.3
Same TAM [0.47] 17.5 119 60.7
Range 7
Ambiguity
Ambiguous [0.55] 22.4 34 13.6
Unambiguous 0.49 18.8 182 86.4
Range 6
Total N 1121
Corrected Mean .175
Log likelihood −513.921
Significance .000

Although previous research has claimed that bilingualism may weaken the
speakers’ sensibility to pragmatic factors, such as switch reference in subject
pronoun expression in Spanish HSs (Montrul, 2004), for this group of bilinguals,
switch reference is a strong predictor of subject pronoun expression, with overt
pronominal subjects being favored in contexts of different reference and nulls in
contexts of same reference. With respect to verb type, recall that previous studies
do not always return this variable as significant but, when they do, stative and
mental verbs favor overt pronominal subjects. External action predicates,
however, favor the omission of the pronoun. In our group of speakers, stative
and external action predicates exert the same effect as in previous studies.
Mental predicates, on the other hand, disfavor the use of overt pronominal
subjects. It is possible that this is due to the specific verbs included in the
category (Orozco, 2015).
This variable remains highly ranked in the lower proficiency group. Switch
reference, however, does not reach significance (Table 8). Instead, TAM
continuity is selected as significant. Verb form ambiguity is not significant in
this group either.

Table 8. Subject pronoun expression in the lower proficiency group

Multivariate regression analysis of the contribution of internal factors to the
probability of producing a 1SG overt pronominal subject vs. a null subject in the
lower proficiency group; non-significant factor groups in square brackets.

Factor weight % N % of data

Verb type
Mental 0.64 51 106 22.3
Stative 0.50 36.8 148 43.1
External 0.41 28.2 91 34.6
Range 23
Tam continuity
Same TAM 0.53 40.1 224 59.8
Different TAM 0.45 32.3 121 40.2
Range 8
Switch reference
Different referent [0.55] 41.7 126 32.4
Same referent [0.48] 34.7 219 67.6
Range 7
Ambiguity
Ambiguous [0.53] 40 66 17.7
Unambiguous 0.49 36.3 279 82.3
Range 4
Total N 933
Corrected Mean .366
Log likelihood −598.184
Significance .028

For the lower proficiency group, the statistical analysis returns results for verb
type which are consistent with previous research; mental and stative predicates
favor overt pronominal subjects, whereas predicates expressing external actions
do not. TAM continuity exhibits the opposite trend attested in previous studies;
same TAM favors overt pronominal subjects. It is important to point out, though,
the small magnitude of effect of this variable (range: 8). Switch reference
exhibits the same trend as in the higher proficiency group. This group of
bilinguals, however, does not exhibit sensitivity to the variable switch reference.
In general, in these two groups, the difference in proficiency manifests itself
not only in overt pronominal rates but also in the strength of the variable switch
reference, which tends to be the highest ranked variable in monolingual varieties
of Spanish once the variable person is excluded from analysis (Alfaraz, 2015,
2017; Martínez Sanz, 2011; Orozco, 2015; Prada Pérez, 2015).

5. Discussion
The purpose of this paper was to examine language contact in a novel
community to better understand the effect of language contact with English on
Spanish subject pronoun expression. In previous studies, there was conflicting
evidence regarding the effect of language contact on the rate and patterns of use
of overt pronominal subjects in comparison to their omission in Spanish. In
NYC, a higher rate of overt pronominal subjects was attested, whereas no
difference was reported in New Mexico speakers. Regarding their distribution, in
New Mexico, no differences, or only minor differences, were attested in
comparing the monolingual and bilingual groups. In contrast, in NYC there were
differences in the strength of the variables.
Our first research question inquired about differences in rates of overt
pronominal subjects across different bilingual communities. In particular, the
comparison focuses on NYC, New Mexico, and Florida. In order to discuss
whether these bilingual communities use a higher rate of overt pronominal
subjects than their monolingual counterparts, it is necessary to establish a
comparison group. In the current study there is no monolingual control group, so
we examine language contact by comparison with results of previous studies and
by comparing a higher and a lower proficiency group within the sample.
Although comparing rates across studies is problematic (Silva-Corvalán, 2001),
it seems that our participants’ rates are among the lowest reported for Caribbean
speakers, including monolingual speakers. As shown in Table 1, in previous
research the lowest percentage reported for 1SG subjects is 24% in Cuba (Ortiz
López et al., 2017) and the highest is 59.8% in the Dominican Republic
(Martínez Sanz, 2011). The higher proficiency group produces overt pronominal
subjects 19.3% of the time, a percentage that is lower than that of Cuba. Our
lower proficiency group produces 37% overt pronouns, a low percentage,
compared to monolingual varieties of Caribbean Spanish. This percentage is
similar to that for Otheguy and Zentella’s (2012) Caribbean Newcomers (36%)
and lower than that for the New York Raised (NYR) group (44%). The
participants in this study, however, are closer in their linguistic background to
those in the NYR group, as they are all dominant in English. Thus, it seems that
even the lower proficiency group exhibits a low rate of overt pronominal
subjects, a result consistent with the finding of Torres-Cacoullos and Travis’
(2010) conclusion that language contact with English does not result in a higher
rate in overt pronominal subjects. In conclusion, in response to our first research
question, our data does not offer evidence of an increase in overt pronominal
subjects in Caribbean HSs in Florida in comparison with monolingual Caribbean
speakers. This result indicates that the differences in previous research between
NYC and New Mexico data cannot be attributed to the linguistic composition of
speakers in these communities.
It may merit further discussing the lower rate of overt pronominal subjects
for bilinguals with higher proficiency in Spanish than for monolingual speakers.
This result may be due to differences in the data collection process. In the
present study, the interlocutor’s variety of Spanish was Peninsular Spanish.
Additionally, the data produced were largely in same referent contexts (70.2% of
the higher proficiency data and 67.6% of the lower proficiency data). Lastly,
there is potentially contact with non-Caribbean varieties. All these facts are
different in this study and, separately or together may have had an effect, an
issue that we leave for future research.
Overall, comparing the rates in both groups of speakers in this study with
rates in monolingual studies in the Caribbean region offers no evidence of a
higher rate in overt pronominal expression due to contact with English.
Comparing both groups, though, reveals a higher rate of overt pronominal
subjects with lower proficiency in Spanish. It is possible, then, that more than
contact with English, the higher rate of overt pronominal expression in bilinguals
only occurs at lower levels of proficiency in Spanish. Thus, speakers of higher
proficiency, who are more balanced in their proficiency in both languages, may
have been capable of maintaining referents in memory longer. Future research
can examine this hypothesis by including the distance between the target subject
and its antecedent, measuring intervening subjects, and comparing its effect
across different groups of speakers.
In addition to rates, patterns were explored in our second research question
through the application of four variables commonly used in the field: switch
reference, TAM continuity, verb from ambiguity, and semantic verb type. An
initial analysis including only the variable proficiency revealed that these two
groups were statistically different. Thus, separate regressions for each group
were performed to examine the role of these variables in these two groups of
speakers. For the higher proficiency group, switch reference and verb type were
significant, with patterns similar to those attested in other communities. As in
monolingual groups of speakers reported elsewhere in the literature, switch
reference exhibited a large effect in this group of bilinguals (contrary to Otheguy
& Zentella’s, 2012 results), as indicated in the range of 29. For comparison,
Alfaraz (2015) reports a range of 27 for monolingual speakers in the Dominican
Republican and Orozco (2015) a range of 32 for monolingual speakers in
Colombian Costeño. In the lower proficiency group, on the other hand, switch
reference is not significant, with a range of 7. Thus, in this group of bilinguals
there is an effect of language contact in the strength (and significance) of the
variable switch reference, as predicted by Otheguy and Zentella (2012) and in
contrast with the results of Torres-Cacoullos and Travis (2010). Our results seem
contradictory since the higher proficiency group patterns more similarly to those
in the U.S. Southwest (i.e. no evidence of contact effects), while the lower
proficiency group’s patterns are closer to those in NYC (i.e. evidence of contact
effects). Since the previous research did not include proficiency, it is possible
that the participants in Torres-Cacoullos and Travis (2010) were of higher
proficiency in Spanish than those in Otheguy and Zentella (2012). If that were
the case, the data seems to indicate that it is proficiency, more than language
contact, which can explain differences between monolingual and bilingual
speakers in previous research and in the lower proficiency group in this project.
Rather surprisingly, the variable verb type was significant for both
proficiency groups. In previous studies there is variability in the relevance of this
factor (Orozco, 2015). In fact, Orozco (2015) proposes that frequency of the
verb may interact with the semantic classification of verbs such that some of the
higher frequency verbs (e.g. tener ‘to have’ among stative verbs) behave
differently from other verbs in the same class. Future research, then, should
examine these high frequency verbs. In the meantime, it seems that this lexical
variable is rather stable in bilinguals in NYC, New Mexico, and Florida.

6. Conclusion
This study contributes to current debates in the language contact literature on the
effect of language contact, in particular with English, on the expression of overt
pronominal subjects in Spanish. Previous literature on Spanish language contact
with English in the U.S. seems to return conflicting results. Bayley & Pease-
Álvarez (1996, 1997), Flores-Ferrán (2004), Silva-Corvalán (1994), Torres-
Cacoullos and Travis (2010), and Travis (2007) reveal no higher rate of overt
pronominal subjects in Spanish in the U.S, as compared to monolingual
varieties. Lapidus and Otheguy (2005a, b), Lipski (1996), Montrul (2004),
Otheguy and Zentella (2012), Otheguy et al. (2007), and Toribio (2004), on the
other hand, report a higher rate. Considering some differences across these
studies, this paper aimed to test the possibility that the dialectal distribution in
the community had an effect: NYC, for instance, has a slightly higher presence
of Caribbean than non-Caribbean Spanish speakers (57.8% Caribbean), whereas
New Mexico has an almost exclusively non-Caribbean Spanish-speaking
population (1.55% Caribbean). Florida exhibits similar demographics to NYC
(55.3% Caribbean). Regarding subject expression, Otheguy & Zentella (2012)
reported a higher rate of overt pronominal subjects in the Caribbean NYR group
(44% overt pronominal subjects) than in the Caribbean Newcomer group (36%).
Torres-Cacoullos & Travis (2010), in contrast, reported the same rate (38%) for
both groups of speakers (monolingual and bilingual) in New Mexico. Given the
similarities in the communities in Florida and NYC, we anticipated finding a
higher rate of overt pronominal subjects in our data than in monolingual data,
given the effect of contact with English in the comparable community of NYC.
Our results indicate that this is not the case. In fact, both higher and lower
proficiency groups report rates within (and even lower) than some of the rates
reported in the literature on monolingual Caribbean varieties of Spanish. Thus,
our results are more consistent with the lack of a contact effect on rates of overt
pronominal subjects. Our examination does reveal, though, an effect of
proficiency in Spanish, where lower proficiency speakers use overt pronominal
subjects significantly more than higher proficiency speakers. We acknowledge
that proficiency in Spanish can be difficult to measure in HS participants. Since
we examined the speech of instructed HSs we were able to use a section of the
DELE as an independent measure of proficiency. While none of the speakers
scored within the advanced range (over 40/50), using a median split we obtained
two differentiated proficiency groups. Thus, even though this task is problematic
for several reasons (written task focused on prescriptive Peninsular Spanish),
using a median split allowed us to use an independent measure of proficiency
normalized for our group of speakers. These results are suggestive of the need to
include proficiency as a factor in future studies. In the absence of a better
measure of proficiency, a median split seems to offer more informative results
than groupings based on preassigned scoring ranges.
Other than rates of overt pronominal subjects, we focused our analysis on
patterns of use of overt and null pronominal subjects as instantiated in the
variables reported as significant in a multivariate regression for each of the
proficiency groups. The results of the higher proficiency group were also
consistent with the results in Torres-Cacoullos and Travis (2010) with respect to
the variables that were found as significant (switch reference and semantic verb
type) and the strength of the variables. For the lower proficiency group,
however, the variable switch reference was not found to be significant, a result
indicative of a language contact effect.
In conclusion, a contact effect manifested at a higher rate in overt
pronominal subjects in bilinguals than in monolinguals and in a weakening of
the effect of variables (instantiated in a loss of statistical significance or a
decrease in the effect size of the variable) is not evident in our data, especially if
we consider the higher proficiency speakers. A proficiency effect, on the other
hand, is attested. The lower proficiency speakers do produce significantly more
overt pronominal subjects and the variable switch reference is not found to be
significant in this group. These results indicate that the differences between
studies in the U.S. Southwest and NYC cannot be attributed to the dialectal
distribution of the communities since our community demographically resembles
the demographics of NYC, yet our results are consistent with those in Torres-
Cacoullos & Travis (2010), with data from the U.S. Southwest. The results
regarding language-external variables are limited by the number of participants
in this study, especially in light of the possible diversity among varieties of
Spanish in the Caribbean region and the effect of proficiency. This study, thus,
invites further research to understand the effects of bilingualism and heritage
language proficiency on the speech of bilinguals. In particular, this study
indicates the need to examine proficiency more closely. Regarding language
contact, its effect does not seem to be persistent throughout speakers, albeit to
different degrees, as widely assumed in the field. Our data is consistent with
Torres-Cacoullos & Travis (2010) in that higher proficiency bilinguals do not
exhibit a simplified or a converged variety of Spanish. In the lower proficiency
group, however, there is ample evidence of differences, not so much in
comparison to the rates, but rather clearly in contrast to patterns in the
monolingual speech described in previous literature. Thus, contact effects were
attested, although not in both groups of participants, and not in all participants in
the lower proficiency group. Even though the field of heritage Spanish
bilingualism has advanced enormously recently, we are still unable to account
for the variation encountered in Spanish subject expression, even when
considering rather homogenous groups of speakers. Thus, further research
examining individual and societal factors, such as speakers’ social networks and
contact with other varieties of Spanish, proficiency in the heritage language and
its adequate measure, variation within Caribbean and within Mainland Spanish,
is needed.

Note
1. This example is in topic continuation, if we only include referential subjects as possible preceding
subjects. ①

References
Alfaraz, G.
(2015.) Variation of overt and null subject pronouns in the Spanish of Santo Domingo. In A. M. Carvalho
R. Orozco, & N. L. Shin (Eds.), Subject pronoun expression in Spanish: A cross-dialectal perspective
 (pp. 3–17). Washington DC: Georgetown University Press. ① ② ③ ④
Alfaraz, G.
(2017.) The variation of subject pronouns over time in Cuban Spanish. In A. Cuza (Ed.), Cuban Spanish
dialectology: Variation, contact, and change (pp. 81–96). Washington, DC: Georgetown University
Press. ①
Ávila-Jiménez, B. I.
(1996.) Subject pronoun expression in Puerto Rican Spanish: A sociolinguistic, morphological, and
discourse analysis, Doctoral dissertation, Cornell University, Ithaca, NewYork. ①
Bayley, R., & Pease-Álvarez, L.
(1996.) Null and expressed pronoun variation in Mexican-descent children’s Spanish. In J. Arnold, R.
Blake, & B. Davidson (Eds.), Sociolinguistic variation: Data, theory, and analysis (pp. 85–99).
Stanford, CA: CSLI. ① ②
Bayley, R., & Pease-Álvarez, L.
(1997.) Null pronoun variation in Mexican-descent children’s narrative discourse. Language Variation
Change, 9, 349–371. ①②
Bentivoglio, P.
(1987.) Los sujetos pronominales de primera persona en el habla de Caracas. Caracas: Central University
of Venezuela. ①
Cameron, R.
(1992.) Pronominal and null subject variation in Spanish: Constraints, dialects, and functional
compensation. Philadelphia: University of Pennsylvania dissertation. ①
Cameron, R.
(1995.) The scope and limits of switch reference as a constraint on pronominal subject expression.
Hispanic Linguistics, 6–7, 1–27. ①
Cameron, R., & Flores-Ferrán, N.
(2004.) Perseverance of subject expression across regional dialects. Spanish in Context, 1(1), 41–65.
①
Carvalho, A. M., Orozco, R., & Shin, N. L.
(2015.) Subject pronoun expression in Spanish: A cross-dialectal perspective. Washington, DC:
Georgetown University Press. ① ② ③ ④
Casanova Seuma, L.
(1999.) El sujeto en catalán coloquial. Revista Española de Lingüística 29(1), 105–131. ①
Cuza, A.
(2013.) Cross-linguistic influence at the syntax proper: Interrogative subject-verb inversion in heritage
Spanish. International Journal of Bilingualism, 17(1), 71–96. ①②
Enríquez, E.
(1984.) El pronombre personal sujeto en la lengua española hablada en Madrid. Madrid: Consejo Superior
de Investigaciones Científicas. ① ②
Erker, D., & Guy, G. R.
(2012.) The role of lexical frequency in syntactic variability: Variable subject personal pronoun expression
in Spanish. Language, 88(3), 526–557. ①②
Erker, D., & Otheguy, R.
(2016.) Contact and coherence: Dialectal leveling and structural convergence in NYC Spanish. Lingua,
172–173, 131–146. ①
Flores, N., & Toro, J.
(2000.) The persistence of dialect features under conditions of contact and leveling. Southwest Journal of
Linguistics, 19, 31–42. ① ②
Flores-Ferrán, N.
(2002.) Subject personal pronouns in Spanish narratives of Puerto Rican in New York City: A
sociolinguistic perspective. Munich: Lincom. ① ②
Flores-Ferrán, N.
(2004.) Spanish subject personal pronoun use in New York City Puerto Ricans: Can we rest the case of
English contact? Language Variation and Change, 16, 49–73. ①②③④
Hochberg, J.
(1986.) Function compensation for /s/ deletion in Puerto Rican Spanish. Language, 62, 609–21.
Holmquist, J.
(2012.) Frequency rates and constraints on subject personal pronoun expression: Findings from the Puerto
Rican highlands. Language Variation and Change, 24, 203–220. ①②③④
Lapidus, N., & Otheguy, R.
(2005a.) Contact induced change? Overt nonspecific ellos in Spanish in New York. In L. Sayahi, & M.
Westmoreland (Eds.), Selected proceedings of the 2nd Workshop on Spanish Sociolinguistics (pp. 67–
75). Somerville, MA: Cascadilla Press. ①
Lapidus, N., & Otheguy, R.
(2005b.) Overt nonspecific ellos in Spanish in New York. Spanish in Context, 2, 157–174.
Lipski, J. M.
(1994.) Latin American Spanish. London: Longman. ①
Lipski, J. M.
(1996.) Patterns of pronominal evolution in Cuban-American bilinguals. In A. Roca & J. B. Jensen (Eds.),
Spanish in contact: Issues in bilingualism (pp. 159–186). Somerville, MA: Cascadilla. ① ②
Martínez Sanz, C.
(2011.) Null and overt subjects in a variable system: The case of Dominican Spanish (Unpublished doctoral
dissertation). University of Ottawa. ① ② ③ ④ ⑤
Montrul, S.
(2004.) Subject and object expression in Spanish heritage speakers: A case of morphosyntactic
convergence. Bilingualism: Language and Cognition, 7, 125–42. ①②③
Montrul, S., & Slabakova, R.
(2003.) Competence similarities between native and near-native speakers: An investigation of the
preterite/imperfect contrast in Spanish. Studies in Second Language Acquisition, 25(3), 351–98.
①②
Morales, A.
(1997.) La hipótesis funcional y la aparición del sujeto no nominal: El español de Puerto Rico. Hispania
80, 153–165. ①②③④⑤
Orozco, R.
(2015.) Pronominal variation in Colombian Costeño Spanish. In A. M. Carvalho, R. Orozco, & N. L. Shin
(Eds.), Subject pronoun expression in Spanish: A cross-dialectal perspective (pp. 17–37). Washington
DC: Georgetown University Press. ① ② ③ ④ ⑤ ⑥ ⑦ ⑧
Orozco, R., & Guy, G.
(2008.) El uso variable de los pronombres sujetos. ¿Qué pasa en la costa Caribe colombiana? In M.
Westmoreland & J. A. Thomas, Selected proceedings of the Fourth Workshop on Spanish
Sociolinguistics (70–80). Sommerville, MA: Cascadilla Proceedings Project. ①
Ortiz López, L. A.
(2009.) Pronombres de sujeto en el español (L2 vs L1) del Caribe. In M. Lacorte, & J. Leeman (Eds.), El
español en Estados Unidos y otros contextos de contactos: sociolingüística, ideología y pedagogía (pp.
85–110). Frankfurt & Madrid: Vervuert & Iberoamericana. ①
Ortiz López, L. A., Dauphinais, A., & Aponte Alequín, H.
(2017.) Cuban Spanish: Is it a null subject parameter dialect? In A. Cuza (Ed.), Cuban Spanish
dialectology: Variation, contact, and change (pp. 97–117). Washington, DC: Georgetown University
Press. ① ② ③ ④
Otheguy, R., & Zentella, A. C.
(2012.) Spanish in New York. Language contact, dialectal leveling, and structural continuity. Oxford:
Oxford University Press. ①②③④⑤⑥⑦⑧⑨⑩⑪⑫⑬⑭⑮⑯⑰⑱
Otheguy, R., Zentella, A. C., & Livert, D.
(2007.) Language and dialect contact in Spanish in New York: Towards the formation of a speech
community. Language, 83, 1–33. ① ②
Prada Pérez, A. de
(2015.) 1sg subject pronoun expression in Spanish in contact with Catalan. In A. M. Carvalho, R. Orozco
& N. L. Shin (Eds.), Subject pronoun expression in Spanish: A cross-dialectal perspective (121–142).
Washington, DC: Georgetown University Press. ①
Prada Pérez, A. de, & Gómez Soler, I.
(2019.) The Effect of Person on the Subject Expression of Spanish Heritage Speakers. In W. Valentín-
Márquez & M. González-Rivera (Eds.), Dialects from Tropical Islands. Caribbean Spanish in the
United States (pp. ???). Abingdon-on-Thames, UK: Routledge. ①
Ranson, D. L.
(1991.) Person marking in the wake of /s/ deletion in Andalusian Spanish. Language Variation and
Change, 3(2), 133–152. ①
Shin, N. L.
(2012.) Variable use of Spanish subject pronouns by monolingual children in Mexico. In K. Geeslin, &
Díaz-Campos (Eds.), Selected proceedings of the 2010 Hispanic Linguistics Symposium (pp. 130–141).
Somerville, MA: Cascadilla Proceedings Projects. ① ②
Shin, N. L. & Otheguy, R.
(2013.) Social class and gender impacting change in bilingual settings: Spanish subject pronoun use in New
York. Language in Society, 42, 429–452. ①②
Silva-Corvalán, C.
(1982.) Subject expression and placement in Mexican-American Spanish. In J. Amastae & L. Elías-
 Olivares (Eds.), Spanish in the United States: Sociolinguistic aspects (pp. 93–120). Cambridge:
Cambridge University Press. ① ②
Silva-Corvalán, C.
(2001.) Sociolingüística y pragmática del español. Washington, DC: Georgetown University Press. ①
Silva-Corvalán, C.
(1994.) Language contact and change: Spanish in Los Angeles. Oxford: Clarendon. ① ② ③ ④ ⑤
Toribio, A. J.
(2004.) Convergence as an optimization strategy in bilingual speech: Evidence from code-switching.
Bilingualism: Language and Cognition, 7, 165–73. ①②③
Torres-Cacoullos, R., & Travis, C. E.
(2010.) Variable Yo expression in New Mexico: English influence? In S. Rivera-Mills, & D. Villa (Eds.),
Spanish of the U.S. Southwest: A language in transition (pp. 185–206). Madrid & Frankfurt:
Iberoamericana & Vervuert. ①②③④⑤⑥⑦⑧⑨⑩⑪⑫⑬⑭⑮⑯⑰
Travis, C. E.
(2007.) Genre effects on subject expression in Spanish: Priming in narrative and conversation. Language
Variation and Change, 19, 101–135. ①②③④⑤⑥
U.S. Census Bureau
(2010.) Hispanic or Latino by type. Retrieved from [Link] ①
CHAPTER 7

Use of the Present Perfect Indicative in New

York Dominican Spanish
Cecily Corbett, Juanita Reyes & Lotfi Sayahi
University at Albany, SUNY

Abstract
Several studies have examined the functional distribution of the
Present Perfect (PP) in Peninsular and Latin American Spanish. This
study contributes to this line of research through examination of the
morphology and function of PP in New York Dominican Spanish. The
objective is to examine the degrees of diffusion and frequency of use
of this tense in the speech of bilingual speakers. We also examine the
functions of all PP occurrences in interviews with 23 speakers and
explore any innovative uses. Results show that the majority of
participants use PP especially for events that occurred in the indefinite
past or continue to have current relevance, which is the normative
context in which PP is used (Comrie, 1976).

Keywords
Dominican Spanish; Present Perfect; Preterite; Spanish in the US

1. Introduction
The present paper aims to describe the uses of the Present Perfect Indicative (PP)
in Dominican Spanish spoken in New York. We determine the contexts in which
the PP surfaces in this contact variety, in contrast to the uses of the Preterite
(PRET) attested in the same corpus. While the PP has been extensively studied
in Peninsular and Latin American Spanish, there has been less empirical
exploration into its use in Caribbean varieties and much less in the United States.
As such, an additional aim of ours is to expand the discussion of PP form and
function to include attestations in Dominican Spanish, particularly in a context
of intense and extended contact in New York.

2. Background

2.1 Present Perfect in Spanish

In broad terms, and as illustrated in Spanish grammar references (for example

Gómez Torrego, 2002; Moliner, 1984) the Present Perfect (PP) is used for
anterior events that have current relevance as in (1), or events that took place in
the past but without being located in a specific timeframe (Comrie, 1976) as in
(2).

(1)
Yo he vivido mi vida feliz con mi familia ¿no? Y a la edad mía de ochenta
años yo no he ido nunca para el cuartel como arrestado ni he estado preso
nunca, nunca, nunca. [101/250]
I have lived my life happy with my family, have I not? And at my age of
eighty years I have never been to jail, like arrested, nor have I ever been
incarcerated, never.

(2)
Pero el dialecto que yo he escuchado muchos de mis amigos hablar es un
dialecto bonito. [016/340]
 But the dialect I have heard many of my friends speak it’s a pretty dialect.

Gómez Torrego (2002, p. 151) offers the following description of the use of the
PP in Spanish, among a few other uses: “[c]on esta forma verbal nos referimos
también a hechos pasados pero que tienen relación con la zona temporal en la
que se encuentra el hablante.” This is in contrast with the Preterite (PRET)
which is used for telic perfective events that have a specific time reference in the
past. Regarding the use of the PRET, María Moliner’s Diccionario de uso del
español states: “la acción expresada por el [pretérito] indefinido no sólo está
completamente terminada en el momento de enunciarla, sino que su tiempo se
fija generalmente con toda precisión,” (Moliner, 1984, p. 1472). This allows for
the common use of the PRET in narrative discourse as illustrated in (3).

(3)
Entonces decidimos que íbamos a salir a caminar y a, no sé, a comer mangos.
Entonces comenzamos y nos fuimos bien lejos, unas cuantas lomas,
montañitas pequeñas, hasta que llegamos a esta hacienda, encontramos una
mata de mangos enorme, y nosotras queríamos mangos de esta, esta mata.
[038/36]
Then we decided that we were going for a walk and to, I don’t know, eat some
mangos. Then we started and we went very far, a few hills, small mountains,
until we arrived at this ranch, we found a huge mango tree and we wanted
mangos from this, this tree.

However, after stating the main uses for each form, Gómez Torrego
(2002, p. 151) cautions that “[e]sta distinción entre el pretérito indefinido y el
pretérito perfecto no es uniforme en todas las zonas de habla hispana. Por
ejemplo, en Galicia, Asturias, Canarias y gran parte de Hispanoamérica la
forma del pretérito perfecto es muy escasa.” In fact, several studies have
examined variation in the functional distribution of the PP and the PRET in
Peninsular and Latin American varieties of Spanish. It has been widely claimed
that in Peninsular Spanish the PP is showing strong signs of grammaticalization
through its extended use into the perfective domains of the PRET (Schwenter &
Torres Cacoullos, 2008; Howe & Schwenter, 2008; among others).
In her comparative study of the use of the PP and PRET in the Canary
Islands and Madrid, Serrano (1994, 1996) finds that in Madrid the use of the PP
is very high for hodiernal (relating to the present day) and prehodiernal (relating
to a past time earlier than today) events, with younger generations in particular
showing higher rates of use than those of the older generation, especially for
prehodiernal events (Serrano, 1994, p. 48). In the Canary Islands, however,
Serrano (1994) argues that PP use is not spreading to domains of the PRET at
any comparable rates and its use in prehodiernal contexts remains limited. In a
study of the use of the PP in the dialect of Alicante, Schwenter (1994) also
shows that younger speakers have a higher rate of use of the PP in prehodiernal
contexts. He posits that “there exist concrete indications that Alicante Spanish,
like French, will continue to extend the semantic domain of the PP until it
ultimately displaces the Preterite in spoken discourse,” (Schwenter, 1994, p. 80).
In Latin American Spanish, on the other hand, it was claimed earlier on that
the use of the PP is restricted in terms of frequency and domains (Lope Blanch,
1972; Westmoreland, 1988), but corpus-based studies have shown that in some
areas it is in fact gradually spreading to domains of use of the PRET as well
(DeMello, 1994; Howe & Schwenter, 2003, 2008). Howe and Schwenter
(2008, p. 104) compare the rates of use of the PP and PRET in different corpora
of Spanish spoken in Madrid, Lima, and Mexico City and come to the
conclusion that the overall distribution of the PP is much higher in Madrid
(53.6%) than it is in Lima (26.4%) or Mexico City (14.8%). In a different study,
Escobar (2007) cites as possible Quechua influence on Andean Spanish the use
of an evidential function for the PP in contrast with the Pluperfect.
2.2 Present Perfect in US Spanish

While variation in the use of the PP in both Spain and Latin America has been
relatively well-documented, in the case of the PP in Spanish in the United States
there is overall limited research. In the studies that do consider PP in the speech
of Spanish-English bilinguals in the US, there are claims that compound tenses
get reduced in favor of simple tenses. For instance, Zentella (1997, p. 193)
reports on cases where the PRET replaces the PP in the speech of Puerto Rican
speakers in New York. Additionally, in her book chronicling the development of
a range of grammatical features in two siblings’ bilingual development, Silva-
Corvalán (2014) proposes four degrees of tense complexity ranging from 1 (least
complex) to 4 (most complex). She places the PP and the other perfect tenses in
level 4, and reports that in her study there is only one instance of use of the
Spanish PP by either of her subjects by age six. Along with ranking the PP as
one of the more complex structures that her bilingual subjects had to contend
with, she finds no evidence of the possible effect that English might have on the
use of the PP in the Spanish of bilinguals in the United States (Silva-Corvalán,
2014, p. 354).

2.3 Present Perfect in Dominican Spanish

With regards to the PP in Dominican Spanish, not many details are available
about innovative domains of use or any possible extension to the perfective
domains of the PRET. In fact, Alba (2004) clearly delineates a use of the PP that
is in line with the normative description presented above:
Igual que en el resto del Continente, en todos los niveles sociales del país es
notoria la tendencia al uso del pretérito simple (llegó, comiste) en contextos en
los que el español peninsular selecciona el compuesto (ha llegado, has comido).
[…] Lo anterior no significa que el tiempo compuesto del pasado sea
desconocido por los hablantes dominicanos. En varios contextos esa forma
constituye incluso la opción obligatoria, como en las oraciones en las que la
acción iniciada en el pasado se mantiene y continúa durante el momento en que
el hablante la dice,(Alba, 2004, p. 136)
An additional, and notable, feature of the PP in Dominican Spanish that we
address in this paper is the use of the third person form of the auxiliary verb
haber with first person singular instead of the corresponding form: yo ha + past
participle versus yo he + past participle, as shown in (4) and (5).

(4)
Yo ha hablado con los primos míos allá.
I has spoken with my cousins there. [018/121]

(5)
Y yo ha ido allá mucho y a mí me encanta Puerto Plata.
And I has gone there a lot and I really like Puerto Plata. [004/464]

This was reported by Bullock and Toribio (2009, p. 54) as one of the
distinguishing features of Spanish in rural areas of the Dominican Republic.

2.4 The community

According to a Pew Research Center analysis of the Census Bureau’s American

Community Survey (ACS), in 2013 approximately 1.8 million Hispanics of
Dominican origin lived in the United States. This figure demonstrates a
significant growth in the Dominican population in the US – by 2013 the
Dominican-origin population had increased by more than three-fold since 1990.
Along with the overall growth in the Dominican population in the US over the
past few decades, the foreign-born Dominican population in particular grew by
171%. As of 2013, more than half (specifically, 55%) of Dominicans in the US
were foreign-born, which is a relatively high figure compared to 35% of all US
Hispanics and 13% of the overall US population (López, 2015).
With regards to the regional dispersion of Dominicans living in the US,
nearly eight-in-ten are living in the Northeastern United States and 47% live in
New York (López, 2015). In New York City in particular, according to a 2014
report analyzing ACS data and put out by the Center for Latin American,
Caribbean and Latino Studies at the City University of New York, the
Dominican population has surpassed the Puerto Rican population in the City and
Dominicans are now New York City’s largest Latino national origin (Begard,
2014).
Researchers studying Dominican Spanish have observed high degrees of
language loyalty among Dominicans in the US (notably Toribio, 2000a; Toribio,
2000b) and data from the ACS analyzed by the Pew Research Center confirm
these observations – 88% of Dominicans in the US report speaking Spanish at
home, which is a relatively high rate compared to all Hispanic Americans
surveyed, of whom 73% speak Spanish at home. Moreover, 78% of second
generation Dominicans and 96% of adult immigrants of Dominican origin speak
Spanish at home in the US (López, 2015).
The burgeoning Dominican population in the United States, in particular in
New York, bolstered by new arrivals from the homeland coupled with the
formidable language loyalty exhibited by the community in the US makes for a
compelling case study for the use and maintenance of the PP in New York
Dominican Spanish.

3. Methods
Our research objectives are to: (a) explore maintenance versus innovation in the
form of the PP – both in terms of the auxiliary verb haber and the past
participle – in New York Dominican Spanish, (b) describe the contexts in which
the PP surfaces in New York Dominican Spanish, and (c) compare the findings
of (a) and (b) to the results of a parallel analysis of PRET data extracted from the
same corpus. The following subsections detail the speakers included in our data
sample, the data collection and extraction procedures, and the variables
considered and coding criteria employed to meet said objectives.

3.1 Participants

In total, the data analyzed here are extracted from sociolinguistic interviews with
23 participants. All participants speak both English and Dominican Spanish and
at the time of recording were residing in New York State. The sample comprises
data from 10 men and 13 women split into three age groups: 18–24, 25–29, and
30+. Table 1 displays the divisions between these three age groups across both
genders.

Table 1. Age group by gender

Men Women Total

Age 18–24 6 5 11
Age 25–29 2 4 6
Age 30+ 2 4 6
Total 10 13 23

So that we can more aptly examine the effect of contact with English on the
behavior of the PP, we have also divided our sample into groups based on the
participants’ age when they first arrived to the United States, as displayed in
Table 2.
Table 2. Participant age upon arrival to the United States

NY-born Ages 2–12 Ages 13–18 Age 18+ Total

3 12 6 2 23

Roughly half (12/23) of the participants arrived to New York between their early
infancy years and the age of 12, six participants arrived between ages 13 and 18,
and two participants arrived later into their adulthood after age 18. Three of the
23 participants were born in New York and have not lived or attended school in
the Dominican Republic.

3.2 Data collection and extraction

Naturalistic, conversational speech was gathered via sociolinguistic interviews

with English-Dominican Spanish bilinguals living in New York. All of the
interviews lasted between 30 and 60 minutes, were conducted in Spanish, and
were semi-structured – meaning that while the interviewers had a list of
questions to pose to the participants, the participants were able to elaborate as
much or as little as they liked on a given topic. Thus, the resulting recordings
have much more the feeling of a natural conversation rather than a structured
interview. The interview-based conversations centered on daily life in the
Dominican Republic and/or New York, childhood memories, general life
experiences, and hypothetical situations.
Interviews were transcribed and first, all instances of the PP were extracted
for analysis. Each PP token was coded (using the criteria listed in the following
subsection) in order to explore when the PP surfaces in Dominican Spanish in
New York, and to examine maintenance of the PP form in this contact variety.
Additionally, a maximum of 20 tokens of the PRET were extracted from all
speakers who used the PP in their interviews in order to compare the uses of the
PP to those of the PRET in New York Dominican Spanish. All PRET tokens
were coded using the exact same criteria as the PP tokens to facilitate the parallel
analyses.

3.3 Variables and coding criteria

To describe the contexts in which the PP surfaces in New York Dominican

Spanish in contrast to the PRET, the following linguistic variables are
considered: person/number, lexical content of the verb (Otheguy & Zentella,
2012), and evidentiality. We are also including data coding criteria that were
established in Schwenter and Torres Cacoullos (2008), such as the temporal
reference, Aktionsart lexical classes of predicate (based on descriptions detailed
in Comrie, 1976), the presence and type of temporal adverbials in the utterance,
the polarity of the verb, and the type of clause in which the verb appears.
Temporal reference, as described by Schwenter and Torres Cacoullos
(2008, pp. 17–19) is considered as a means to measure temporal distance, which
could play a role in the use of the PP versus the PRET in specific functional
domains. In the present paper, five temporal reference points are considered:
hodiernal (today), hesternal (yesterday), prehesternal (before yesterday),
irrelevant, and indeterminate. While hodiernal, hesternal, and prehesternal
references describe situations that are “temporally anchored to past time
reference points located with respect to utterance time” (2008, p. 18), irrelevant
and indeterminate temporal references do not. An irrelevant temporal reference
refers to a temporal location in the past that cannot be determined by asking the
question: ¿cuándo? Irrelevant temporal references can in many cases correspond
to perfective actions where there is a relational link from the past to the present.
Examples of irrelevant temporal references extracted from our data sample are
displayed in Example (6).

(6)
1. La bandera esa es la más bonita que hay que yo he visto. [009/390]
That flag it’s the most beautiful one there is that I have seen.
2. Muchas personas que viven en la capital tanto como los otros municipios
del país también hablan como muchas personas que viven en el Cibao y
estas personas sin embargo nunca han estado en el Cibao. [013/181]
Many people that live in the capital as well as other municipalities of the
country also speak like people that live in the Cibao and these people
have never been in the Cibao though.
3. Allá, se celebra más allá, por lo entendido, porque no lo he pasado una
navidad allá. [010/447]
There, it’s celebrated more there, as I understand, because I have not
spent Christmas there.

Indeterminate temporal references are found in utterances where the reader or

addressee cannot resolve the temporal distance between the utterance time and
the past situation. Unlike irrelevant temporal references, however, indeterminate
temporal references can be resolved by asking the speaker ¿cuándo? (Schwenter
& Torres Cacoullos, 2008, p. 18). Examples of indeterminate temporal references
we extracted are seen in Example (7).

(7)
1. Y no, era mucho más cómoda, era un mundo en donde cual yo he nacido
y me sentía cómodo. [013/27]
And no, it was a lot more comfortable, it was a world in which I have
been born and I felt comfortable.
2. …también se identifican cada persona de cada región del país, con
 ciertas pronunciaciones, pero ha cambiado yo creo… [012/109]
…also each person is identified from each region of the country, with
certain pronunciations, but it has changed I think.

Aktionsart lexical classes of predicate were coded according to binary

oppositions regarding the stativity, telicity, and punctuality of each token.
Stativity distinguishes states from dynamic situations. As described by Comrie,
“[t]o remain in a state requires no effort, whereas to remain in a dynamic
situation does require effort…” (1976, p. 49). In other words, a dynamic situation
is one that involves some form of movement or change while a stative situation
would not. The quality of telicity differentiates accomplishments from activities.
According to Comrie, telic situations (that is: accomplishments) have an
eventual terminal point whereas atelic situations (or activities) have “no such
terminal point, and can be protracted indefinitely or broken off at any point,”
(1976, p. 44). Regarding punctuality, Comrie describes a durative situation to be
one that “lasts for a certain amount of time (or at least, is conceived of lasting for
a period of time),” whereas a punctual situation is one that “does not last in time
(is not conceived of lasting in time), one that takes place momentarily,”
(1976, pp. 41–42). These predicate lexical classifications were determined based
on the lexical information of the infinitive form, separate from the context in
which each token appeared.
Our analysis considers the use of the PP and PRET with four categories of
temporal adverbials as detailed by Schwenter and Torres Cacoullos (2008).
These categories are proximate adverbs, adverbs of frequency, specific adverbs,
and connective adverbs. Examples of each of these categories are displayed in
Table 3.

Table 3. Examples of adverbials by type (based on Schwenter and Torres

Cacoullos, 2008)

Proximate Frequency Specific Connective

ahora a veces ayer primero

últimamente en ocasiones cuando… antes
esta semana cada año [calendar dates] después
este mes ___ veces [clock times] luego
siempre entonces
nunca al final

Proximate adverbials are those that refer to the “current temporal frame,” or
spans of time that extend until the moment of speech (Dahl, 1985, p. 137).
Frequency adverbials also indicate a situation that has relevance in the present.
According to Schwenter and Torres Cacoullos, proximate and frequency
adverbials should appear in contexts where the PP is the preferred form
(2008, p. 15). Specific and connective adverbials should instead be found in
contexts where the PP is used innovatively in the domain of an alternate tense
because “temporal specification or anchoring to another situation presumably
detracts from (focusing on the result associated with) a current relevance
interpretation,” (Schwenter & Torres Cacoullos, 2008, p. 16, cf. Dahl & Hedin,
2000, p. 395 and Fleischman, 1983, p. 199).
The polarity of the verb was coded with binary options: affirmative or
negative. Negation of a verb has been reported to atelicize the situation, resulting
in one that is continuative (Squartini & Bertinetto, 2000, p. 412). With regard to
the type of clause in which the target structure appears, we coded three
categories: relative clauses, yes/no questions, and all other clause types.
Each instance of the PP and PRET extracted from our sample was coded
according to all of the independent linguistic variables listed, and the
extralinguistic variables corresponding to the speaker who uttered the given
token. The extralinguistic factors that we have coded for include the age, gender,
and place of birth of the participants. We also coded for level of educational
attainment (primary, secondary, tertiary, or post-graduate) and self-reported
Spanish proficiency for all participants, as well as the age at the time of arrival to
the United States. After our data were coded according to dependent, linguistic,
and extralinguistic variables, we ran descriptive analyses using the Statistical
Package for Social Sciences (IBM Corporation, 2013).

4. Results
Extraction yielded 224 instances of PP across 21 of the 23 participants – the PP
was not attested in interviews with two of the participants in our sample. For
comparative purposes, from each of the 21 PP-productive speakers, we extracted
a maximum of 20 instances of the PRET which yielded 402 PRET tokens. First,
we examine the diffusion of PP use across the participant sample. Next, we
examine maintenance of the PP form in New York Dominican Spanish and
explore the contexts in which PP surfaces in this variety. Finally, we compare
attested PP functions to those of the PRET.

4.1 Diffusion of PP use

During the sociolinguistic interviews, each participant had comparable

opportunities to discuss past events, but there exist in the sample quite varied
patterns of use of the PP among each of the participants considered. Figure 1 is a
graphic representation of the diffusion of the PP use across all of the productive
speakers.

Figure 1. The diffusion of PP use across all of the productive participants, split by
participant gender
The bars represent the total number of PP utterances by each speaker, and the
two unproductive speakers are not included. The white bars represent men and
the black bars represent women. The most prolific speaker produced 24 PP
tokens during his interview and the least productive speaker produced only two
during hers. The average use rate for the sample was 10.6 cases per speaker;
6.4/speaker for women and 15.1/speaker for men.

4.2 PP form

Our description of the use of the PP in New York Dominican Spanish considers
not only function but also form. We focus on the morphology of the PP, taking
into account both the auxiliary verb haber conjugation and the realization of the
past participle.

4.2.1 Auxiliary verb

The results from our analysis show that in 92.41% of the cases (207/224), the
auxiliary verb agreed with the subject in person and number. In the remaining
7.58% of all utterances, the third person singular form of haber (ha) was used
with the first person singular subject pronoun, yo. This combination of yo + ha is
attested in the homeland variety of Dominican Spanish (Bullock & Toribio,
2009), and examples of its use in our sample are found in (8) and (9). (8) was
uttered by a 22-year-old female, NY-born participant and (9) was produced by a
26-year-old male participant who arrived to the United States when he was nine
years old.

(8)
Yo dije que- yo nunca ha ido para Puerto Rico. [037/218]
I said that- I has never been to Puerto Rico.

(9)
Y yo siempre ha querido viajar para ver la casa y todo pero nunca tuve la
oportunidad. [016/563]
And I has always wanted to travel to see the house and all but I never had the
opportunity.

The lack of subject-verb agreement was only found in first person singular
utterances and the yo + ha realization occurred in 17 out of the 101, first person
singular utterances in our sample.

Table 4. Frequency in yo + ha realizations out of all first person singular PP uses

NY-born Arrived 2–12 Arrived 13–18 Arrived 18+

4/4 100% 13/68 19.12% 0/23 0/6

As seen in Table 4, despite the low number of tokens, the yo + ha realization was
categorical with our NY-born participants. Although the yo + ha realization is
also attested in the group of participants who arrived to NY between the ages of
two and 12, it is much less frequent at 19.12%.

4.2.2 Past participle

Along with person/number agreement in the auxiliary verb, we examined the
formation of the past participle. Our sample contained 185 utterances of regular
participles and 39 instances of irregular participles. Two irregular participles
were uttered with an incorrect morphology, and both instances were found in the
speech of NY-born participants. These two utterances account for 0.89% of our
entire PP sample. Example (10) is one of these cases.

(10)
…le preguntó si alguien ha morido y ha dejado a ella. [025/510]
…he asked her if someone had died and left her.

Here, the participant regularized the past participle, resulting in “si alguien ha
morido” instead of using the irregular muerto.1 In sum, considering that the
auxiliary verb disagreement phenomenon has been attested in the reference
dialect, and that only two past participles were problematic, it is fair to argue that
the morphology of the PP in Dominican Spanish in New York is preserved
almost intact.

4.3 PP function

Looking at the factors that condition the use of the PP by our informants, we
note that in 57.58% of the PP utterances, speakers do not use a temporal adverb.
But in the cases where they do, 78.57% (77/99) of the time they are using the PP
with an adverb of proximity or frequency. Regarding temporal reference, we find
that 15.62% (35/224) of all PP utterances were used in either a hodiernal or
prehesteral context. In turn, the results for the PRET show that 34.9% (140/402)
of the tokens were used to refer to hodiernal or prehesteral events. On the other
hand, when the time reference is irrelevant, the PP is used in 74.6% (167/402) of
the cases while use of the PRET, as expected, reaches only 20.9% (84/402). This
indicates that Dominican Spanish speakers in New York show some extension of
the PP to hodiernal and prehesternal contexts in which one could anticipate
PRET as the unmarked form in general Spanish. In terms of other linguistic
factors, we also find that the PP was used twice as often with durative situations
than with punctual ones. Finally, regarding polarity we find that both the PP and
the PRET were used more frequently in affirmative sentences than negative
ones.
Considering extension of the PP to other contexts, our results show that the
older speakers were at their time of arrival to the United States, the less they
extend the PP to other contexts (including non-PRET contexts) as displayed in
Table 5. Results from a Pearson Chi-Square test demonstrate that there is a
statistically significant relationship between informants’ age of arrival and the
extension of the PP into contexts of innovative use (p = .014 when p < .05 is
considered to be significant).

Table 5. Frequency of expected uses versus innovative extension by age of arrival

group

Age of arrival Expected use Innovative use Total

NY-born 11/17 64.71% 6/17 35.29% 17/17 100%

Age 2–5 35/45 77.78% 10/45 22.22% 45/45 100%
Age 6–12 80/87 91.95% 7/87 8.05% 87/87 100%
Age 13–18 43/51 84.31% 8/51 15.69% 51/51 100%
Age 18+ 23/24 95.83% 1/24 4.17% 24/24 100%

An example that we determined was extension to the domain of the PRET is

displayed in (11). Here, the speaker is using the PP with a perfective event in a
specific timeframe in the past.

(11)
Porque es una forma de expandir tu cultura y mientras más interacciones tú
 tienes con diferentes países – como yo, he ido a Nicaragua el año pasado.
[030/140]
Because it is a way to expand your culture and the more interactions with
different countries you have – like me, I have gone to Nicaragua last year.

Other cases of extension include examples such as (12), where the speaker, a
New York-born participant, uses the PP instead of the Pluperfect. Further
extension was noted in cases where the PP is used in a context where the Present
Indicative would be the unmarked choice, as in (13).

(12)
…mi abuelo ha muerto hace dos años, pero mi Mamá estaba muy cerca de él,
y estaba muy triste siempre… [025/505]
…my grandfather has died two years ago but my mother was very close to
him, and she was always very sad.

(13)
…es muy difícil para ellos verdaderamente absorber el idioma inglés en un
cien por ciento o en un nivel en el cual ellos puedan decir todas las palabras
que ellos han conocido desde que nacieron. [013/267]
…It is very difficult for them to truly absorb the English language one
hundred percent or to a level in which they can say all the words they have
known since they were born.

5. Conclusion
The principal goals of this paper were to describe the overall use of the PP by
New York Dominican Spanish speakers in the context of contact and then to
determine if use of the PP is, as seen in other varieties of Spanish, extended to a
non-sequential perfective function at the expense of the PRET. In terms of
overall use of the PP in New York Dominican Spanish, we found that men
tended to use the form more than women in conversations about their life history
and personal experiences, which is further confirmed by our finding that the
unproductive speakers in our sample are women. We also find that the PP is
most frequently used in perfective clauses that contain proximate or frequency
adverbs. With regards to extension to domains of the PRET and to sequential
discourse, we find that such extension is limited. Cases where the PP was used
instead of other tenses by New York Dominican Spanish speakers represent only
4.46% (10/224) of the tokens produced.
Earlier claims that all perfective tenses should be placed at the highest
degree of morphosyntactic complexity, such as the proposal of Silva-Corvalán
(2014), do not seem to imply a more drastic reduction of the PP in the speech of
New York Dominican Spanish speakers than that of other tenses. That said, those
informants born in New York did tend to on average use the PP with relatively
less frequency than their later-arriving counterparts. As noted above, interviews
with NY-born informants yielded on average 5.66 utterances of the PP per
speaker, while interviews with other bilingual speakers who moved to NY later
in their childhood provided on average 200% more PP utterances. Since the PP is
listed as one of the most complex structures one is faced with during acquisition
and since Silva-Corvalán (2014) reports only one instance of PP use by her
informants before age six, it stands to reason that our informants who remained
in monolingual Spanish settings (i.e. the Dominican Republic) for longer have
perhaps a more precise command of the use of this form. Notwithstanding that
the NY-born informants tended to use the PP with much lower frequencies and
demonstrated minor morphological variation, our collective findings confirm
that speakers of New York Dominican Spanish maintain not only the PP
morphology, but also its canonical use in the majority of the cases.
Note
1. The only other instance of a missed target in the formation of past participles was a regularization of the
infinitive verb decir expressed as decido instead of the irregular past participle dicho. ①

References
Alba, O.
(2004.) Cómo hablamos los dominicanos: Un enfoque sociolingüístico. Santo Domingo: Grupo León
Jiménes. ① ②
Begard, L. W.
(2014.) Have Dominicans surpassed Puerto Ricans to become New York City’s largest Latino nationality?
An analysis of Latino population data from the 2013 American community survey for New York City
and the Metropolitan Area. New York, NY: Center for Latin American, Caribbean, and Latino Studies.
①
Bullock, B. E., & Toribio, A. J.
(2009.) Reconsidering Dominican Spanish: Data from the rural Cibao. Revista Internacional De
Lingüística Iberoamericana, 7.2(14), 49–73. ① ②
Comrie, B.
(1976.) Aspect. Cambridge: Cambridge University Press. ① ② ③ ④ ⑤ ⑥
Dahl, Ö.
(1985.) Tense and aspect systems. Oxford: Blackwell. ①
Dahl, Ö. & Hedin, E.
(2000.) Current relevance and event reference. In Ö. Dahl (Ed.), Tense and aspect in the languages of
Europe (386–401). Berlin: Mouton de Gruyter. ①
DeMello, G.
(1994.) Pretérito compuesto para indicar acción con límite en el pasado: Ayer he visto a Juan. Boletín de
Real Academia Española, 4, 611–633. ①
Escobar, A. M.
(1997.) Contrastive and innovative uses of the present perfect and the preterite in Spanish in contact with
Quechua. Hispania, 80(4), 859–870. ①
Fleischman, S.
(1983.) From pragmatics to grammar: Diachronic reflections on pasts and futures in Romance. Lingua, 60,
183–214. ①
Gómez Torrego, L.
(2002.) Gramática didáctica del español. Madrid: SM Ediciones. ① ② ③
Howe, C., & Schwenter, S.
(2003.) Present perfect for preterite across Spanish dialects. University of Pennsylvania Working Papers in
Linguistics, 9(2), 7. ①
Howe, C. & Schwenter, S.
(2008.) Variable constraints on past reference in dialects of Spanish. In M. Westmoreland & J. A. Thomas
(Eds.) Selected proceedings of the Fourth Workshop on Spanish Sociolinguistics (100–108).
Somerville, MA: Cascadilla Proceedings Project. ① ② ③
IBM Corporation
(2013.) IBM SPSS statistics for Windows, Version 22.0. [Computer program]. Armonk, NY: IBM Corp.
Lope Blanch, J.
(1972.) Estudios sobre el español de México. Mexico City: Universidad Nacional Autónoma de México.
López, G.
(2015.) Hispanics of Dominican origin in the United States 2013 Washington, DC: Pew Research Center.
①②③
Moliner, M.
(1984.) Diccionario de uso del español, Tome II. Madrid: Gredos. ① ②
Otheguy, R., & Zentella, A. C.
(2012.) Spanish in New York: Language contact, dialectal leveling and structural continuity. Oxford:
Oxford University Press. ①
Serrano, M. J.
(1994.) Del pretérito indefinido al pretérito perfecto: un caso de cambio y gramaticalización en el español
de Canarias y Madrid. LEA: Lingüística Española Actual, 16(1), 37–58. ① ② ③
Serrano, M. J.
(1996.) Accounting for morpho-syntactic change in Spanish: The present perfect case. University of
Pennsylvania Working Papers in Linguistics, 3(1), 51–61. ①
Schwenter, S.
(1994.) The grammaticalization of an anterior in progress: Evidence from a peninsular Spanish dialect.
Studies in Language, 18(1), 71–111. ①②
Schwenter, S., & Torres Cacoullos, R.
(2008.) Defaults and indeterminacy in temporal grammaticalization: The ‘perfect’ road to perfective.
Language Variation and Change, 20, 1–39. ①②③④⑤⑥⑦⑧⑨
Silva-Corvalán, C.
(2014.) Bilingual language acquisition: Spanish and English in the first six years. Cambridge: Cambridge
University Press. ①②③④
Squartini, M., & Bertinetto, P. M.
(2000.) The simple and compound past in Romance languages. In Ö. Dahl (Ed.), Tense and aspect in the
languages of Europe (pp. 403–439). Berlin: Mouton de Gruyter. ①
Toribio, A. J.
(2000a.) Language variation and the linguistic enactment of identity among Dominicans. Linguistics,
38(5), 1133–1159. ①
Toribio, A. J.
(2000b.) Nosotros somos dominicanos: Language and self-definition among Dominicans. In A. Roca (Ed.),
Research on Spanish in the United States. Linguistic issues and challenges (pp. 252–270). Somerville,
MA: Cascadilla Press. ①
Westmoreland, M.
(1988.) The distribution and the use of the present perfect and the past perfect forms in American Spanish
Hispania, 71(2), 379–384. ①
Zentella, A. C.
(1997.) Growing up bilingual: Puerto Rican children in New York. Malden, MA: Blackwell. ①
CHAPTER 8

Transfer and convergence between Catalan

and Spanish in a bilingual setting
Amelia Jiménez-Gaspar1, Acrisio Pires2 & Pedro Guijarro-
Fuentes1
1
Universitat de les Illes Balears | 2 University of Michigan

Abstract
We present results from a production study of forty-five bilingual
Catalan-Spanish speakers (26 women and 19 men, aged 16 to 65 years
old), to determine whether there is transfer or convergence between the
two languages in a bilingual setting. The participants are residents of
the capital and several villages of Majorca (Spain).
We provide evidence that the production of third-person clitics in
Majorcan Catalan is clearly affected by bilingualism with Spanish.
First, there is a pattern of partial transfer in the use of the neutral clitic
ho in Majorcan Catalan, which shows a semantic extension matching
Spanish lo. Second, the bilinguals show a pattern of syllabic (CV(C))
forms of pronominal clitics, favored by a similarity with Spanish
forms. This provides evidence of both morphophonological and
semantic transfer in the knowledge and production of bilinguals who
acquired both languages in childhood.
 Keywords
pronominal clitics; bilingualism; language contact and change;
Majorca; social variables

1. Introduction
In this study, we examine the production of clitics by bilingual speakers of two
closely related romance languages, Catalan and Spanish. Pronominal clitics have
not raised a lot of interest in the Spanish-Catalan bilingual context (but see
Jiménez-Gaspar, Pires, & Guijarro-Fuentes, 2017; Perpiñán, 2017) especially in
the case of Majorcan (or Balearic varieties), despite the extensive literature
related to their syntactic or morphophonological features (e.g. Bonet, 1995;
Fernández-Ordóñez, 2001; Larrañaga & Guijarro Fuentes, 2012).
Blas Arroyo (2011) states that in bilingual situations the majority language
often influences the minority one. We attempt to determine whether this is the
case between Spanish (the majority language in this contact situation) and
Catalan (the minority language), in the case of adults who became bilingual in
childhood. This has also been extensively observed in the context of heritage
bilingualism and when one of the languages is identified as the weaker language
(e.g. Guijarro-Fuentes, Pires, & Nediger, 2017; Meisel, 2011; Montrul, 2004,
2008; Pires & Rothman, 2009; Silva-Corvalán, 1994). More specifically, if there
are linguistic transfers between Spanish and Catalan so that the two languages
become more structurally similar in one or more linguistic domains, Blas Arroyo
(2011: 379) explains that “the process of linguistic integration would mean that
Catalan forms would have to adapt, to a greater or lesser extent, to the
morphophonological and grammatical rules of Spanish.” We also aim at
determining whether these effects arise in the opposite direction, from Catalan to
Spanish.
The bilingual participants in the study lived in Majorca, one of the Balearic
Islands in the Mediterranean. Therefore, we will refer to their dialects of the two
languages as Majorcan Catalan and Majorcan Spanish (henceforth MS and MC).
The focus of this paper is the investigation of the speakers’ production of third-
person non-reflexive pronominal clitics in the two languages (e.g. le ‘him/it’, la
‘her’ in Spanish). This empirical scope is due to the fact that the grammatical
properties of third person non-reflexive clitics set them apart from first, second
person and reflexive clitics both in Catalan and Spanish. For instance, only third-
person non-reflexive clitics show distinctive forms based on gender (e.g. lo/la in
Spanish and el/la in Catalan).
Even though Catalan and Spanish are closely related Romance languages,
their pronominal systems differ in various respects regarding the properties of
third-person pronominal clitics. For instance, Catalan shows a wider range of
pronominal clitic forms than Spanish, mostly depending on the position that the
pronominal clitics occupy in relation to the verb (proclitics or enclitics). In our
analysis of the production data by bilingual speakers of Catalan and Spanish, we
try to determine whether there is a transfer of features between the Catalan and
Spanish clitic systems due to bilingualism between the two languages in
Majorca. Besides, we attempt to specify the direction of such effects. If the
transfer is unidirectional, the evidence of interference will occur only in one of
the two languages, which will show a similar property to the other language.
Alternatively, both languages may show evidence of transfer from each other,
which would then constitute a case of convergence (see e.g. Thomason,
2001, p. 89). In order to determine whether Majorcan Catalan and Spanish show
evidence of such changes specifically in the context of Majorca, we analyze the
data from our participants of Majorca in comparison with core properties of the
Spanish and Catalan varieties from Barcelona (part of Oriental Catalan, like
MC), also taking into account previous studies on the clitic system of Catalan
(Bonet & Lloret, 2005; Bonet, 2002; Perea, 2012; Batllori et al., 2004, among
others). If the bilingual data representing MS and MC show different properties
from Peninsular Spanish and Catalan, respectively, we will then address whether
such differences in MS/MC could represent instances of language change
resulting from interference between the two languages in Majorca.
We also attempt to determine whether the linguistic preference of the
speakers for either language (for Spanish, Catalan or a lack of preference), and
social variables (gender, age, education level, and area of residence in Majorca)
affect the properties of clitics in their production of MC and MS.
This paper is organized as follows: in Section 2, we present some historical
information regarding Catalan and Spanish in Majorca. In Section 3, we briefly
review some aspects of the linguistic analysis adopted in this study for third-
person pronominal clitics in both languages, considering Peninsular Central
Catalan as spoken in Barcelona (henceforth Peninsular Catalan/CC), and
Peninsular Spanish. These properties are then used later as the baseline to which
we compare Majorcan Catalan and Spanish to determine whether they show
evidence of dialectal differences that could result from changes affecting only
the bilingual context of Majorca, but not necessarily the Peninsula. Section 4
briefly discusses previous studies regarding these clitics in Catalan and Spanish
in the Balearic Islands, considering in particular Majorca. Section 5 addresses
our research questions and hypotheses. Sections 6 to 8 present our methodology
and analysis of experimental results, determining to which extent there is
evidence from Majorca for transfers between Catalan and Spanish in the 3rd-
person pronominal clitic system. The last section summarizes the main
contributions of this paper.
2. Language contact in Majorca: Majorcan Catalan and
Spanish
The historical record indicates that (Old) Catalan was spoken in the Balearic
Islands as early as the 13th century, when people from areas around Barcelona
and Girona occupied the islands under King Jaime I, as part of the Christian
Reconquest of Spain from the Muslims. From 1229 onward, Catalan went on to
become the main language of the islands, although other languages such as
Arabic, Hebrew and Aragonese continued to be spoken there. Spanish arrived on
the islands in the 15th century, due to the union of the dynasties of Castile and
Aragon, after which “Catalan-speaking territories fell under increasing cultural
pressure and underwent a gradual process of linguistic castilianization”
(Enrique-Arias, 2010, p. 103), although Catalan never ceased to be spoken in
Majorca. Enrique Arias suggests that the situation of contact between Catalan
and Spanish in Majorca further changed since the eighteenth century, due to the
Bourbon Dynasty, and since then Spanish became the language of prestige,
replacing Catalan in all administrative, educational, religious and commercial
areas. Nevertheless, Catalan gained official status thanks to the Law of
Linguistic Normalization (Llei de normalització lingüística) of 1986. Amengual
(2011a, p. 216) argues that the situation of Catalan has been different in Majorca
in comparison with other regions where Catalan is spoken. He argues that the
speaker’s place of origin and place of residence are specially important variables
to explain the use of Catalan:
Census data (INE, 2001) shows that Catalan has not extended to all the linguistic
spheres in the Balearic Islands; the Spanish-dominant group, and especially the
immigrant group composed of speakers born outside of Spain, have the lowest
percentages of Catalan use, which has also been supported by recent linguistic
work (Blas-Arroyo, 2007; Melià, 2002). In addition, the data collected by the
Government of the Balearic Islands also shows that the use of Spanish or
Catalan varies depending on the specific island and that Catalan is present and is
used at varying degrees in urban versus rural areas.
Amengual (2011a) adds that, according to the data collected by the Government
of the Balearic Islands, Catalan is more used in Majorca and in Minorca, than in
Ibiza; and in Majorca it is more used in inland/rural areas than in Palma, the
capital. This is the reason why, in the current study, we consider data from Palma
(the capital of Majorca), but also from the villages of Llucmajor, Capdepera,
Llubí and Soller, with the goal to assess whether the speakers’ linguistic
preference and their place of residence play a significant role in their production
of 3rd person object pronominal clitics.

3. Spanish and Catalan pronominal clitics

In this section we review some of the core properties of the clitic systems of
Catalan and Spanish that will be relevant as background for our analysis of the
production of clitics by Catalan-Spanish bilinguals in Majorca. On the basis of a
comparison between these core properties and our data from bilinguals, we will
determine whether the properties of third-person clitics in Majorcan Catalan and
Spanish show evidence of transfer.
Pronominal clitics in Catalan and Spanish represent the grammatical direct
or indirect object of the verb and appear next to the verb, either as proclitics,
before the verb, or as enclitics, after the verb.1 Unless otherwise specified, this
same general analysis and other features below are shared between Spanish and
Catalan pronominal clitics.
The clitic position depends on verb morphosyntax, that is, if the verb is
inflected (finite), clitics must appear as proclitics, as shown in (1) for Spanish
and for Catalan; however, if the verb is uninflected, that is, an infinitive (2) or a
gerund, or if it is an affirmative imperative (4), the clitic normally appears after
these forms; if the embedded verb is uninflected, the clitic may raise to a
proclitic position to a higher verb in the clause, in cases of restructuring (3):

(1)
a. Quiero el coche = lo quiero / *quiero lo (Spanish)

b. Vull el cotxe = el vull / *vull lo (Catalan)

[Link] . the car = [Link] [Link]

‘I want the car = I want it’

(2)
a. Voy a comprarlo (el coche) (Spanish)

b. Vaig a comprar-lo (el cotxe) (Catalan)

[Link] to [Link]

‘I am going to buy it’

(3)
a. Lo voy a comprar (el coche) (Spanish)

b. El vaig a comprar (el cotxe) (Catalan)

[Link] [Link] to buy

(4)
a. Compra el coche = cómpralo / *lo compra (Spanish)

b. Compra el cotxe = compra’l / *el compra (Catalan)

[Link] . the car =[Link] .-[Link]

 ‘Buy the car = buy it

Pronominal clitics can be divided into distinct sets of forms. Third-person clitics
include separate reflexive and non-reflexive forms. They also show distinct
accusative (5) and dative clitics forms in Spanish and Catalan (6). Third-person
non-reflexive accusative clitics show morphological number (and so do the
dative ones) and gender distinctions (see Sections 3.1 and 3.2 for details,
especially Table 1 for Catalan, and Tables 2a and 2b for Spanish):2

(5)
a. Ayer lo compré (lo = tu regalo) (Spanish)

b. Ahir el vaig comprar (el = el teu regal) (Catalan)

Yesterday [Link] bought (it= your present)

‘I bought it yesterday.’

(6)
a. Le di un regalo a él (Spanish)

b. Li vaig donar un regal a ell (Catalan)

[Link] [Link] a gift to him.

‘I gave a present to him.’

3.1 Third-person clitics in Catalan

For Catalan, we will consider Peninsular Catalan3 (spoken in Barcelona) as a

comparative baseline. Table 1 presents the different forms of third-person non-
reflexive clitics in Peninsular Catalan, adapted from Bonet & Lloret (2005). A
relevant property of these clitics are their alternative forms, which may also
follow different patterns of syllabication in the language (Bonet, 2002; Bonet &
Lloret, 2005): Bonet and Lloret (2005, p. 40) explain that CC(i)-clitics are clitics
composed by [at least] “one consonant, and variable appearance of schwa plus
variable consonant deletion”. V-clitics are the forms with only one vowel, such
as ho ([u], neuter), or hi ([i], locative). CV(z)-clitics are formed by one
consonant, “followed by a vowel representing a morpheme, with an optional
additional plural morph”. Finally, 3rd person masculine accusative clitics allow
allomorphy encompassing both CC(i)-clitics and CV(z)-clitics.

Table 1. Third-person (non-reflexive) Barcelonan Catalan clitics (Adapted from

Bonet & Lloret, 2005, p. 40)

Clitic type Label Phonetic alternations Citation form

CC(i)-clitics 3rd dat. pl. [elzi], [lzi] elzi

V-clitics neuter [u] ho
CV(z)-clitics 3rd dat. sing. [li] li
3rd acc. fem. sing. [lə], [l] la
3rd acc. fem. pl. [ləz] les
3rd masc. acc. clitics 3rd acc. masc. sing. [əl], [lu], [l] el
3rd acc. masc. pl. [əlz], [luz], [lz] els

In our analysis of the Majorcan production data, we will investigate to which

extent Majorcan Catalan clitics are distinct from the Peninsular Catalan forms
and will attempt to determine whether those differences are due to interferences
from Spanish in the Majorcan bilingual context.

3.2 Third-person clitics in Spanish

Regarding Spanish non-reflexive pronominal clitics in the third-person, there are

two pronominal systems, a so-called etymological system that is characterized
by a differential case (accusative and dative), and a referential system, which
does not show a distinction of case in the form of the pronoun, but of gender
(masculine and feminine). The etymological and the referential pronominal
systems of Spanish are shown in Tables 2a and 2b, respectively:

Table 2a. Etymological 3rd-person, non-reflexive pronominal system in Spanish*

Case Masculine Feminine Neutral Syntactic function

Accusative lo(s) la(s) lo direct object

Dative le(s) indirect object

* Tables 2a and 2b adapted from Fernández-Ordóñez (2001, p. 10, 12). ①

Table 2b. Referential 3rd-person, non-reflexive pronominal system in Spanish

Case Masculine Feminine Neutral Syntactic function

Accusative and Dative le(s)/ lo(s) la(s) lo direct object or indirect object

The referential system is found primarily in the Central and the Western parts of
Castile, more specifically from the south of the Cantabrian Mountains to La
Mancha (Fernández-Ordóñez, 2001), whereas the etymological system appears
to be used in the rest of Spain, and therefore in the Balearic Islands as well.
However, the properties of these clitics have not been analyzed in depth in the
Balearic Islands (see Section 4).
The referential system of Spanish can also show the phenomena well-
known as leísmo, laísmo, and loísmo. Leísmo is the linguistic process by which
speakers use the pronominal clitics le/les to refer to the direct object, instead of
lo/los (in fewer cases la/las). Likewise, there are two types of leísmo, one of
them involves animate reference (7) and the other involves inanimate reference
(8). In laísmo and loísmo speakers use the pronominal clitics la/las or lo/los (this
change is more unusual) instead of le/les to refer to the indirect object, when the
antecedent is feminine or masculine, respectively, as in (9) and (10)):4

(7)
(*)Le veo todos los días
[Link] see every day
‘I see him every day’

(8)
El coche, siempre le aparca en el garaje
the car, always [Link] park in the parking lot
‘The car, (s)he always parks it in the parking lot.’

(9)
(*)La dije a ella la verdad
[Link]-FEM told to she the truth

(10)
(*)Lo dije a él la verdad
[Link]-MASC told to him the truth
‘I told her/him the truth.’

Although the etymological pronominal system is also taken to be the one used in
Majorca, there are no previous studies that investigate in detail Spanish-Catalan
bilingual production in Majorca, to determine whether it shows the same
properties of either pronominal system of Peninsular Spanish. By using the
pronominal system properties above as a comparative baseline, our analysis will
determine whether Majorcan Catalan-Spanish bilinguals show different
properties in their Spanish and Catalan production, and to which extent such
different properties may result from interferences or transfers.
4. Previous studies
Although there is extensive descriptive and theoretical research on the clitic
system of Spanish (e.g. Bosque & Demonte, 1999; Uriagereka, 1995), there has
been less theoretical research on the formal properties of clitics in Catalan and
its different dialects (but see e.g. Gavarró, 1991; Martín, 2012; Perea, 2012;
Todolí, 2002; Viaplana, 1980; Wexler et al., 2004; Perpiñán, 2017; Batllori et al.,
2004).5 Bonet (1991, 1995) proposes a more extensive formal analysis of
different clitic forms.
Roca (1992) indicates that there is substantial interest in the third-person
non-reflexive pronominal clitics, given that they vary regarding various features,
including case, gender, and number, as discussed in Section 3. However, to the
best of our knowledge, there has been no previous detailed study of the third-
person pronominal clitics in the context of adult Catalan-Spanish bilingualism.
Jiménez-Gaspar, Pires, & Guijarro-Fuentes (2017) analyze other pronominal
clitics in adult Catalan-Spanish bilingualism, and Perpiñán (2017) focuses on the
expression of Catalan adverbial clitics en ‘of’ and hi ‘locative’ in the adult
grammar of Catalan-Spanish bilinguals. However, as Perpiñán indicates, the
Catalan adverbial clitics have no grammatical equivalent in Spanish, at least as
overt clitic elements. Therefore, they differ from the pronominal clitics
considered in the present study and in Jiménez-Gaspar et al. (2017), which have
overt counterparts in both Catalan and Spanish.
Perea (2012) proposes an analysis of Catalan clitics considering dialectal
differences between Balearic Catalan (BC) and Peninsular Catalan (CC).
However, she relies primarily on data collected and described by Alcover (1916)
and other work from the same period, which lacks details about the properties of
different dialects as currently spoken. Among the properties she identifies in
MC, she explains that les (third-person DO plural feminine pronominal clitic)
can be used for both accusative and dative cases;6 speakers use the feminine
accusative pronominal clitic les instead of the dative els to refer to the indirect
object, representing a counterpart to Spanish laísmo, as shown in the example
below:

(11)
Les ne duràs dues (a elles, dues pomes) (BC)
[Link] [Link] .partitive [Link] two (to them, two apples)
Els ne duràs dues (CC)
‘You will bring them two apples’. (Example adapted from Perea 2011: 124)

Another relevant pattern of variation that will be relevant for the analysis of our
data concerns the variant forms elze [əlzə]/elzi [əlzi], as in (12). Perea (2012)
takes this pattern to involve an epenthetic vowel after the plural form, a common
view to earlier analyzes.

(12)
Elze/elzi agrada el xocolata7 (MC, bilingual subject)
[Link] please the chocolate
els agrada la xocolata8 (CC)
‘They like chocolate.’

Such an analysis also relates to work by Bonet (1991, 1995, and 2002) and
Bonet & Lloret (2005), who consider extensively the analysis of third-person
pronominal clitics in different dialects of Catalan. In particular, Bonet (1991)
proposes that the dative singular clitic /li/ takes the form /i/ when it is combined
with an accusative clitic (els [əlz]), as indicated in (13)–(14) (from Bonet,
1995, p. 641):

(13)
 Ses pomes, a s’al·lot [əlzi] donaré més tard
The apples, to the child [Link] -[Link] [Link] later
‘The apples, I will give them to the child later. ’

(14)
Els llibres, a en Quim [əlzi] donaré
The books, to the Quim [Link] -[Link] [Link]
‘The books, I will give them to Quim.’

Later, Bonet (1995, 603) considers similar examples and states:

Even though the output clitic form looks like the third person plural dative clitic
in isolation, /lzi/ (the schwa being epenthetic), the antecedent of the dative clitic
is singular, not plural (a en Quim). […] the plural marker /z/ in the output form
has to come from the accusative (plural) source (the antecedent being plural: els
llibres ‘the books’).
Other authors adopt an analysis in which an epenthetic vowel is added to the
dative clitic, to yield [əlzi]. For Boeckx & Martín (2013, p. 13):
Speakers do not use the normative els as the third person plural dative clitic.
Rather, speakers add a vowel [i] to the normative form, perhaps by analogy with
the dative singular l-clitic that ends in [i]. Thus, the dative plural l-clitic ends up
pronounced as [əlzi].
However, Wheeler et al. (1999) argue that the use of [əlzi] is a reanalysis of the
singular indirect object pronoun li, and they specify that this use differentiates
the third-person plural masculine direct object els from the third-person plural
indirect object [əlzi]. Seguí Trobat (2014) also argues that these uses are a
reanalysis in analogy to the singular li, for the same reason.
In our analysis of the production data in this paper we will determine which
ones of the MC third person clitic forms match the forms from Peninsular
Catalan presented above and in Section 3.1. In particular, our data regarding the
occurrence of plural dative/accusative plural clitic els and its variants (e.g. elzi,
elze) will also support an analysis in which the epenthetic vowel [i] or [ə] is
added.
Bonet (1995, p. 603) adds that the third person accusative clitic may not
show feminine agreement, although a feminine marker is present when this clitic
appears in isolation and the antecedent is feminine. She also argues that, since
Catalan has an independent neuter clitic ho /o/ (or /u/), the third person
accusative clitic can take the form of the neuter clitic through a specific
morphological process. For Bonet, this accounts for parallel occurrences such as
[əlzo] in (15):

(15)
Ses pomes, a els al·lots [əlzo] donaré més tard
The apples, to the children [them] [Link] later
‘The apples, I will give them to the children.’

In the next sections, we investigate whether there are any effects of Catalan-
Spanish bilingualism on the production of third-person non-reflexive clitics by
Spanish and Catalan bilinguals in Majorca. In particular, in Section 7 we will
discuss the properties of third-person non-reflexive clitics in Catalan as spoken
by the Catalan-Spanish bilinguals in Majorca, considering results related to the
properties discussed in this and the previous section.

5. Research questions and hypotheses

Previous studies on bilingualism and language contact (e.g. Blas Arroyo, 1998,
2011; Seib 2001; Sinner & Wesch, 2008; Silva-Corvalán, 1994; Poplack, 1993;
Thomason, 2001; Thomason & Kaufman 1988) address the possibility of
transfer or convergence between two languages in a bilingual setting. In this
context, we explore two research questions:
1. Are there distinct properties or variation in the Majorcan Spanish (MS) and
Majorcan Catalan (MC) clitic systems regarding the third-person (non-
reflexive) clitics that can be evidence of transfer or convergence between
the two languages, as produced by bilingual Catalan-Spanish speakers in
Majorca?
2. If such distinctions or variation arise, do extralinguistic/social variables,
such as the speakers’ linguistic preference or zone of residence, affect clitic
production by these bilingual speakers?

With respect to the first question, to the best of our knowledge there are no
previous studies that have explored in detail the existence of transfers or
convergence between Catalan and Spanish in the pronominal system (but see
Jiménez-Gaspar, Pires, & Guijarro-Fuentes, 2017 and references therein).
Transfer usually corresponds to individual features that are transferred
unidirectionally from one language to the other, whereas convergence
corresponds to cases in which both languages show some innovation and
become more similar to each other (see e.g. Blas Arroyo, 1998, 2004, 2011;
Thomason, 2001). We treat the process of transfer more precisely as involving
reanalysis by learners in their grammar of one language as the result of their
acquisition/knowledge of the other language.
If there is evidence of transfer or convergence, it is possible that it will take
place in the morphology and syntax, with the use of the same form (in this case,
the same pronominal clitic) in both languages due to their influence on each
other, but it may also involve semantic features. For example, we will consider
whether there is an influence from Spanish to Catalan regarding the neutral third
person clitic, since in Spanish there is only one form (lo) for two different
referents: (i) the singular masculine (16) and (ii) the neutral ones (17).
(16)
He visto el coche que quieres comprar
1. [Link] the car which [Link] buy
‘I have seen the car that you want to buy.’
Lo he visto
2. [Link] . [Link] [masculine]
‘I have seen it.’

(17)
Juan me ha dicho que no puedes venir a mi fiesta
1. John [Link] have-told that not [Link] came to my party
Juan me lo ha dicho
2. John [Link] [Link] [Link] [neutral]
John has told it to me

Nevertheless, Catalan has distinct pronominal clitics for these two options:

(18)
Compraré aquest cotxe
1. [Link] this car
I will buy this car
El compraré
2. [Link] [Link] [masculine]
I will buy it

(19)
El Joan ha dit que no vindrà demà
1. John [Link] that not [Link] tomorrow
‘John has said he won’t come by tomorrow.’
 El Joan ho ha dit
2. John [Link] [Link] [neutral]
‘John has said it.’

Regarding the second question, if there is evidence of transfer, we expect the

linguistic preference – that is, if participants prefer using Spanish or Catalan –
and the zone of residence of speakers to affect the production of third-person
object clitics and the occurrence of transfers between Majorcan Catalan and
Majorcan Spanish, in line with what we found in our previous study about the
use of first and second person clitics by Catalan-Spanish bilinguals (Jiménez-
Gaspar et al., 2017). Nevertheless, we do not expect other external variables
(such as speakers gender, age or level education) to affect the production of these
clitics, given that we did not find this effect with the same participants in
Jiménez-Gaspar et al. (2017).

6. Methodology

6.1 Participant’s selection

Forty-five Catalan-Spanish adults participated in this study. More specifically,

we present data from 26 women and 19 men, aged between 16 and 65 years old
(SD = 7.124; range: 46.3). Most of them, 83%, were born in Majorca; the rest
(four participants, 17%) moved to the island 15 to 40 years before their
participation in the study.9
The participants are residents of several major geographic areas of Majorca:
the capital, Palma, and the areas corresponding to the villages of Llucmajor,
Capdepera, Soller and Llubí. This division is important because the speakers’
dominant L1, i.e., either Catalan or Spanish (if speakers show dominance), tends
to change depending on where they reside. This information was collected via an
ethnolinguistic questionnaire (see Table 3). In Table 3, we can see the different
social and extra-linguistic variables represented by speakers from each zone of
residence. The number of subjects in each sub-group (each variant) is indicated
after the colon on each cell in Table 3.

Table 3. Social variables*

Speakers Area of Gender Age Education Linguistic

(45) residence level preference

Capital of Majorca 24 Palma F: 14 15– Minim.:2 Catalan:7

(n = 24) M:10 20: 1 Secund.:7 Spanish:10
21– Medium:4 Both:7
30: 9 Superior:11
31–
40: 8
41–
50: 3
51–
60: 0
+60:
3
Villages of Majorca 10 Llucmajor F: 4 15– Minim.:3 Catalan:9
(n = 21) M: 6 20: 0 Secund.:3 Spanish: 0
21– Medium:3 Both:1
30: 2 Superior:1
31–
40: 0
41–
50: 6
51–
60: 0
+60:2
8 Capdepera F: 6 15– Minim.:4 Catalan:6
 M: 2 20: 1 Secund.:4 Spanish:2
21– Medium: 0 Both: 0
30: 3 Superior: 0
31–
40:0
41–
50: 1
51–
60: 2
+60:
1
2 Soller F: 2 15– Minim.: 0 Cat.:2
M: 0 20: 0 Secund.: 0 Spanish: 0
21– Medium: 2 Both: 0
30: 1 Superior: 0
31–
40: 1
41–
50: 0
51–
60: 0
+60:
0
1 Llubí F: 0 15– Minim.:0 Cat.:1
M: 1 20: 0 Secund.:0 Spanish: 0
21– Medium: 0 Both: 0
30: 1 Superior: 1
31–
40: 1
41–
50: 0
51–
60: 0
+60:
0

* Subjects were classified in six age groups (15–20, 21–30, 31–40, 41–50, +60 years old). There were three
groups regarding linguistic preference (Spanish, Catalan; or both, to indicate no preference between the two
languages), and four groups regarding level of education: minim (that is, primary school), secondary (high
school), medium (if the speakers have obtained non-degree education beyond high school, e.g. professional
education) and superior (undergraduate or graduate degree). ①

Regarding linguistic preference, participants were asked to indicate whether they

preferred using either language or whether they were equally comfortable using
both languages.10 More precisely, the subjects were asked which language they
preferred to use with their family, their friends, at work, while shopping, or when
they talked to a doctor. There were 11 sequential bilinguals in Spanish and
Catalan among the participants, that is, they acquired Spanish from birth, but
they learned Catalan after they were 6 years old. These subjects indicated that
they preferred to use Spanish in their daily life.

6.2 Data collection

The synchronic data that we analyzed came from oral, spontaneous production
by all forty-five participants in the study. We recorded each participant twice,
once in Spanish and once in the Majorcan Catalan variety. Most of the
recordings lasted approximately 10 to 15 minutes (but recordings in which more
than one of the speakers participated lasted between 30 and 40 minutes), and
involved a variety of topics, mainly related to family, hobbies, work and
individual experiences. The interviews were conducted in the homes of the
participants in a very spontaneous and relaxing atmosphere.

Table 4. Participants‘ ethnolinguistic information

Palma Capdepera Llucmajor Soller Llubí

(n = 24) (n = 8) (n = 10) (n = 2) (n = 1)

Time in Majorca Always 96% 87.5% 90% 100% 100%

 +10 years 4% 0% 0% 0% 0%
+20 years 0% 12.5% 10% 0% 0%
Language at home Catalan 62.5% 62.5% 80% 100% 100%
Spanish 37.5% 12.5% 20% 0% 0%
Both 0% 25% 0% 0% 0%
Language at work Catalan 8% 25% 60% 50% 100%
Spanish 50% 12.5% 40% 50% 0%
Both 42% 62.5% 0% 0% 0%
Mother L1 Catalan 62.5% 87.5% 90% 100% 100%
Spanish 37.5% 12.5% 10% 0% 0%
Father L1 Catalan 54% 87.5% 90% 100% 100%
Spanish 46% 12.5% 10% 0% 0%

Additionally, in response to the ethnolinguistic questionnaire, participants

provided information not only on the social variables mentioned above, but also
regarding their native language, profession, their parents’ education and their
own, the origin of their parents, the native language of their parents, and so on.
For instance, these questions were worded as follows (either in Catalan or in
Spanish, if the subjects preferred to answer these questions in Spanish): If you
were not born in Majorca, how long have you lived on the island? What
language do you use at home, with your friends and at work? What formal
education have your parents received? Table 4 summarizes additional results
from the ethnolinguistic questionnaire.
As we can observe in Table 4, nowadays Catalan is present not only at
home, but also at work, both in the capital, Palma, and the villages far away
from Palma. There is an important difference between the numbers of
participants in each zone. It is relevant to explain that in Capdepera there is a
large population of foreigners and people from the Peninsula, therefore, it is
expected to see that participants from there used more Spanish than Catalan, in
comparison with other villages, but not as much as in Palma. In the other
villages (Llucmajor, Soller and Llubí), the situation is different since the
habitants from there are often simultaneous bilingual speakers and tend to use
Catalan much more than Spanish.

7. Analysis and results

We have analyzed 1111 clitic tokens in the data (623 in Catalan and 488 in
Spanish). We divided the third-person non-reflexive pronominal clitics
depending on their case (accusative/dative) and gender
(masculine/feminine/neutral).11 This yielded three data sets in Spanish:
dative/indirect object le/les (n = 159 tokens), feminine accusative la/las (n = 85)
and masculine accusative lo/los (n = 244). In Catalan, the analysis yielded a set
with the dative/indirect object li/els (240 tokens), and three other sets based on
gender in the accusative/direct object: masculine el/els (113), feminine la/les
(104) and neutral ho (166). Additionally, we considered the position of the
clitics, because proclitic and enclitic forms can be distinct from each other in
Catalan (see Section 3).
For statistical analysis of all the data, we used the SPSS (Statistical Package
for the Social Sciences) program and, more specifically, we applied a logistical
regression analysis targeting the dependent variable, that is, the production of the
3rd person clitics, to determine whether the clitic forms produced by the subjects
matched the corresponding Peninsular Catalan or Peninsular Spanish, vs.
Majorcan Catalan (MC) or Majorcan Spanish (MS) forms, respectively, as
summarized in Section 3. The independent variables are gender, age, level
education, the zone of residence and linguistic preference of the subjects, as
identified in Table 3.
The production of third-person, non-reflexive clitics, in both languages
interests us regarding their morphophonological and syntactic features. First, we
aim at understanding which clitic forms are produced by bilingual Catalan-
Spanish speakers in Majorca, and whether they matched the Peninsular
Spanish/Catalan forms presented in Section 3, or whether they represent forms
unique to MC that show evidence of transfers from MS and vice-versa. Finally,
we aim at determining whether the external variables affect the production of
clitic forms in Catalan and Spanish by bilingual speakers in Majorca.

7.1 Clitic production in Majorcan Catalan

Let us consider first the production of third-person non-reflexive clitics in

Catalan in Majorca. We can observe (in Table 5) that there is a variable number
of “non-peninsular” uses, that is, specific uses from Majorcan Catalan which are
different from Peninsular Catalan (see Table 1). We treated as “non-peninsular”
uses in Table 5 (and elsewhere) the tokens that did not match the clitic forms
from Peninsular Catalan (spoken in Barcelona) presented in Table 1. Overall,
79% (494 tokens) of the clitic uses matched peninsular forms, and 21% (129
tokens) were distinct forms that we categorized as non-peninsular, uniquely
Majorcan Catalan uses. In our analysis of clitic forms (either proclitics or
enclitics), as summarized in Table 5, we observe that two types of clitics show a
higher frequency of non-peninsular (uniquely Majorcan) uses, dative (li/els) and
neutral (ho) clitics:

Table 5. The use of the third personal clitics in Majorcan Catalan

Third-person clitics Tokens

Peninsular uses Non-peninsular uses

Dative li/els 199/240 83% 41/240 17%

Acc. fem. la/les 97/104 93% 7/104 7%
Acc. masc. el/els 98/113 87% 15/113 13%
Neutral ho 100/166 60% 66/166 40%

Considering the set of non-peninsular forms that appear in our production data,
the distribution of third-person non-reflexive clitics in MC shows the forms in
Table 6. The forms that are boldfaced in the table are forms found in MC that do
not match the Peninsular Catalan forms summarized in Table 1:

Table 6. Non-peninsular uses in Majorcan Catalan (production data)

Before the verb After the verb Case
sing. masc./fem. li -li dative
masculine el /ho l’ -lo accusative
feminine la l’/la -la
Third person
neutral ho/(lo) -ho
pl. masc./fem. əls/əlzi /lis/əlzə/ -los dative
masculine əls/əlzi /lis/əlzə -los accusative
feminine les/ ləzə -les

In the case of the dative clitics, all the Majorcan (non-peninsular) uses
correspond to plural forms. More often, they appear to present an epenthetic
vowel (elze [əlzə]). Whereas most Catalan varieties present the epenthetic vowel
([i]) (see alternative analysis in Section 4) in the dative form, the most important
difference between MC and the peninsular Catalan dialects has to do with the
presence of another epenthetic vowel, [ə], in a phenomenon that also extends to
accusative forms.
We reviewed different proposals for the existence of such forms (elzi/elze)
in Section 4, mostly in connection with an epenthetic vowel analysis going back
to work by Bonet (1991, 1995). For instance, Seguí-Trobat (2014) argues that
these uses are a reanalysis in analogy to the singular form li (see also Boeckx &
Martín, 2013, p. 13), and that a change in the epenthetic vowel would make it
possible to distinguish the direct object (using elzə) from the indirect object
(using elzi). Our data does show that uses of elzə/elzi allow reference to both
cases (accusative and dative). This contrast is partially supported by our
examples from MC in (20) to (23). As we can see in these examples, not only is
the epenthetic vowel [i] added to the form els (20)–(21), which different authors
viewed as analogous with the dative singular li, but the vowel [ə] can also be
appended, as in the accusative forms in (22)–(23). Both forms can occur as
plural accusative forms in MC (see Table 6). These data support analyses that
proposed an epenthetic vowel after the form els (yielding elzi or elze).

(20)
a ells, [elzi] pagues 20 euros (MC)
to them, [Link] [Link] 20 euros [Female from Sóller, age 30]
‘(You) pay them 20 euros.’

(21)
No elzi diu res (MC)
not [Link] said nothing [Female from Llucmajor, age 63]
‘He doesn’t say anything to them’

(22)
Els xots, [elzə] va Øgafar (MC)
the lambs, [Link] [Link] [Male from Llucmajor, age 67]
‘The lambs, he catched them.’

(23)
No elze vol (les camisetes) (MC)
not [Link] want (elze = the shirts) [Male from Llubí, age 29]
 ‘He doesn’t want them’

Moreover, we consider that if the Majorcan (non-peninsular) plural dative uses

are indeed a reanalysis in agreement with li, the form used could also be lis;
there were three occurrences with this form, as shown with the dative plural
clitic use in (24).

(24)
Perquè lis donguis coses gratis
because [Link] give things free [MC, male from Llucmajor, 42
years old]
‘Because you give them free things.’

Finally, as summarized in Table 6, in the production of non-peninsular forms it is

primarily the accusative masculine plural clitic that takes a similar form to the
dative plural clitic [elzə/i]. Perea (2012) previously argued that there is an
additional variation in the choice of clitic pronouns in Majorcan Catalan:
feminine les can also be used for both accusative and dative cases:

(25)
Les/l[ə]zi han deixat els llibres
[Link]. have left the books
‘They have left the books to them.’

We also observed gender variation in our production. First, a non-peninsular

(Majorcan) masculine plural accusative clitic (elze) can be used with a feminine
referent, as in Example (26). Second, we also observed the opposite pattern in
the production data – in exceptional cases, the feminine plural accusative clitic is
used for a masculine referent (27):

(26)
 Elz[ə] necessita (doscentes mil pessetes)
1. [Link] [Link] ([Link] thousand [Link]) [MC,
female from Palma, 61]
2. Les necessita (Peninsular Catalan counterpart)
‘he needs them’ (two hundred thousand pesetas)

(27)
Les matem (els porcs)
1. [Link] [Link] (the [Link]) [MC, Male from Llucmajor, 43]
2. Els matem (Peninsular Catalan counterpart)
‘We kill them’ (the pigs)

The preference for syllabic pronominal clitics could be the explanation for this
pattern in Majorcan Catalan, that is, the use of the epenthetic vowels [i] or [ə], as
well as the production of the feminine form les instead to els could be strategies
to obtain syllabic forms, ending in CV(C). The fact that speakers also
sporadically use the plural dative (or masculine accusative) clitic lis in analogy
with dative singular li may add further evidence to this adoption of CV(C)
patterns for third-person clitics in Majorcan Catalan.
Interestingly, in addition to the observations above regarding the uses of the
plural forms əlzi/əlzə/lis, there is also a widespread regularization of enclitic
forms, so that all the enclitic forms in MC show a syllabic CV(C) pattern (i.e. the
elided accusative sing. masc. form ‘l from Peninsular Catalan corresponds to lo,
and the dative/accusative plural ‘ls corresponds to los, in our MC data), as
shown in Table 6. This pattern represents one aspect of convergence with
Spanish, which only has CV(C) clitic forms. In Jiménez-Gaspar et al. (2017) we
argued that there was a similar convergence with Spanish for 1st and 2nd person
accusative clitics in Majorcan Catalan (e.g. me, nos, vos…replace the variants em
‘me’, ens ‘we’, us ‘you’, respectively), and the same group of bilinguals
considered in this study showed a clear preference for the use of the CV forms in
Majorcan Catalan, matching the form of Spanish clitics.12
Turning to the neutral clitic ho in MC, it seems to show a semantic
extension in its reference that matches the Spanish clitic (lo), in that it can be
used in reference to a masculine noun (28)–(29). In addition, there are a few
examples in which the Spanish clitic lo is used for ho in MC (30).These are
possibilities that are unique to the MC data:

(28)
No ho pots sembrar (el gra)
not [Link] can [Link] . ([Link] seed). [Female from
Llucmajor, 63]
‘You cannot sow it’

(29)
M’ho deixas? (el compressor)
to [Link] lend [Link] . ([Link] compressor) [Female from
Llucmajor, 63]
‘Do you lend it to me?’

(30)
Lo vaig saber (això)
[Link] [Link] (this) [Male from Llucmajor, 43]
‘I knew it’

The difference between the production of DO masculine and DO neutral clitics

in MC is statistically significant (p-Value: 0.001). Speakers produce only 15 non-
peninsular DO masculine uses out of 113 tokens (most of these non-peninsular
forms involve clitic omission), instead of what occurs with the DO neutral ones.
Unique Majorcan/non-peninsular uses of neutral clitics rise to 66 out of 166
tokens (see Table 5), corresponding to the changes to ho (lo) indicated above,
which we argue also involve a pattern of transfer from Spanish.
In sum, the analysis of the Majorcan Catalan third person clitic data shows
a pattern of partial divergence between MC and Peninsular Catalan regarding the
plural accusative and dative forms, although this is partially independent of
interferences from Spanish (and in fact shows up in the Majorca villages where
influence from Spanish may be lower, as we will see in Section 8.1). In addition,
we identified two specific patterns of convergence between Majorcan Catalan
and Spanish regarding third person clitic forms, in the production of CV(C) clitic
forms (lo/los), and in the production of a neutral/masculine clitic forms (ho/lo).

7.2 Clitic production in Majorcan Spanish

Turning to the analysis of the production data by the bilinguals in Spanish, a

comparison of their production to Peninsular Spanish shows that the variation
summarized in Table 7 is not significant, given the low frequency of (non-
Peninsular) uses that are unique to MS (29 out of 488 tokens, 5.8%). Most of
these MS uses are omissions of the accusative clitics lo/los and la/las in
sentences in which they should appear in Peninsular Spanish. There are also only
a few instances of leísmo and laísmo, in which dative clitics are replaced by
accusative clitics (and vice-versa). This confirms that, in Majorca, the
pronominal clitics of the etymological system (corresponding to the peninsular
Spanish forms in Table 2a) are widely used, as shown in Table 7.
These results indicate that there are less substantial effects in the production
of third-person pronominal clitics in Spanish by Catalan-Spanish bilinguals in
Majorca, unlike what we observed for Majorcan Catalan in the previous section.
Therefore, this supports a pattern of transfer, with the observation that the
majority language (Spanish) has more influence in the minority language
(Catalan) regarding the 3rd personal pronominal clitics in the Majorcan dialects.

Table 7. The use of the third personal clitics in Majorcan Spanish

Tokens

Peninsular uses (PS) Non-peninsular uses (MS)

Dative le/ les 146/159 91.8% 13/159 8.2%

Acc. la/las 82/85 96.5% 3/85 3.5%
Acc. lo/los 231/244 94.7% 13/244 5.3%

8. Analysis across extra-linguistic variables

8.1 Social (extra-linguistic) variables in Majorcan Catalan

In Table 8, we can observe the number of tokens and their corresponding

percentages in the production of third person pronominal clitics, distinguishing
Peninsular from Non-Peninsular (unique to Majorca) uses. We have divided the
uses between dative (Indirect Object) and accusative (Direct Object). In addition,
we have separated accusative clitics depending on their grammatical gender
(masculine, feminine or neutral). As we consider the role of extra-linguistic,
social variables, we can observe in Table 8 that they have limited influence on
the use of third-person non-reflexive clitics in MC, except for a few specific
patterns in boldface.

Table 8. The production of third person Object clitics in MC across extra-

linguistic variables

Peninsular uses Non-peninsular uses

 IO DO DO DO IO DO DO DO
(masc.) (fem.) (neut.) (masc.) (fem.) (neut.)

Gender Women 123/147 63/67 60/65 58/103 24/147 4/67 5/65 45/103
84% 94% 92% 56% 16% 6% 8% 44%
Men 76/93 35/46 37/39 42/63 17/93 11/46 2/39 21/63
82% 76% 95% 67% 18% 24% 5% 33%
Age 15–20 5/5 3/3 0/0 0% 0/0 0% 0/5 0/3 0/0 0/0 0%
100% 100% 0% 0% 0%
21–30 77/89 32/33 28/28 25/41 12/89 1/33 0/28 16/41
87% 97% 100% 61% 13% 3% 0% 39%
31–40 41/43 22/29 17/20 32/59 2/43 7/29 3/20 27/59
95% 76% 85% 54% 5% 24% 15% 46%
41–50 36/52 13/15 26/27 16/28 16/52 2/15 1/27 12/28
69% 87% 96% 57% 31% 13% 4% 43%
51–60 7/7 7/7 2/2 3/4 75% 0/0 0/0 0/2 1/4 25%
100% 100% 100% 0% 0% 0%
+60 33/44 21/26 24/27 24/34 11/44 5/26 3/27 10/34
75% 81% 89% 71% 25% 19% 11% 29%
Ed. Minim 48/61 17/18 13/14 13/20 13/61 1/18 1/14 7/20
Level 79% 94% 93% 65% 21% 6% 7% 35%
Secund 68/78 38/41 27/31 25/44 10/78 3/41 4/31 19/44
87% 93% 87% 57% 13% 7% 13% 43%
Med. 43/53 19/22 23/23 22/37 10/53 3/22 0/0 15/37
81% 86% 100% 59% 19% 14% 0% 41%
Sup. 40/48 24/32 34/36 40/65 8/48 8/32 2/36 25/65
83% 75% 94% 62% 17% 25% 6% 38%
Zone Palma 101/109 48/59 56/62 57/97 8/109 11/59 6/62 40/97
93% 81% 90% 59% 7% 19% 10% 41%
Llucm. 54/80 26/29 34/35 27/47 26/80 3/29 1/35 20/47
67% 90% 97% 57% 33% 10% 3% 43%
Capdep. 30/31 16/17 5/5 8/9 89% 1/31 1/17 0/5 1/9 11%
97% 94% 100% 3% 6% 0%
Soller 12/17 7/7 2/2 4/8 50% 5/17 0/7 0/2 4/8 50%
71% 100% 100% 29% 0% 0%
 Llubí 2/3 67% 1/1 0/0 0% 4/5 80% 1/3 0/1 0/0 1/5 20%
100% 33% 0% 0%

More specifically, the variable gender seems to play a role in the production of
accusative (direct object) masculine clitics; women omitted this type of clitic
less often than men:

Table 9. Effect of gender and the use of 3rd person masculine DO pronominal
clitics in MC

Model Summary

Model R R² Adjusted R² RMSE R² change F change df1 df2 p

1 0.260 0.067 0.059 0.331 0.067 8.035 1 111 0.005

Still, considering the significant difference regarding gender in the production of

accusative masculine clitics in Table 8, we can observe in Figure 1 that there are
few accusative masculine tokens considered exclusively non-peninsular (i.e. at
least not matching peninsular clitic forms), either in men’s or in women’s
production.

Figure 1. Peninsular and non-peninsular uses of accusative masculine DO clitics

in MC across gender
Most of these non-peninsular masculine uses in Figure 1 involve omission of the
relevant accusative clitic, as illustrated in (31)–(32). Although men produce less
accusative masculine singular clitics than women (n = 46 vs. 67), the fact that
they produce peninsular forms less often than women do is still significant.

(31)
Mos Ø du en la caseta (el porc)
Ens el du a la caseta (el porc)
[Link] Ø bring in the hut (the pig) [Majorcan Catalan Male, Llucmajor,
42]
‘They bring (it) to us in the hut’

(32)
Jo Ø Ø explicava (a sa gent que m’havia equivocat d’avió)
I Ø Ø explained (to the people that [Link] confused of plane) [Male
from Capdepera, 29]
‘I explained (it) (to them) (to the people that I was wrong about the plane)’

Regarding the use of dative (indirect object) clitics, there is a statistically

significant contrast by area of residence (p-Value: 0.001):

Table 10. Liner regression across speaker’s zone of residence and the use of 3rd
person IO pronominal clitics in MC

Model Summary

Model R R² Adjusted R² RMSE R² change F change df1 df2 p

1 0.321 0.103 0.099 0.358 0.103 27.28 1 238 < .001

Considering the area of residence of our subjects, speakers from the villages of
Llucmajor and Soller present more variation in the use of the dative clitics in
Catalan (using different forms such as les, els, elzi). Most of their dative uses
matching Peninsular Catalan forms involve the singular form /li/, which does not
present variation in this Catalan variety. Most of their non-peninsular dative uses
are plural forms. These uses are primarily due to the use of an epenthetic vowel
in the plural form (elzi/elze), as we discussed before. Less often, speakers use the
form lis (plural form) as analogous with li (singular form) (see Section 7.1 for
analysis of these patterns).
In Figure 2, we can observe the number of non-peninsular (Majorcan)
tokens of dative forms by each area of residence. This shows a substantial lack
of non-peninsular dative forms in Capdepera and Palma in comparison with
Llucmajor or Soller, where there are more MC occurrences that do not match
Peninsular Catalan (in Llubí there are too few tokens). This indicates that the
villages of Soller and Llucmajor more clearly show evidence of non-peninsular
forms, at least in the production of third person dative clitics.

Figure 2. MC dative clitics across area of residence

We have also found interesting results in the correlation between the zone of
residence and the production of the DO neutral clitic, ho, which is characterized
by a meaning extension in MC, that is, it can refer to masculine referents, rather
than only neutral ones (see Section 7.1 for details). Although there are not
statistically significant results by area of residence, many of the non-peninsular
uses of clitic ho correspond to masculine referents, and they are common for
speakers across different areas of residence, as shown in Figure 3:

Figure 3. MC accusative neutral clitic, ho, across area of residence

In sum, we observed precise effects from extra-linguistic variables (gender and
area of residence) on a couple of properties of third-person clitics that show
unique features in MC (plural dative clitics, accusative neuter clitic ho).
However, neither age nor the education level of subjects were significant for the
production of third person clitics in Majorcan Catalan, and there was no overall
pattern of effects that could be linked to potential effects of extra-linguistic
variables in Majorca.

8.2 Social variables in Majorcan Spanish

We have tested for Majorcan Spanish the same correlations we had considered
for Majorcan Catalan data, but we have not found statistically significant results
in the linear regressions analysis across neither gender, age, zone of residence
nor linguistic preference. This is likely the case because, as shown in Table 11,
there are few non-peninsular tokens in MS, reducing group differences across
social variables.
Notice also that the sub-groups by linguistic preference, age, zone of
residence and level of education were not balanced, limiting the possible
conclusions we could reach regarding some of the differences across these
groups. Despite that, our results regarding the production of third person
pronominal clitics in Majorcan Spanish show that participants sometimes
omitted these clitics or produced tokens indicating leísmo and laísmo. (33)
shows an occurrence of so-called leísmo, the use of the dative clitic instead of
the accusative clitic, and (34) shows an occurrence of laísmo, in which the
feminine accusative clitic is used to refer to the indirect object. More
specifically, speakers who produced IO pronominal clitics instead of DO ones, or
vice versa, were simultaneous bilinguals whose linguistic preference is Catalan.
However, the frequency of this distribution was not high enough to yield a
statistically significant pattern in MS. These replacements, which we briefly
discussed in Section 3, are very uncommon, although they match the referential
clitic system used in a large part of the Peninsula (see Table 2b). A more
common occurrence among the few non-peninsular uses counted in Table 11 was
the omission of direct object clitics (feminine or masculine).

(33)
La nevera, siempre le tenía llena
the fridge always IO-CL . had full [MS, Female from Llucmajor, 63]
‘He had always the fridge full’

(34)
La pedí para casar
[Link] asked to marry [MS, Male from Llucmajor, 67]
‘I asked her for getting married’
Table 11. The production of 3rd person object clitics in MS across social variables

Peninsular uses Non-peninsular uses

IO DO (masc.) DO (fem.) IO DO DO
(masc.) (fem.)

Gender Women 83/91 91% 124/133 42/44 95% 8/83/ 9/133 7% 2/44 5%
93% 9%
Men 63/68 93% 107/111 40/41 98% 5/68 7% 4/111 4% 1/41 2%
96%
Age 15–20 2/2 100% 0/1 0% 1/1 100% 0/2 0% 1/1 100% 0/1 0%
21–30 52/59 88% 89/95 94% 29/30/ 7/59 6/95 6% 1/30 3%
97% 12%
31–40 24/24 51/52 98% 19/19 0/24 0% 1/52 2% 0/19 0%
100% 100%
41–50 33/34 97% 46/47 98% 16/16 1/34 3% 1/47 2% 0/16 0%
100%
51–60 16/19 84% 24/27 89% 0/1 0% 3/19 3/27 11% 1/1
16% 100%
+60 19/21 90% 21/22 95% 17/18 94% 2/21 1/22 5% 1/18 6%
10%
Ed. Minim 35/38 92% 35/38 92% 25/26 96% 3/38 3% 3/38 8% 1/26 4%
level Secund. 44/47 94% 85/93 91% 26/27 96% 3/47 6% 8/93 9% 1/27 4%
Medium 36/37 97% 71/72 99% 18/18 1/37 3% 1/72 1% 0/18 0%
100%
Superior 31/37 84% 40/41 98% 13/14 93% 6/37 1/41 2% 1/14 7%
16%
Zone Palma 66/72 92% 121/126 49/50 98% 6/72 8% 5/126 4% 1/50 2%
96%
Llucmaj 30/33 91% 50/51 98% 22/23 96% 3/33 9% 1/51 2% 1/23 4%
Capdep. 38/42 90% 42/49 86% 9/10 90% 4/42 7/49 14% 1/10
10% 10%
Soller 10/10 7% 14/14 1/1 100% 0/10 0% 0/14 0% 0/1 0%
100%
 Llubí 2/2 100% 4/4 100% 1/1 100% 0/2 0% 0/4 0% 0/1 0%

Regarding area of residence, Amengual (2011a, see also 2011b) argued that there
are dialectal differences between the areas of predominant use of Spanish or
Catalan in Majorca, and we wanted to determine whether these could affect the
use of clitics in Majorcan Spanish, similarly to what we showed for Majorcan
Catalan. However, we did not find the area of residence (or other social factors)
to be significant regarding the properties of Majorcan Spanish. This was partially
due to the limited variation we found regarding these differences in MS, as
shown in Table 11.

8.3 Analysis across linguistic preferences

Finally, let us consider the effect of linguistic preference in our results. As we

discussed in Section 6.1, we determined a speaker’s linguistic preference based
on which language they preferred to use more often in different contexts of
interaction. We classified speakers into three different groups based on their
linguistic preference: Catalan, Spanish or equal preference for both languages.
Although we did not have an independent measure of speaker’s proficiency or
dominance in either language, we took linguistic preference to be indicative of
their language dominance. In fact, as we indicated in Section 6.1, all subjects
who were sequential bilinguals and acquired Catalan later indicated a clear
preference for Spanish.
In Catalan, speakers’ difference in linguistic preference had a significant
effect only on the production of dative clitics: speakers who indicated a
linguistic preference for Catalan produced more vernacular MC forms that did
not match Peninsular Catalan (such as [elze/elzi] rather than els), compared to
speakers who indicated a linguistic preference for Spanish. The latter group
produced forms that are closer to Peninsular Catalan. One explanation for this
takes into account the fact that all subjects who had a linguistic preference for
Spanish were sequential bilinguals, therefore they typically learned Catalan
primarily at school, and were exposed primarily to a register that was closer to
Peninsular Catalan, instead of acquiring the vernacular MC forms from birth.13
In addition, as shown in Table 12, there was a high frequency of MC uses of
the DO neutral clitic ho that did not match Peninsular Catalan. These non-
peninsular uses were more frequent among speakers who had Catalan as their
language of preference, even if participants in general showed a high frequency
of these uses, involving the extension of the use of ho for masculine referents.
Interestingly, we can observe that the frequency of non-peninsular uses
decreased for speakers who had Spanish as their language of preference,
indicating that they tended to give preference to uses of ho that were closer to
Peninsular Catalan, and that would arguably be the ones that they had more
extensive exposure by acquiring Catalan primarily in school.

Table 12. Linguistic preference in MC

Peninsular uses Non-peninsular uses

IO DO DO DO IO DO DO DO
(masc.) (fem.) (neutral) (masc.) (fem.) (neutral)

Linguistic Catalan 109/140 62/74 44/47 58/107 31/140 12/74 3/47 49/107
preference 78% 84% 94% 54% 22% 16% 6% 46%
Spanish 44/48 21/22 17/18 28/41 4/48 1/22 1/18 13/41
92% 95% 94% 68% 8% 5% 6% 32%
Both 46/52 15/17 36/39 14/18 6/52 2/17 3/39 4/18
88% 88% 92% 78% 12% 12% 8% 22%
In Majorcan Spanish, as we observed in previous sections, there are only a few
differences in comparison to the peninsular variety (see Table 7). In parallel,
there were no significant effects of a linguistic preference for Catalan or Spanish
on the production of MS:

Table 13. Linguistic preference in MS

Peninsular uses Non-peninsular uses

IO DO DO (fem.) IO DO DO
(masc.) (masc.) (fem.)

Linguistic Catalan 64/70 96/101 31/33 93% 6/70 9% 5/101 2/33 6%

preference 91% 95% 5%
Spanish 49/55 66/70 94% 31/31 6/55 4/70 6% 0/31 0%
89% 100% 11%
Both 33/34 69/73 95% 20/21 95% 1/34 3% 4/73 5% 1/20 5%
97%

As we can see in Table 13, there are very low frequency and variation in the uses
in MS that did not match Peninsular forms, therefore linguistic preference does
not lead to any significant difference in the production of third person clitics in
Majorcan Spanish.

9. Final remarks
In this paper, we have presented a description and analysis of third-person
pronominal clitics of Catalan and Spanish as produced by Catalan-Spanish
bilinguals in Majorca. We considered two questions at the outset: (i) Are there
distinct properties or variation in the production of third-person (non-reflexive)
clitics that can be evidence of transfer or convergence between Catalan and
Spanish in Majorca? (ii) Do extralinguistic/social variables, such as the speakers’
linguistic preference, gender and their zone of residence, affect these properties
in either language?
Regarding the first question, we expected the occurrence of transfers from
one language to the other. However, unlike what Jiménez et al. (2017) found in
their analysis of the bilingual production of first- and second-person pronominal
clitics in Majorca, in the current investigation there is no evidence of substantial
convergence between the two systems.
However, regarding Majorcan Catalan, there are specific instances of
transfer or interference from Spanish, as in the case of the neutral clitic ho,
which has partially developed a semantic extension to masculine forms, sharing
this underlying semantic feature with lo in Spanish. That is, the bilingual MC
speakers show evidence of having acquired the neutral clitic ho with part of the
semantic properties of the Spanish masculine/neutral clitic, lo, using it in MC to
refer to both neutral and definite masculine referents (in the singular).
Speakers also show evidence of using the accusative singular masculine
forms lo and los in Majorcan Catalan (matching Spanish) in different contexts,
showing a pattern of regularization of enclitic forms to a CV(C) pattern. We took
this regularization to represent one aspect of convergence with Spanish, which
only has CV(C) clitic forms, matching the CV regularization we found in
Jiménez-Gaspar et al. (2017) for 1st and 2nd person accusative clitics in
Majorcan Catalan (e.g. me, nos, vos…replace the variants em ‘me’, ens ‘we’, us
‘you’, respectively).
Concerning the third-person dative plural masculine clitic els in Catalan
(also used for accusative masc/fem), a clear pattern we found was the insertion
of an epenthetic vowel, which can be not only [i] but also [ə]: [elzə] and [elzi],
supporting a range of analyses proposing an epenthesis analysis to the
production of these forms. The fact that speakers add an epenthetic vowel in
forms like els ([elzi][elzə]), or use the dative clitic lis in analogy with li may add
further evidence to the adoption of a regular (X)CV(C) pattern for third-person
clitics in Majorcan Catalan.
Regarding third-personal clitics in Majorcan Spanish, their uses do not vary
greatly from the uses that are considered peninsular. However, we note that,
unlike what happens with the dative le/les pronouns, almost all of the few non-
peninsular occurrences of pronominal clitics in Majorcan Spanish (less than
8.5%) correspond either to omissions of accusative clitics in sentences in which
they should appear, or to instances of leísmo and laísmo, in which dative clitics
are replaced by accusative clitics.
Concerning the social variables (gender, age, area of residence and level
education) that we correlated with the third-person object clitics, we did not find
significant correlations regarding Majorcan Spanish. However, we observed that
speakers who preferred using Catalan to Spanish produced more non-peninsular
forms which involve leísmo and laísmo, even if there were not enough
occurrences to confirm the significance of this or other patterns of variation in
MS.
We found a couple of relevant effects of extra-linguistic variables on
Majorcan Catalan, such as the influence of gender in the production of
accusative masculine clitics, since men omitted them more often than women.
Besides, the area of residence of our participants seems to influence their
production of dative clitics in MC, given that speakers from Llucmajor and
Soller (where Catalan use is more widespread than in Palma or Capdepera)
present more variation. The non-peninsular uses of these participants often
involved the production of the epenthetic vowels [i] and [ə] in the plural form. In
a few cases participants also used lis in analogy with the singular form li.
Given our results, we conclude that the production of third-person clitics in
Majorcan Catalan is clearly affected by bilingualism with Spanish, as in the use
of the neutral clitic ho in MC with a semantic extension matching Spanish lo. In
addition, the evidence suggests that our Majorcan Catalan-Spanish bilinguals
prefer syllabic (CV(C)) forms of pronominal clitics, favoring a convergence with
Spanish in this respect, as it was also observed in Jiménez-Gaspar, Pires and
Guijarro-Fuentes (2017) regarding first and second person pronominal clitics in
Majorcan Catalan. In parallel, in Majorcan Spanish there are some sporadic
transferences from Catalan when speakers have Catalan as their linguistic
preference, but they were not robust enough in our production data to indicate a
consistent pattern of transfer or interference from Catalan in Spanish. These
results provide evidence of a primary pattern of unidirectional transfer from
Spanish to Catalan in Majorca, which may be due to the status of Spanish as the
dominant language in this bilingual setting.

Notes
1. As it is well known, pronominal clitics are unstressed pronouns. They cannot occur alone in an utterance
(i), since they need a host (a verb in Spanish and Catalan). They cannot be focused or topicalized (ii), nor
conjoined (iii) or modified (Eisenchlas, 2003; Wheeler et al., 1999; Bosque & Demonte, 1999).
1.
a. Lo veo./ *lo. (Spanish)

b. El veig /*lo. (Catalan)

[Link] [Link]/it-CL

‘I see him./ Him.’

2.
a. *Quiero verLO (Spanish)

b. *Vull veure’L (Catalan)

‘I want to see HIM’

3.
a. *Lo y la vi (Spanish)

b. *El i la vaig veure (Catalan)

[Link] and [Link] [Link]-PAST

 ‘I saw him and her.’

2. First and second person clitics (e.g. me ‘me’, te ‘you’) lack a gender distinction between masculine and
feminine and can be both reflexive and non-reflexive. The third-person reflexive clitic se ‘self’ lacks both
morphological number and gender distinctions. These two types of clitics also lack a case distinction
between accusative and dative forms, unless a preposition is added (See Jiménez-Gaspar, Pires, & Guijarro-
Fuentes, 2017, for an investigation of these two sets of clitics in Majorcan Catalan and Majorcan Spanish).
①

3. We consider Barcelonan (Peninsular) Catalan as a baseline for comparison because it is well documented
and is closely related to the dialects spoken in the Balearic Islands, including Majorcan Catalan. ①

4. These are examples produced by different speakers and are not from our bilingual subjects’ production
data. ①

5. There are some descriptive studies on the use of clitics in Majorcan Catalan, such as Alcover (1916),
Fischer (2003), and Perea (2012). Other studies, such as Seguí-Trobat (2014) and Wheeler et al. (1999),
tend to focus on the different forms from a normative point of view. ①

6. As discussed in Section 3.1, the third-person, dative case forms do not have gender differences across
dialects, unlike the non-reflexive third-person accusative case forms. ①

7. This is a spontaneous example form one of our bilingual subjects. One of the reviewers indicates that
this example is odd for them; it could be because in MC speakers use the masculine determiner with a
feminine noun here. ①

8. Notice that Peninsular Catalan also allows at least the form [elzi], according to Table 1 (see Bonet,
1995). See also (13)–(14). ①

9. These four participants were originally from Seville, Madrid, Badajoz (in Extremadura) and Barcelona.
①

10. We consider linguistic preference as a factor that relates to speakers’ dominance in each language,
although we did not carry out any tests to estimate language dominance independently or in connection to
language preference. ①

11. In Catalan there are two separate clitic forms to refer to masculine or neutral referents, instead of a
single form in Spanish (lo) which is used for either masculine or neutral referents, as we discussed before.
①

12. However, Jiménez-Gaspar et al. (2017) also showed that the CV pattern matching Spanish for 1st and
2nd person forms (e.g. me ’me’, te, vos ‘you’, se ‘self’) is not an innovation in contemporary Majorcan
Catalan, but actually it maintains a pattern that was predominant both in Peninsular Central and Balearic
Catalan between the 13th and 17th centuries (preceding the period when Spanish became the dominant
language in the Balearic Islands, see Section 2). Nevertheless, as also argued by Jiménez-Gaspar et al.
(2017), the CV pattern is now arguably favored in contemporary MC also due to Catalan-Spanish
bilingualism. ①

13. This outcome would have similarities to a case study in Pires & Rothman (2009), who explored the
possible effect that schooling in a multidialectal domain had on late childhood language acquisition; they
argued that learners who have more exposure to formal schooling in late childhood acquire certain variants
that are not acquired or are acquired later by less educated speakers. ①

References
Alcover, A. M.
(1916.) Estudis de Sintaxis dialectal catalana. Bolletí del Diccionari de la Llengua Catalana, 9, 11–63.
②
Amengual, M.
(2011a.) Spanish and Catalan in Majorca: Are there contact-induced changes in the Catalan vowel system?
In L. A. Ortiz-López (Ed.), Selected proceedings of the 13th Hispanic Linguistics Symposium (pp.
214–223). Somerville, MA: Cascadilla Proceedings Project. ① ② ③
Amengual, M.
(2011b.) Verbal morphology and identity in Majorca: The manifestation of attitudes in writing. In J.
Michnowicz & R. Dodsworth (Eds.). Selected proceedings of the 5th Workshop on Spanish
Sociolinguistics, (pp. 26–39). Somerville, MA: Cascadilla Proceedings Project. ①
Batllori, M., Iglesias, N., & Martins, M.
(2004.) L’ordre de clitics en català antic: Morfologia i sintaxi dels clítics pronominals en català medieval
In Contribucions a congressos (D-FL) (pp. 1–68). Girona: Universitat de Girona. ① ②
Blas Arroyo, J. L.
(1998.) Las comunidades de habla bilingües: Temas de sociolingüística española. Zaragoza: Libros
Pórtico. ① ②
Blas Arroyo, J. L.
(2004.) El español actual en las comunidades del ámbito lingüístico catalán. In R. C. Aguilar (Ed.),
Historia de la lengua española, 2 (pp. 1065–1086). Barcelona: Ariel. ①
Blas Arroyo, J. L.
(2007.) Spanish and Catalan in the Balearic Islands. International Journal of the Sociology of Language
184, 79–93. ①
Blas Arroyo, J. L.
(2011.) Spanish in contact with Catalan. In M. Diaz-Campos (Ed.), The handbook of Hispanic
sociolinguistics (pp. 374–394). Malden, MA: Blackwell. ①②③④
Boeckx, C., & Martín, T.
(2013.) The dative clitic is more than a clitic, 1. Lleida: Universitat de Lleida. ① ②
Bonet, E.
(1991.) Morphology after syntax: Pronominal clitics in Romance (Unpublished doctoral dissertation). MIT.
①②③④
Bonet, E.
(1995.) Feature structure of Romance clitics. Natural Language & Linguistic Theory 13(4), 607–647.
①②③④⑤⑥⑦⑧
Bonet, E.
(2002.) Cliticització. In J. Solà, M. R. Lloret, J. Mascaró, & M. Pérez Saldanya (Eds.), Gramàtica del
català contemporani, Vol. 1 (pp. 933–989). Barcelona: Empúries. ① ② ③
Bonet, E., & Lloret, M. R.
(2005.) More on alignment as an alternative to domains: The syllabification of Catalan clitics. Probus,
17(1), 37–78. ①②③④⑤⑥
Bosque, I., & Demonte, V. (eds.) (1999.) Gramática descriptiva de la lengua española. Vol. 3.
Madrid: Real Academia Española – Espasa Calpe. ① ②
Eisenchlas, S.
(2003.) Clitics in Child Spanish. First Language, 23(2), 193–211. ①
Enrique-Arias, A.
(2010.) On language contact as an inhibitor of language change. In A. A. Breitbarth, C. Lucas, S. Watts
D. Willis. Continuity and change in grammar (pp. 97–117). Amsterdam: John Benjamins.
①
Fernández-Ordóñez, I.
(2001.) Hacia una dialectología histórica. Reflexiones sobre la historia del leísmo, el laísmo y el loísmo
Boletín de la Real Academia Española, 81(284), 389–464. ① ② ③
Fischer, S.
(2003.) The Catalan clitic system: A diachronic perspective on its syntax and phonology. Berlin: Mouton:
de Gruyter. ①
Gavarró, A.
(1991.) A note on Catalan clitics. Catalan Working Papers in Linguistics, 1, 65–73. ①
Guijarro-Fuentes, P. A. Pires, & W. Nediger
(2017.) Delay in the acquisition of Differential Object Marking by Spanish monolingual and bilingual
teenagers. International Journal of Bilingualism, 21(2), 159–77. ①
Jiménez-Gaspar, A., Pires, A., & Guijarro-Fuentes, P.
(2017.) Bilingualism and language change: the case of pronominal clitics in Catalan and Spanish.
International Journal of Bilingual Education and Bilingualism, 22, 1–19. ①②③④⑤⑥
⑦⑧⑨⑩⑪⑫
Larrañaga, M. P., & Guijarro-Fuentes, P. (Eds.) (2012.) Pronouns and clitics in early language. Berlin:
De Gruyter. ①
Martín, F. J.
(2012.) Deconstructing Catalan object clitics (Unpublished doctoral dissertation). New York University.
Meisel, J. M.
(2011.) Bilingual language acquisition and theories of diachronic change: Bilingualism as cause and effect
of grammatical change. Bilingualism: Language and Cognition, 14(2), 121–145. ①
Melià, J.
 (2002.) La situació lingüística a les Illes Balears: Comentaris al voltant d´una enquesta. llengua i ús:
Revista Tècnica de Politica Lingüística, 24, 61–64. ①
Montrul, S.
(2004.) Subject and object expression in Spanish heritage speakers: A case of morphosyntactic
convergence. Bilingualism: Language and Cognition, 7(2), 125–142. ①
Montrul, S.
(2008.) Incomplete acquisition in bilingualism. Re-examining the age factor. Amsterdam: John Benjamins.
①
Perea, M. P.
(2012.) Les Combinacions de clítics pronominals en els dialectes catalans. Estudis Romànics, Institut
d’Estudis Catalans, 34, 99–143. ① ② ③ ④ ⑤ ⑥
Perpiñan, S.
(2017.) Catalan-Spanish bilingualism continuum: The expression of adverbial clitics in different types of
early bilinguals. Linguistic Approaches to Bilingualism, 7(5), 477–513. ①②③
Pires, A., & J. Rothman
(2009.) Acquisition of Brazilian Portuguese in late childhood: Implications for syntactic theory and
language change. In A. Pires, & J. Rothman (Eds.), Minimalist inquiries into child and adult language
acquisition: Case studies across Portuguese (pp. 129–154). Berlin: Mouton de Gruyter. ①
②
Poplack, S.
(1993.) Variation theory and language contact. In D. Preston (Ed.), American dialect research (pp. 251–
286). Amsterdam: John Benjamins. ①
Roca, F.
(1992.) Object clitics in Spanish and Catalan. Catalan Working Papers in Linguistics, 2, 245–280. ①
Seguí Trobat, G.
(2014.) La llengua pròpia de les Balears: Morfosintaxi Catalana. Mallorca: Edicions Documenta Balear.
①②③
Seib, J.
(2001.) La variedad bilingüe del español hablado en Cataluña caracterizada por interferencias y
convergencias con el catalán. Mannheim: Universität Mannheim. ①
Silva-Corvalán, C.
(1994.) Language contact and change: Spanish in Los Angeles. New York, NY: Oxford University Press.
①②
Sinner, C., & Wesch, A.
(2008.) El castellano en las tierras de habla catalana. Madrid & Frankfurt: Iberoamericana & Vervuert.
①
Thomason, S. G., & Kaufman, T.
(1988.) Language contact, creolization, and genetic linguistics. Berkeley, CA: University of California
Press. ①
Thomason, S. G.
(2001.) Language contact: An introduction. Edinburgh: Edinburgh University Press. ① ② ③
Todolí, J.
(2002.) Els pronoms. In J. Solà, M. R. Lloret, J. Mascaró, & M. Pérez Saldanya (Eds.), Gramàtica del
català contemporani, 3 (1341–1433). Barcelona: Empúries. ①
Uriagereka, J.
(1995.) Aspects of the syntax of clitic placement in Western Romance. Linguistic Inquiry, 26(1), 79–123.
①
Viaplana, J.
(1980.) Algunes consideracions sobre les formes pronominals clítiques del barceloní. Anuario de Filología
6, 459–484. ①
Wexler, K., Gavarro, A., & Torrens, V.
(2004.) Feature checking and object clitic omission in child Catalan and child Spanish. In J. Berns, H.
Jacobs, & D. Nouveau (Eds.), Romance languages and linguistic theory: Selected papers from Going
Romance, Groningen 28–30 November 2002 (pp. 253–268). Amsterdam: John Benjamins.
①
Wheeler, M., Yates, A., & Dols, N.
(1999.) Catalan: A comprehensive grammar. London: Routledge. ①②③
Syntax
CHAPTER 9

The distribution and use of present and past

progressive forms in Spanish-English and
Spanish-Brazilian Portuguese bilinguals
Julio César López Otero & Alejandro Cuza
Rutgers University | Purdue University

Abstract
This study examines the distribution and use of simple and progressive
forms in two groups: English-speaking heritage speakers of Spanish in
the U.S. (n = 9) and Brazilian Portuguese-speaking heritage speakers
of Spanish in Brazil (n = 15). We hypothesized that the groups would
show different crosslinguistic influence from their dominant languages
in their choice of verb forms. We collected semi-spontaneous
production data via oral narratives and analyzed group differences in
verb form, either simple or progressive, in activity and
accomplishment verbs (Vendler, 1967). The results show a main effect
for group, confirming that English-Spanish bilinguals favor
progressive verb forms in such contexts, while Brazilian Portuguese-
Spanish bilinguals opt for simple verb forms. We discuss our findings
following previous work by Jiang (2000) and Putnam & Sánchez
(2013).

Keywords
heritage language; verb morphology; progressive verb form
1. Introduction
The present study examines the use and distribution of present and past tense
progressive forms in Spanish as a heritage language among Spanish/English and
Spanish/Brazilian Portuguese bilinguals born and raised in the US and Brazil
respectively (Cuza, 2010; Cuza & López Otero, 2016; Geeslin & Fafulas, 2012;
Klein, 1980; Sánchez-Muñoz, 2004). The term heritage speaker refers to second-
generation immigrants or early arrivals exposed to a minority language during
early age in a naturalistic context where a majority language was also spoken
(Montrul, 2004; Polinsky, 2011; Valdés, 2001).
Previous work on the acquisition of past and present tense aspectual
differences in Spanish has shown significant difficulties among Spanish heritage
speakers and L2 learners, especially in the acquisition past tense aspectual
features (Cuza & Miller, 2015; Montrul, 2002a; 2008; Montrul & Slabakova,
2003). L2 learners and heritage speakers don’t seem to fully acquire preterite vs.
imperfect aspectual distinctions, and overextend the preterite form to contexts
where the imperfect should be used. We add to this previous work in two crucial
ways: First, we examine present and imperfect progressive forms, an area of
research still underexplored (Cuza, 2010; Cuza & López Otero, 2016; Geeslin &
Fafulas, 2012; Sánchez-Muñoz, 2004). In regard to the acquisition of present
tense aspectual properties, previous work documents difficulties in the
acquisition of the ongoing value of the present form, and overextension of the
present progressive in English-speaking heritage speakers and L2 learners of
Spanish (Cuza & López Otero, 2016). Second, we investigate and compare
Spanish heritage speakers exposed to English as a dominant language in the US
with speakers exposed to Brazilian Portuguese in Brazil, a language pair so far
unexplored as far as the present progressive and past tense progressive forms are
concerned.
 This language pair is interesting because BP and Spanish are closely related
languages and share similar morphological and lexical features. However, BP
behaves similarly to English as opposed to Spanish in regard to the selectional
properties of the present tense (Schmitt, 2001): the simple present selects only a
habitual meaning (i.e., O Paulo estuda espanhol, ‘Paulo studies Spanish’), while
the present progressive selects ongoing readings (i.e., O pai está assistindo o
jogo, ‘The father is watching the game’). In the past, on the other hand, the
imperfect progressive is the only option in English when expressing imperfective
ongoing readings (i.e., I was walking when the wolf approached me), whereas
both Spanish and BP have an imperfect simple form and an imperfect
progressive form (i.e., Juan dormía/estaba durmiendo cuando el ladrón entró, O
João dormia/estava dormindo quando o ladrão entrou, ‘John was sleeping when
the burglar broke in’). No previous research to our knowledge has examined the
acquisition of Spanish aspectual values among heritage speakers of Spanish with
BP as dominant language.
The study is organized as follows: Section 2 discusses the differences
between the simple and progressive forms in the present and the past tenses in
Spanish, English and BP. Section 3 presents our research questions and
hypotheses. Section 4 describes our study and methodology. Section 5 shows the
results, followed by the discussion and conclusions in Section 6.

2. Theoretical background

2.1 Tense and aspect

Tense is a deictic feature that connects the time of the referred situation with
another moment, usually with the moment of speaking (Comrie, 1976). It can be
present, past, and future. Aspect, on the other hand, is not deictic and refers to
the different ways to see the internal constituency of a situation (Comrie, 1976).
Tense and aspect refer to time, but tense involves an external relation in time,
whereas aspect informs about the internal temporal structure of the situation.
Lexical aspect has been defined as the aspectual information provided by
the lexical properties of the verbs and their predicates (Colomé, 2013; Vendler,
1967): punctuality, telicity, and dynamicity. According to their lexical aspect,
Vendler (1967) classifies verbs and their predicates into four categories: states,
activities, accomplishments, and achievements. Table 1 presents the four
categories in which verbs can be classified according to their semantic features:
punctuality, telicity, and dynamicity:

Table 1. Lexical aspect and semantic features (Vendler, 1967)

Features States Activities Accomplishments Achievements

Punctuality − − − +
Telicity − − + +
Dynamicity − + + +

2.2 Semantic constraints in the selection of the Spanish present tense

The Spanish present tense allows for a wide spectrum of aspectual values that
may precede or follow the speech act (Alarcos Llorach, 1994; Yllera Fernández,
1999). The Spanish simple present can adopt a habitual meaning, an ongoing
meaning, and a historical present interpretation, among others. The Spanish
present progressive, on the other hand, can have an ongoing meaning and allows
for a temporary-habitual meaning (Schmitt, 2001; Yllera Fernández, 1999).
In contrast with Spanish, the simple present in both English and BP do not
allow an ongoing interpretation, as only the present progressive has an ongoing
reading (i.e., Julia is playing/*plays now; A Júlia está brincando/*brinca
agora). However, the present progressive can have a temporary-habitual
meaning in both English and BP. Table 2 summarizes the aspectual values of the
simple present and the present progressive in the three languages:

Table 2. Aspectual differences in the present tense: Spanish, English, and BP

Aspectual values Spanish English BP

Simple present [+ongoing] Julia juega ahora * *

[+habitual] Julia juega todos los Julia plays A Júlia brinca todos os días
días everyday
Present [+ongoing] Julia está jugando ahora Julia is playing A Júlia está brincando agora
progressive now
[+habitual] Julia está jugando Julia is playing A Júlia está brincando
últimamente lately ultimamente

As shown in Table 2, the present progressive can select both an ongoing and a
temporary-habitual reading in all three languages. Additionally, the Spanish
simple present allows for an ongoing interpretation.

2.3 Semantic constraints on the selection of the imperfect progressive

in Spanish and BP
In both the Spanish and the BP past tenses, there are two different simple forms
with different aspectual values: the preterit, which depicts a completed event or
state, and the imperfect, which selects a habitual or ongoing interpretation. On
the other hand, English only has one simple past form, which does not select for
any specific aspectual reading. Furthermore, an imperfective interpretation can
be reached in English by using the periphrases such as used to or would (e.g.,
The neighbor used to/would visit the grandmother every evening). An
imperfective ongoing reading can be expressed in the three languages with the
use of a progressive form. In Spanish and BP, the imperfect progressive, as well
as the imperfect, can select for imperfective ongoing meanings, whereas in
English the only option is the past progressive, as represented below:

(1)
1. Spanish
Mi hermano estaba cantando cuando nuestra hermana llamó.
2. Brazilian Portuguese: (imperfective ongoing)
Meu irmão estava cantando quando a nossa irmã ligou.
3. English:
My brother was singing when our sister phoned.

In Spanish and BP, the imperfect can select for the readings above in addition to
its habitual reading:

(2)
1. Spanish
Mi hermano cantaba cuando nuestra hermana llamó.
‘My brother was singing when our sister (imperfective ongoing)
arrived.’

2. Brazilian Portuguese
Meu irmão cantava quando a nossa irmã ligou.
‘My brother was singing when our sister phoned.’

The Spanish and BP imperfect tense selects both habitual and ongoing
imperfective readings. On the other hand, the imperfect progressive in BP and
Spanish as well as the past progressive in English select for ongoing
imperfective readings, although both Spanish and BP, as opposed to English,
also have simple imperfect morphologies to express ongoing imperfective
readings. This study focuses on the acquisition of these aspectual differences in
Spanish, crucially ongoing readings. Table 3 summarizes the aspectual values in
the past in Spanish, English, and BP:

Table 3. Aspectual differences in the past tense: Spanish, English, and BP

Aspectual values Spanish BP English

Perfective [completed] Ana leyó un libro A Ana leu um livro Ana read a book
Imperfective [habitual] Ana siempre leía un A Ana lia um livro Ana always read a
libro sempre book
* * Ana used to/would
read a book.
[ongoing] Ana leía un libro A Ana lia um livro *
cuando llamé. quando liguei.
Ana estaba leyendo un A Ana estava lendo um Ana was reading a
libro cuando llamé. livro quando liguei. book when I phoned.

As shown in Table 3, Spanish and BP have two options to express ongoing

readings in the past: imperfect and imperfect progressive. English, in contrast,
only has the past progressive.

3. Previous acquisition research

3.1 The Spanish present progressive

The aspectual distribution of the present progressive in Spanish as a heritage

language has been previously examined in various studies (Cuza, 2010; Cuza &
López Otero, 2016; Geeslin & Fafulas, 2012; Klein, 1980; Sánchez-Muñoz,
2004). Klein (1980) compares two populations of Puerto Rican immigrants
living in New York City: heritage speakers and L1 Spanish late bilinguals. Using
semi-spontaneous conversation and a picture description task, the author
investigated the role of transfer from English into Spanish with regard to
aspectual selection in the present tense and found a narrowing of the simple
present towards an exclusively habitual reading and a spread of the use of the
present progressive when expressing all ongoing readings. This was confirmed
by the results, which show that the heritage group used the present progressive
significantly more than the L1 Spanish late bilingual group when expressing
ongoing readings instead of using the simple present with an ongoing value. A
similar question motivated Sánchez-Muñoz’s (2004) study. The author examined
the role of cross-linguistic influence from English present progressive into
Spanish. She implemented a picture description task and an interview among
Spanish-English bilinguals living in Los Angeles. The author concluded that the
bilinguals, particularly the heritage speakers, overextend the use of the Spanish
present progressive in contexts where monolinguals would use the simple
present with an ongoing value.
Recently, Cuza and López Otero (2016) examined the acquisition of the
aspectual values of the simple present and the present progressive among
Spanish heritage speakers and L2 learners. The authors implemented an elicited
production task, an acceptability judgment task, and a forced preference task.
The findings indicate that the experimental groups, especially the L2 learners,
overextend the scope of the simple present in contexts where native speakers
preferred the present progressive. The authors conclude that these findings may
be the result of a simpler aspectual configuration where the less marked form,
the simple present, has increased its scope at the expense of the present
progressive.
Furthermore, Cuza (2010) and Geeslin and Fafulas (2012) also examined
the aspectual distribution of the present progressive in Spanish and its
acquisition. However, their experimental groups did not include heritage
speakers. Instead, they focused on L2 learners and long-term immigrants. Cuza
(2010) is the first study to investigate the acquisition of the selectional properties
of the Spanish simple present in L2 speakers and Spanish long-term immigrants
in New Jersey and Toronto. The author implemented a written acceptability
judgment task, a truth-value judgment task and an oral narrative based on the
wordless book Frog Story to examine four conditions: both the simple present
and the present progressive with their respective ongoing and habitual
interpretations. The author followed de Swart’s (1998) selectional approach to
aspectual variation to account for the differences in the selectional properties of
the simple present and the present progressive of the bilinguals. The results are
consistent with those found among heritage speakers (Klein, 1980; Sánchez-
Muñoz, 2004): that is, reduction of the selectional scope of the simple present to
a habitual reading and a more categorical use of the present progressive in
ongoing contexts.
More recently, Geeslin and Fafulas (2012) studied the linguistic variables
that would constrain the use of the simple present or the present progressive,
such as lexical aspect, animacy of the subject, and clause type, among others.
They implemented a multiple-choice test and a video-narration activity to a
group of L2 speakers and native speakers serving as control baseline. Both
groups behaved similarly, although the use of the present progressive was more
common in the L2 learners. However, in general terms, both groups produced
the simple present in most of the instances. The authors argue that L2 learners
can acquire the constraints that rule the use of simple present and the present
progressive.
To summarize, most of previous research indicates overextension of the
present progressive to express ongoing meanings as well as reduction of the
semantic values of the simple present as a result of cross-linguistic influence
from English. However, other studies have found that Spanish-English bilinguals
produced the simple present more. The present study further examines the
distribution of these forms in English-Spanish bilinguals, but also compares the
production of these bilinguals with that of BP-Spanish bilinguals.

3.2 The Spanish imperfect progressive

The aspectual distribution of the imperfect progressive and the imperfect in

Spanish as a heritage language remains underexplored. Most previous research
has documented semi-spontaneous speech in heritage speakers of Spanish in
Texas (Chaston, 1987; Lavandera, 1981; Solé, 1977). Lavandera’s (1981)
seminal study documents an increase in the use of the imperfect progressive in
heritage speakers of Spanish. The author analyzed the speech of nine heritage
speakers of Spanish in a family gathering. The results show that, besides the fact
that most productions included instances of code-switching, the imperfect
progressive was more frequent than the imperfect. Lavandera argues that the
increase in the use of auxiliary verb forms (i.e., progressive forms) to the
expense of more morphologically simple forms (i.e., the imperfect) is the result
of contact with English, as auxiliaries are more frequent in English and they
express tense and aspect. Furthermore, the author indicates that code-switching
scenarios can trigger such morphological distribution.
Chaston (1987) examines the speech of 18 heritage Spanish-speaking
college students. Specifically, the author looks at both the preterit and imperfect
production and correlated their performance with their overall proficiency and
sociolinguistic factors, such as usage and attitudes. The findings indicate that,
contrary to previous research (Lavandera, 1981; Solé, 1977), the imperfect
progressive is not gaining ground over the imperfect. However, the author notes
that, in ongoing actions in the past, over half of the events are expressed with the
imperfect progressive. Following Solé (1977), Chaston argues that the
progressive form may be used to emphasize the ongoingness of the event. The
author does not attribute this phenomenon to language contact or to cross-
linguistic effects from English and calls for further investigation. On the other
hand, Mrak (1998) looks at the imperfect forms in the narratives of nine heritage
speakers of Spanish living in Houston, TX and finds that the third generation
uses the imperfect progressive more than the first and second generations.
More recently, this phenomenon has been examined in bilingual
populations. Lamanna (2008) examines the distribution of the imperfect and the
imperfect progressive in ongoing contexts, where they are interchangeable.
Lamanna examines written data from a corpus of U.S. Spanish and from corpora
of Mexican, Cuban and Puerto Rican Spanish in order to control for potential
differences in the U.S. variety. In absolute terms, the results are inconclusive, but
by taking a closer look the findings indicate that ongoing and continued actions
are more usually expressed with the imperfect progressive in the U.S. variety
than in the monolingual corpora. Finally, the author suggests that there may be a
relation between the choice of verb form and the lexical frequency.
To summarize, most previous research documents an overextension of the
use of the imperfect progressive in heritage Spanish in the United States.
However, Dumont and Wilson (2016) found different results. They examined the
use of the imperfect and the imperfect progressive by comparing two corpora of
spoken Spanish from Spanish-English bilinguals in New Mexico and
monolinguals from Ecuador. The authors found out that the distribution of the
imperfect forms is not changing in New Mexican Spanish to become structurally
more similar to English. However, the constraints ruling the use of each form
seem to be weakening in the New Mexican data more than in the monolingual
variety. The present study examines the use and distribution of the Spanish
imperfect progressive in English-Spanish bilinguals, but also compares their
production to BP-Spanish bilinguals.

3.3 Research question and hypothesis

As mentioned earlier, the goal of the present study is to examine the use and
distribution of present and imperfect progressive forms in English-speaking and
BP-speaking heritage speakers of Spanish. More specifically, we are interested
in examining to what extent heritage speakers of Spanish with BP and English as
their dominant language use present and imperfect progressive forms, instead of
simple present and imperfect forms with activity and accomplishment verbs.
And if difficulties are found, whether they can be accounted for in terms of
crosslinguistic influence from their dominant language. We pose the following
research question:

RQ:
Do heritage speakers of Spanish show crosslinguistic influence form their
dominant language when facing the possibility of using two different forms
that have the same semantic reading in a given context?

We hypothesize that both English-speaking and BP-speaking heritage speakers

of Spanish will show crosslinguistic influence from their dominant languages
when facing the possibility of using either the imperfect or the imperfect
progressive with activity and accomplishment verbs. We also expect to find
differences in verb form as a function of tense across the groups. Specifically,
the use and distribution of the past tense forms will be different across groups
due to different selectional properties of past tense forms in English and BP. On
the other hand, the use and distribution of the present tense forms will be similar
across groups due to the fact that the readings of the present tense forms, both
simple and progressive, are similar in English and BP.
4. The experiment

4.1 Participants

Twenty-four participants took part in the study: 9 English-speaking heritage

speakers of Spanish and 15 BP-speaking heritage speakers of Spanish. The
participants completed a language history questionnaire about their linguistic
background, their patterns of language use and their self-assessment of their
skills in their languages (Cuza, 2013; Cuza & López Otero, 2016): their heritage
language, Spanish, and in their dominant languages. Furthermore, the
participants completed a modified version of the DELE language proficiency test
for Latin American Spanish (Cuza, Pérez-Leroux & Sánchez, 2013) in order to
guarantee that all participants were similar in terms of proficiency.1 The English-
speaking participants were tested at the principal investigator’s language
acquisition lab, whereas the BP-speaking participants were tested in public
places in São Paulo, Brazil.
The English-speaking group consisted of Spanish heritage speakers born
and raised in the U.S., except for two (mean age at testing, 19 years old; age
range, 18–22).2 Their parents were born in Mexico, Argentina, the U.S. and Peru.
Their mean score in the DELE proficiency test was 41/50. Regarding their
patterns of language use, 56% (5/9) of them reported speaking ‘Spanish’ or
‘mostly Spanish’ at home, 33% (3/9) reported speaking ‘equal English and
Spanish’ and 11% (1/9) reported speaking ‘mostly English’ or ‘slightly more
English’. Most of the participants reported using more English at school, work,
and social situations, and 66% (6/9) indicated feeling comfortable in both
English and Spanish; the other 33% (3/9) indicated feeling more comfortable in
English. Their reported self-proficiency was almost native-like (3.7/4) in English
and almost good/fluent (3.1/4) in Spanish.
 The BP-speaking group consisted of Spanish heritage speakers born and
raised in Brazil, except for two (mean age at testing, 32 years old; age range, 21–
55). Their parents were born in Chile, Argentina, Spain, Paraguay, Bolivia and
El Salvador. Their mean score in the DELE proficiency test was 45/50.
Regarding their patterns of language use, 73% (11/15) reported speaking ‘mostly
Portuguese’ or ‘only Portuguese’, while 20% (3/15) reported speaking ‘equal
Portuguese and Spanish’ or ‘slightly more Portuguese’ and only 7% (1/15) of
them reported speaking ‘Spanish’ or ‘mostly Spanish’ at home. Portuguese is
also more present in other contexts: most of the participants reported using more
Portuguese at school, work, and social situations, and 87% (13/15) indicated
feeling more comfortable in Portuguese; the other 13% (2/15) indicated feeling
comfortable in both Portuguese and Spanish. They reported having almost
native-like proficiency in Portuguese (3.7/4) and almost good/fluent proficiency
in Spanish (3/4).

4.2 Methods and structures under analysis

The goal of the present study is to examine the use and distribution of present
and imperfect progressive forms in English-speaking heritage speakers of
Spanish and BP-speaking heritage speakers of Spanish. Specifically, we examine
contexts with activity and accomplishment verbs:
1. Activity verbs:

(5)
El lobo estaba dormiendo.
‘The wolf was sleeping.’

(6)
Mientras el cazador pasaba cerca de la casa.
 ‘While the hunter was passing by the house.’

2. Accomplishment verbs:

(7)
Y cuando estaba agarrando flores por su abuelita, el lobo se fue.
‘And when she was picking up flowers for her grandma, the wolf left.’

(8)
Se acercó porque le preguntaba cosas.
‘She got closer because he was asking her things.’

The verbal forms analyzed in this study were extracted from a semi-spontaneous
task (oral narrative). Following previous research (Cuza, 2010; Montrul, 2002b;
Montrul & Potowski, 2007), the participants were asked to narrate The Little Red
Riding Hood in Spanish. The participants were presented with a wordless
storybook of The Little Red Riding Hood. The book had a total of 10 pages that
the participants could freely flip. Before starting the oral production, the
participants were told to take a look at the wordless storybook to become
familiar with the story. The participants were asked to narrate the story freely.
They were not asked to produce any specific form or to narrate the story in a
certain tense. This oral narrative was the first task to be implemented from a
series of other tasks that examined other phenomena. The narratives were
digitally recorded for later transcription and analysis. Each verb in the narratives
was coded for verb form (simple, progressive), lexical aspect (state, activity,
accomplishment, achievement), tense (present, past), and participant group
(English-Spanish, BP-Spanish).

5. Results
The participants produced a total of 430 verbs in the forms under examination:
simple present (SP), present progressive (PP), imperfect (IM), and imperfect
progressive (IP). In both the present and past tenses, both groups used the simple
forms more than the progressive forms. Specifically, the BP-speaking heritage
speakers of Spanish showed a tendency to use the simple present and the
imperfect more than the progressive forms (SP = 98.2% vs PP = 1.8%;
IM = 93.5% vs IP = 6.5%). As shown in Figure 1, this tendency was stronger
than in their English-speaking counterparts, who, although they also used the
simple forms more, did so in a less categorical fashion (SP = 92.7% vs PP 7.3%;
IM = 85.4% vs IP = 14.5%).

Figure 1. Overall tendency of use of verb forms under examination per group
(n = 430)
The English-speaking group used simple forms 85.5% of the time and
progressive forms in 14.5% of the instances. On the other hand, the BP speakers
showed a more categorical pattern: they produced simple forms 94.7% of the
time, while they only used progressive forms 5.3% of the time. However, these
overall tendencies do not reflect the context in which the simple and the
progressive forms are interchangeable. If we take a closer look to those contexts,
the tendencies of use appear more clearly. Specifically, the contexts in which the
simple and progressive forms are interchangeable are those in which their lexical
aspect is either activity or accomplishment. In the narratives, there were 70
instances of activity or accomplishment verbs. Figure 2 shows the tendencies of
use of the verb forms under examination when expressing activities or
accomplishments.

Figure 2. Tendency of use of verb forms under examination per group: Activity
and accomplishment verbs (n = 70)
The results for activity and accomplishment verbs showed clear tendencies in
both groups. The English-speaking group produced more progressive tokens in
both the present and past tenses (SP = 33% vs PP = 67%; IM = 36% vs
IP = 64%), whereas the BP-speaking group produced more simple forms in both
tenses (SP = 62% vs PP = 38%; IM = 60% vs IP = 40%). In addition, the data
were analyzed using generalized linear models with a binomial link function in
order to assess the predictability of the simple or progressive forms. For the
models, participants were random intercepts, and group (English-Spanish, BP-
Spanish), tense (present, past) and lexical aspect (activity, accomplishment) were
fixed factors. All models included form by group interaction. The statistical
significance of form, group, and the “verb form” by group interaction were
assessed using hierarchical partitioning of variance via nested model
comparisons.
Figure 3. Probability of using progressive forms as a function of group: English-
Spanish, BP-Spanish
The panel in Figure 3 plots the probability of using progressive forms, in
opposition to simple forms, as a function of group: either English-speaking or
BP-speaking heritage speakers of Spanish. The function represents the rate of
change from one form to the other in the probability space. The results for the
contrast between simple and progressive forms revealed a main effect of group
(χ(1) = 4.08; p = 0.04), but not of tense (χ(1) = 0.07; p = 0.78) or lexical aspect
(χ(1) = 1.32; p = 0.24). Therefore, the use of simple or progressive forms was
determined only by the language group of the participants, and not by the tense
or the lexical aspect of the verb. The English-Spanish bilinguals showed a
tendency to prefer the progressive forms, while the BP-Spanish bilinguals used
more simple forms.
Overall, in the narratives the simple forms were more frequent than the
progressive. However, a closer look was necessary to perceive the tendencies
within groups. The results reveal that, in activity and accomplishment verbs, the
two groups of heritage speakers of Spanish are significantly different: the
English-speaking group has a tendency to use progressive forms, whereas the
BP-speaking group uses more simple forms.
6. Discussion and conclusions
The goal of the present study was to unveil the tendencies of use of progressive
versus simple forms in two groups of Spanish heritage speakers: an English-
dominant group and a BP-dominant group. For this purpose, we analyzed their
oral production via a semi-spontaneous production task. We took a closer look at
activity and accomplishment verbs and at the forms employed in such verbs (i.e.,
simple or progressive). We predicted differences between the two groups
regarding the use of the present or progressive forms in activity and
accomplishment verbs. Furthermore, we expected the experimental groups to
behave differently due to crosslinguistic influence from their dominant
languages.
As shown in the results section, the two groups display a statistically
significant difference in their tendencies of use when contrasting the progressive
forms with their simple counterparts in both the present and past tenses.
Specifically, with activity and accomplishment verbs, the English-speaking
group used progressive forms 65% of the time, whereas the BP-speaking group
did so only 39% of the time. Furthermore, the overall results, which include
verbs other than activity and accomplishment verbs, are also consistent with this
tendency. Indeed, the English-speaking group used progressive forms 14.5% of
the time, whereas the BP-speaking group did so only 5.3% of the time. The
overall tendencies of the groups lead to believe that there might be
crosslinguistic effects from their dominant languages. This view is supported by
the results from the analysis of the narratives in their native language. BP
speakers, whose language also allows for both simple and progressive forms in
ongoing activities and accomplishments, produced a similar pattern with these
verbs. They produced simple forms 58% of the time, in opposition to 42% of
progressive forms. Below are some examples of the production of participants
from both groups:

(1)
English-speaking heritage speakers of Spanish:
1. El lobo le preguntó dónde estaba yendo, qué estaba haciendo, por qué
iba por el bosque sola…
‘The wolf asked her where she was going, what she was doing, why she
was going through the forest alone…’
2. Cuando se estaba comiendo a Caperucita, justo había un leñador que
estaba pasando y escuchó los gritos.
‘While he was eating Little Red Riding Hood, there was a lumberjack
passing by and heard the screams.’
3. pero después un hombre afuera de la casa vio que la estaba atacando y
fue y la rescató.
‘But later a man from outside the house saw that he was attacking her
and rescued her.’
4. Entonces, ella estaba caminando por el bosque y se encontró con un
lobo.
‘Then, she was walking through the forest and encountered a wolf.’
5. El lobo estaba hablando con ella y le dijo al lobo que iba a ir a casa de
su abuela.
‘The wolf was talking to her and she told the wolf that she was going to
her grandma’s house.’

(2)
BP-speaking heritage speakers of Spanish:
1. En el camino, encontrose un lobo, un lobo malo que preguntó a ella o
 que ella estaba haciendo.
‘On the way, she encountered a wolf, an evil wolf that asked her what she
was doing.’
2. Mientras ella tomaba flores en la floresta, el lobo fue a la casa de su
abuela.
‘While she was picking flowers in the forest, the wolf went to her
grandma’s house.’
3. Era una vez, Caperucita Roja, que estaba paseando por la floresta…
‘Once upon a time, Little Red Riding Hood, who was walking through
the forest…’
4. El lobo también se traga a la nena, pero pasaba por allí un señor…
‘The wolf swallows up the girl too, but a gentleman was passing by..’
5. …un cazador, y se da cuenta de lo que está pasando.
‘…a hunter, and he notices what is happening.’

The results of the generalized linear models indicate that neither tense nor lexical
aspect is significant in relation to the use of one form or the other, in contrast to
what was expected. However, our results do confirm that English-speaking
heritage speakers of Spanish use progressive forms more than their BP-speaking
counterparts. This corroborates previous research documenting overextension of
the present progressive in heritage speakers of Spanish with English as dominant
language (Klein, 1980; Sánchez-Muñoz, 2004). Our data suggest that the
English-speaking group may experience crosslinguistic effects from their
dominant language. In the past tense, on the other hand, our results are consistent
with those found for the present tense, as well as with those found in previous
research (Chaston, 1987; Dumont & Wilson, 2016; Lamanna, 2008; Lavandera,
1981; Mrak, 1998). Progressive forms are more used by English-speakers, while
BP-speakers employ more simple forms.
 Previous research discussing the high use of the imperfect progressive in
English-speaking heritage speakers of Spanish claim that it may be triggered by
code-switching (Lavandera, 1981) or by aspectual differences with the imperfect
(Chaston, 1987). However, we did not find any case of code-switching in our
data, and further research documenting this phenomenon in different generations
of bilinguals (Mrak, 1998) as well as in monolinguals (Dumont & Wilson, 2016;
Lamanna, 2008) have not found any aspectual difference between the imperfect
and the imperfect progressive in bilingual populations.
These results can be accounted for by Putnam and Sánchez’s (2013) work.
Putnam and Sánchez (2013) argue that the differences between heritage speakers
and monolinguals can be explained in terms of frequency of activation. Heritage
speakers, usually dominant in the majority language, do not activate their
heritage language for production and comprehension purposes as frequently as
monolinguals do, which weakens the associations existing between lexical items
and their syntactic and semantic features, as well as their morphology. This view
is consistent with Lardiere’s (2008, 2009) feature re-assembly hypothesis, which
argues that, although bilinguals may have the syntactic knowledge of a certain
structure, their production of such structure might due to morphological
competence. Additionally, Jiang (2000) argues that morphology is the most
challenging lexical component to acquire. The differences between English-
speaking and BP-speaking heritage speakers may be due to morphological
competence. Specifically, when presented with two possible verb forms in
activity and accomplishment verbs, the English-speaking group tends to use the
form that is morphologically closest to the progressive forms in English. Note
that the use of the Spanish imperfect progressive in activity and accomplishment
verbs is grammatical and that there was not a case of ungrammatical use of the
imperfect progressive in any other context in the data set. The BP speakers, on
the other hand, used simple forms more than the progressive forms, and an
analysis of their narratives in BP showed that they follow a similar verb form
distribution in their native language. This might be an indicator that BP-Spanish
bilinguals also experience crosslinguistic effects. Additionally, this group
difference may also be explained in morphological terms if we consider that
Spanish and BP share a considerable part of their morphology, unlike English.
Therefore, the different distribution in the groups might also be the result of
structure avoidance due to lack of morphological competence of the imperfect
tense among the English-speaking heritage speakers. Finally, we did not find
significant differences in the use of a certain verb form as a function of tense,
against our expectations. We expected to find differences across groups and
tense because the selectional properties of English and BP present tense are
similar, whereas they differ in the past tense.
To conclude, our findings suggest that in contexts with activity and
accomplishment verbs, these two groups of Spanish heritage speakers experience
crosslinguistic effects from their dominant languages in different ways. The
English speakers overextend the progressive forms, while the BP speakers
overextend use simple present verb forms. We argue that these speakers,
particularly the English speakers, when presented with two grammatical options
in the input, tend to use the form that is morphologically closer to their dominant
language. In addition to crosslinguistic influence effects, we argue that the
patterns of overextension are also conditioned by the avoidance of marked
forms. The imperfect, in opposition to the imperfect progressive, is a marked
form that presents difficulties in heritage speakers, as it is learned later than the
preterit in both L1 and L2 acquisition (Montrul, 2002b). The imperfect
progressive, despite being a compound verb form, is less challenging than the
imperfect.
Notes
1. This proficiency test is composed of a vocabulary task from the MLA Foreign Language Test and cloze
test from the Diploma de Español como Lengua Segunda (DELE) test (Bruhn de Garavito, 2002; Duffield
& White, 1999; Montrul & Slabakova, 2003). We implemented Cuza, Pérez-Leroux & Sánchez’s (2013)
modified version of the original test, which has adapted some lexical items in the vocabulary section and
uses a completely different cloze section. ①

2. Among the English-speakers, one participant was born in Mexico and came to the USA before the age of
one; another participant was born in Argentina and immigrated to the USA at the age of ten. In the BP-
speaking group, one participant was born in Chile and went to Brazil at the age of two; another participant
was born in Argentina and went to Brazil at the age of nine. ①

References
Alarcos Llorach, E.
(1994.) Gramática de la lengua española. Madrid: Espasa Calpe. ①
Bruhn de Garavito, J.
(2002.) Verb raising in Spanish: A comparison of early and late bilinguals. In Proceedings of the 26th
annual Boston University conference on language development (pp. 84–94). ①
Chaston, J. M.
(1987.) Aspect choice in preterit and imperfect usage in the speech of Mexican-American college students
in Texas: a sociolinguistic approach (Unpublished doctoral dissertation). University of Texas at Austin.
①②③④
Colomé, L. C.
(2013.) Tense and aspect in second language Spanish. In K. L. Geeslin (Ed.), The handbook of Spanish
second language (pp. 235–252). Chichester, UK: John Wiley & Sons. ①
Comrie, B.
(1976.) Aspect: An introduction to the study of verbal aspect and related problems. Cambridge: Cambridge
University Press. ① ②
Cuza, A.
(2013.) Crosslinguistic influence at the syntax proper: Interrogative subject–verb inversion in heritage
Spanish. The International Journal of Bilingualism, 17, 71–96. ①
Cuza, A.
(2010.) The L2 acquisition of aspectual properties in Spanish. Canadian Journal of Linguistics, 55(2),
1001–1028. ①②③④⑤⑥
Cuza, A., & López Otero, J. C.
(2016.) The acquisition of the semantic values of the Spanish present tense in L2and heritage Spanish
Revista Española de Lingüística Aplicada, 29(2), 462–486. ①②③④⑤⑥
Cuza, A., & Miller, L.
(2015.) The protracted acquisition of past tense aspectual values in child heritage Spanish. In R. Klassen
 J. M. Liceras, & E. Valenzuela (Eds.), Hispanic linguistics at the crossroads: Theoretical linguistics,
language acquisition and language contact (pp. 211–229). Amsterdam: John Benjamins.
①
Cuza, A., Pérez-Leroux, A. T., & Sánchez, L.
(2013.) The role of semantic transfer in clitic-drop among simultaneous and sequential Chinese-Spanish
bilinguals. Studies in Second Language Acquisition, 35(1), 93–125. ①②
De Swart, H.
(1998.) Aspect shift and coercion. Natural Language and Linguistic Theory, 16, 347–85. ①
Dumont, J., & Wilson, D. V.
(2016.) Using the variationist comparative method to examine the role of language contact in synthetic and
periphrastic verbs in Spanish. Spanish in Context, 13(3), 394–419. ①②③
Geeslin, K., & Fafulas, S.
(2012.) Variation of the simple present and present progressive forms: A comparison of native and non-
native speakers. In K. Geeslin & M. Díaz-Campos (Eds.), Selected proceedings of the 14th Hispanic
Linguistics Symposium (pp. 179–196). Somerville, MA: Cascadilla Press. ① ② ③ ④ ⑤
Jiang, N.
(2000.) Lexical representation and development in a second language. Applied Linguistics, 21(1), 47–77.
①②
Klein, F.
(1980.) A quantitative study of syntactic and pragmatic indicators of change in the Spanish of bilinguals in
the U.S. In W. Labov, (Ed.), Locating language in time and space (pp. 69–82). New York, NY:
Academic Press. ① ② ③ ④ ⑤
Lamanna, S.
(2008.) Usage of imperfect and imperfect progressive verb forms in Spanish as a majority and minority
language: Is there an effect for language contact? In J. Bruhn de Garavito & E. Valenzuela (Eds.),
Selected proceedings of the 10th Hispanic Linguistics Symposium (pp. 251–264). Somerville, MA:
Cascadilla Proceedings Project. ① ② ③
Lardiere, D.
(2008.) Feature-assembly in second language acquisition. In J. Liceras, H. Zobl & H. Goodluck (Eds.),
role of formal features in second language acquisition (pp. 106–140). Mahwah, NJ: Lawrence
Erlbaum Associates. ①
Lardiere, D.
(2009.) Some thoughts on the contrastive analysis of features in second language acquisition. Second
Language Research, 25, 173–227. ①
Lavandera, B. R.
(1981.) Lo Quebramos, but only in performance. In R. P. Durán (Ed.), Latino language and communicative
behavior (pp. 49–67). Norwood, NJ: Ablex. ① ② ③ ④ ⑤
Montrul, S.
(2004.) Subject and object expression in Spanish heritage speakers: A case of morpho-syntactic
convergence. Bilingualism: Language and Cognition, 7, 125–142. ①
Montrul, S.
 (2008.) Incomplete Acquisition in Bilingualism: Re-examining the Age Factor. Amsterdam: John
Benjamins. ①
Montrul, S.
(2002a.) Competence and performance differences between monolinguals and 2nd generation bilinguals in
the tense/aspect domain. In J. F. Lee, K. L. Geeslin, & J. C. Clements (Eds.), Structure, meaning, and
acquisition in Spanish (pp. 93–114). Somerville, MA: Cascadilla. ①
Montrul, S.
(2002b.) Incomplete acquisition and attrition of Spanish tense/aspect distinctions in adult bilinguals.
Bilingualism: Language And Cognition, 5(1), 39–68. ①②
Montrul, S. & Potowski, K.
(2007.) Command of gender agreement in school-age Spanish bilingual children. International Journal of
Bilingualism, 11(3), 301–328. ①
Montrul, S., & Slabakova, R.
(2003.) Competence similarities between native and near-native speakers: An investigation of the preterite-
imperfect contrast in Spanish. Studies in Second Language Acquisition, 25(3), 351–398. ①
②
Mrak, A.
(1998.) El discurso del pasado en el español de Houston: Imperfectividad y perfectividad verbal en una
situación de contacto. Southwest Journal of Linguistics, 17, 115–128. ① ② ③
Polinsky, M.
(2011.) Reanalysis in adult heritage language: New evidence in support of attrition. Studies in Second
Language Acquisition, 33(2), 305–328. ①
Putnam, M., & Sánchez, L.
(2013.) What’s so incomplete about incomplete acquisition? – A prolegomenon to modeling heritage
language grammars. Linguistic Approaches to Bilingualism, 3, 478–508. ①②③
Sánchez-Muñoz, A.
(2004.) Transfer in the Spanish progressive constructions in Los Angeles. USC Working Papers in
Linguistics, 2, 16–29. ① ② ③ ④ ⑤ ⑥
Schmitt, C.
(2001.) Cross-linguistic variation and the present perfect: The case of Portuguese. Natural Language &
Linguistic Theory, 19(2), 403–453. ①②
Solé, Y.
(1977.) Continuidad/descontinuidad idiomática en el español tejano. The Bilingual Review, 4, 189–199.
②③
Valdés, G.
(2001.) Heritage language students: Profiles and possibilities. In J. Peyton, D. Ranard, & S. McGinnis
(Eds.), Heritage languages in America: Preserving a national resource (pp. 37–80). McHenry, IL:
Center for Applied Linguistics. ①
Vendler, Z.
(1967.) Linguistics in philosophy. Ithaca, NY: Cornell University Press. ① ② ③ ④
Yllera Fernández, M. A.
(1999.) Las perífrasis verbales de gerundio y participio. In I. Bosque & V. Demonte (Eds.), Gramática
descriptiva de la lengua española (pp. 3391–3441). Madrid: Real Academia Española, Colección
Nebrija y Bello, Espasa Calpe. ① ②
CHAPTER 10

Portuguese-Spanish contacts in Misiones,

Argentina
Probing (for) code-switching constraints
John M. Lipski
The Pennsylvania State University

Abstract
The present study reports on spontaneous Portuguese-Spanish mixing
as produced by L2 and balanced bilinguals in various communities
along the Brazilian border, as well as three interactive tasks conducted
with fluent bilinguals in Misiones province, Argentina. Taken together
the results reveal a residue of permeable but grammatically-grounded
constraints even between morphosyntactically and lexically cognate
sibling languages and among individuals who do not routinely code-
switch. These constraints are more robustly maintained among fluent
bilinguals but are sometimes contravened by L2 speakers. The data
also demonstrate that code-switching does not emerge automatically in
bilingual settings involving highly congruent languages.

Keywords
Portuguese; Spanish; codeswitching constraints; bilingualism;
language processing

1. Introduction: Grammatical constraints on intra-

sentential code-switching(?)
Many bilingual speech communities are characterized by frequent language
switching, which may take the form of intra-sentential switches that smoothly
blend elements of both languages in a fashion that differs substantially from the
halting mixtures produced by second-language learners. Although in principle
fluent bilingual code-switching can occur at any juncture, cross-linguistic
observations covering a variety of bilingual environments converge on the
existence of configurations that either favor or inhibit fluid switching. Early
attempts to account for these observed tendencies focused on overall
grammatical congruence of the languages involved, including the need to avoid
grammatical violations in either language and to ensure that post-switch
syntactic patterns do not conflict with logical extensions of the pre-switch
language (e.g. Lipski, 1978, 1982, 1985; Poplack, 1980; Wakefield et al., 1975).
These early findings were expanded to include formal notions of linguistic
structure, first within classic transformational grammar (e.g. Rivas, 1981;
Sankoff & Poplack, 1981), then the principles and parameters framework (e.g.
Belazi, Rubin, & Toribio, 1994; Bentahila & Davies, 1983; D’Introno et al.,
1991; DiSciullo, Muysken, & Singh, 1986; Doron, 1983; Dussias, 2003;
Halmari, 1997; Klavans, 1985; Toribio, 2001a, 2001b; Woolford, 1983), and
more recently in the Minimalist model (e.g. Jake, Myers-Scotton, & Gross,
2002; MacSwan, 1999, 2000, 2004, 2005; van Gelderen & MacSwan, 2008;
Hebblethwaite, 2007, 2010 offers a divergent opinion). Also figuring in the
earliest studies – many of which involved Spanish-English code-switching in the
United States – were claims that switches were inhibited following certain
grammatical categories: between subject pronouns and verbs (Gumperz,
1977, p. 26; Koronkiewicz, 2014; Timm, 1975, p. 477); after (fronted)
interrogative words (D’Introno et al., 1991, p. 403; Ebert, 2014, pp. 192–198;
Peñalosa, 1980, p. 58; Woolford, 1983, p. 531; 1984); between negative
adverbials and verbs (Timm, 1975, p. 479); between object clitic and verb (Pfaff,
1979); between auxiliary verb and infinitive (Muysken, 2000, pp. 12–13;
Peñalosa, 1980, p. 65; Timm, 1975, pp. 478–479). Also implicated are closed-
class items (Azuma & Meijer, 1997; Doron, 1983; Jake, 1994; Joshi, 1985;
Prince & Pintzuk, 2000), including system morphemes (Myers-Scotton, 1993).
Examples of nominally disallowed Spanish-English code-switches are given in
(1).

(1)
BETWEEN SUBJECT PRONOUN AND PREDICATE:

*nosotros are traveling to Detroit tomorrow.

*We viajamos a Detroit mañana.

BETWEEN FRONTED INTERROGATIVE ELEMENT AND THE REMAINDER OF THE

SENTENCE:

*Where viven tus padres?

*Dónde do your parents live?

BETWEEN NEGATIVE ELEMENT AND VERB:

*My family does not hablar español

*Mi familia no speaks Spanish.

BETWEEN AUXILIARY VERB AND INFINITIVE:

*No te puedo help with your homework.

*I can’t ayudarte con tu tarea

Many of these putative restrictions appear to be inextricably linked to language-

specific syntactic properties, for example null- vs. overt-subject status, existence
or non-existence of preverbal object clitics, verb raising and (English) do-
support, V-O vs. O-V word order and head directionality, etc., and
counterexamples from other bilingual environments have been pointed out,
inevitably intertwined with issues of language typology (e.g. Berk-Seligson,
1986; Clyne, 1987; Mahootian & Santorini, 1996; Prince & Pintzuk, 2000;
Woolford, 1983). This raises the question of whether these restrictions loosen or
disappear as typological differences between the languages diminish, or whether
some intrinsic traits set these grammatical elements apart.
A related question involves differences in intra-sentential code-switching
behavior between fluent bilinguals and less proficient L2 or heritage bilinguals.
There are studies that examine L2 speakers’ perception and acceptance of
various code-switched configurations (e.g. Giancaspro, 2013; Grabowski, 2011;
Guzzardo Tamargo & Dussias, 2013; Potowski & Bolyanatz, 2012);
comparatively fewer studies examine spontaneous code-switches as produced by
L2 bilinguals, both deliberate and unintended (Lipski, 2014, 2016; Bartlett,
Ebert, & Vergara, 2012; Potowski, 2009; also Liebscher & Dailey-O’Cain,
2004). Whether intra-sentential code-switching by fluent bilinguals is
qualitatively different from language mixing by L2 speakers remains an open
question, as does the possibility for a proficiency-related gradient correlated with
code-switching types.
A number of studies have suggested that convergence in bilingual contact
environments may be facilitated by code-switching (e.g. Burt, 1992; Poplack,
Zentz, & Dion, 2012; Salmons, 1990; Savić, 1995; Toribio, 2004; Torres
Cacoullos & Travis, 2011, 2015). It has also been claimed (e.g. Sebba, 1998) that
code-switching itself is facilitated by language congruence. The latter research
embodies the assumption that code-switching is already a common strategy in
the speech communities being studied and cannot be uncritically extrapolated to
situations in which code-switching is induced or presented to bilinguals who do
not routinely code-switch.
The present study examines various language contact environments
involving Portuguese and Spanish in communities near the Brazilian border. The
massive degree of similarity between Spanish and Portuguese makes this
language dyad a propitious environment for examining the relationships between
language typology and code-switching, and the correlations between code-
switching restrictions and language proficiency, giving rise to the following
research questions:
How robust are proposed restrictions on intra-sentential code-switching
when highly cognate and congruent languages are involved?
How does adherence to such constraints vary as a function of bilingual
proficiency?
Can code-switching constraints be reliably tested with bilinguals who do
not routinely code-switch?
To what extent is code-switching an expected outcome among bilingual
speakers of congruent languages?

Full answers to these questions will require massive cross-linguistic

explorations; the present study is limited to a specific Spanish-Portuguese
microcosm in which some initial results can be obtained. Expanding on previous
research, the focus is on speech communities located in Spanish-speaking
countries near the border of Brazil, and which reflect two different bilingual
profiles. The first set of data represents spontaneous language mixing in the L2
Portuguese of Spanish speakers. Although these individuals do not engage in
deliberate code-switching, some produce Spanish-Portuguese configurations that
appear to violate the previously mentioned code-switching constraints. The
second data set is derived from both a sociolinguistic corpus and a series of
interactive tasks carried out among balanced Portuguese-Spanish bilinguals in
communities along the Argentina-Brazil border. Like the L2 Portuguese speakers
in other border regions, these speakers do not routinely engage in intra-sentential
code-switching, but unlike in the case of L2 speakers, those mixed utterances
that do occur rarely contravene proposed constraints. At the same time the
reactions of these fluent bilinguals to code-switched utterances reveal a residue
of grammatically-based constraints even between these sibling languages. Taken
together, these data are suggestive of qualitative differences between L2 speakers
and balanced bilinguals. They also indicate that whereas code-switching may be
facilitated by congruence, it is not a necessary consequence.

2. A corpus of L2 Portuguese along the Brazilian border1

In order to compare intra-sentential language mixing between L2 speakers and
fluent balanced bilinguals, the behavior of L2 Portuguese speakers in daily
contact with both Portuguese and Spanish will be examined first. Along the
extensive Brazilian border with neighboring Spanish-speaking nations there are
a number of border-straddling communities in which Portuguese is used as a
second language within the respective Spanish-speaking countries. In the
majority of instances the language-contact configurations are asymmetrical:
Spanish is not spoken in Brazilian border communities and Brazilians tend to use
Portuguese when traveling to neighboring Spanish-speaking communities. In a
series of previous studies a corpus of L2 Portuguese as spoken by L1 Spanish
speakers (and a few examples of L2 Spanish produced by Portuguese speakers)
has been collected from the following border communities:

ARGENTINA :

Paso de los Libres (bordering on Uruguaiana, Brazil);

BOLIVIA :

Cobija (bordering on Brasiléia, Brazil); Guayaramerín (bordering on Guajará-

Mirim, Brazil); Bolpebra (bordering on Iñapari, Peru and Assis, Brazil);
COLOMBIA :
 Leticia (bordering on Tabatinga, Brazil);
PARAGUAY :

Pedro Juan Caballero (bordering on Ponta Porã, Brazil; Zanja Pytã (bordering
on Sanga Puitã, Brazil); Capitán Bado (bordering on Coronel Sapucaia,
Brazil); Bella Vista Norte (bordering on Bela Vista, Brazil);
PERU :

Iñapari (bordering on Assis, Brazil); Santa Rosa (bordering on Tabatinga,

Brazil);
URUGUAY :

Chuy (bordering on Chuí, Brazil);

VENEZUELA :

Santa Elena de Uairén (bordering on Pacaraima, Brazil)

The corpus consists of fifty sociolinguistic interviews, each approximately one

half hour in length, conducted by the author. Participants were recruited
informally in each community, using the snowball technique of personal
references, based on the criteria of having acquired Portuguese as a second
language in late adolescence or early adulthood (with no family history of
Portuguese or extended residence in Brazil), and using Portuguese only when
speaking with Brazilians. All participants had completed at least secondary
education (in Spanish), and none had studied Portuguese. No additional
measures of proficiency were taken, since the purpose was to collect samples of
the full range of L2 Portuguese along the Brazilian border. Although each
interview was nominally in Portuguese, the mutually known fact that both
speaker and listener knew Portuguese and Spanish (which emerged during the
informal recruitment process) and the absence of a sociolinguistically
intimidating environment did not serve as impediments to spontaneous language
mixing, deliberate as well as apparently unnoticed. All of the interviews were
conducted in Portuguese except for a group of Brazilian students in the Bolivian
city of Cobija, who had acquired Spanish as a second language and who were
interviewed in Spanish.

2.1 Apparent code-switching restrictions in L2 Portuguese

In none of the border communities in which L2 Portuguese data were collected

does “classic” intra-sentential code-switching occur, an assertion confirmed by
the author’s informal observations spanning at least a week in each of the
communities. Brazilians speak only Portuguese, L2 Portuguese speakers use
only Spanish with one another, and Spanish speakers’ L2 Portuguese – which is
only used with Brazilians – varies depending on the nature of their daily
activities, with the greatest knowledge of Portuguese found among individuals
engaged in commerce with Brazilian customers.
In these communities, native Spanish speakers who had acquired
Portuguese informally as a second language produced some combinations that
contravene the aforementioned grammatical infelicities of intra-sentential code-
switching. Despite the presence of elements from both languages, the explicit
assumption was that the conversations were conducted in Portuguese (or Spanish
for the Brazilian L2 Spanish speakers), and responses were coded accordingly.
When dealing with close cognates, spectrographic analysis (vowel length,
diphthongs, nasalization) was used for additional verification, e.g. Sp. quién ~
Ptg. quem ‘who’ or the negators no (Sp.) ~ não (Ptg., often realized as [ũ] in
atonic position). Some examples of combinations produced during conversations
presumed to be entirely in Portuguese (except for L2 Spanish speakers in Cobija,
Bolivia) are in (2).

(2)
BETWEEN SUBJECT PRONOUN AND PREDICATE:
 SEI LÁ yo
know-1s LOC I

‘I don’t know’ (Cobija, Bolivia)

ellos JÁ MISTURAM

they now mix-3PL

‘Now they mix [languages]’ (Guayaramerín, Bolívia)
Ella MORA aquí
she live-3s here
‘she lives here’ (Capitán Bado, Paraguay)
EU puedo TRABALHAR LÁ E TRABALHAR acá
I can-1s work there and work here
‘I can work there and work here’ (Leticia, Colombia)
ÊLES tienen que saber
they (m.) have COMP know
‘they have to know’ (Iñapari, Perú)
EU conohco MUITO Macuchí y cuando he FALADO con ello
I know-1s much Macuchí and when have-1s speak-PART with them
ello FALAM portuguéh
they speak-3PL Portuguese
‘I know many Macuchí and when I have spoken with them they speak
Portuguese’ (Santa Elena de Uairén, Venezuela)
que yo SAIBA parece que VAI ser por su cuenta
COMP I know-subj seem COMP go be for POSS account
‘As far as I know it seems that it will be on [their] own’ (Cobija, Bolivia,
Brazilian’s Spanish)

BETWEEN NEGATIVE ELEMENT AND VERB:

 ¿MAS VAI o no VAI ?

but go-3s or NEG go-3s

‘Are [you] going or not?’ (Cobija, Bolivia)
NÃO sabría decirle
NEG know-3s-cond say+DAT -3s
‘I wouldn’t know how to tell you’ (Paso de los Libres, Argentina)
él JÁ no PODE MAIS

he now NEG able-3s more

‘he can’t do any more’ (Pedro Juan Caballero, Paraguay)
não quieren falar a língua di nois
NEG want-3PL speak ART language of us
‘They don’t want to speak our language’ (Capitán Bado, Paraguay)
NÃO tiene LOJA de ARTESANATO

NEG have-3s store of handicrafts

‘There are no handicrafts stores’ (Leticia, Colombia)
NÃO hay nada pa este POVO

NEG exist nothing for this town

‘There is nothing for this town’ (Iñapari, Peru)

BETWEEN FRONTED INTERROGATIVE ELEMENT AND THE REMAINDER OF THE

SENTENCE:

VOCÊ qué cosa QUER ?

you what thing want-3s

‘What do you want?’ (Iñapari, Peru)
mozo ONDE queda Boacha?
waiter where stay-3s Boacha
‘Waiter where is Boacha [a discotheque]?’ (Santa Elena de Uairén,
Venezuela)
 ¿quién QUER TER mah conocimiento?
who want-3s have more knowledge
‘Who wants to know more?’ (Cobija, Bolivia, Brazilian L2 Spanish
speaker)

BETWEEN AUXILIARY VERB AND INFINITIVE:

porque NÃO TEM , cómo le puedo FALAR , vitrina

because NEG have-3s how DAT -3s can-1s speak window
‘Because there is no, how can I say it to you, display window”
(Guayaramerín, Bolivia)
não quieren falar a língua di nois
NEG want-3PL speak ART language of us
‘They don’t want to speak our language’ (Capitán Bado, Paraguay)
EU puedo TRABALHAR LÁ E TRABALHAR acá
I can-1s work there and work here
‘I can work there and work here’ (Leticia, Colombia)
NOI hemos PRECISADO de aprender un POQUINHO

we have-1PL need-PART of learn ART little-DIM

‘we have needed to learn a little bit’ (Santa Elena de Uairén, Venezuela)
yo quiero BRINCAR

I want-1s play
‘I want to play’ (Chuí Brazil, L2 Spanish speaker)

ALTERNATIONS THAT DO NOT RESPECT CONSTITUENT BOUNDARIES (“RAGGED

MIXING”)

OS BRASILEIROS FALAM portugueh y algunos preguntan

ART Brazilians speak-3PL Portuguese and some-m-pl ask-3PL
QUALQUER cosa

anything
 ‘The Brazilians speak Portuguese and some of them ask [for] whatever’
(Leticia, Colombia)
hay temporada que A GENTE FAIS compra SÓ NO BRASIL

exist season COMP people make purchase only in-ART Brazil

porque conviene
because suit-3s
‘There are times when people only buy in Brazil because it’s favorable for
them’ (Capitán Bado, Paraguay)
LÁ TEM merienda, desayuno, ALMORÇO , entonce OS gurí va
there have-3s snack breakfast lunch therefore ART kid go-3s
LÁ

there
‘There they give breakfast, lunch, snack, so the kids go there’ (Bella Vista
Norte, Paraguay)
en loh libros de historia SEMPRE está el NOME D’ÊLE

in ART books of history always be-3s ART name of him

‘His name is always in the history books’ (Santa Elena de Uairén,
Venezuela)

All of the examples in (2) were produced without obvious disfluencies in

circumstances in which the speakers were attempting to use only Portuguese,
which suggests that this type of “code-switching” is tacitly accepted by second-
language speakers of both Portuguese and Spanish. In order to determine
whether balanced bilinguals exhibit similar behavior patterns with respect to
putative code-switching restrictions, attention will turn to a series of speech
communities in northeastern Argentina.

3. Portuguese-Spanish bilingualism in Misiones province,

2
Argentina: Corpus data2
In the extreme northeastern Argentine province of Misiones, Portuguese is
spoken as a first language in several rural communities and small towns in the
eastern half of the province, which borders on Brazil. The sociolinguistically
permissive environment and the existence of large numbers of fluent Portuguese-
Spanish bilinguals makes this an ideal region from which to extract a corpus of
Portuguese as spoken by balanced bilinguals. Misiones is separated from Brazil
by the narrow Uruguay River and in a few places only by small creeks; in one
area (Bernardo de Irigoyen Argentina, Dionisio Sequeira and Barracão, Brazil)
there is a land border with no natural demarcation. The Portuguese spoken in
Misiones bears the mark of rural vernacular Brazilian Portuguese of the
neighboring states of Rio Grande do Sul and Santa Catarina, and bears little
resemblance to urban standardized Portuguese (Maia, 2004; also Daviña, 2003;
Sturza, 1994, 2004, 2005). Despite the fact that most residents of the chacras
‘small farms’ of eastern Misiones listen to Brazilian Portuguese media, there is
no attempt to emulate prestigious Brazilian varieties, such as is occurring in
northern Uruguay (e.g. Carvalho, 2004).
In order to study the Portuguese-Spanish interface in Misiones, a corpus of
sixty-two sociolinguistic interviews of approximately thirty minutes each with
adult bilinguals was collected. The format was free conversation in both Spanish
and Portuguese (usually on separate occasions) on a range of topics that included
agricultural practices, community events, and personal histories. All participants
were native speakers of vernacular Misiones Portuguese, none had lived
extensively in Brazil or studied Portuguese, and all spoke Spanish fluently
although not always with fully native-like proficiency. The interviews were
conducted in the following communities, including outlying agricultural
colonias: Puerto Iguazú, Comandante Andresito, San Antonio, Bernardo de
Irigoyen, Pozo Azul, El Soberbio, Barrerinho, Colonia Alicia, Santa Rita, Alba
Posse, 25 de Mayo, Panambí, and San Javier).

3.1 Putative code-switching constraints and the Misiones corpus

In most instances, Portuguese-Spanish bilinguals in Misiones do not mix

languages, as indicated by the interviews in the corpus and informally verified
by the author’s numerous extended visits to the region. There are occasional
reflexes of Portuguese-like clitic placement in Spanish (e.g. Juan quiere se casar
con María ‘Juan wants to marry María’), and some lexical borrowings from
Spanish in Portuguese (e.g. aduana for alfândega ‘customs post,’ lechuga for
alface ‘lettuce,’ durazno for pêssego ‘peach’), all falling into the category of
“insertion” (Muysken, 2000), but few intra-sentential switches (e.g. Muysken’s
“alternation”) or “ragged” mixes (Muysken’s “congruent lexicalization”). Even
fewer wae cases in which one of the aforementioned code-switching constraints
is contravened. Among the scarce examples found in the Misiones Portuguese
corpus are those in (3).

(3)
SWITCHES BETWEEN SUBJECT PRONOUN AND PREDICATE:

mi IRMÃ se CHAMA GISELA E yo me CHAMO Cristian

my sister REFL call-3s Gisela and I REFL call-1s Cristian
My sister’s name is Gisela and my name is Cristian’ (San Javier)
EU TENHO que CONTÁ e voh TEIN que ESCONDÉ

I have COMP count and you have COMP hide

‘I have to count and you have to hide’ (Alba Posse)
ÊLE QUER MOSTRÁ el amor que él TEM PRA O NETO

he want show ART love COMP he have for ART grandchild

‘He wants to show the love he has for his grandchild’ (Comandante
 Andresito)
él TINHA pensado fazer REVISTA

he have-IMP thought make magazine

‘He was thinking about starting a magazine’ (El Soberbio)

SWITCHES BETWEEN NEGATIVE ELEMENT AND VERB:

por aquí NÃO hay MAIS ÔNIBUS

for here NEG exist more bus

‘There are no buses around here’ (25 de Mayo)
no ME LEMBRO mucho
NEG 1s-REFL remember-1s much
‘I don’t remember much’ (Bernardo de Irigoyen)

SWITCHES BETWEEN AUXILIARY VERB AND INFINITIVE:

no se puede FAZER MUITO

NEG REFL can-3s do much

‘Not a lot can be done’ (Colonia Alicia)

ALTERNATIONS THAT DO NOT RESPECT CONSTITUENT BOUNDARIES (“RAGGED

MIXING”)

NA CIDADE había como una CERTA resistência

in+ART city exist-imp like ART certain-f resistance
‘In the city there was like a certain resistance; (San Javier)
había un TEMPO cuando EU era RAPAZ

exist-imp ART time when I be-IMP boy

‘There was a time when I was a boy’ (25 de Mayo)

3.2 Do Misiones Portuguese speakers really code-switch?

The Misiones bilingual Portuguese-Spanish speech communities offer a

promising environment to test reactions to putative code-switching constraints
by non-code switchers and with closely related languages. To this end, three
interactive tasks were conducted. The goal was to (literally) triangulate possible
code-switching configurations without explicitly requiring acceptability
judgments, which are notoriously unreliable when dealing with non-canonical
varieties. The tasks included translation of unexpectedly mixed utterances
(implicitly requiring the assignment of a base language), language classification
(identification of language switching), and elicited repetition (using cognitive
distractions to evoke preferred configurations). All participants were raised
speaking Misiones Portuguese but none had lived in Brazil or received formal
instruction in Portuguese; some also acquired Spanish at home while others
spoke no Spanish until beginning school. All had completed at least the first year
of secondary education in Misiones and were completely literate in Spanish.
The language referred to in Misiones as brasilero ‘Brazilian’ is a highly
vernacular offshoot of southern Brazilian Portuguese, but residents of Misiones
often assert that they speak Portuñol, a term frequently reinforced in regionalist
literature, newspaper articles, and printed and on-line tourist information.
Although few examples of “classic” intra-sentential code-switching can be heard
in Misiones, the all-pervasive “Portuñol” sentiments suggest that bilingual
speakers’ intuitions about language mixing might not coincide with canonical
grammatical descriptions of Portuguese and Spanish. The interactive tasks
described in the following sections provided objective measures of perceived
language mixing.

4. First interactive task: Speeded translation

A multi-faceted speeded translation task (details reported in Lipski, 2017a) was
administered in order to probe for various features of vernacular Portuguese and
Spanish in Misiones, as well as to test bilingual speakers’ reactions to putatively
mixed utterances.

4.1 Participants

A total of fifty participants (ages 18–45) were recruited, all natives of Pozo Azul,
El Soberbio, Colonia Alicia, Colonia Aurora, and surrounding agricultural
colonias including El Fisco, Barrerinho, Paraíso, and Puerto Londero.

4.2 Materials

Ninety utterances were created using various text-to-speech programs with

female voices in Latin American Spanish and Brazilian Portuguese, and
modified phonetically in PRAAT (Boersma & Weenink 1999–2005) and
structured morphosyntactically to approximate local varieties of Portuguese and
Spanish. Thirty utterances were entirely in Spanish, thirty were entirely in
Portuguese, and thirty contained language switches in various configurations
(fifteen beginning in Portuguese and fifteen beginning in Spanish). The switches
occurred after subject pronouns, interrogative words, negators, between auxiliary
verbs and infinitives, and after other lexical items (verbs and nouns). Each
stimulus sentence was followed by a 500 ms. gap and an audible beep. The
stimuli were randomized and presented on a portable computer. Examples are in
(4).

(4)
SPANISH TO PORTUGUESE

dónde fica a casa do prefeito

where stay ART house POSS mayor
‘Where is the mayor’s house?
 voh tem que perguntar na escola
you have COMP ask LOC school
‘You have to ask at the school’
Cuando ellos hablan misturam as língua
when they speak-3PL mix-3PL ART -F-PL language
‘When they speak [they] mix language(s)’

PORTUGUESE TO SPANISH

alguns brasileiro entendem o que nosotros hablamos

some(pl.) Brazilian understand-3PL ART +COMP we speak-1PL
‘Some Brazilians don’t understand what we say’
faz muito tempo que no hablo portugués
make-3S much time COMP NEG speak-1S Portuguese
‘I haven’t spoken Portuguese for a long time’
que eu saiba os uruguaios não podem hablar guaraní
COMP I know ART -PL Uruguayans NEG able-3PL speak Guaraní
‘As far as I know Uruguayans cannot speak Guarani.’

4.3 Procedure

Participants listened to the stimuli sequentially through headphones; repetition

was not permitted. They were told that they would hear some sentences in
Spanish and some in Portuguese and were instructed to rapidly translate from the
language of the stimulus into the “other” language after the beep. It was not
disclosed that some utterances contained both Spanish and Portuguese elements.
Respondents were timed out if a response was not initiated within two seconds
following the beep. In those (quite infrequent) cases where participants
responded to a mixed stimulus by asking which language it was in or into which
language it should be translated, the responses were not included in the analysis.
Stimuli and responses were digitally recorded.

4.4 Results

Table 1 gives sample results for the translation of mixed stimuli (results for all-
Spanish and all-Portuguese stimuli are presented in Lipski 2017a).3 Very few
mixed translations resulted from the mixed stimuli, and none contained
monotonic shifts (beginning in one language and continuing in the other). There
was a general tendency to translate mixed utterances into Spanish, perhaps
because the participants were aware of the author’s interest in Misiones
Portuguese. The data in Table 1 represent translations that emerged in a single
language, in effect neutralizing the code-switching in the stimuli, and those
instances in which the translation maintained a language shift.

Table 1. Results of translation task: Mixed Spanish ~ Portuguese stimulus

utterances

Switch type % switched to monolingual % unchanged

PRO + (5) 97.6% 2.4%

Q + (5) 99.1% 0.9%
AUX + infinitive (5) 97.5% 2.5%
lexical (15) 92.9% 7.1%

In order to determine the degree to which different code-switched stimuli were

translated, a generalized linear mixed-effects model (logistic regression) was
fitted in R (R Core Team 2014, version 3.3.1) using the lme4 package (Bates
et al., 2014), with participant and stimulus as random intercepts, retention of the
switch during translation vs. translation to a monolingual utterance as response
variable, and switch type as fixed effect. P-values were approximated with the
lmerTest (Kuznetsova et al., 2014) and car (Fox & Weisberg, 2011) packages. In
comparison to the null model (no fixed effects), the mixed-effects model
accounted for significantly more of the variance: χ2 (3) = 21.2; p < .0001. In
particular there were significant differences between switches following subject
pronouns, fronted interrogative words, and between auxiliary verbs and
infinitives, all in comparison to switches after lexical items taken as reference
level, as in Table 2.

Table 2. Results of the analysis of translation of code-switched stimuli

Category z-score estimate std. error p<

PRO + 2.091 1.5240 0.7288 .04

Q+ 1.984 2.0176 1.0168 .05
AUX + infinitive 2.775 1.3166 0.4744 .006

These results provide a first indication that even in the lexically and
grammatically cognate Portuguese-Spanish bilingual environment certain
grammatical categories appear to disfavor intra-sentential switches more than
others.

5. Second interactive task: Language classification

A second interactive task, administered a year after the first, was designed to
probe for participants’ explicit awareness of language switching as opposed to
the identification of local vernacular varieties of Portuguese and Spanish.

5.1 Participants
This task was performed by seventy-one adult Portuguese-Spanish bilinguals
(ages 18–52) from the communities of El Soberbio, Colonia Alicia, Santa Rita,
Alba Posse, and surrounding agricultural zones, with linguistic and demographic
profiles similar to those in the first task (details in Lipski, 2017a).

5.2 Materials

Seventy-five utterances were created using various Portuguese and Spanish text-
to-speech programs with female voices, fifteen in Spanish, fifteen in Portuguese,
and forty-five containing language switches at various points as in the examples
shown in (4).

5.3 Procedure

The stimuli were loaded onto a portable computer and were presented using a
script written for the PEBL experiment-building platform (Mueller & Piper,
2014), which randomized the stimulus presentation order for each participant.
Participants were told that they would hear sentences in Spanish, in Portuguese,
and containing mixtures of the two languages, and instructed to classify each of
the stimuli as all-Portuguese by pressing the left shift key, all-Spanish by
pressing the right shift key, or mixed by pressing the space bar. Colored dots on
the keys were reinforced by screen icons, augmented by images of the Argentine
flag, the Brazilian flag, and (for mixed utterances) an image of a blend of the
two flags. The program recorded responses and reaction times.4

5.4 Results

Participants correctly classified all-Spanish (93%) and all-Portuguese (77%)

utterances at high rates. The lower score for Portuguese stimuli can perhaps be
attributed to local residents’ belief that they speak “Portuñol.” The classification
results for mixed stimuli are given in Table 3.

Table 3. Classification experiment; results for mixed stimuli (N = 71)

Switch type % classified mixed

PRO + (5) 82.1%

Q + (5) 58.2%
NEG + (5) 80.6%
AUX + infinitive (5) 83.6%
lexical (25) 75.7%

A generalized linear mixed-effects model was fitted with response (correct or

incorrect) to the language classification task as the response variable, participant
and stimulus as random intercepts, and switch-type as fixed effect. The mixed
model accounted for a significantly greater proportion of the variance than the
null model: χ2 (5) = 506.48; p < .0001. Compared to utterances with switches
following lexical items (verbs and nouns) taken as the reference level, switches
after pronouns, negators, and between auxiliary verb and infinitive were
classified as mixed at significantly higher rates, as shown in Table 4. The
unexpectedly lower rate of mixed classification for switches after interrogative
words may be due to the considerable phonetic similarity between the cognate
Spanish and Portuguese items (e.g. Ptg. quem-Sp. quién ‘who’).

Table 4. Results of the analysis of classification of code-switched stimuli

Category z-score estimate std. error p<

PRO + 4.061 1.2470 0.3071 .0001

NEG + 3.049 1.0686 0.3504 .003
Q+ 6.274 2.0089 0.3202 .0001
AUX + infinitive −2.800 −0.9002 0.3215 .006

As with the translation task, the results of the language classification task
indicate that not all switch-types are created equal, and that negators, pronouns,
and auxiliary verbs stand out from nouns and main verbs in triggering awareness
of a language shift.

6. Third interactive task: Memory-loaded repetition

A third task involving elicited repetition was designed to probe Portuguese-
Spanish bilinguals’ implicit processing of code-switched utterances. The
rationale is that “when listeners hear a sentence that exceeds the capacity of their
short-term memory, they will pass it through their own grammar before
repeating it” (Gullberg, Indefrey, & Muysken, 2009, p. 34). In many cases
respondents’ errors frequently reflect their own grammars, i.e. what they WOULD
HAVE SAID instead of what was actually said. Repetition with concurrent memory
loading can provide insights into preferred and dispreferred switch-types,
bypassing explicit judgments and nudging participants toward producing their
own preferred configurations without revealing the true nature of the data being
sought.

6.1 Participants

This task was performed by sixty-seven adult bilingual participants (ages 18–52)
from the towns of El Soberbio, Colonia Alicia, Santa Rita, and the surrounding
rural areas (details in Lipski, 2018b). All had previously participated in the
language-classification task on a different day during the same visit.

6.2 Materials
Twenty utterances were created using various Spanish and Portuguese text-to-
speech programs with female voices, of which eight were presented in
vernacular Misiones Portuguese and twelve contained Portuguese-Spanish code-
switches similar to the examples in (4). Each stimulus consisted of a test
utterance accompanied by a seven-second video clip (with sound removed),
taken from popular cartoon programs (e.g. Loony Tunes, Batman, Superman,
Green Lantern, Flash, and Spider Man). This use of video material as a distractor
is similar to the technique described by Norcliffe and Jaeger (2014), and has
been applied successfully by the present author in other bilingual settings
(Lipski, 2017b, 2018a). Each stimulus began with the target utterance
(accompanied an image of a listening ear), followed by a cartoon video clip, a
ten-second countdown video, and finally the image of a speaking mouth.

6.3 Procedure

The stimuli were presented on a portable computer using the PsychoPy

experiment-building platform (Pierce, 2007), which randomized the order of
presentation for each participant. Participants were asked to retain the stimulus
utterance in memory when it was first presented. During the ten-second
countdown after the video clip they were to describe the cartoon as accurately
possible in the language of their choice. When the image of the mouth appeared
they were to repeat the stimulus utterance exactly as heard. Stimuli and
responses were digitally recorded. It was not disclosed that some of the stimuli
contained both Portuguese and Spanish elements.

6.4 Results

The all-Portuguese stimuli were repeated without changing language (verbatim

or with very small within-language modifications) at the rate of 98.8% across all
participants. The results for mixed stimuli are given in Table 5, where it can be
seen that many participants “followed instructions” and repeated mixed
utterances verbatim, while others’ “repetitions” achieved monolingual status.

Table 5. Results of memory-loaded repetition experiment (N = 67)

Switch type % switched to monolingual % unchanged

PRO + (3) 75.6% 24.4%

Q + (3) 59.5% 40.5%
AUX + infinitive (3) 58.6% 41.4%
lexical (3) 3.6% 95.4%

A generalized linear mixed-effects model was fitted with speaker and stimulus as
random intercepts and retention of language switch vs. “correction” to
monolingual status as response variable, and switch type as fixed effect. A
likelihood comparison with the null model revealed that the model with switch
type as fixed effect accounted for a significantly greater proportion of the
variance: χ2 (3) = 12.197; p < .007. Compared to utterances with switches
following lexical items taken as the reference level, switches after pronouns,
interrogatives, and between auxiliary verb and infinitive were altered to achieve
monolingual status at significantly higher rates, as shown in Table 6.

Table 6. Results of the memory-loaded repetition of code-switched stimuli

Category z-score estimate std. error p<

PRO + 4.271 5.388 1.262 .0001

Q+ 3.308 4.278 1.293 .001
AUX + infinitive 2.796 4.187 1.498 .006
7. General discussion
In the research reported here, corpus data from L2 speakers and balanced
bilinguals have been supplemented by structured interactive tasks carried out in
a fluent bilingual setting where two cognate languages are in sociolinguistic
freefall. The data collected for the present project provide preliminary responses
to the research questions posed at the outset.
A comparison of L1 and L2 Portuguese corpus data has shown that fluent
Portuguese-Spanish bilinguals in Misiones produce few intra-sentential switches,
and those that do occur rarely violate proposed grammatical constraints. On the
other hand, data collected from a broad cross-section of L2 speakers reveal
repeated instances where these same constraints have been overridden in mixed
utterances produced in nominally monolingual conversations. The data sets are
not sufficiently fine-grained to identify the point(s) on the L2 proficiency
continuum where these constraints become effectively operative, but the
qualitative differences are clear.
Although Spanish and Portuguese are highly congruent both lexically and
structurally, the lack of significant code-switching among both L2 speakers and
fluent bilinguals in the border regions under study is not surprising, given the
nature of the respective speech communities. The L2 speakers use Portuguese
only with Brazilians and speak only Spanish to fellow citizens. Fluent bilinguals
in Misiones, Argentina use only Portuguese within their rural communities and
in some of the province’s smaller towns, and Spanish with interlocutors who are
known or suspected non-Portuguese speakers. In all of border communities
described here, language choice is not an affirmation of identity, but simply a
consequence of personal background. Ethnic and cultural characteristics are
virtually identical on both sides of the border, and the complex interplay of
language mixing, ethnicity, and identity such as found, e.g. in border
communities in the U. S. (e.g. Rangel, Moyna, & Rodriguez Loureiro, 2015) or
in Belize (Balam, 2015; Balam, Prada Pérez, & Mayans, 2014) is not found.
Despite popular notions about “Portuñol,” the linguistic environment is almost
completely without prescriptivist strictures; it is the natural separation of the
domains in which Spanish and Portuguese are used that effectively precludes
intra-sentential code-switching.
The results of the three interactive tasks conducted with fluent Portuguese-
Spanish bilinguals in Misiones demonstrate that consistent reactions to intra-
sentential code-switching can be obtained in speech communities where such
code-switching is infrequent, with outcomes that converge with reported data
from habitual code-switchers in other bilingual environments. The experimental
data also provide evidence that certain grammatical categories are implicitly
regarded as more felicitous than others in intra-sentential language switches even
between two languages that are not separated by any major morphosyntactic
differences. The behavior of Portuguese-Spanish bilinguals in Misiones suggests
that the proposed code-switching constraints reviewed earlier may at least
partially transcend language-specific bilingual environments. At the same time,
it has been seen that these constraints are far from absolute, since many of the
putatively anomalous combinations were implicitly accepted during
classification, translation, and repetition. The “soft” nature of the constraints can
plausibly be attributed at least in part to the highly cognate lexicons as well as to
the morphosyntactic congruence between Spanish and Portuguese, although
even between typologically more distinct languages such as Spanish and English
these constraints evoke only partial explicit rejection, e.g. Aguirre (1981, p. 304;
1985, p. 65) and Koronkiewicz (2014, pp. 83f.) for Spanish-English switches
after subject pronouns, Ebert (2014, pp. 192–198) for Spanish-English switches
after WH-interrogatives, D’Introno et al. (1991) for switches after negative
elements, and Giancaspro (2013) for Spanish-English auxiliary-VP switches.
Relatively low levels of metalinguistic experience may also contribute to higher
levels of implicit or explicit acceptance of code-switches that are rejected by
speakers with more formal training or practice (Bradac, Martin, Elliott, & Tardy,
1980). Additional research in a variety of bilingual environments is required in
order to put forth an explanation for the differential behavior of these
grammatical elements (Lipski, 2017c, 2019 offers some preliminary ideas); the
fact that the present inquiry has demonstrated these tendencies between two
highly cognate languages suggests a possible intersection of language-specific
and language-independent factors. Taken in the larger context of language
contact and code-switching research, the Portuguese-Spanish data hint at the
eventual feasibility of establishing a reciprocal relation between typological
distance and the firmness of intra-sentential switching restrictions. The present
study is offered as a preliminary step in that direction.

Notes
1. I offer my gratitude to the following individuals, who greatly assisted my field work: Ingard Miauchi
(Cobija); Kelly Gamboa (Guayaramerín); Derlis Torres (Pedro Juan Caballero, Zanja Pytã, Capitán Bado,
Bella Vista Norte); María Silvia Chichizola de Ezama (Paso de los Libres); Pedro González Segura
(Leticia); Celso Curi Paucarmaita, Alberto Cardozo, César Ochoa, Jorge Quispe, Narciso Paricahua
(Iñapari); Celia Cisneros and Elba Wolf (Santa Elena de Uairén); Graciela Barrios, Selva Chiricó, Adolfo
Elizaincín, Magdalena Coll (Rivera); Lidia Vidal, Raquel Puig, Mirta Costa Fernández, Alice Lucas, María
Irene Moyna (Chuy). ①

2. I gratefully acknowledge the assistance of the following individuals, without whose help this research
would not have been possible: Liliam Prytz Nilsson and Viviana Eich (Ministry of Education, Posadas),
Liliana Daviña, and Ivene Carissini da Maia (Universidad Nacional de Misiones, Posadas), Sergio
Chajkowski and Rosendo Fuchs (Panambí), Hugo Cámara Robles (Comandante Andresito), Sandra Grabe
(Puerto Iguazú), Nélida Aguerre (Alba Posse), Norma Ramírez, Justino Steinhaus, and Ricardo Leiva (El
Soberbio), Darío Miranda and Manglio Vargas (Colonia Alicia), Elsa Rodríguez de Olivera (25 de Mayo),
Daniel Ziemann and Carlos Knoll (Santa Rita), Fátima Zaragoza and Juan Carlos Morínigo (Bernardo de
Irigoyen), Isabelino Fonseca (San Antonio), Roberto Pinto (San Javier), Marcelo Ragotín, Cristina Barchuk,
and Luis Alberto Vogeli (Pozo Azul). Thanks are also due to the dozens of Misioneros who have graciously
welcomed me into their homes and lives and have shared their languages with me. ①

3. Negation was not included in this analysis since in Misiones vernacular Portuguese não is phonetically
very close to Spanish no. ①

4. Reaction times were significantly faster for all-Spanish stimuli than for all-Portuguese or all-mixed, but
there were no significant differences between the latter two categories, nor among the various mixed types.
①

References
Aguirre, A.
(1981.) Toward an index of acceptability for code alternation: An experimental analysis. Aztlán, 11, 297–
322. ①
Aguirre, A.
(1985.) An experimental study of code alternation. International Journal of the Sociology of Language
59–82. ①
Azuma, S., & Meier, R.
(1997.) Open class and closed class: Sentence-imitation experiments on intrasential code-switching.
Applied Psycholinguistics, 18, 257–276. ①
Balam, O.
(2015.) Code-switching and linguistic evolution: The case of ‘Hacer + V’ in Orange Walk, Northern Belize
Lengua y Migración, 7, 83–109. ①
Balam, O., Prada Pérez, A., & Mayans, D.
(2014.) A congruence approach to the study of bilingual compound verbs in Northern Belize contact
Spanish. Spanish in Context, 11(2), 243–265. ①
Bartlett, L., Ebert, S., & Vergara, D.
(2012.) Linguistic competence in heritage speakers and second language learners: Evidence from code
switched wh questions. Presented at 2012 UIC Bilingualism Forum, University of Illinois Chicago. ①
Bates, D., Maechler, M., Bolker, B. & Walker, S.
(2014.) lme4: Linear mixed-effects models using Eigen and S4. R package version 1.1–12. [Link]
[Link]/package=lme4 ①
Belazi, H., Rubin, E., & Toribio, A. J.
(1994.) Code-switching and X-bar theory: The Functional Head Constraint. Linguistic Inquiry, 25, 221–
238. ①
Bentahila, A., & Davies, E.
(1983.) The syntax of Arabic-French code-switching. Lingua, 59, 301–330. ①
Berk-Seligson, S.
(1986.) Linguistic constraints on intrasentential code-switching: A study of Spanish/Hebrew bilingualism
Language in Society, 15, 313–348. ①
Bradac, J., Martin, L., Elliott, N., & Tardy, C.
(1980.) On the neglected side of linguistic science: Multivariate studies of sentence judgment. Linguistics
 18, 967–995. ①
Burt, S. M.
(1992.) Codeswitching, convergence and compliance: The development of micro community speech
norms. Journal of Multilingual & Multicultural Development, 13(1–2), 169–185. ①
Carvalho, A. M.
(2004.) I speak like the guys on TV: Palatalization and the urbanization of Uruguayan Portuguese.
Language Variation and Change, 16, 127–151. ①
Clyne, M.
(1987.) Constraints on code switching: How universal are they? Linguistics, 25, 739–764. ①
Daviña, L. S.
(2003.) Fronteras discursivas en una región plurilingüe: Español y portugués en Misiones (Unpublished
MA thesis). Universidad de Buenos Aires. ①
D’Introno, F., Fiallo, A. T., Ram, K., & Sicilia, D.
(1991.) Condiciones gramaticales sobre la alternancia. Hispania, 74(2), 400–408. ①②③
DiSciullo, A. M., Muysken, P., & Singh, R.
(1986.) Government and code-mixing. Journal of Linguistics, 22, 1–24. ①
Doron, E.
(1983.) On a formal model of code-switching. Texas Linguistic Forum, 22, 35–59. ① ②
Dussias, P.
(2003.) Spanish-English code mixing at the Auxiliary Phrase: Evidence from eye-movement data. Revista
Internacional de Lingüística Iberoamericana, 1(2),7–34. ①
Ebert, S.
(2014.) The morphosyntax of Wh-questions: Evidence from Spanish–English code-switching (Unpublished
doctoral dissertation). University of Illinois, Chicago. ① ②
Fox, J., & Weisberg, S.
(2011.) An {R} companion to applied regression (2nd ed.). Thousand Oaks CA: Sage. [Link]
.[Link]/jfox/Books/Companion ①
Giancaspro, D.
(2013.) L2 learners’ and heritage speakers’ judgments of code-switching at the auxiliary-VP boundary. In
Cabrelli Amaro, G. Lord, A. de Prada Pérez, & J. E. Aaron (Eds.), Selected proceedings of the 16th
Hispanic Linguistics Symposium (pp. 56–69). Somerville, MA: Cascadilla Proceedings Project. ① ②
Grabowski, J.
(2011.) Stability of grammaticality judgments in German-English code-switching (Unpublished MA
thesis). Arizona State University. ①
Gullberg, M., Indefrey, P., & Muysken, P.
(2009.) Research techniques for the study of code-switching. In B. Bullock & A. J. Toribio (Eds.), The
Cambridge handbook of linguistic code-switching (pp. 21–39). Cambridge: Cambridge University
Press. ①
Gumperz, J.
(1977.) The sociolinguistic significance of conversational code-switching. RELC Journal, 8, 1–34.
 ①
Guzzardo Tamargo, R., & Dussias, P.
(2013.) Processing of Spanish-English code-switches by late bilinguals. In S. Baiz, N. Goldman, & R.
Hawkes (Eds.), BUCLD 37: Proceedings of the 37th annual Boston University Conference on
Language Development (Vol. 1, pp. 134–146). Somerville, MA: Cascadilla Proceedings Press. ①
Hebblethwaite, B. J.
(2007.) Intrasentential code-switching among Miami Haitian Creole-English bilinguals (Unpublished
doctoral dissertation). Indiana University. ①
Hebblethwaite, B. J.
(2010.) Adverb code-switching among Miami’s Haitian Creole-English second generation. Bilingualism:
Language and Cognition, 13, 409–428. ①
Jake, J.
(1994.) Intrasentential code-switching and pronouns: On the categorical status of functional elements.
Linguistics, 32, 271–298. ①
Jake, J., Myers-Scotton, C., & Gross, S.
(2002.) Making a minimalist approach to codeswitching work: Adding the Matrix language. Bilingualism:
Language and Cognition, 5, 69–91. ①
Joshi, A.
(1985.) Processing of sentences with intrasentential code switching. In D. R. Dowty, L. Karttunen, & A. M.
Zwicky (Eds.), Natural language parsing: Psychological, computational and theoretical perspectives
(pp. 190–205). Cambridge: Cambridge University Press. ①
Klavans, J.
(1985.) The syntax of code-mixing: Spanish and English. In L. King & C. Maley (Eds.), Proceedings of the
Linguistic Symposium on Romance Languages (pp. 213–231). Amsterdam: John Benjamins. ①
Koronkiewicz, B.
(2014.) Pronoun categorization: Evidence from Spanish/English code-switching (Unpublished doctoral
dissertation). University of Illinois at Chicago. ① ②
Kuznetsova, A., Bruun Brockhoff, P., & Haubo Bojesen Christensen, R.
(2014.) lmerTest: Tests for random and fixed effects for linear mixed effect models (lmer objects of lme4
package). R package version 2.0–6. [Link] ①
Liebscher, G., & Dailey-O’Cain, J.
(2004.) Learner code-switching in the content-based foreign language classroom. The Canadian Modern
Language Review / La Revue Canadienne des Langues Vivantes, 60, 501–525. ①
Lipski, J.
(1978.) Bilingual language switching and the problem of linguistic competence. In M. Paradis (Ed.),
Aspects of bilingualism (pp. 250–264). Columbia, SC: Hornbeam Press. ①
Lipski, J.
(1982.) Spanish English language switching in speech and literature: Theories and models. Bilingual
Review, 9, 191–212. ①
Lipski, J.
(1985.) Linguistic aspects of Spanish English language switching. Tempe, AZ: Arizona State University,
 Center for Latin American Studies. ①
Lipski, J.
(2006.) Too close for comfort? the genesis of “portuñol/portunhol”. In T. L. Face & C. A. Klee (Eds.),
Selected proceedings of the 8th Hispanic Linguistics Symposium (pp. 1–22). Somerville, MA:
Cascadilla Press.
Lipski, J.
(2008.) Searching for the origins of Uruguayan Fronterizo dialects: Radical code-mixing as “fluent
dysfluency.” Journal of Portuguese Linguistics, 8, 5–46.
Lipski, J.
(2009.) “Fluent dysfluency” as congruent lexicalization: A special case of radical code-mixing. Journal
Language Contact, 2, 1–39.
Lipski, J.
(2011a.) Um caso de contato de fronteira: O sudoeste. In H. Mello, C. Altenhofen, & T. Raso (Eds.), Os
contatos linguísticos no Brasil (pp. 349–368). Belo Horizonte: Editora UFMG.
Lipski, J.
(2011b.) Encontros lingüísticos fronteiriços. Ideação, 13(2), 83–100. [Link]
/ideacao/article/viewArticle/6109
Lipski, J.
(2014.) Spanish-English code-switching among low-fluency bilinguals: Towards an expanded typology
Sociolinguistic Studies, 8. 23–55. ①
Lipski, J.
(2015.) Portuguese/portuñol in Misiones, Argentina: Another Fronterizo? In S. Sessarego & M. González
(Eds.), New perspectives on Hispanic contact linguistics (pp. 253–281). Frankfurt & Madrid: Vervuert
& Iberoamericana..
Lipski, J.
(2016.) The role of unintentional/involuntary code-switching: Did I really say that? In R. Guzzardo, C.
Mazak, & M. C. Parafita Couto (Eds.), Spanish-English codeswitching in the Caribbean and the US
(pp. 139–168). Amsterdam: John Benjamins. ①
Lipski, J.
(2017a.) Portuguese or Portuñol? Language contact in Misiones, Argentina. Journal of Linguistic
Geography, 4, 47–64. ①②③
Lipski, J.
(2017b.) Ecuadoran Media Lengua: More than a “half”-language? International Journal of American
Linguistics, 82, 233–262. ①
Lipski, J.
(2017c.) Language switching constraints: More than syntax? Data from Media Lengua. Bilingualism:
Language and Cognition, 20(4), 722–746. ①
Lipski, J.
(2018a.) Can agreement be suppressed in second-language acquisition? Data from the Palenquero-Spanish
interface. Second Language Research, 34(3), 275–308. ①
Lipski, J.
(2018b.) La interfaz portugués-castellano en Misiones, Argentina: zona de prueba para la alternancia de
lenguas. Estudios Filológicos, 60, 169–190. ①
Lipski, J.
(2019.) Field-testing code-switching constraints: A report on a strategic languages project. Languages, 4, 7.
①
MacSwan, J.
(1999.) A minimalist approach to intrasentential code switching. New York, NY: Garland Press. ①
MacSwan, J.
(2000.) The architecture of the bilingual language faculty: Evidence from intrasentential code switching
Bilingualism: Language and Cognition, 3, 37–54. ①
MacSwan, J.
(2004.) Codeswitching and linguistic theory. In T. Bhatia & W. Ritchie (Eds.), Handbook of bilingualism
(pp. 283–310). Malden, MA: Blackwell. ①
MacSwan, J.
(2005.) Codeswitching and generative grammar: A critique of the MLF model and some remarks on
“modified minimalism.” Bilingualism: Language and Cognition, 8, 1–22. ①
Mahootian, S., & Santorini, B.
(1996.) Code switching and the complement/adjunct distinction. Linguistic Inquiry, 27, 464–479. ①
Maia, I. C. da
(2004.) Intercambios lingüísticos de frontera: Incidencia el el hablar de los alumnos del profesorado en
portugués de la UNAM (Unpublished MA thesis). Universidad Nacional de Misiones, Posadas. ①
Mueller, S. T., & Piper, B. J.
(2014.) The psychology experiment building language (PEBL) test battery. Journal of Neuroscience
Methods, 222, 250–259. ①
Muysken, P.
(2000.) Bilingual speech: A typology of code-mixing. Cambridge: Cambridge University Press. ① ②
Peirce, J. W.
(2007.) PsychoPy – Psychophysics software in Python. Journal of Neuroscience Methods, 162(1–2), 8–13.
Program available at: [Link] ①
Peñalosa, F.
(1980.) Chicano sociolinguistics: A brief introduction. Rowley, MA: Newbury House. ① ②
Pfaff, C.
(1979.) Constraints on language mixing: Intrasentential code-switching and borrowing in Spanish/English
Language, 55, 291–318. ①
Poplack, S.
(1980.) Sometimes I’ll start a sentence in Spanish y termino en español. Linguistics, 18, 581–618.
①
Poplack, S., Zentz, L., & Dion, N.
(2012.) Phrase-final prepositions in Quebec French: An empirical study of contact, code-switching and
 resistance to convergence. Bilingualism: Language and Cognition, 15(2), 203–225. ①
Potowski, K.
(2009.) Forms and functions of code-switching by dual immersion students: A comparison of heritage
speaker and L2 children. In M. Turnbull & J. Dailey-O’Cain (Eds.), First language use in second and
foreign language learning (pp. 87–114). Bristol, UK: Multilingual Matters. ①
Potowski, K., & Bolyanatz, M.
(2012.) Reactions to (in)felicitous codeswitching: heritage speakers vs. L2 learners. In K. Geeslin & M.
Díaz-Campos (Eds.), Selected proceedings of the 14th Hispanic Linguistics Symposium (pp. 116–129).
Somerville, MA: Cascadilla Proceedings Project. ①
Prince, E. & Pintzuk, S.
(2000) [1983]. Bilingual code-switching and the open/closed class distinction. University of Pennsylvania
Working Papers in Linguistics (PWPL), 6(3), 237–257. ① ②
Rangel, N., Moyna, I. M., & Rodriguez Loureiro, V.
(2015.) Is that what I sound like when I speak? Attitudes toward Spanish, English, and code-switching in
two Texas border towns. Spanish in Context 12(2), 177–198. ①
Rivas, A.
(1981.) On the application of transformations to bilingual sentences (Unpublished doctoral dissertation).
University of Massachusetts, Amherst. ①
Salmons, J.
(1990.) Bilingual discourse marking: Code switching, borrowing, and convergence in some German-
American dialects. Linguistics, 28, 453–480. ①
Sankoff, D., & Poplack, S.
(1981.) A formal grammar for code-switching. Papers in Linguistics, 14, 3–46. ①
Savić, J. M.
(1995.) Structural convergence and language change: Evidence from Serbian/English code-switching.
Language in Society, 24(4), 475–492. ①
Sebba, M.
(1998.) A congruence approach to the syntax of codeswitching. International Journal of Bilingualism, 2,
1–19. ①
Sturza, E. R.
(1994.) O espanhol do cotidiano e o espanhol da escola: um estudo de caso na fronteira Brasil-Argentina
(Unpublished MA thesis). Universidade Federal de Santa Maria, Brazil. ①
Sturza, E. R.
(2004.) Fronteiras e práticas lingüísticas: Um olhar sobre o portunhol. Revista Internacional de Lingüística
Iberoamericana, 2, 151–160. ①
Sturza, E. R.
(2005.) Línguas de fronteira: o desconhecido território das práticas lingüísticas nas fronteiras. Ciência e
Cultura, 57(2), 47–50 [Link] ①
Timm, L.
(1975.) Spanish-English code-switching: El porque y how not to. Romance Philology, 28, 473–482. ①
③
Toribio, A. J.
(2001a.) Accessing bilingual code-switching competence. International Journal of Bilingualism, 5, 403–
436. ①
Toribio, A. J.
(2001b.) On the emergence of bilingual code-mixing competence. Bilingualism: Language and Cognition
4, 203–231. ①
Toribio, A. J.
(2004.) Convergence as an optimization strategy in bilingual speech: Evidence from code-switching.
Bilingualism: Language and Cognition, 7(2), 165–173. ①
Torres Cacoullos, R., & Travis, C. E.
(2011.) Testing convergence via code-switching: Priming and the structure of variable subject expression
International Journal of Bilingualism, 15(3), 241–267. ①
Torres Cacoullos, R., & Travis, C. E.
(2015.) Gauging convergence on the ground: Code-switching in the community. International Journal of
Bilingualism, 19, 365–386. ①
Van Gelderen, E., & MacSwan, J.
(2008.) Interface conditions and code-switching: An F-movement analysis of pronouns and lexical DPs
Lingua, 118, 765–776. ①
Wakefield, J. Jr., Bradley, P., Lee Yom, B.-H., & Doughtie, E.
(1975.) Language switching and constituent structure. Language and Speech, 18, 14–19. ①
Woolford, E.
(1983.) Bilingual code-switching and syntactic theory. Linguistic Inquiry, 13, 519–535. ① ② ③
Woolford, E.
(1984.) On the application of WH movement and inversion in code-switching sentences. Revue Quebecoise
de Linguistique, 14, 77–86. ①
Language variation, linguistic perceptions
and attitudes
CHAPTER 11

Real perception or perceptive

accommodation?
The Dominirican ethnic-dialect continuum and
sociolinguistic context
Luis A. Ortiz López & Cristina Martínez Pedraza
Universidad de Puerto Rico

Abstract
Dominicans in Puerto Rico have opted for closed spaces due to their
social marginalization. Nonetheless, their interactions with Puerto
Ricans has facilitated a Domini-Rican continuum. This research
examines speech data, to compare dialect recognition and linguistic
perceptions of Puerto Ricans toward the varieties in this continuum.
During a verbal-guise task, 24 Puerto Ricans listened to five stimuli of
the Domini-Rican continuum and answered items about extralinguistic
variables: nationality, social class, educational level, intelligence and
speech affability. Results support that linguistic production (i) allows
individuals to distinguish dialects and (ii) facilitates the acceptance
criteria of immigrants and their descendants. The data suggests that
perceptions and attitudes are correlated with the characteristics of both
stimuli and participants, mainly the perception of nationality.
 Keywords
dialectal contact; migration; Dominican Spanish; Puerto Rico;
linguistic perceptions and attitudes

1. Introduction
Languages and dialects are markers of identity and social alliance. Speakers
conceive themselves as members of a certain speech community, while being
aware of the differences of their community in relation to others. (Wright, 2014).
In situations of language contact, dialect contact, and even in marked social
stratification – as its the case with African-American Vernacular English – the
differences between these groups determine the power relations of society
(Woodlard, 1985). In these contexts, the language becomes an instrument of the
dominant group to exclude, devaluate, and even obscure minority groups
through expression – such as, i/we vs. you/them, those, others; what is right vs.
what is incorrect; the good vs. the bad, etc. – that stratify the different groups
and the linguistic characteristic associated with them, defining which ones carry
social prestige and which don’t (Lippi-Green, 1997). Through these social and
psycholinguistic mechanisms, the groups in contact generate and promote –
consciously or unconsciously – sociolinguistic perceptions, attitudes and
stereotypes towards members outside of their group and their linguistic
characteristics, as a result of their ideologies. In this sense, the language, dialect,
color of skin, ethnic origin, economic status, sexual orientation, habits, etc.
acquire positive or negative values in light of the community’s ideology
(Álvarez, 2009; Toribio, 2007).
The dominant group assumes control and dictates the social and linguistic
patterns of a community, while minority groups react and contribute to the social
dynamics that are generated; often through a process of manipulation called false
consciousness, in which the minority groups validate the discourse of the
dominant group as a result of the linguistic insecurity of its members (Blas
Arroyo, 1999; Lippi-Green, 1997). At the same time, these minority groups put
into practice negotiation mechanisms with others and themselves, such as the
change of language and the de-dialectalization of their linguistic variant (Blas
Arroyo, 1999). The phenomenon of migration exemplifies these socio-
psycholinguistic processes that promote social, cultural and linguistic contact,
including dialectal variations and language negotiations between communities.
The Caribbean – seen as a place of great movement, of multiple coexisting races,
of immigrants and transmigrants, and of linguistic creolization (Duany, 2011;
Mintz, 1996; Ortiz López, 2015) – appears to be an ideal space to examine the
translinguistic behaviors that have been little discussed so far (Bullock &
Toribio, 2009; Fúster, 2012; Mejía Pardo, 1993; Suárez Büdenbender, 2009).
For example, the Hispanic Antilles – although geographically separated –
share ethnic, cultural and linguistic characteristics that hinder clear phenotypic
and linguistic distinctions among the members of these communities; they
coexist. Consequently, the dialectal differences between the islands – within the
Spanish macrosystem – are somewhat blurred and problematic for dialectal
zoning proposals (Lipski, 1994; López Morales, 1992; Ortiz López, 2000;
Zamora & Guitart, 1988), but ideal for studies of linguistic perception, attitudes
and dialectal, intradialectal and intrafamilial stereotypes (Potowski, 2011). In the
case of the Dominican immigration to Puerto Rico, a minority marginalized and
segregated group is enclosed in urban spaces within Puerto Rican society
(Duany, 2005; Mejía Pardo, 1993; Suárez Büdenbender, 2009), but due to the
continuous interactions, including those within the same family, ethnic
continuums have developed: the Dominicans – sons and daughters of Puerto
Ricans and Dominicans – and the Dominican Puerto Ricans – sons and
daughters of Dominicans born in Puerto Rico – (see Diagram 1). These groups
are situated between the Dominicans or the newly arrived Dominicans) –
corresponding to the first generation of immigrants – and Puerto Ricans in situ.

Diagram 1. Dominican-Rican ethnic-dialect continuum

This situation of dialectal contact – between different types of immigrants and
ethnic continuum – is the focus of the research carried out in the Metropolitan
Area of San Juan, Puerto Rico: the sociolinguistic contact of Dominiricans in
San Juan. In this first approach, the perceptions, linguistic attitudes and
stereotypes of Puerto Ricans in relation to the various groups in the ethnic and
linguistic continuum are documented, analyzed and evaluated. Specifically, the
research explores the sociolinguistic knowledge of Puerto Ricans towards the
Puerto Rican-Dominican dialect continuum, through auditory and visual stimuli.
The linguistic perceptions of these are correlated with the inherent and social
characteristics – such as ethnic group, social class, educational level, skin color,
intelligence and the correctness and acceptance of its linguistic variant – of the
members of this Dominican-Puerto Rican continuum. Finally, the involuntary
and voluntary perceptive behaviors of the speakers and their implicit and explicit
attitudes are analyzed in order to account for the social and psycholinguistic
relationships between the Puerto Ricans and other individuals in the continuum.
 The following questions guide this study: (1) Can Puerto Ricans identify
the dialectal differences between Dominicans and Puerto Ricans within
Caribbean Spanish? (2) Can Puerto Ricans distinguish the ethnic-dialectal
continuums (i.e., Domirican and Puerto Rican-Dominican)? (3) Are there
processes of bidialectalization in Dominicans and Puerto Ricans-Dominicans,
and of de-dialectalization in Dominican immigrants that make it hard to
distinguish their origin? (4) What are the dialectal, ethnic, social and
psychological attitudes generated by each of these ethno-dialectal groups? To
accomplish the objectives of this study, this research follows the following
structure: the second section offers a socio-historical overview of Dominicans in
Puerto Rico; the third section exposes the theoretical approach that will be used
to explain the data; the fourth section describes the methodology of the study, the
fifth section exposes and analyzes the most relevant findings of the research, and
the sixth and last section discusses the findings and offers some conclusions and
recommendations for future investigations.

2. Migratory context: Dominicans in Puerto Rico

There is a long history of migratory flow between the Dominican Republic and
Puerto Rico that goes back to the colonial era. Factors such as the short distance
between the islands, the importance of maritime transportation to the indigenous
people, and the shared historical and cultural events have contributed to the
bidirectionality of this migration. At the end of the 18th century, the migratory
dynamics tended to be more unilateral and restricted. The transfer of Hispaniola
to France (1795), the subsequent Haitian Revolution (1804) and the Haitian
occupation of Santo Domingo (1822–1844) favored the migration to Puerto Rico
of Dominican exiles, mostly white owners with their black slaves and mulatto
workers. However, at the end of the 19th century, the migratory direction
changed as the development of the Dominican sugar industry drove the Puerto
Rican labor force to migrate to the Dominican Republic. This migration
decreased between 1930 and 1960, as Puerto Ricans began to emigrate en masse
to the United States, and Dominicans were prevented from traveling to Puerto
Rico due to strict migratory restrictions during the Trujillo dictatorship (Duany,
2010; Hernández, 1994).
In the 1960s, a massive emigration of Dominicans to Puerto Rico begins as
a result of the political destabilization of the Dominican Republic after the
assassination of the dictator Rafael Leónidas Trujillo (1961), from the coup of
state against president Juan Bosch (1963) and the second US military occupation
of the Dominican Republic in 1965 (Duany, 2010; 2005; Graziano, 2013;
Pascual Morán & Figueroa, 2000). During this time, the Dominican migration to
the United States increased from approximately 9,915 people in 1950 to 84,065
in 1960 (Duany, 2010). The first wave of emigration was characterized by the
legal migration of members of the urban middle class with a high level of
education and income (Hernández, 1994; Pascual Morán & Figueroa, 2000). The
second wave commenced during the 1980s, as an effect of the economic crisis
generated by inflation of the late 1970s, when the Dominican Republic moved
from an agricultural to an export-oriented economy, which helped accentuate
class differences. The economic crisis that characterized this era led to a massive
emigration of the less advantaged classes that sought better opportunities. By
that time, Puerto Rico was already an industrialized urban country, which
impelled that Dominican labor migratory flow to be pointed towards the
neighboring island (Duany, 1995; Hernández, 1994). Restrictions in the US
immigration laws and difficulties in obtaining visas motivated many Dominicans
to emigrate illegally. During the 1990s and early 2000s, there was constant
immigration to Puerto Rico. Dominicans became the most representative ethnic
minority (De Toro-Alfonso et al., 2012; Duany, 2005). Certainly, the movement
of Dominicans to the Island has grown exponentially over the years. While the
1990 Census reported 37,505 Dominicans in Puerto Rico, the 2010 Census
recorded 68,036. However, these figures are very conservative, since they do not
include the undocumented population, which is estimated at between 20,000 and
150,000 Dominicans (Duany, 2010, 2005; Pascual Morán & Figueroa, 2000).
The motivations for the Dominican emigration are fundamentally
economic. As Duany (2010) asserts, Dominicans travel in search of a better life
that provides political and economic stability, allows them to satisfy essential
needs and gives way to the realization of personal aspirations and family
projects. Given this panorama, three structural elements can be pointed out in
Dominican emigration: economic considerations, extreme poverty and the
benefits of migration for the Dominican government. Puerto Rico offers a
favorable destination for many Dominican emigrants and serves as a bridge for
the United States (Duany, 2010, 2005; Graziano, 2013; Pascual Morán &
Figueroa, 2000). Unlike the United States, Puerto Rico and the Dominican
Republic share the same language, a similar cultural history and geographic
proximity that – together with the economic opportunities and living conditions
of the country – motivate immigration to Puerto Rican soil.
On average, Dominican emigrants are men between 18 and 40 years old,
with few formal studies, and working women with low level of education
(Duany, 2010; Graziano, 2013). Most live in San Juan – especially in Santurce,
Río Piedras and Hato Rey – and work informally in agriculture, construction,
maintenance, domestic services and street commerce. However, in recent years,
there has been an increase in the establishment of small Dominican-owned
businesses, such as wineries, supermarkets, bars, pharmacies, bakeries,
restaurants, beauty salons, radio and television programs, newspapers,
orchestras, legal and professional services, travel agencies, moving companies,
remittance business, car dealerships, gardening and construction companies
1
(Duany, 2010).1
This immigrant population is also associated with a darker skin color than
the general Puerto Rican community. This perception has generated negative
perceptions among Puerto Ricans (Suárez Büdenbender, 2009), which are
manifested in jokes, parodies seen in the general media (Duany, 2010, 2005;
Martínez, 2003; Rosa Abreu, 2002), and in daily social (Mejía Pardo, 1993) and
linguistic (Suárez Büdenbender, 2013) interactions. Despite previous studies on
this community, there are few sociolinguistic studies (Mejía Pardo, 1993; Suárez
Büdenbender, 2009, 2013) and very few experimental works that account for the
perceptions, attitudes and explicit and implicit stereotypes within an ethnic and
linguistic continuum (see Diagram 1). The inquiry in the perceptions, attitudes
and stereotype will serve as conceptual support to address the findings of this
research.

3. Linguistic attitudes, perceptions and stereotypes

The linguistic attitudes have been studied mainly from two theoretical
approaches: behaviorist and mentalist. While behavioral studies evaluate
attitudes based on answers to the linguistic behavior of the speakers in a natural
setting, mentalist approaches conceive them as a complex inner mental state that
can be accessed through stimulus (González, 2008). Attitudes are often elicited
through direct or indirect stimuli. In the field of sociolinguistics there has been a
preference for mentalist approaches, which are more predictive and do not
require direct observation. Questionnaires, interviews and surveys have been
widely used instruments to measure opinions and, therefore, the linguistic
attitudes of speakers. However, due to the degree of awareness that motivates a
question, this type of methodology has been replaced by more experimental,
implicit techniques, such as auditory perception techniques and matched guise or
hidden pairs (Lambert et al., 1960). By means of a short recording, or auditory
stimulus, the speaker is exposed to implicit information that provokes a series of
attitudes, beliefs and stereotypes about what he or she hears (González, 2008).
This procedure gave way to the verbal guise technique, which starts with more
natural and spontaneous stimuli from different speakers of the same linguistic
variety. To both techniques other strategies have been incorporated – such as the
use of photos, eye movements, etc. – to expand on the collection of data
(Campbell-Kibler, 2006).
Social psychologists have recognized that attitudes can be understood as
evaluative responses to an object (Vogel & Wänke, 2016), either impulsive or
through a reflexive process, that is, implicit or explicit attitudes. Explicit
attitudes are those that can be reported or constructed after an introspective
process, through techniques of direct questioning about the object of study. This
technique has certain limitations, because often the participants realize what the
researchers are trying to measure and may choose to modify their answers. On
the other hand, indirect techniques – such as verbal guidance – try to prevent
these conscious states of the participants, although this objective is not always
achieved (Pantos, 2010; Vogel & Wänke, 2016). Hence, psychologists have
engaged in implicit or unconscious access to the attitudes of the participants,
recognizing, in principle, that attitudes influence the behavior of individuals and
can be measured indirectly, either by physiological reactions (Pantos, 2010;
Vogel & Wänke, 2016) or other mechanisms. Therefore, it is pertinent not only
to access what the subjects explicitly express about the object of study, but also
to investigate other features of the object, implicitly through eye movements,
gestures, body language, laughter, among others (Vogel & Wänke, 2016). For
example, time, frequency, intensity, directionality of a look, laughter, and
movements of a person’s body can reveal their internalized attitudes about the
object of study. This study will use some of these experimental techniques, as
discussed in the next section.

4. The study

4.1 Participants

This first approach investigates, explicitly and implicitly, the perceptions,

attitudes and stereotypes of Puerto Ricans towards the Dominican immigrants
and their descendants in the dialectal continuum. With this goal in mind, a
sample of Puerto Ricans was selected, composed of 24 participants, all of them
residents for most of their lives in the metropolitan area of Puerto Rico, divided
according to the following variables: age (18 to 75 years): 6 participants from
among 18 and 25 years old; 7, between 30 and 45 years old; 6, between 50 and
65 years old, and 5, aged 70 or older; sex: 12 women and 12 men; education: 1
with less than high school; 2 with associate degree; 8 with baccalaureate and 13
with graduate studies; social class (according to income): 8 from low social
class; 9 of the middle class and 7 of the upper class (Table 1). This sample,
recruited through the snowball technique, is small, but represents the Puerto
Rican community of San Juan.
The geographical distancing factor of the participants was also taken into
account with respect to the Dominican neighborhoods in San Juan. Three criteria
were established within this variable: near, intermediate and far. This variable
attempt to measure the effect of the geographic distance of the community with
respect to the perceptions, attitudes and stereotypes of Puerto Ricans towards
Dominicans, Dominiricans and others in the dialectal continuum.

Table 1. Participants – Evaluators (24 Puerto Rican residents of San Juan)

Variable Categories

Age 18–25 years 30–45 years 50–65 years 70 years or

older
6/25% 7/29.3% 6/25% 5/20.8%
Gender Female Male
12/50% 12/50%
Education High school or Associate Baccalaureate Graduate Graduate
less degree student studies
1/4.2% 2/8.3% 8/33.3% 3/12.5% 10/41.7%
Social class Low Middle Upper
8/33.3% 9/37.5% 7/29.2%
Geographical Near Intermediate Far
distancing
5/20.8% 9/37.5% 10/41.7%

4.2 Participants of the stimulus: Verbal guise

A stimulus sample representative of the dialectal continuum (Diagram 1) was

chosen, in which homogeneity was sought, so that age (28–32 years), gender
(male), level of education (all with university studies, but without completing the
BA degree), and the topic of conversation of the participants of the stimulus (the
economic situation in Puerto Rico) were controlled. On the far left side of the
ethnic-dialectal continuum is the Dominican immigrant – with one year or less
of having arrived in Puerto Rico –, followed by the established Dominican –
with more than 10 years residing in Puerto Rico –, the son or daughter of
Dominican parents – those born in Puerto Rico, called Puerto Rican
Dominican –, the Dominirican – the son or daughter of a Puerto Rican or
Dominican father and a Dominican or Puerto Rican mother –, and, finally, on the
far right, the Puerto Rican. The five participants-stimuli, recruited by reference
of acquaintances, were interviewed for a space of 45 minutes to 1 hour, by one
of the principal investigators. From each of the recordings, a fragment of
approximately 1 minute was taken, devoid of any lexical contamination and of
ideological positions or direct prejudices on the subject.

4.3 Instruments and type of analysis

Once the five stimuli were recorded, a questionnaire was prepared with direct
and indirect questions about these ethnic and linguistic groups, and the
perceptions, attitudes and stereotypes of the Puerto Ricans towards them. The
questionnaire included open questions about nationality or ancestry, level of
education and social class, and questions – within a Likert scale of four grades:
strongly agree, agree, disagree, strongly disagree – about intelligence and the
correctness or affability of speech, in order to correlate the perceptions reported
by speakers with involuntary perceptions, attitudes, stereotypes and social
prejudices. It also included some questions that served as distractors and others
that obtained data for other variables that are not contemplated in this chapter.
The questionnaire was complemented with a test of four groups of four photos
each, which tried to present the Puerto Rican participants with a particular photo
within a continuum of race or skin color, level of education, type of profession
(level of education) and place of residence (social class). The participant would
select four of twelve photos, four for each of the categories mentioned. The
photos chosen for this test, which did not vary in order or by stimulus, were
validated by thirteen Puerto Rican participants who were not part of the main
study.
Both the questionnaire and the photo test were administered in a face-to-
face session of approximately 45 minutes as a semi-spontaneous interview,
guided and recorded by one of the two principal investigators, a 26-year-old
woman and a 48-year-old man. The tests were carried out in the homes or places
of work of the participants, in universities and schools, and in the Consulate of
the Dominican Republic in Puerto Rico, according to the convenience of the
participant. As part of the procedure, one of the investigators explained the
consent sheet and instructions for the task. Then, the participants were asked to
listen to the stimuli and to answer the questions of the questionnaire and any
other that would arise naturally as part of the exchange with the interviewer.
Subsequently, the participant chose four photos that identified physical and
social aspects of the stimuli. All the participants listened to the stimuli in the
same order. In the same way, the order of the questions and images in the photo-
questionnaire was not altered. At the end of each interview, the participant
completed an online questionnaire about their sociodemographic data.
The answers to the questions and the photos were collected in an online
instrument and coded according to a coding table that correlated each of the
variables with the five stimuli that were part of the research, with the support of
SPPS. Through this program, statistical analysis was performed – including
frequencies, percentages, cross-tabulations and Chi-square tests – as discussed in
the next section.

5. Results and analysis

In this section we disclose the findings from the interviews made to twenty-four
Puerto Ricans about the five stimuli under evaluation, specifically the ethnic and
dialectal identification, the physical qualities, such as skin color, social
characteristics, such as social class, education, profession, place of residence,
among others, and evaluative judgments, such as the degree of intelligence and
the perceptive affability of his speech. Regarding the ethnic-dialectal continuum,
the data in Figure 1 show that Puerto Ricans have a very precise dialectical
awareness of the opposite degrees of the continuum: they recognize the stimulus
of the Puerto Rican with 100% assertiveness, and the newly arrived Dominican
very accurately (91.7%). This perception does not vary according to the gender
of the Puerto Rican evaluators when it comes to the Puerto Rican stimulus. Both
women and men perceive linguistically, without any doubt, the Puerto Rican. As
far as Dominicans are concerned, although there are high degrees of auditory
dialectal perception, the recognition varies according to the gender of the
evaluators: Puerto Rican women perceive Dominicans more precisely.
Dominirican, a Puerto Rican of mixed parents, almost always of Puerto
Rican father and Dominican mother, is highly recognized by the Puerto Rican
evaluators as Puerto Rican (87.5%); his Dominican background is barely
perceived (8.4%). For this stimulus, men were more skilled at perceiving it
(45.8%) than Puerto Rican women (41.7%). On the other hand, the Puerto
Rican-Dominican, the son or daughter of Dominican parents and born in Puerto
Rico, is very difficult to perceive linguistically. The perception fluctuates
between Puerto Rican (58.3%), Dominican (29.2%) and Domirican (12.5%). The
X2 tests prove this last finding with a very high statistical significance (0.03).
Finally, the Dominican linguistic variant is hardly perceived in the established
Dominican, whom the evaluators mistakenly identified as Puerto Rican (91.7%)
with the same high percentage that they identified the newly arrived Dominican.

Figure 1. Perception of ethnic group or nationality according to Puerto Ricans

By correlating stimuli with physical qualities, such as skin color, social traits,
such as social class, education, profession, place of residence, among others; and
other more subjective qualities, such as intelligence and perceptive affability, the
social perception appears to be more complex. From this range of complexities,
varied attitudes and stereotypes are given to the various ethnic-dialectal
continuum groups, as shown in the following sections. Before proceeding with
the analysis, it must be made clear that all the stimuli of this study are part of a
fairly homogeneous sociodemographic group (Section 4.1). Regarding the skin
color, the polarity of the continuum together with the intermediate groups were
subjected to an association of perceptions, by means of the photo test, within a
scale of three values: white, brown and black (Figure 2). Puerto Ricans perceive
the established Dominican as whiter (45.8%), followed by the Puerto Rican
Dominican (33.3%) and the Puerto Rican (29.2%), and identified as darker
skinned the newly arrived Dominican (54.3%). In spite of the fact that the
category of trigueño characterizes all the groups, except the newly arrived
Dominicans (perceived as darker skinned), the continuum closest to the Puerto
Rican – that is, the Puerto Rican, the Dominirican and the Puerto Rican
Dominican – received most consistently the racial attribute of trigueño; without
there being marked differences according to the gender of the Puerto Rican
participants. In other words, they perceive as the whitest the established
Dominican, as the blackest the newcomer Dominican, and as trigueño – more
leaning towards white –, the groups closest to the Puerto Rican ethnic
continuum. These data dialogue with the findings of the social class variable
(Figure 3).

Figure 2. Perception of skin color, according to Puerto Ricans

All the members in the ethnic-dialectal continuum were perceived as members
of the middle class, although this perception is stronger as the continuum gets
closer to the Puerto Rican. Puerto Ricans were perceived within a middle to
upper social scale, a scale that fluctuates between 75% within the middle class
and 25% in the upper; without the variable of gender affecting this perception.
The Dominirican, although more associated with the Puerto Rican (see
Diagram 1/ Figure 1), is more linked to the middle class. Only 12.5% of Puerto
Ricans, mainly men, perceived the stimuli as a member of the lower class. In
contrast, the closer to Dominican a person was in the continuum – newly arrived
or established – the more he was associated to the lower class. The data, based
on questions from the questionnaire, was supported by the photo test that inquire
into the perceived level of education, type of profession and place of residence
of the participants of the stimulus (see Figure 4 and 5).
The evaluators situated the Puerto Rican stimulus in between the middle
and upper social class (93.3%) at a rate of 70.8% in the middle class and 12.5%
in the upper class. On the other hand, the closest to Dominican a stimulus was in
the continuum, the higher the social valorizations he received since they were
identified as upper-middle class: Dominican newcomers received an 83.3% and
the established Dominican a 91.7%. They were both mostly perceived as
member of the middle class in the semi-spontaneous interviews. The
Dominirican and Puerto Rican Dominican were perceived by more than 30% of
the Puerto Rican evaluators as members of the lower class; that is, they are
associated more as members of the lower middle class than the two extremes of
the dialectal continuum – Puerto Ricans and Dominicans, both established and
newcomers –, although the latter is still associated in some degree to the lower
class (16.7%).

Figure 3. Perception of social class, according to Puerto Ricans

To indirectly investigate the social judgements that arise from the stimuli, the
participants were exposed to a series of photos related to a place of residence,
type of profession and workplace. In regard to the place of residence, the two
extremes of the dialectal-ethnic continuum – Puerto Rican and Dominican, both
the newly arrived and the established one – received the best residential
evaluations. These dialectal groups were placed in residents of the upper-middle
class (83.3% and 83.3 and 91.7%, respectively). On the other hand, the
Dominirican and the Puerto Rican-Dominican, although they are still considered
members of the middle class, one third of the Puerto Rican participants place
them in low-class residences (33.3%). These findings seem to coincide with
perceptions about type of occupation seen in Figure 4.

Figure 4. Place of residence, according to Puerto Ricans

As for the profession/occupation criteria (Figure 5), the groups at opposite sides
of the continuum acquire once more the best evaluations: the Puerto Rican and
the established Dominican were perceived as professionals by 50% of
participants, the newcomer Dominican was also perceived as a professional by
41.7%. In contrast, the Dominican and Puerto Rican Dominican were perceived
as less educated, they tended to be associated with technical (58.3%) and non-
technical services (50%).

Figure 5. Perception of the type of profession/occupation, according to Puerto

Ricans
Lastly, this study examined more implicit evaluations, such as the perceived
level of intelligence and affability of the stimuli (Figure 6). These perceptions
could be correlated with the physical characteristic and social attributes analyzed
in the previous sections. Regarding the level of intelligence, the data
corroborates the previous conclusions. Those at opposite side of the ethnic-
dialectal continuum – the Puerto Rican and Dominican, both newly arrived and
established – received the best evaluations of intelligence (54.2% of perceived
intelligence, and between 75% perceived intelligence and 87.5% average
intelligence). Those groups in the middle of the continuum – the Dominirican
and Puerto Rican Dominican – were perceived with average levels of
intelligence (70.8% and 54.2%).

Figure 6. Intelligence level, according to Puerto Ricans

The affability trait received evaluations similar to that of perceived intelligence
(Figure 7). The members at opposite side of the continuum – the Puerto Rican
(54.2%) and the Dominican, newcomer (58.3%) and established (41.7%) – were
perceived as pleasant and very pleasant people; while the Dominirican and
Puerto Rican Dominican were perceived as neutral or unpleasant (62.5%). In the
next section, these findings are discussed in light of the debate on attitudes and
the socio-historical Puerto Rican-Dominican context.

Figure 7. Level of affability, according to Puerto Ricans

6. Discussion and conclusions
This investigation aimed to examine, through a test of verbal guise, the direct
and implicit attitudes of Puerto Ricans towards the Puerto Rican-Dominican
ethnic-dialect continuum, represented by five auditory stimuli (Diagram 1).
Following Vogel & Wänke (2016), through direct and indirect questions and a
series of photographs, physical, social and psychological properties associated
with the stimuli – such as ethnic and linguistic origin, color of skin, social class,
place of residence, level of education, type of profession or occupation, level of
perceived intelligence and affability of speech – were measured. This
investigation sought to implicitly evaluate other perceived characteristics of the
stimulus, such as physiological, social and psychological traits. Both instruments
produced very systematic responses, which confirm that auditory perceptions
can be linked to linguistic, physical, social and psychological judgments (Blas
Arroyo, 1999; Díaz-Campos & Navarro Galisteo, 2009).
In this way, the research finds positive evidence for the question: Can
Puerto Ricans identify the dialectal differences between Dominicans and Puerto
Ricans, as different dialects within Caribbean Spanish? The data strongly
suggests that Puerto Ricans have a very assertive perceptive dialect
consciousness about their own and the Dominican dialects. They identified,
without being mistaken, both the Puerto Rican and the Dominican, the two
extremes of the ethnic-dialectal continuum. However, the barriers begin to be
confused with the Dominicans who established themselves for a long time in the
Puerto Rican community, they identify them as a member of the receiving
community.
Regarding the distinction and associations of the Puerto Rican Dominicans
and Dominiricans in the ethnic-dialectal continuum, the results prove that these
groups, although perceived as Puerto Ricans by the majority of the evaluators,
project certain linguistic hybridity. That is, although both groups acquire features
of the socially dominant dialect – Puerto Rican –, they retain certain features of
the Dominican dialect; which makes them linguistically and socially more
complex, more blurred and more difficult for Puerto Ricans to perceive. As
members who share ethno-sociolinguistic features with those closer in the
ethnic-dialectal continuum, Puerto Ricans perceived Dominiricans as Puerto
Rican (87.5%), and Puerto Rican Dominican as both Puerto Rican (58.3%) and
Dominican (30%). These proportions seem to make sense if one considers that
the Puerto Rican Dominican receive more input from the Dominican community
as both parents are Dominican, than the Dominirican, who has only one parent
with said ancestry.
The established Dominican, Dominirican and Puerto Rican Dominican
stimuli present linguistic accommodations that could be interpreted, in the case
of the established Dominican, as a process of de-dialectalization from their
native dialect and adoption of the dominant one, and in the case of the other two,
as a process of bidialectalization. Both processes are the result of the forces and
tensions produced by the Puerto Rican-Dominican ethno-sociolinguistic context
in Puerto Rico (Duany, 2010, 2011). These findings help answer the question:
what are the linguistic, ethnic, social and psychological attitudes generated by
these ethnic-dialectal groups?
This research found that the linguistic perception towards Puerto Ricans
oscillate between positive or very positive in almost all the categories. Puerto
Ricans are associated with a trigueño skin tone – more leaning towards white –
as belonging to the middle class, with mobility towards the upper class, with a
middle-class residence, as professional, intelligent and affable. This positive
perception of Puerto Ricans is contrasted with the real characteristics of the
stimulus: a dark-skinned, middle-class working man, resident in a lower-middle
class area, with only a few years of higher studies. Therefore, these positive
attitudes towards the Puerto Rican stimulus seem to respond more to a pattern of
solidarity, empathy and identification with the Puerto Rican community. These
positive attitudes contrast, at the same time, with the reported negative attitudes
of Puerto Ricans towards their own dialectal variety (López Morales, 1992).
This finding requires further explanation. Therefore, the results of the other
stimuli must be examined.
The positive perception of Puerto Ricans also extended to the two
representatives of the Dominican community in the continuum – the newly
arrived and established Dominican. The established Dominican seems to be
valued more positively than other groups, as he was identified by the evaluators
as Puerto Rican. The newcomer Dominican was perceived as black or trigueño,
but never white, middle class, as a resident of the same class sector, as
professional, intelligent and affable. The established Dominican, in turn, was
perceived as white or trigueño, of middle and upper class, as professional,
intelligent and affable, always in degrees superior to the newcomer Dominican
and even to the Puerto Rican. In other words, among Puerto Ricans, there is an
assertive ethnolinguistic perception of the two extremes of the ethnic-dialectal
continuum – Puerto Ricans and newly arrived Dominicans –, but not for the
established Dominican. However, both extremes, represented by these three
stimuli, generated favorable and very favorable attitudes. The reported attitudes
towards Puerto Ricans could be interpreted as contradictory if compared with
other sociological studies of discrimination against Dominicans in Puerto Rico
(Duany, 2005, 2011; Rosa Abreu, 2002; Suárez Büdenbender, 2009). Later on,
these supposed perceptual contradictions will be discussed.
If the opposite groups of the continuum are perceived with positive
attitudes, then, it should be expected that the hybrid groups in the ethnic-
dialectal continuum – the Dominiricans and the Puerto Rican Dominicans –
would preserve some or all of the positive associations that the evaluators
attributed to the Puerto Rican and the Dominican stimuli, especially to the
established Dominican. The data did not confirm these intuitions. The
Dominiricans and the Puerto Rican Dominicans were judged less positively, if
not negatively, in comparison with their ethnolinguistic ascenders. Both groups
receive more consistently the racial attribute of trigueño. Within the racial scale,
the group perceived as the whitest was the established Dominicans; as the
blackest, the newcomer Dominicans; and as darker but leaning towards white
were the groups closest to the Puerto Ricans in the continuum. The Dominirican
and the Puerto Rican Dominican received lower social evaluations, as was
anticipated by the questionnaire data. The questionnaire and photo test results
suggest a greater tendency to perceive these intermediate groups as members of
the lower class, with low class residences and with jobs based on technical and
non-technical services. Also, they are perceived as individuals with average or
neutral intelligence and as less pleasant than others.
The findings of this investigation seem to confirm that attitudes influence
the judgments and evaluations of an individual, either indirectly or through
physiological (Pantos, 2010; Vogel & Wänke, 2016), social and psychological
reactions. The Puerto Rican evaluators have mentally constructed a physical,
social and psychological model of themselves, of the Dominican otherness and
of the descendants of both groups. Based on these qualities, they judged,
grouped and stereotyped, directly and indirectly, consciously and unconsciously,
the stimuli. On a less conscious level (Pantos, 2010; Vogel & Wänke, 2016), the
participants displayed a greater emotional, affective, solidary and identity
connection towards Puerto Ricans. On a more conscious level, the evaluation
and judgment of the otherness of the Dominican immigrant was influenced by
stereotypes of poverty and discrimination.
The explicit attitudes of Puerto Ricans towards their own and the
Dominican dialects, as extremes of the ethnic-dialectal continuum, respond to a
revision and affirmation of their implicit attitudes, as theorized by Pantos (2010).
Despite the fact that the attitudes displayed by Puerto Ricans could contradict
other attitudes towards themselves and Dominicans; such attitudes should not be
seen as cancelling out the others, but as two attitudes that coexist within one
individual or one community. They should be seen as part of the complex
context of the Puerto Rican-Dominican community. Therefore, by expressing
positive attitudes towards one’s group – represented by the Puerto Rican
stimulus, identified as such by 100% of the participants –, and for the Dominican
otherness – represented by the newly arrived Dominican stimulus, identified as
such by more than 90% of the participants –, the Puerto Rican subjects
interviewed unconsciously and consciously position themselves in a continuous
and somewhat mobile terrain as the result of socio-historical ties with the
Dominican immigrant community.
It should be noted that this community has been more and more present in
different areas of Puerto Rican society, mainly in the metropolitan area.
Although the community has been stereotypically associated with individuals of
darker skin tones, mainly men between 18 and 40 years with few formal studies,
and working women with low level of education (Duany, 2010; Graziano, 2013),
residents of San Juan – especially in sectors of Santurce, Río Piedras and Hato
Rey – and carrying out informal jobs dedicated to agriculture, construction,
maintenance, domestic service and street commerce, this stereotype is becoming
increasingly complex and being challenged. Although stereotypes continue to
portray immigrants as socially stagnant, which help generate negative
perceptions (Suárez Büdenbender, 2009) through jokes and parodies in the
media (Duany, 2010, 2005; Martínez, 2003; Rosa Abreu, 2002) and in everyday
social (Mejía Pardo, 1993) and linguistic interactions (Suárez Büdenbender,
2013), the stimuli require a different perceptual analysis. It could be due to real
perceptions and/or perceptive accommodations. The stimuli on both sides of the
ethnic-dialectal continuum projected positive physical, social and psychological
characteristics to Puerto Ricans that could be counteracted by the ethno-
sociolinguistic stereotypes of both groups. In addition, the fact that the
interviews were conducted by two Puerto Rican interviewers – a university
professor and a graduate student – could have revealed to some degree the
ulterior motive of the research, although it was designed to mitigate its impact on
the participants. Also, the fact that Puerto Ricans are now more aware of the
ethno-sociolinguistic diversity of their community further complicates the
investigative panorama of attitudes.
There is no doubt that dialects are markers of identity and that in contact
situations, they are socially negotiated. Consequently, speakers may choose to be
part of a speech community or to stay out of said community (Wright, 2014). In
the Puerto Rican Dominican context, where there is a marked social and
linguistic stratification, the differences between groups intensify. Both the
Dominiricans and Puerto Rican Dominicans, as well as the established
Dominicans project themselves as part of the Puerto Rican community; they
have bidialectalized and de-dialecticized, respectively, in favor of the Puerto
Rican dialect. In the case of Puerto Ricans with some type of Dominican
ancestry, their ethno-sociolinguistic hybridity is not lost on Puerto Ricans.
Perhaps this hybridity – as part of the ethno-dialectal continuum – is the
mechanism that Puerto Ricans use to exclude, devaluate, and hierarchize them
inferiorly, according to concealed social prestige (Lippi-Green, 1997). This
research demonstrated that there are social and psycholinguistic mechanisms
among ethnolinguistic groups in contact, that generate and promote consciously
or unconsciously sociolinguistic perceptions, attitudes and stereotypes towards
others and their linguistic varieties. In this way, the dialectal or sociolect variety,
color of skin, ethnic origin, socioeconomic status, profession, intelligence, etc.
acquire positive or negative values in light of the community’s prevailing
ideology (Álvarez, 2009; Suárez Büdenbender, 2009, 2013; Toribio, 2007).
To further study these perceptions, attitudes and stereotypes, future
investigations must consider the perception of Dominicans, residents of Puerto
Rico and the Dominican Republic, and the perception of Puerto Ricans
Dominicans and Dominiricans. Likewise, the stimuli should be broadened by
including women and a more heterogeneous social representation of this ethno-
dialectal continuum. In addition, future studies should employ other
experimental methodologies from psycholinguistics, which can test the implicit
behavior of participants when evaluating, judging and stereotyping the stimuli,
such as reaction time, frequency, intensity, directionality of the look, laughter,
movements of the body, captured by an eye-tacker. These methodologies have
still to be applied in the majority of sociolinguistic studies.

Note
1. This description contrasts with that of Dominican in the United States, where there is more legal
immigration, with more formal education and more specialized jobs (Suárez Büdenbender, 2009). ①

References
Álvarez, A.
(2009.) Sobre la construcción discursiva del país: Actitudes lingüísticas en Venezuela. Presente y pasado-
Revista de Historia, 14(27), 87–106. ① ②
Blas Arroyo, J.
(1999.) Las actitudes hacia la variación intradialectal en la sociolingüística hispánica. Estudios Filológicos
34, 47–72. ①②③
Bullock, B. E., & Toribio, A. J.
 (2008.) Kreyol incursions in Dominican Spanish: The percept of Haitianized Spanish. In M. Niño-Murcia
& J. Rothman (Eds.), Linguistic identity and bilingualism (pp. 175–198). Amsterdam: John Benjamins.

Bullock, B. E., & Toribio, A. J. (Eds.) (2009.) The Cambridge handbook of linguistic code-switching.
Cambridge, UK: Cambridge University. ①
Campbell-Kibler, K.
(2006.) Listener perceptions of sociolinguistic variables: The case of (ING) (Unpublished doctoral
dissertation). Stanford University. ①
Díaz-Campos, M. & Navarro Galisteo, I.
(2009.) Perceptual categorization of dialect variation in Spanish. In J. Collentine, M. García, B. Lafford
F. M. Marin (Eds.), Selected proceedings of the 11th Hispanic Linguistic Symposium (pp. 179–195).
Somerville, MA: Cascadilla. ①
Duany, J.
(2011.) Blurred borders: Transnational migration between the Hispanic Caribbean and the United States
Chapel Hill, NC: The University of North Carolina Press. ①②③
Duany, J.
(2010.) La nación en vaivén: Identidad, migración y cultura popular en Puerto Rico. San Juan, Puerto
Rico: Callejón. ① ② ③ ④ ⑤ ⑥ ⑦ ⑧ ⑨ ⑩ ⑪ ⑫
Duany, J.
(2005.) Dominican migration to Puerto Rico: a transnational perspective. CENTRO Journal, 17(1), 242–
268. ① ② ③ ④ ⑤ ⑥ ⑦ ⑧
Duany, Jorge, Luisa Hernández Angueira, and César A. Rey
(1995.) El Barrio Gandul: economía subterránea y migración indocumentada en Puerto Rico. Caracas:
Nueva Sociedad. ①
Fúster, N.
(2012.) Variación y actitudes lingüísticas en Cuba (Unpublished doctoral disseration). Universidad de
Puerto Rico, Río Piedras. ①
González Martínez, J.
(2008.) Metodología para el estudio de las actitudes lingüísticas. In Actas del XXXVII Simposio
Internacional de la Sociedad Español de Lingüística (SEL). Retreived from [Link]
/simposiosel/act ① ②
Graziano, F.
(2013.) Undocumented Dominican migration. Austin, TX: University of Texas Press. ① ② ③ ④
Hernández, J. (Ed) (1994.) Corrientes migratorias en Puerto Rico/ Migratory trends in Puerto Rico. San
Germán, Puerto Rico: CISCLA- Revista interamericana. ① ② ③
Lambert, W., Hodgson, R., Gardener, R., & Fillenbaum, S.
(1960.) Evaluational reactions to spoken language. Journal of Abnormal and Social Psychology, 60, 44–51.
①
Lippi-Green, R.
(1997.) Language ideology and the language subordination model. In English with an accent: Language,
ideology and discrimination in the United States (pp. 63–73). New York, NY: Routledge. ① ② ③
Lipski, J.
(1994.) Latin American Spanish. London: Longman. ①
López Morales, H.
(1992.) El español del caribe. Madrid: Editorial Mapfre. ① ②
Martínez-San Miguel, Y.
(2003.) Caribe Two Ways: Cultura de la migración en el Caribe insular hispánico. Río Piedras, Puerto
Rico: Ediciones Callejón. ① ②
Mejía Pardo, D.
(1993.) Macroestructuras, superestructuras y proposiciones de opiniones en 17 relatos de puertorriqueños
acerca de dominicanos (Unpublished MA thesis). Universidad de Puerto Rico, Río Piedras. ① ② ③ ④
⑤
Mintz, S. W.
(1996.) Enduring substances, trying theories: The Caribbean Regios as Oikoumenl. Journal of the Royal
Anthopological Institute, 2(2), 289–311. ①
Ortiz López, L.
(2000.) El español de Puerto Rico en el contexto caribeño: El debate sobre su genesis. Revista de Estudios
Hipánicos, XXVII(1), 99–118. ①
Ortiz López, L.
(2015.) Transnacionalidad e identidades lingüísticas en la frontera dominico-haitiana. In J. R. Valdez (ed.),
International Journal of the Sociology of Language, 15(233), pp. 15–39. Berlin & New York: Mouton
de Gruyter. ①
Pantos, A. J.
(2010.) Measuring implicit and explicit attitudes toward foreign-accented speech (Unpublished doctoral
dissertation). Rice University, Houston, Texas. ① ② ③ ④ ⑤
Pascual Morán, V., & Figueroa, D. I.
(2000.) Islas sin fronteras: Los dominicanos indocumentados y la agricultura en Puerto Rico. San Germán,
Puerto Rico: CISCLA-Revista Interamericana. ① ② ③ ④
Potowski, K.
(2011.) Intrafamilial dialect contact. In M. Díaz Campos (Ed.), The handbook of Hispanic sociolinguistics
(pp. 579–597). Hoboken, NJ: Blackwell. ①
Rosa Abreu, A.
(2002.) La identidad cultural de la mujer dominicana en clase trabajadora en Puerto Rico: su articulación
en la comedia televisiva (Unpublished MA thesis). Universidad de Puerto Rico, Río Piedras. ① ② ③
Suárez Büdenbender, E. M.
(2013.) “Te conozco bacalao”: Investigating the influence of social stereotypes on linguistic attitudes.
Hispania, 96(1), 110–134. ①②③④
Suárez Büdenbender, E. M.
(2009.) Perceptions of Dominican Spanish and Dominican self-percepction in the Puerto Rican diaspora
(Unpublished doctoral dissertation). Pennsylvania State University, University Park, PA. ① ② ③ ④ ⑤
⑥⑦⑧
Toribio, A.
(2007.) Language attitudes and linguistic outcomes in Reading, Pennsylvania. In R. Salaberry (Ed.),
 Language allegiances and bilingualism in the United States (pp. 35–61). Clevedon, UK: Multilingual
Matters. ① ②
Toro-Alfonso, J., López Ortiz, L., & Nieves Lugo, K.
(2012.) Sexualidades migrantes: La Emigración de hombres Dominicanos gay. Caribbean Studies, 40(1),
59–80. ①
Vogel, T. K., Wänke, M., & Bohner, G.
(2016.) Attitudes and attitude change. London: Routledge, Taylor & Francis Group. ①②③
⑤⑥⑦
Woolard, K.
(1985.) Language variation and cultural hegemony: Toward and integration of sociolinguistic and social
theory. American Ethnologist, 12(4), 738–748. ①
Wright, S.
(2014.) The map, the group and language ideology. Journal of World Languages 1(2), 81–98.
②
Zamora, J., & Guitart, J.
(1988.) Dialectología hispanoamericana. Salamanca: Publicaciones del Colegio de España. ①
CHAPTER 12

Andean Spanish and Provinciano identity

Language attitudes and linguistic ideologies
towards Andean Migrants in Lima, Peru
Daniela Salcedo Arnaiz
Princeton University

Abstract
The present perception study provides an approach to the language
attitudes and ideologies that both Limeños and Andean migrants have
towards a selection of (stigmatized) morphosyntactic features from
Andean Spanish, in order to propose a definition for Provinciano
identity. I collected data using a Matched-Guise task. The findings
show that people perceive Andean Spanish as an ‘incorrect’ Spanish
spoken mostly by Andean migrants, which allegedly share
stereotypical characteristics such as a low level of education,
indigenous race and poverty. Both the relevant literature and my
fieldwork research indicate that (Andean) identity is undeniably
conditioned by location, ethnicity, level of education, and race, as well
as all the processes, relationships and social interactions that take place
in Lima.

Keywords
Andean Spanish; perception; language attitudes;
linguistic ideologies; Provinciano identity
1. Introduction
Language attitudes and ideologies are an inherent aspect of every multilingual or
multidialectal society, and this is no exception in Lima, where Limeño and
Andean Spanish coexist due to a massive internal migration from the Andes to
Peru’s capital over the last forty years. This investigation presents a quantitative
and qualitative approach to the attitudes and ideologies that the residents of Lima
have towards Andean Spanish (AS) and its speakers in order to mark out the
intersection between power, status, and its association with language use
perceived by Lima’s inhabitants.
When compared to Limeño Spanish (LS),1 AS presents differences at every
grammatical level. Nevertheless, in this study, I only focus on the
morphosyntactic features which make this contact variety different from the
standard used in Lima, where my research is conducted. The reason for this
decision is that the focus of my work is not a description or analysis of AS per se
but a perception study on how certain features of AS morphosyntax might index
Provinciano identity. The expression provinciano refers, in a broad sense, to any
person who was born in the provinces as opposed to being born in the city.
Nevertheless, it is generally inferred that provincianos are of Andean origin.
Therefore, when I use the term, I am alluding to a provinciano identity or group
of identities and, within it, a provinciano way of speaking that might resemble
AS but, as I will explain throughout this work, it is not exactly the same.
According to Campbell-Kibler (2006), “understanding the structure of
sociolinguistic variation requires understanding what information it conveys to
listeners and how. Variation not only correlates with social structures but carries
social meaning, influencing listener perceptions and, through them, social
structures”. This is clearly the case of Peru’s capital, a city which grows
unstoppably. Lima’s society is stratified in various ways, and “power and status
are often able to translate social difference into social deficiency” (Edwards,
1999). As in other situations, such social inequalities are reinforced by or
reflected in ideologies of language. Therefore, when a participant listens to a
given variety, this works as a trigger that evokes prejudices or stereotypes about
the speech community in question, i.e. AS evokes language attitudes and
linguistic ideologies because it indexes a particular social meaning.
From the point of view of ‘social meaning’, a term coined by Silverstein
(2003) and later developed by Eckert (2008), meaning is not only what the
speaker wants to convey through the choice of a particular style (linguistic
features), but also the meaning assigned to the speaker’s stylistic choices by
his/her interlocutor. In other words, speakers project and listeners interpret a
specific social meaning determined by the perspective they take and the situation
at hand. Throughout this paper, I show how this concept is relevant to my
research in that I am investigating the perception of the social meaning of
speakers’ use of language. Moreover, both Eckert and Silverstein make it clear
that listeners’ perceptions are closely bound up with a person’s or community’s
ideology of language.
The present perception study traces the reactions that both Limeños and
Andean migrants have towards a selection of (stigmatized) morphosyntactic
features from AS. In contrast to much of the previous work that has been done
on AS speakers and Quechua-Spanish bilinguals, this study focuses on
perception instead of production. This focus makes an important contribution to
existing literature because participants/listeners include not only speakers of
Limeño Spanish (LS), the most prestigious variety spoken in Lima, but also AS
speakers, both monolinguals and Quechua-Spanish bilinguals. Strand and
Johnson (1996) and Niedzielski (1999) have demonstrated that speech
perception does not depend only on physical factors, but also on listeners’
expectations based on sociological factors only indirectly related to language
(Thomas, 2002).
The role of perception is central when talking about language/dialect
contact situations due to migration, especially in the case of the first migrating
generations. When this contact takes place in an environment of social conflict,
it has repercussions for perception because there is an inequality between
varieties, as is the case between AS and LS. Caravedo (2014) proposes two
different points of reference for new migrants in the city: analytic and synthetic
perception. Analytic perception consists of identifying isolated features present
in the target variety (e.g. the palatal lateral of AS), while synthetic perception
consists of recognizing a variety in a ‘general’ way; the speaker will perceive,
even in a vague or imprecise way, characteristics that will be recognized as part
of a certain dialect. Since every generalization is a consequence of analysis,
synthetic perception presupposes the previous identification of some features
through analytical perception, so that we can say that it is the former type of
perception which determines the identification of a variety as a block. The
analytical perception concept is an accurate way of explaining how AS is
perceived as a representation of Andean identity, even though AS is not
considered an actual variety but an incorrect Spanish spoken by people with a
very specific profile. Caravedo agrees with many other researchers (Le Page &
Tabouret-Keller, 1985; Milroy, 2004) on that there are certain linguistic features
that are associated with specific groups, the socio-indexical features.
Scholars have addressed related topics such as AS features, discrimination
towards this variety, Andean migrants in Lima and Quechua-Spanish contact,
etc. For example, Klee and Caravedo (2005) discuss the negative image that
migrants have of their dialect due to Limeños’ discrimination. However, they are
more focused in analyzing how different degrees of stigmatization of features
affect their spread and acceptance when in contact with LS. Godenzzi (2008) has
another interesting study which focuses on similar topics: he interviewed three
different inhabitants of different neighborhoods in Lima in order to analyze both
their linguistic and discursive features. Each of the three neighborhoods selected
are typical areas where either Limeños or Andean migrants live; in this way, the
author wants to find out each participant’s perception of themselves within Lima.
Furthermore, linguists have studied AS grammar in Peru in relative depth, both
as the variety of Quechua-Spanish bilinguals as well as the variety of
monolinguals, product of historical contact. However, there are only a few
studies dealing with the linguistic identity of AS speakers in the specific context
of migration, at least in the case of this Andean country, hence there are still
many gaps to be filled.2
Migration as a social phenomenon plays a central role in the construction of
linguistic identities, as Hazen and Hamilton (2008) show in their article about
Appalachian English in the diaspora. Currently, there are no other studies which
deal with the linguistic identity of AS speakers in the specific context of
migration – at least not in Peru – and my research brings to light important
aspects of this topic. In this way, we can better understand to what extent
Andean migrants’ identity is re-shaped when they migrate to Lima and become
provincianos.
Arellano and Burgos (2004) describe contemporary Lima as a city of
migrants and informales, who live mostly in the urban peripheries located north,
east and south of central Lima. Furthermore, they describe the parallel existence
of two Limas: the ‘classic Lima’, which dates from that historical city described
as La Ciudad de los Reyes, an urban and organized center exclusive to
aristocratic and colonial society; and the “New Lima”, constituted by migrants
and their descendants, which is consolidating itself as a multicultural and
cosmopolitan space. According to the authors, peripheral areas in Lima have
been associated with the stereotypical image of the rural, poor and marginalized
migrant, while the classic or traditional Limeño is associated with economic
success, sociopolitical power and national culture.
In that classic Lima live those who Smith (2008) and Godenzzi (2008) call
‘traditional Limeños’, whereas ‘new Limeños’ are the descendants of migrants or
provincianos. Being born in Lima is not enough to be considered Limeño, as the
data I collected will show.3 The descendants of many migrants who arrived in the
capital looking for a brighter future have improved their economic status, have
moved to better neighborhoods, and have access to better education.
Nevertheless, new and classic Limeños still hold on to certain stereotypes,
including linguistic ones, which remain in the social structure of the city.

2. Language attitudes and linguistic ideologies: About

(in)correctness
According to Milroy (2004, p. 161), language attitudes are “manifestations of
locally constructed language ideologies”, that is to say that language attitudes are
in a sense particular manifestations of language ideologies. I find this definition
adequate because it shows the link between language attitudes and linguistic
ideologies, two terms that are sometimes dissociated because of the different
theoretical frameworks in which they are embedded: the study of language
attitudes started in the field of social psychology and linguistic ideologies have
been analyzed by linguistic anthropologists such as Kroskrity (2004). In fact, the
combination of both is the best way I have found to outline provincialism in
Lima, as I explain in the following paragraphs.
A simple way to define language attitudes is as the reactions, beliefs, or
values that people have about language and language use. Speakers and listeners
have a host of attitudes that range from positive to negative, although they may
“not always be simply positive or negative, but may subsume both positivity and
negativity” (Haddock & Maio, 2004). Attitudes models, based on Plato, are
generally held to have three components: affect (feelings towards an object,
person or situation), cognition (thoughts and beliefs) and behavior (reactions)
(Hernández-Campoy, 2004). As a result of the combination of these components,
attitudes are fluid and even sometimes contradictory.
The field of social psychology has explored attitudes in depth, trying to
develop a better understanding of the mental processes that are involved in
human interaction, including language. Language attitudes have also been
studied by those branches of sociolinguistics that focus on the structure of
language and its relationship to social constructs and processes, such as language
contact variationism, perceptual dialectology or discourse analysis. According to
Campbell-Kibler (2006), “although these two threads of research have slightly
different perspectives, they share a great deal of common interest”. I have
selected the concept of language attitudes, then, because it can be applied to the
phenomena of dialect contact addressed here. I am interested in understanding
the coexistence of difference (rural vs. urban, indigenous vs. white, uneducated
vs. educated, etc.) in the specific location of Lima and what language attitudes
have developed due to the interaction of LS speakers and AS speakers.
We can find some examples of language attitudes towards Andean migrants
in Lima in Klee and Caravedo’s (2005) article, “Andean Spanish and the Spanish
of Lima”:
Andean groups exhibit distinguishing sociocultural and linguistic characteristics
that receive subjective negative evaluation from native inhabitants of Lima, or
Limeños. Specifically, they often speak Quechua or Aymara as a mother tongue
and have learned Spanish as a second language, frequently without systematic
instruction and often in adulthood. In addition, Andean migrants have a minimal
level of schooling, and some have had no formal instruction and arrive in Lima
in conditions of extreme poverty.
Interestingly, the authors seem to be reproducing the same attitudes and
ideologies that lead to the stigmatization that AS speakers suffer every day. In
my opinion, their perspective is biased and it depicts only a group of Andean
migrants. Not all AS speakers who arrive in Lima are extremely poor or
uneducated. People in the Andes can have a higher level of education too, just as
in Lima, or they can be rich, and many of them do not speak an indigenous
language. It is true that these cases are not as frequent, but I believe that
researchers should be careful in order to avoid adopting as their own, the
attitudes that pertain to the other groups they study or, in any case, make sure to
describe our positionality in the study; Andean migrants are not one single and
homogeneous group.4 Furthermore, we cannot ignore all the social and economic
changes that the Peruvian capital is going through due to internal migration and
the so called “emerging classes”.
On the other hand, linguistic ideologies can be described as the perceptions
held by people about language and, more importantly, how those perceptions are
projected onto speakers. In this way, it is not hard to find similarities between
this theory and the language attitudes framework explained above. Irvine (1989)
defines a language ideology as “the cultural system of ideas about social and
linguistic relationships, together with their loading of moral and political
interests”. I do not entirely agree with this definition because it suggests that
ideologies are shared culturally determined communal beliefs, while attitudes are
more individual in nature and, as I stated before, I am more comfortable
affirming that language attitudes portray one particular worldview of particular
individuals in a particular context. An alternative is to understand language
ideologies as “ingrained, unquestioned beliefs about the way the world is, the
way it should be, and the way it has to be with respect to language”(Wolfram &
Schilling-Estes, 2006).
Due to the fact that linguistic ideologies are perceived as natural, they tend
to be considered as unquestionable truths. In general terms, ideologies refer to
relevant aspects of a group or a characteristic that makes it different from the
others (Van Dijk, 2003). Furthermore, they belong to the social and collective
memory of that group (Kroskrity, 2004; Van Dijk, 2003); thus, within any
society, there are often ideologies in conflict (de los Heros, 2008).
My interest in the notions of language attitudes and linguistic ideologies for
addressing my research topic can be explained by the relationship that exists
between linguistic ideologies and ontology (Howard, 2007). According to
Howard (2007), the study of linguistic ideologies enlightens us about the role of
language in ontology, in who we are. She claims that language is not a neutral
communicative code, but a means to express our feelings uniquely linked to our
sense of self. To recognize this emotional characteristic of language helps us to
also recognize the ways in which linguistic matters are sensitive for language
users. For example, this is the case of Andean migrants in Lima, who suffer
discrimination because of the way they speak. As De los Heros (1999) claims,
Limeños tend to consider AS as incorrect or inferior Spanish.
Rosaleen Howard (2007), when talking about linguistic ideologies in the
Andes, claims that even though one can observe a hybrid culture in this area, the
discourse about identity tends to be dichotomous. For instance, my informants
from Andean origin would differentiate between an Ayacuchano person and a
Cajamarquino in terms of race, beauty, and dialect, among others.5 Therefore, we
cannot conceive Andean migrants as one homogeneous group nor can we think
of AS as a uniform variety shared throughout the Andes. The AS- LS and
Limeños-Provincianos oppositions I refer to in this text are a mere investigative
tool but, by no means, an exact description of Lima’s reality because both
language and identity, are in reality much more complex and fluid, and should be
understood as continua, not as well-defined entities. Therefore, I propose to
focus on a group of linguistic features that index Provinciano identity – a sub-
group of what researchers call AS. We can find an example of the dichotomy
Howard refers to in Mick’s (2011) article about domestic workers in Lima,
where participants use words such as acá ‘here’ vs. allá ‘there’ to represent the
opposition between Lima (here) and the Andean provinces (there).6 My
recordings present the same kind of oppositions, as if it didn’t matter how much
effort you make to belong here, you will always be considered an outsider.
According to Mick (2011), this transposition of spatial categorization into a
social one can be explained by the semiotic process of ‘erasure’ (Irvine, 2001), a
discursive simplification which reduces the complex sociolinguistic situation of
‘Peru’s capital to a binary opposition (Mick, 2011, p. 192).
Within the framework of language attitudes and linguistic ideologies, I want
to discuss two concepts that are central in this study: dialect awareness and
linguistic correctness. Yiakoumetti, Evan, & Esch (2005) conducted research in
Cyprus, where Greek Cypriots are bidialectal: they speak the Cypriot Dialect
(CD) and Standard Modern Greek (SMG). As expected, the standard is spoken
in public (more prestigious) domains while CD is spoken in private domains and
is usually their mother tongue. The latter is not officially recognized at school
and, as a consequence, the standard is the only variety used in formal learning,
and Cypriots are treated as if they were monolinguals. This situation, which
affects speakers’ attitudes and performance in the standard variety, is similar in
many ways to the one that takes place in Lima, between LS, the prestigious
variety, and AS. The coexistence of these two seems to be patent only for
linguists and other researchers. In practical terms, however, AS is not recognized
as a dialect, not even by its speakers. It is as if it did not exist other than as a
group of disconnected marked features in opposition to the standard (unmarked)
features of LS. These features are not perceived as constitutive pieces of a
certain variety but as manifestations of an incorrect usage of Spanish due to lack
of proficiency (Limeños generally assume that Provincianos always speak either
Quechua or Aymara and thus they don’t speak ‘good’ Spanish, which is not true
for all cases) or lack of proper education. Here, linguistic correctness is
associated with the prestige and status that are usually attributed to standards.

3. Andean Spanish: Towards a definition

Currently, it is generally agreed that AS is a contact variety that emerged
because of the contact of Quechua and Aymara with Spanish after the arrival of
Spanish conquistadors in Andean countries. Rivarola (2000) defines it as a
geographically limited variety that is spoken in the Andean areas proper (as
opposed to the coastal or Amazonian varieties of Spanish) of Andean countries,
areas where Spanish has coexisted with Quechua and Aymara for centuries.
Zavala (1999) affirms that the AS that is spoken today is very similar to the
variety learnt by indigenous people during Colonial times; it is a linguistic
variety characterized by a variation of the noun phrase due to the reduction of
categories and the omission of connectives, as well as the reinterpretation and
the extension of Spanish functions and categories through Quechua notions.
In other words, there is still a great number of speakers who keep learning
Spanish with plenty of Quechua characteristics. The author also states that AS
cannot be defined as one uniform and homogeneous variety, but a linguistic
continuum explained by different social factors. Finally, Escobar (1994) makes a
distinction between AS and Bilingual Spanish as two “varieties resulting from
contact between Spanish and Quechua, although the former is a native variety
and the latter is not”. The lack of this distinction “has led to confusion between
the linguistic characteristics which distinguish Andean Spanish and Bilingual
Spanish, and it often seems to be assumed that the two varieties are identical.”
The difficulty in defining AS resides, in a way, in the confusion between the
AS of bilinguals and the AS of monolinguals.7 Escobar (1994) explains that
“similarities between AS and Bilingual Spanish are due to the fact that both
varieties of Spanish are products of a language contact situation, in this case of
Quechua and Spanish”. However, she is convinced that it is possible to
determine which grammatical constructions are characteristic of each variety.
Furthermore, she argues that “constructions found in Andean Spanish are also
found in Bilingual Spanish, but the inverse does not stand”. For example, de mi
mamá su casa ‘of my mother her house’, that is analytical genitive constructions
with the possessor + possessee order, is a structure that belongs to Bilingual
Spanish but not to Andean Spanish. The author claims that the difference
between the Spanish of bilinguals and monolinguals is that they are different in
the nature of their contact. “While Andean Spanish is the product of contact at
the social level, Bilingual Spanish is the product of contact at the individual
level”. According to Rivarola (1990), AS became a monolingual variety as early
as the end the 18th century, but not all researchers agree. For example, Escobar
claims that AS emerged in the 20th century as L1, and its emergence cannot be
dissociated from the particular scenario which allowed its existence, which is
migration.8
Following Winford’s (2003, p. 24) table on the possible outcomes of contact
situations and the types of cross-linguistic influence they involve, I would say
that AS is a case of language shift with moderate to heavy substratum
interference; a result of group second language acquisition that started more than
five hundred years ago and that is still a very vital process. Nevertheless, talking
about the monolingual AS, we cannot talk about substratum influence in the
same way because there are many areas of the Andes where Quechua has not
been heard for centuries. It is the trace of its presence that remains in the shape
of linguistic traits such as SOV word order; discursive markers like pues and
nomás; and the evidential use of the present perfect and pluperfect. As a matter
of fact, even the Spanish spoken in Lima (Coastal, Ribereño or Limeño
Spanish), considered the most prestigious variety in this area, shows the
influence of AS features (Cerrón Palomino, 2003) that have transcended the
filters of stigmatization. For instance, the postposition of the verb in sentences
like María linda es ‘María beautiful is’ shows an SOV order that, although not
canonical, is grammatical in standard Spanish for certain pragmatic functions
such as topicalization. I will dare to say, though, that if we collect data to
conduct a quantitative analysis, we will find out that the frequency of the OV
order would be much higher in LS than in dialects that historically have not been
in contact with Quechua.
In fact, nowadays, there are many areas, Andean and non-Andean (due to
waves of migration), where monolingual varieties of AS have been documented
(Rivarola, 1986). In other words, AS can include not only the L2 variety of
Spanish spoken by Quechua-Spanish bilinguals, but also a contact variety
spoken by monolinguals. Therefore, we cannot speak of AS as a homogeneous
and uniform dialect (Zavala, 1999); currently, it is a more of a linguistic
continuum that includes Quechua monolinguals with some understanding of
Spanish (a minority, in most places), Quechua and Spanish bilinguals (with a
wide range of differences in their proficiency of each language), and Spanish
monolinguals. And all the variation derived from this situation can be found not
only among different Andean countries but even among different areas of the
same country; this is the case of Peru, where the Quechua family presents the
widest set of dialects. In Babel’s (2010) words referring to the Andean country
of Bolivia, “it is prestigious to be a Spanish speaker. However, not all Spanish
speakers speak alike, nor do they speak the same way in all situations”.
Zavala (1999) states that if researchers want to understand AS in all its
complexity, we should not define it as an individual psycholinguistic
phenomenon, reduced to the characteristics of bilingualism, but as a widely
extended variety that emerged because of the substratal effects of Quechua and
Aymara in (Peruvian) Spanish. Another important fact that she claims is that
according to the 1994 census, the tendency of the population that speaks
vernacular varieties is to shift from being Quechua monolinguals to being
Spanish monolinguals, without experiencing a mediating bilingualism. This
assumption is by all means questionable, but can be interpreted as the reluctance
to speak the L1 until some proficiency in the L2 has been achieved, i.e. the
stages of bilingualism are probably more abrupt than in other processes of
second language learning. In any case, as a result, new and particular varieties of
AS are emerging among groups.
When compared to LS, AS presents differences at every grammatical level.
Nevertheless, in this study, I will only focus in the morphosyntactic features
which make this contact variety different from the standard used in Lima, where
my research is conducted. The reason for this decision is that, as stated before,
the focus of my study is not a description or analysis of AS per se but a
perception study on how certain features of AS morphosyntax might index a
provinciano identity. Thus I will only mention and describe those features that
were part of my fieldwork research as representative of stigmatized AS or
provinciano speech. The excerpts containing the morphosyntactic features that I
have utilized in my research were obtained from different sources: prior work
with AS speakers, the fieldwork I conducted with bilinguals from an Andean
region in Peru, the features I have collected in the speech of comedians imitating
Andean ‘accent’, and my own intuitions as a LS speaker. Furthermore, I will
base my explanation of these features on the work of Zavala (1999), Escobar
(2000), Rivarola (2000), Granda (2001), and Cerrón-Palomino (2003).
It is important to note that not all morphosyntactic features of AS are
perceived in the same way or have the same social meaning; thus I have only
selected a sub-set of features that are usually associated with Andean migrants.
For instance, leísmo, the use of indirect object pronouns (le, les) where direct
object pronouns are the norm in LS, is claimed to be an AS feature that does not
necessarily index provinciano speech (Caravedo, 1996; Klee & Caravedo, 2005).
This claim is based on the fact that, because of a lack of stigma, it is being
spread in Lima, specifically by the descendants of migrants.
Cerrón Palomino (1990) proposes a tentative hierarchy of AS features
according to their degree of stigmatization. According to him, features such as
monophthongization of diphthongs, vowel variability, omission of articles, and
lack of grammatical concord are the most salient. However, there are other
features that might not be perceived by LS speakers as ‘incorrect’ or provinciano
and that are being integrated into the standard and are reshaping it unnoticeably,
what Cerrón Palomino calls “the revenge of Quechua against diglossic
discrimination” (1990, p. 169).
After decades of coexistence in Peru’s capital, contact between AS and LS
has had sociolinguistic repercussions such as the construction of stereotypes, and
the consequent emergence of stigmatization and discrimination among Lima’s
inhabitants because of the way they speak (among other factors). Thus this city
represents a unique environment for the study of AS. Lima, after the rural
exodus from the Andes, has become an urban field in which convergences and
confrontations are multiethnic and multicultural (Godenzzi, 2008).
LS is not a monolithic variety; its variation is conditioned by different
sociolinguistic factors. In the same way, there is not one single AS, especially if
we take into consideration that Andean migrants come from many different
towns and cities. However, in terms of language attitudes and linguistic
ideologies, perceptions tend to be black or white: any marked feature (i.e. not
LS, the standard in Lima, which is considered the unmarked dialect), especially
when combined with certain social cues such as ethnicity, origin or
socioeconomic status, can index provinciano and, therefore AS.
There is a fair amount of literature on (Peruvian) AS that ranges over
grammatical descriptions, variationist studies of particular traits, and even
peculiar pragmatic uses found in this variety when compared to the standard(s).
Also, another matter of interest in Andean Linguistics is language attitudes, and
even linguistic ideologies or linguistic policies, but mostly in relation to Spanish
and Andean languages (Quechua and Aymara). However, there are still many
missing gaps in the literature such as the study of the complex sociolinguistic
situation of Lima, which tends to be simplified in the discourse into a binary
opposition between Andean migrants and Limeños that does not represent all the
different nuances actually involved in the reality of Peru’s capital. This
complexity has been documented only in a few articles and almost always
related to critical discourse analysis (Godenzzi, 2008; Smith, 2008; Mick, 2011).

4. The study: Design and methods

The data presented here was collected during the summer of 2012, as a
complement of a previous field work conducted the previous year. The data from
2011 was originally collected with two main goals in mind. The first one was to
find out the role of morphosyntax in the definition of Andean Spanish/identity: I
wanted to test if the morphosyntactic component of grammar was as salient as
phonology in terms of recognizing a person’s speech as provinciano. The
question arises from the fact that researchers interested in language attitudes
towards AS tend to focus their attention on phonological features as the most
salient ones. Secondly, I wanted to find out if when compared to the highly
salient phonological features of AS, morphosyntactic features were less salient
in terms of stigmatization.
The first hypothesis, however, was contradicted because, even though the
participants recognized the selected morphosyntactic features as provinciano
(Andean) when isolated, that didn’t happen most of the times when a native
speaker of LS would use AS features in her speech, with a LS phonology. In
other words, it seems that when there are no other hints but language to
determine somebody’s (Andean) identity, morphology and syntax might not be
‘enough’ to consider somebody as provinciano or indigenous.
After conducting a first study, I realized that the participants’ reactions to
differences in phonology among the stimuli presented to them might have played
a role in their evaluations of the scripts. Even though I recorded speakers of both
dialects using both standard and Andean features on purpose in order to have a
variety of inputs, in the end it was unclear if participants were also guided by
phonological cues when expressing their attitudes towards the recorded speakers.
For instance, participants who considered the LS speaker using Andean
morphosyntax as “capitalina” would agree that the morphosyntax she used,
when isolated, was mostly “provinciana”.
In the current study, I did not use speakers of different Spanish dialects but
only LS speakers because I wanted to report how my informants perceived AS
morphosyntactic variants when they were not combined with that variety’s
characteristic phonological, suprasegmental or discursive elements. I utilized
similar semantic scales, with a much smaller and simpler task in order to obtain
a more precise control so as to exclude factors that are not related to the
morphosyntactic variables I wished to investigate.
I already stated that, even though language attitudes towards AS met my
expectations, it is undeniable that phonology plays a more important role than
morphosyntax when determining Provinciano identity. Isolated morphosyntactic
features were considered provincianos but when participants listened to the
speech samples, the LS speaker using AS morphology was not considered
provinciana. Only when they saw the features in paper, without a specific
phonological context,9 informants recognized them as Provincianas, incorrect,
etc. On the other hand, when the AS speaker read the LS script, the informants
reacted in a different way: many of them considered her “less provinciana” (not
the highest number in the scale), but only a few considered her Limeña.
Thus, the main questions that I wanted to answer were (a) What happens
when an AS morphosyntactic feature is inserted in a non-AS grammatical
context? (b) What do these features inserted within the speech of a Limeño
speaker mean? (c) Do they still refer to provinciano/ serrano/ incorrect/ etc.?10 (d)
Are the AS morphosyntactic features enough to mark speakers as Andean or
Provincianos?

4.1 Matched-Guise task

I employed the Matched-Guise Technique (MGT), a methodological tool, which

has proven to be succesful in identifying and eliciting stereotypes and attitudes
toward particular sociolinguistic groups such as Appalachian people (Luhman,
1990): “The matched-guise technique offers scholars of language, particularly
those who study language ideologies, a unique lens into those unconscious ideas
that may not be apparent in interviews or participant observation” (Booth, 2009).
According to Campbell-Kibler (2006), “listener perceptions are perhaps most
thoroughly studied within the covert study of language attitudes, using the
Matched Guise Technique developed by Lambert and his colleagues” (Lambert,
Hodgson, Gardner & Fillenbaum, 1960; Lambert, 1967). These researchers
believed that, since linguistic prejudice is many times more socially sanctioned
than other forms, listeners would express their opinions more openly if the
responses were triggered by linguistic performances. It has proven to be
succesful in identifying and eliciting stereotypes and attitudes toward particular
sociolinguistic groups such as Appalachian people (Luhman, 1990) or, in this
case, Andean migrants in Lima. Also, as Kroskrity (2004, p. 505) states,
“members may display varying degrees of awareness of local language
ideologies” and thus my research had to be able to gather evidence regarding
these more or less unconscious ideas about language.
The design of this MGT was based on the findings obtained in the first
study, thus it also has a somewhat different structure when compared to the most
paradigmatic MGT tasks. It was a much simpler version, because I needed a
much more controlled task to confirm or contest the findings of the previous
stage of data collection. It consists of pairs of sentences, each read by a different
LS speaker. This allowed me to test informants’ language attitudes toward AS
morphosyntax in a more controlled way, by keeping the phonology constant, and
changing only the morphosyntactic features. Most of the AS sentences I utilized
in the design of this task were extracted from natural speech that I had recorded
in prior research in order to make it as real as possible for participants.
Each session with the participants was divided into three sections: first,
after listening to each speech sample of the MGT, they had to fill in a table with
the group of semantic differential scales. Second, I discussed with participants
some of the sentenced played for them, and asked them if they considered them
correct or incorrect in order for them to express conscious grammatical
judgments. Third, they had to answer a demographic survey with information
such as age, gender, origin, level of education and occupation.
I recruited 6 Limeño women between the ages of 23 and 55 to record a pair
of sentences each. In this way, the same voice had to read both the Limeño and
the Andean version of the sentence that was modified in one single
morphosyntactic feature in the two guises. This strategy contributed to avoid the
interference of other grammatical or suprasegmental features when participants
were making their judgments about the speaker that could have been misleading.
I did not use speakers of different Spanish dialects but only LS speakers because
I wanted to report how my informants perceived AS morphosyntactic variants
when they were not combined with that variety’s characteristic phonological,
suprasegmental or discursive elements.
4.2 Stimuli

Six of the sentences contained morphosyntactic features attributed to AS (one

feature per sentence) that I had extracted from natural speech recordings I
collected for different research projects. Even though all of these features usually
work in AS in particular contexts – especially in terms of emphasis, topic,
evidentiality – their markedness make them indexes of Provinciano identity even
when taken in isolation, for both Limeño speakers and speakers of this dialect.
The other six worked as a counterpart to the first ones: they were almost
identical but the AS morphosyntactic features were substituted by more standard
(i.e., non-Andean) morphosyntax. Finally, I added four filler sentences that
worked as distracters.
1. Los pobres eso comen (GB) ‘Poor people eat that’ SOV
2. Poco plata debe tener (ASU) ‘She probably doesn’t have a lot of money’
NO GENDER AGREEMENT11
3. La forma como lo trata a su hijo (ASU, modified) ‘The way he treats his
son’ DO DOUBLING
4. Las peras lo ponía en su mandil (GB) ‘He put the pears on his apron’
NEUTRAL LO (agreement)
5. Yo me he nacido aquí en Lima ‘I (me) was born here in Lima’
EMOTIONAL COMMITMENT REFLEXIVE (Zavala, 1999)/
REFLEXIVE PRONOUN REDUNDANCY IN NON-REFLEXIVE
VERBS (Escobar, 2000)
6. Me caí del tercer piso de su casa de mi tía (Godennzi, 2011) ‘I fell down
from the third floor of my aunt’s house’ POSSESIVE MARKER
DOUBLING

1. Los pobres comen eso

 2. Poca plata debe tener
3. La forma como trata a su hijo
4. Las peras las ponía en su mandil
5. Yo he nacido aquí en Lima12
6. Me caí del tercer piso de la casa de mi tía

1. La niña no dejaba de mirarlos ‘The girl wouldn’t stop looking at them’

2. La niña no dejaba de mirarla ‘The girl wouldn’t stop looking at her’
3. No me di cuenta ‘I didn’t realize it’
4. No me he dado cuenta ‘I haven’t realized it’

4.3 Participants

My strategy for selecting the participants was to have a smaller (compared to the
49 participants from the previous experiment), but representative, sample of
people. When recruiting participants, I decided that only traditional Limeños, the
ones who are born in Lima with no direct connection to internal migration from
the Andes, would be considered Limeños. However, people who were born in
Lima but were the children or grandchildren of Andean migrants were treated as
belonging to the group of Provincianos, with whom they share many
characteristics. Because there are different varieties of AS, I opted to include
migrants from the same area, who share most of the morphosyntactic features
being evaluated. Therefore, I disregarded speakers of Northern Andes varieties
and focused instead on those from the Central and Southern Andes.
The selection of participants followed general criteria:
1. age 18 or older
2. either born in Lima or living in the city for at least two years (to consider
them actual immigrants, and not just visitors);
 3. other general characteristics: both sexes, different levels of education and
various occupations.

The participants in the study included 11 women and 12 men, all adults (18 and
older), and their ages ranged from 21 to 70.13 16 people were born in Lima
(traditional Limeños), 7 born in different provinces (mostly from Central
Peruvian Andes). All of them spoke Spanish, and 6 of them spoke Quechua with
different degrees of proficiency.

4.4 Status vs. solidarity

Just as important as the stimuli used in the tasks are the methods used to report
and analyze participants’ answers. Following prior studies on language attitudes
(Woolard, 1984; Ryan, Giles, & Hewstone, 1988; Woolard & Gahng, 1990;
Baker, 1992; Edwards, 1999; Yiakoumetti et al., 2005; Campbell-Kibler, 2006;
Booth, 2009), I used the dimensions of status and solidarity, which are said to
determine the condition of different linguistic varieties (Giles, Hewstone, Ryan,
& Johnson, 1988).
Thus, based on the vertical axis of status and the horizontal axis of
solidarity, I developed a set of evaluative scales, specifically semantic
differential scales, which were developed by psychologist Charles Osgood and
his associates to measure the meaning of concepts (Osgood, May, & Miron,
1957). The respondent is asked to rate an object or a concept, in this case certain
morphosyntactic features, along a series of bipolar scales defined with
contrasting adjectives at each end.
The variables related to status were:
Provinciana-capitalina
There is a geographical opposition between the capital and the
province, especially since Peruvians usually use the word ‘provinciano’ to
refer to Andean provincianos, while Lima is located, in some sense, in
opposition to it because it is right next to the ocean, at sea level.
Furthermore, this geographical opposition entails other oppositions that are
rarely questioned in Peru: the contrast between urban (Lima) and rural (the
Andes), and the contrast between modernity (Lima) and backwardness (the
Andes).
Pobre-rica
Poverty is another characteristic that, although not exclusive of
Andean migrants, is related to the Andes because of the stereotypical image
of Andean people being poor peasants living in rural areas. That stereotype
is reproduced in the city, since provincianos generally live in peripheral
areas associated to people with a low income. These days, however, New
Limeños, the descendants of Provincianos, are considered an economically
emergent group which is contributing enormously to Peru’s improvement.
Therefore, we can witness how some of the participants did not necessarily
associated being migrant with being poor, especially the ones that belonged
to the group of New Limeños.
Indígena-blanca
Although an arbitrary opposition at first sight, the contrast between
white (Lima) and indigenous (Andes) is the most salient one, even though
Peru’s capital also has an important group of African descendants, Chinese,
Japanese, and so on. Moreover, it was the one I was interested in because of
the topic of my study. It was interesting to notice that participants almost
automatically associate the indigenous characteristic with an Andean origin,
even though we can also find indigenous population in the Amazon area.14
Bonita-fea
I was particularly interested in reporting the attitudes that participants
had towards AS regarding the beauty of their speakers because, Lima’s
 inhabitants tend to relate beauty to whiteness (and ugliness to
indigenousness). Furthermore, this situation is not unusual throughout Latin
America after Spanish colonization: there, the association between beauty
and whiteness is unconscious and barely contested; on the contrary,
indigenous race is perceived as ugly (Silvestre, 1994; Casaús, 1998).

The semantic scales related to solidarity were:

Simpática-antipática
Buena-mala

Solidarity categories were selected mostly following the previous literature

(Giles, Hewstone, Ryan, & Johnson, 1988) because likeability or kindness are
general characteristics that work in many groups such as in the case of the
inhabitants of Lima.
The following is a table that summarizes the information explained above:

Table 1. Summary of features included in the Matched-Guise test

Feature Values Approximate English gloss Dimension

Positive Negative

Niceness Simpático Antipático ‘nice’-‘not nice’ Solidarity

Provincialism Capitalino Provinciano ‘from the capital’-‘from the Status
province’
Poverty/Wealth Rico Pobre ‘rich’-‘poor’ Status
Whiteness Blanco Indígena ‘white’-‘indigenous’ Status
Beauty Guapo/bonita Feo ‘good-looking’-‘ugly’ Status/(Solidarity?)
Kindness Bueno Malo ‘kind’-‘mean’ Solidarity
Literacy Analfabeto Sabe ‘illiterate’ – ‘literate’ No differential semantic
leer/escribir scale
Participants had to listen to 16 sentences read by at least six LS speakers with
similar backgrounds (gender, age, origin, education, etc.), as I explained before,
and had to evaluate them using a table with semantic scales. The trickiest part of
this kind of research is to select the right terms for the evaluative scales.
Sometimes, one has to adjust the terms depending on how useful and clear
informants’ judgments are. It is always difficult to tell whether terms like “rich”,
“indigenous” or “trustworthy” mean the same thing to all informants. That is
why I opted to fill in the scales myself instead of giving the participants a sheet
so they can write down their evaluations; in that way, I could monitor what they
were understanding by terms such as “educated” (in Spanish, it could refer not
only to the level of education of a person but also to a well-mannered person).
Also, to counterbalance the order of positive and negative adjectives, I began
some scales with the positive term (nice-not nice) and others with the negative
term (poor-rich). In this way, I prevented the respondent from falling into a fixed
pattern of always checking to the right or left.
According to the population I was interviewing, I selected a group of
general traits associated with status and solidarity, appropriate to Lima society.
For example, for the status dimension, I used the adjective “beautiful” and its
counterpart “ugly”, which is not very common in this kind of study. The reason
these seemed appropriate was that the stereotype of the person from the Andes is
that she/he is of indigenous race (i.e. non-white), and indigenous people are
almost always ugly from a westernized perspective of beauty, I thought it was
important to include that pair since white people in Lima tend to be considered
more beautiful.15 Between each pair of adjectives, there was a 1 to 4 scale, and
the participant needed to choose, for example, if the speaker was “very
beautiful” (1), “a little beautiful” (2), “a little ugly” (3) or very “ugly” (4). The
choice of an even number was made on purpose because relevant literature has
shown that when the number is uneven the informants have the tendency to
choose the number in the middle of the scale. In fact, just as an anecdote, many
of the participants in my research asked me if there was not a 2.5.
Once I finished the data collection stage, I organized the results of the
Matched-Guise tests in an Excel spreadsheet in order to find out whether my
participants’ judgments and the different variables differ significantly when
compared in groups. Then, I applied statistical tests to see if the differences in
the ratings of the various guises, as well as the differences in the evaluations of
Limeños and Provincianos were significant. For measuring the ratings of the
variables obtained from these tasks, I have used two different statistical
techniques: the t-Test, when only two samples were compared and the ANOVA
(analysis of variance), when the comparison was between more than two
samples. The p-value I have taken into consideration to determine the
significance of an analysis is p < 0.05.

5. Results and analysis

The central goal of this study is to explore how Provinciano identity is shaped
through language when in contact with a more prestigious variety (LS) and its
speakers. Therefore, my fieldwork research was designed to investigate the
language attitudes and linguistic ideologies underlying the ways in which
Limeños and Andean migrants perceive and evaluate features of AS
morphosyntax. My aim was to determine whether it is valid to talk about
Provinciano identity, understanding the term as a homogeneous perception of the
other as, ‘migrant’, non-white, etc. when this other uses certain morphosyntactic
features that belong to the AS inventory in the particular context of Lima.
I first present the results of the MGT task, which reveal general attitudes
toward the various sentences among the participants in the study. Then, I
compare the ratings of those who were born in Lima with those of Andean

16
origin.16 When playing the recordings to each participant, a fact that struck me as
strange was that many of the participants, no matter their origin, could not
perceive the difference in the contrasting elements in the different pairs of
sentences. At the beginning, I wondered if it had to do with technological
problems so I had to replay the phrase, but afterwards I realized that a more
likely explanation was that people were unconsciously hearing what they wanted
to hear, neutralizing the slight differences between the pairs of sentences. As
Campbell-Kibler (2006) states: “listeners showed agency selecting what cues to
attend to and how to interpret them”.

5.1 Andean Spanish versus Limeño Spanish morphosyntax

Despite its superficial image as a ‘multicultural’ and ‘mestizo’ city, Lima still
puts strong social pressure so that newcomers integrate and assimilate to the
hegemonic Limeño culture. Within the context of prosperity that is being
experienced in the urban peripheries, considered before as the place of the poor,
the way people speak Spanish has become an important element to identify and
differentiate Limeños (Arellano & Burgos, 2004). In other words, Lima’s
inhabitants have found and reinterpreted traditional linguistic and spatial
strategies to express difference among Classic Limeños, migrants and New
Limeños.
The results shown in the following figures are only the ones from the
comparisons that turned out to be significant: provincialism, poverty, and
whiteness. In Figures 1 to 3, the results show the answers of both migrants and
Limeño participants. For instance, in Figure 2, the horizontal axis indicates how
pobre (1) or rica (4) the speaker enunciating the sentence is considered to be; the
vertical axis indicates the two different groups of sentences that participants
(divided into Limeños and migrants) had to listen to. What turns out to be
significant in this figure (P value) is the different perceptions that participants
had for each group of features.
Figure 1 shows the degree of provincialism perceived in the two groups of
sentences: the one with AS morphosyntactic features and the one which works as
counterpart. I collapsed all my informants’ responses and separated them in two
different groups according to the morphosyntax of one particular trait. This
figure demonstrates how the LS sentences were perceived to be said by a
capitalino more than the AS sentences. The results of the previous study had
lead me to think that phonology was more salient than morphology when
indexing Provinciano identity. However, even though in this study the (LS)
phonology remained the same and only the morphosyntax changed, participants
were able to determine which ones sounded more or less capitalino. This
illustrates the fact that morphosyntax also index Provinciano identity, just not as
definitively as phonology.

Figure 1. Perceptions on provincialism, AS and LS morphosyntax (t = −3.3,

df = 262, p = 0.0011)
Although being Provinciano literally means ‘to be born in a province, not in the
capital’, when talking about Provincianos in Lima that people generally refer to
migrants from rural areas in the Andes, at least in my experience both as Limeña
and as a researcher. The difference in perception of AS and LS sentences
becomes a pattern between the characteristics related to status like this, while
none of the ones related to solidarity turned out significant results. It seems to be
the case that it is more important to establish the difference between groups in
terms of power than in terms of solidarity, which should not be surprising after
all since human societies are generally organized based on who are the most and
the least powerful, whatever the source of that power is.
Figure 2 shows the results of the evaluations on the two groups of sentences
in terms of poverty: the one with AS morphosyntactic features and the one
which works as counterpart. Participants perceived AS morphosyntax as (0.32
out of 4) less typically produced by rich people. The differences were significant
at the p = 0.005) level. In terms of wealth, although we observe the same
tendency shown in Figure 1, the difference between LS and AS sentences is
smaller. Perhaps this has to do with the fact that these days the economic status
is being challenged by new Limeños, the so-called ‘emergent classes’. If a
participant qualified one of the AS guises as coming from a poor person, I would
ask them if they thought that it was common to imagine a provinciano as poor.
Their answers, especially the ones from the Andean migrants and new Limeños
varied, but many of them stated that they were talking about this specific case,
not that they thought Provincianos would necessarily be poor. Their responses
revealed that Limeño society appears to be undergoing important changes.

Figure 2. Perceptions on wealth, AS and LS morphosyntax (t = −3.52, df = 262,

p = 0.0005)
Figure 3 indicates the evaluations on the two groups of sentences in terms of
whiteness. Participants considered AS sentences (2.48 out of 4) significantly less
‘white’ than the LS sentences (2.93 out of 4). In Latin America, the connection
between whiteness and beauty seems almost natural and, therefore, barely
contested. I believe that this is the reason why I have not been able to find much
literature about it, a huge contrast if we review American literature on
white/black comparisons. It seems that, if you are beautiful, it is because you are
white. Unfortunately, this stereotype does not only come from white people but
also from non-white people, a situation that might seem contradictory. Silvestre
(1994), when discussing the analogous situation in the Dominican Republic,
considers prejudice against blacks a ‘contradictory behavior’, since it is a
country with a high percentage of blacks.
During my fieldwork, I asked the participant Borja, an Andean migrant in
his sixties, why he thought that the person in the guise was beautiful (LS
sentence), to which he answered: ‘Well, she must be beautiful because she
“sounds” white’. I then asked him if non-white people could not be considered
beautiful. He gave me a surprised look for a second, as if I were asking him a
tricky question, followed by a smile: ‘Oh, of course, they can be beautiful too’.

Figure 3. Perceptions on whiteness, AS and LS morphosyntax (t = −4.41, df = 262,

p < 0.0001)
5.2 Limeños, Andean migrants and new Limeños: Status versus
solidarity
In the following figures, the answers of migrants and Limeños for the semantic
scales have been divided to see the differences in their evaluations. For instance,
in Figure 6, the horizontal axis indicates how indígena (1) or blanca (4) the
speaker enunciating the sentence is considered to be; the vertical axis indicates
the two different groups of sentences that participants (divided into Limeños and
migrants) had to listen to. Limeños’ evaluations are shown in the lighter tone of
gray and provincianos’ evaluations appear in dark gray. What turns out to be
significant in this figure (P value) is the difference between Limeño and migrant
responses for each group of features.
One of the most interesting findings in these results is the fact that all
significant results come from the status dimension, not from the solidarity
dimension. In other words, Limeños and provincianos have very similar
perceptions about both dialects in terms of solidarity. But what is it about the
traits related to power that generates very different opinions? I believe that it has
to do with the necessity of Limeños to distance themselves from “the other”,
those with less status. Migrants and new Limeños, as well, tend to have a similar
behavior to that of Limeños: they constantly try to assimilate to prestigious
Limeño society in order to be accepted, they also want to distinguish themselves
from those with less status and assimilate to the more prestigious classic
Limeños.
While conducting fieldwork, Provinciano informants constantly told me
how, when they just arrived in the city, they had to learn many things. Migrants
who had been in Lima before them were the ones who explained to them how
things worked, what they could say, how they should speak, what costumes they
should forget and which new ones they should embrace, what they should wear,
and so on because they had already experienced that first cultural shock.
The perceptions of Limeños and migrants are, with some exceptions,
relatively the same. However, Limeños are more extreme in the differences they
perceive between AS and LS sentences. Is it a “gatekeeping” strategy, where
differences in solidarity do not play an important role but where status
differences need to be highlighted in order to establish a “we” vs. “they”? Even
Berta, a provinciana from Huánuco, central Andes, uses the third person plural
when talking about people just like her: “Provincianas get confused with letters,
even when they are literate they get confused”. In other words, many
provincianos kind of distance themselves from the social category of being
provinciano in Lima. Even so, a valid question to ask, is if that distancing
responds to the fact that they do not consider themselves Provincianos. They are
perceived as so, but as I have repeatedly stated, identity is not fixed but
performed in each interaction. They meet most of the stereotypical
characteristics selected for the attitudinal scales, it is true, but that does not
necessarily make them Provincianos in their own perception.
Undoubtedly, the discussion about identity is always an extremely complex
one; so complex, in fact, that the discriminated individuals internalize prejudices
about provinciano inferiority to the extent of projecting them on others of similar
background. When listening to them, I was under the impression that there was
an internal struggle between who they were and who they wanted to be, a
struggle shown in what I have labeled as a somewhat ‘schizophrenic’ discourse.
As an example of these frequent ‘contradictions’, the following is a fragment of
my conversation with Alberto, a gardener from Central Andes: “The one you
interviewed last, she shows a little lack of study…In any case, a provinciano/
(we) make(s) mistakes when vocalizing a word”.17
Figure 4 illustrates the degree of provincialism perceived according to the
morphosyntactic feature under study. Limeños considered speakers uttering LS
sentences more capitalinos than Provinciano participants; in the same vein,
speakers uttering AS sentences were considered more Provincianos by Limeños
than by Provincianos. Again, this supports the idea that Limeños are more
drastic in their perceptions and that they detect AS more easily because they
want to distance themselves from “the other”. Although still understudied,
researchers such as Klee and Caravedo (2005) support the idea that many
grammatical features, specifically the least stigmatized could have been
incorporated in LS; thus it is an illusion to think that there is ONE incorruptible
LS. It is more appropriate to think that the coexistence of Limeño, Andean and
other Spanish varieties in Lima has resulted in a continuum that cannot be easily
segmented in order to say who speaks what.

Figure 4. Perceptions on provincialism (provinciana (1) – capitalina (4))

according to Limeños and migrants, AS and LS morphosyntax
[F(3, 260) = 8.132, p < 0.0001]
Why is it that in Figure 4 Limeños rate LS sentences as more capitalinas than
the migrants do, while the opposite happens with AS sentences? A plausible
explanation for this could be that Limeños want to emphasize the distance
between LS and AS, even if only in terms of morphology. Therefore they might
be more susceptible to linguistic features that differentiate them from groups
with less status. We can perceive this attitude even among provincianos’ children
like Azucena: “Lima has been flooded with people from the provinces (using a
mix of a fatalistic and resigned tone). For example, my house is full of them
now. All my cousins have come from the Sierra (mountains/Andes)”.
Language plays a central role in this “gatekeeping” process because it
contributes to create and shape the image of “the other”, the danger that is
threatening the established division of power. Due to different reasons but also to
internal migration, Lima has changed dramatically in the last fifty years. Just as
an example, let’s think of Andean migrants who arrived in Lima in the seventies,
looking for better opportunities in the city. They settled down in the peripheral
areas of the city, the so called pueblos jóvenes or asentamientos humanos
(shanty towns): sandy and arid hills where access to the most basic needs was
unimaginable. And yet, forty years later, one can find the sons and daughters of
some of those migrants (the new Limeños) having enough money to pay for the
same higher education that many white upper-middle class Limeños take for
granted. Therefore, this “subversion” needs to be constrained so that social
mobility is blocked. In Githinji’s (2008) words, “language as one of the key
markers of social categorization becomes a key target of subjective attitudes and
stereotype towards the unlike others, or the out-group”.
Figure 5 is another example of Limeños expressing more drastic differences
than provincianos between both groups in terms of wealth. It illustrates the
degree that poverty is perceived according to the morphosyntactic feature under
study. The stereotype of provincianos being usually poor (peasants) stills
persists; however, the results shown in the following figure, where the distance
between Limeños and provincianos is smaller, is supported by some of the
testimonies from my recordings. For instance, Adán, a new Limeño cook, son of
Andean migrants, does not agree with this prejudice. According to him, there are
many provincianos that are businessmen, which implicitly means that they have
a good economic level.

Figure 5. Perceptions on poverty (pobre (1) – rica (4)) according to Limeños and
migrants, AS and LS morphosyntax [F(3, 260) = 67.42, p = 0.0013]
Figure 6 illustrates the degree of whiteness associated with the morphosyntactic
feature under study. While participants like Beto showed certain reluctance in
using terms such as Provinciano or indigenous, other Limeños like Barbara had
no problem in expressing highly pejorative terms like serranaza.18 Maybe this
relates to the fact that Beto is much whiter than Barbara, and that is why she
feels the urge to distance herself as much as possible from a person who speaks
in a provinciano fashion. In any case, Beto’s behavior is not very common
among white, rich, educated Limeños, while Barbara’s behavior, unfortunately,
tends to be the norm. Expressions such as Acá todos somos cholos ‘We are all
cholos here’ show how racism is usually erased from the discourse, but not from
reality. Many Peruvians base the idea that we are a homogenous group (of
cholos) in the fact that there has been a deep process of mestizaje since colonial
times. Nevertheless, negative language attitudes towards non-Limeño varieties
are only one way to prove that the country is far from eradicating prejudices
against indigenous people.

Figure 6. Perceptions on whiteness (indígena (1) – blanca (4)) according to

Limeños and migrants, AS and LS morphosyntax [F(3, 260) = 8.532, p < 0.001]
Figure 7 illustrates the degree of beauty perceived according to the
morphosyntactic feature under study. From the beginning, I found the pair
bonita-fea somewhat troubling. I assigned the concept of beauty to the status
dimension because I argue that, in Peru, it is closely associated with the concept
of whiteness. However, it did not turn out to be significant when subjects were
not divided according to their origin, even though all the other status scales were
significant.

Figure 7. Perceptions on beauty (bonita (1) – fea (4)) according to Limeños and
migrants, AS and LS morphosyntax [F(3, 260) = 4.051, p = 0.0077]
Table 2. Literacy percentages according to participants*

% A1 B1 A2 B2 A3 B3 A4 B4 A5 B5 A6 B6

Literate 100 100 69.2 95.4 100 100 81.8 100 40.9 81.8 72.7 77.2
Not sure – – – – – – – – 18.1 – 4.54 –
Illiterate – – 31.8 4.54 – – 18.1 – 40.9 18.1 22.7 22.7

* Table 2 does not show the significance between different averages as all the figures show, but the
percentages of affirmative or negative answers of the participants. ①

Informants had to determine the literacy of the person who uttered each recorded
sentence. Looking at Table 2, it is very interesting to see how certain features
indicate “illiterate” more than others. In this sense, AB1: los pobres eso
comen/los pobres comen eso and AB3: la forma como (lo) trata a su hijo were
perceived as a phrase only a literate, i.e. educated, person would say. However, a
high percentage of participants considered that the people who uttered A5 yo me
he nacido aquí en Lima and A2: poco plata debe tener were illiterate. I believe
that there are two possible explanations for these percentages: first, that there is a
different degree of markedness in terms of Provinciano identity or, as I claimed
before, that some features are more provinciano than others and, therefore, the
speakers sound less educated. The other explanation is that these two particular
features do not belong to AS as a monolingual dialect but as a second language
variety, whose features are only used in the early stages of the L2 learning. For
example, lack of gender concord (as in poco plata) is not exclusive of AS but a
common feature found in L2 varieties (Escobar, 1994).
An unexpected reaction I found among the informants, specifically among
Limeños, was the “confusion” they reported when they heard the voice of a
typical Limeña using a morpshosyntactic feature that didn’t match their
expectations. Therefore, they found it hard to complete the semantic scales
sometimes as in the case of Bernardo, who told me he was puzzled because his
intuitions were under question: his perception led him to call Provinciano
somebody who sounded Limeño but had “something” that made him feel uneasy
about categorizing her as a capitalina. Benito, a Limeño lawyer, also found it
difficult to fill in the scales because he heard a Limeño accent with a
grammatical structure that did not fit. This alleged “confusion” contributes to
answer the question about the degree of markedness of Andean morphosyntax
and if it is marked enough to attribute Provinciano identity to the speaker or does
it only sound “weird”. It is possible that traditional Limeños’ confusion arises
because they recognize that the AS morphosyntax is not their own, while AS
speakers did not notice the contrast in such a drastic way and thus they are able
to associate social meaning to it in an easier way.

6. Conclusions and final remarks

This chapter has presented a quantitative and qualitative approach to the
attitudes and ideologies that the residents of Lima have towards AS and their
speakers. It identifies an understudied area, i.e., that of language attitudes
towards AS of speakers from Lima and speakers from the Andes. There have
been relatively few perception studies of AS, and even fewer studies focusing on
morphosyntactic phenomena. I tracked certain stigmatized grammatical traits
and how they end up consolidated in the figure of the migrant or the Provinciano
identity. Furthermore, the analysis of the data collected has several theoretical
implications which constitute an important contribution to the field of
linguistics.
 My claim is that Provinciano or ‘migrant’ identity is usually portrayed in
mainstream discourses in an impoverished and simplistic manner, even though it
is by no means simply a set of features which constitute a single identity for
Andean migrants. Being Andean in the Andes has completely different
implications than being Andean in Lima and the negative attitudes shown toward
language are a perfect example of this claim; thus both the relevant literature and
my research point to the fact that (Andean) identity is undeniably conditioned by
location, ethnicity, level of education, and race, as well as all the processes,
relationships and social interactions that take place in the Peruvian capital.19
The concept of Provinciano identity as an interaction of different
characteristics together also led to another finding: many Limeño participants
seemed to be unable to specifically associate the AS morphosyntactic features
with a provinciano. Limeño participants looked confused and expressed how
they “knew” the person was white or capitalina but, at the same time, they also
“knew” that no Limeño (probably referring specifically to traditional Limeños,
not new Limeños) would say something like that. In other words, even though
AS phonology was shown to be central for indexing Provinciano identity, AS
morphosyntax also has social meaning, only in a more ambiguous manner. This
is the reason why Limeños recognized a given AS feature as ‘weird’ or not likely
to be said by the person in the recording, but could not necessarily pinpoint why:
the different components of grammar index Provinciano identity to different
degrees.
One of the recurring patterns found through the quantitative analysis is that
it is more important to establish the difference between groups in terms of power
than in terms of solidarity. That is to say that LS obtained higher scores for those
characteristics related to status; simultaneously, it was AS which tended to
receive higher scores for those characteristics related to solidarity. The data
analysis also revealed that Limeños are more polarized in their answers than
Andean migrants. Based on this result, one can conclude that they perceive more
distinction between Provincianos and themselves, than Provincianos perceive
between themselves and Limeños.
Interestingly, I found that these stigmatized morphosyntactic features were
not indexes of Provinciano identity each by itself but mostly only when
combined. Therefore, when a certain feature was interpreted as being said by an
indigenous person, it was usually assumed that this person was also Provinciana,
had a low level of education, had a low income, and so on. In other words
Provincialism, as any other identity, is relative. For example, in Delforge (2012),
we find a definition of Provinciano as the counterpart of an urban Cuzqueño,
while in Lima both rural and urban Cuzqueños might be considered
Provincianos.

Notes
1. For analytical purposes, I rely on these two seemingly autonomous and opposite categories, but the
complexities of reality exceed this dichotomy. ①

2. i.e. an SLA phenomenon ①

3. I will use ‘traditional’ and ‘classic’ indistinctively. ①

4. Positionality: A concept of cultural anthropology, which refers to one’s own social position in relation to
the group under study. “There is no gaze that is not positioned (Irvine & Gal, 2000)”. For example, it is
important to specify the fact that I am Limeña and that I speak LS. ①

5. Ayacucho and Cajamarca are two different departments in the Andes ①

6. Domestic workers in Lima are usually girls of Andean origin, but not necessarily. ①

7. This is not even considered AS by many scholars such as Anna María Escobar herself, and therefore
names it Billingual Spanish (a series of interlanguages with more or less stability and systematicity). ①

8. Personal communication ①

9. But also with a particular social context – of literacy practices and standard language ideologies. ①

10. Serrano is a pejorative term: from the Sierra/Andes ①

11. Sentences b) and d) present also SOV order associated with AS but, in these two cases, standard
Spanish varieties might favor this word order to express focus or topicalization. ①

12. The present perfect here might also be an Andean feature, but one that is not unusual in LS and, in the
appropriate pragmatic context, acceptable in other standard varieties as well. ①

13. See Appendix A for more information on participants ①

14. Indigenous people from the Amazonia are ethnically and culturally different when compared not only
with people from the Andes but also among themselves. However, it is relevant to mention the relatively
small impact of Amazonian peoples in the history, demographics, and mainstream culture of Peru. Thus in
some sense we might say that white and Andean indigenous are considered poles. ①

15. This is, of course, a subjective judgment conditioned by the dominant cultures. ①

16. I have considered second and third generations of migrants, people actually born in Lima, as migrants
or non-Limeños. However, I agree with other researchers such as Anna María Escobar that the children and
grandchildren of migrants have a different social and cultural profile, topics that I will discuss in my
qualitative analysis but not in the MGT responses because their identity is closer to migrants than to
traditional Limeños. ①

17. The English translation does not accurately show how the grammatical person “a provinciano” (3rd,
singular) does not agree with the verb “make” (1st, plural): el que ha entrevistado última, le falta un poco
del estudio…De todas maneras un provinciano tenemos errores en vocalización de una palabra. ①

18. It is as if calling somebody “Serrano” (from the Sierra/Andes and all the negative characteristics
implicit in the term) is not enough and some people need to add the augmentative suffix -azo/-aza. ①

19. This assumption does not mean that AS speakers are not discriminated against in the Andes; however,
the dynamics change with the change of context. For a discussion on this topic, see Zavala, 2011. ①

References
Arellano Cueva, R. & Burgos Abugattas, D.
(2004.) Ciudad de los Reyes, de los Chávez, los Quispe… Lima: Epensa, Arellano Investigación de
Marketing. ① ②
Babel, A.
(2010.) Contact and contrast in Valley Spanish (Unpublished doctoral dissertation). University of
Michigan, Ann Arbor. ①
Baker, C.
(1992.) Attitudes and language. Clevedon, UK: Multilingual Matters. ①
Booth, C.
(2009.) ‘An ocean of culture’: Language ideologies and the social life of language in multilingual
Darjeeling, India. Texas Linguistic Forum 53 (pp. 8–17). Proceedings of the Seventeenth Annual
Symposium about Language and Society, SALSA XVII. ① ②
Campbell-Kibler, K.
 (2006.) Listener perceptions of sociolinguistic variables: The case of (ing) (Unpublished doctoral
dissertation). Stanford University, Palo Alto, CA. ① ② ③ ④ ⑤
Caravedo, R.
(1996.) Pronombres objeto en el español andino. Anuario de Lingüística Hispánica, 12–13, 545–568. ①
Caravedo, R.
(2014.) Percepción y variación lingüística: Enfoque sociocognitivo. Madrid & Frankfurt: Iberoamericana
& Vervuert. ①
Casaús Arzú, M.
(1998.) La metamorfosis del racismo en Guatemala = Uk’exwachixiik ri kaxlan na’ooj pa iximuleew.
Guatemala City: Cholsamaj. ①
Cerrón Palomino, R.
(1990.) Aspectos sociolingüísticos y pedagógicos de la motosidad en el Perú. In R. Cerrón Palomino &
Solís Fonseca (Eds.), Temas de lingüística amerindia (pp. 153–180). Lima: CONCYTEC and GTZ. ①
②
Cerrón Palomino, R.
(2003.) Castellano andino: Aspectos sociolingüísticos, pedagógicos y gramaticales. Lima: Pontifícia
Universidad Católica del Perú, Fondo Editorial and GTZ. ①
Delforge, A.
(2012.) ‘Nobody wants to sound like a provinciano’: The recession of unstressed vowel devoicing in the
Spanish of Cusco, Perú. Journal of Sociolinguistics, 16(3), 311–335. ①
Eckert, P.
(2008.) Variation and the indexical field. Journal of Sociolinguistics, 12, 453–476. ①
Edwards, J.
(1999.) Refining our understanding of language attitudes. Journal of Language and Social Psychology, 18,
101–110. ①②
Escobar, A.
(1994.) Andean Spanish and bilingual Spanish: Linguistic characteristics. In P. Cole, G. Hermon, & D.
Martin (Eds.), Language in the Andes (pp. 51–73). Newark, DE: Center for Latin American Studies,
University of Delaware. ① ② ③
Escobar, A.
(2000.) Contacto social y lingüístico: El español en contacto con el quechua en el Perú. Lima: PUCP. ①
Giles, H., Hewstone, M. & Ryan, E. B.
(1988.) Research on language attitudes. In U. Ammon, N. Dittmar, & K. J. Mattheier (Eds.),
Sociolinguistics: An International Handbook of the Science of Language and Society (Vol. 1, pp. 585–
597). Berlin: Walter de Gruyter. ① ②
Githinji, P.
(2008.) Ambivalent attitudes: Perception of Sheng and its speakers. Nordic Journal of African Studies,
17(2), 113–136. ①
Godenzzi, J. C.
(2008.) Trazas lingüísticas y discursivas de la ciudad: El caso de Lima. Tinkuy: Boletín de Investigación y
Debate, 9, 47–64. ① ② ③ ④
Granda, G. de
(2001.) Estudios de lingüística andina. Lima: Fondo Editorial PUCP. ①
Haddock, G. & Maio, G. R. (Eds.) (2004.) Contemporary perspectives on the psychology of attitudes.
New York, NY: Psychology Press. ①
Hazen, K., & Hamilton, S.
(2008.) A dialect turned inside out: Migration and the Appalachian diaspora. Journal of English
Linguistics, 36(2), 105–128. ①
Hernández-Campoy, J. M.
(2004.) El fenómeno de las actitudes y su medición en sociolingüística. Murcia: Universidad de Murcia,
Tonos Digital. ①
Heros, S. de los
(1999.) Prestigio abierto y encubierto: Las actitudes hacia las variantes del castellano hablado en el Perú
Revista de Humanidades, 6, 13–44. ①
Heros, S. de los
(2008.) Ideologías lingüísticas sobre el estándar y las variantes regionales del español en dos textos
escolares secundarios. Revista de Lenguas para Fines Específicos Ns 13 & 14. ①
Howard, R.
(2007.) Por los linderos de la lengua: Ideologías lingüísticas en los Andes. Lima: IFEA, IEP, Fondo
Editorial PUCP. ①②③
Irvine, J.
(1989.) When talk isn’t cheap: Language and political economy. American Ethnologist, 16(2), 248–267.
①
Irvine, J.
(2001.) “Style” as distinctiveness: The culture and ideology of linguistic differentiation. In J. Rickford &
Eckert (Eds.), Style and sociolinguistic variation (pp. 21–43). Cambridge: Cambridge University
Press. ①
Irvine, J., & Gal, S.
(2000.) Language ideology and linguistic differentiation. In P. V. Kroskrity, (Ed.), Regimes of language:
Ideologies, polities, and identities (pp. 35–83). Santa Fe, NM: School of Advanced Research (SAR
Press). ①
Klee, C. & Caravedo, R.
(2005.) Contact-induced language change in Lima, Peru: The case of clitic pronouns. In D. Eddington
(Ed.), Selected proceedings of the 7th Hispanic Linguistics Symposium (pp. 12–21). Somerville, MA:
Cascadilla Proceedings Project. ① ② ③ ④
Kroskrity, P. V.
(2004.) Language ideologies. In A. Duranti (Ed.), A companion to linguistic anthropology, (pp. 496–517).
Malden, MA: Blackwell. ① ② ③
Lambert, W. E., Hodgson, R. C., Gardner, R. C. & Fillenbaum, S.
(1960.) Evaluational reactions to spoken languages. The Journal of Abnormal and Social Psychology,
60(1), 44–51. ①
Lambert, W. E.
 (1967.) A social psychology of bilingualism. Journal of Social Issues, 23, 91–109. ①
Le Page, R. B., & Tabouret-Keller, A.
(1985.) Acts of identity: Creole-based approaches to language and ethnicity. Cambridge: Cambridge
University Press. ①
Luhman, R.
(1990.) Appalachian English stereotypes: Language attitudes in Kentucky. Language in Society, 19, 331–
348. ①②
Mick, C.
(2011.) Discourses of ‘border-crossers’: Peruvian domestic workers in Lima as social actors. Discourse
Studies, 13, 189–209. ①②③④
Milroy, L.
(2004.) Language ideologies and linguistic change. In C. Fought (Ed.), Sociolinguistic variation: Critical
reflections (pp. 161–177). New York, NY: Oxford University Press. ① ②
Niedzielski, N.
(1999.) The effect of social information on the perception of sociolinguistic variables. Journal of Language
and Social Psychology, 18, 62–85. ①
Osgood, Ch. E., Suci, G. J., & Tannenbaum, P.
(1957.) The measurement of meaning. Urbana, IL: University of Illinois Press. ①
Rivarola, J. L.
(1986.) Bilingüismo histórico y español andino. Actas del IX Congreso de la AIH. Centro Virtual
Cervantes. Retrieved from [Link] ①
Rivarola, J. L.
(2000.) Español andino: Textos de bilingües en los siglos XVI y XVII. Madrid: Iberoamericana.
②
Silverstein, M.
(2003.) Indexical order and the dialectics of sociolinguistic life. Language and Communication, 23, 193–
229. ①
Silvestre, E.
(1994.) Cultural identity and racial self-prejudice. Paper presented at The Ohio State University, Columbus.
①②
Smith, S.
(2008.) Discriminación, integración y el discurso del progreso en la Nueva Lima: Testimonios urbanos de
cuatro jóvenes limeños. Tinkuy: Boletín de investigación y debate, 9, 65–82. ① ②
Strand, E., & Johnson, K.
(1996.) Gradient and visual speaker normalization in the perception of fricatives. In D. Gibbon, Natural
language processing and speech technology: Results of the 3rd KOVENS Conference, Bielefeld,
October, (pp. 14–26). Berlin: Mouton de Gruyter. ①
Thomas, E.
(2002.) Sociophonetic applications of speech perception experiments. American Speech, 77, 115–147.
①
Van Dijk, T.
(2003.) Ideología y discurso: una introducción multidisciplinaria. Barcelona: Ariel. ① ②
Winford, D.
(2003.) An introduction to contact linguistics. Malden, MA: Blackwell. ①
Wolfram, W., & Schilling-Estes, N.
(2006.) American English: Dialects and variation. Malden, MA: Blackwell. ①
Woolard, K.
(1984.) A formal measure of language attitudes in Barcelona: A note from work in progress. International
Journal of the Sociology of Language, 47, 63–71. ①
Woolard, K., & Gahng, T.
(1990.) Changing language policies and attitudes in autonomous Catalonia. Language in Society, 19, 311–
330. ①
Yiakoumetti, A., Evan, M., & Esch, E.
(2005.) Language awareness in a bidialectal setting: The oral performance and language attitudes of urban
and rural students in Cyprus. Language Awareness, 14, 254–260. ①②
Zavala, V.
(1999.) Reconsideraciones en torno al español andino. Lexis, 23, 25–85. ① ② ③ ④ ⑤
Zavala, V.
(2011.) Racialization of the bilingual student in higher education: A case from the Peruvian Andes.
Linguistics and Education, 22, 393–405. ①
CHAPTER 13

On the effects of Catalan contact in the

variable expression of Spanish future tense
A contrastive study of Alcalá de Henares,
Madrid and Palma, Majorca
Andrés Enrique-Arias & Beatriz Méndez Guerrero
Universitat de les Illes Balears | Universidad Autónoma de
Madrid

Abstract
This paper aims to test the influence of Catalan-Spanish bilingualism
in the variable distribution of the morphological future (MF) cantaré
and the periphrastic variant (PF) voy a cantar in Spanish. As Catalan
does not have a PF, it has been suggested that in the areas where
Spanish coexists with Catalan the MF is relatively more used. A
contrastive study of a bilingual community (Palma de Mallorca) and a
Spanish monolingual one (Alcalá de Henares) confirms this
hypothesis: the frequency of use of the MF among bilinguals is double
what it is in a non-contact setting. At the same time, however, the two
speech communities yield similar results in regard to the semantic
values associated to each of the forms.

Keywords
language change; language variation; Spanish–Catalan contact;
future tense
1. Introduction
A considerable number of studies regarding the expression of futurity throughout
the Spanish-speaking world report a major change whereby the territory of the
morphological future (MF) cantaré is being progressively taken over by the
periphrastic future (PF) voy a cantar. This development, which is seen as the
continuation of a historical trend that dates back to the 17th century (Aaron,
2010), has been studied primarily in the Spanish of Latin America: in many of
such varieties the use of the MF has decreased considerably and in some of them
it has even disappeared altogether (for comparative overviews of available
studies see, for instance, Sedano, 2007; Blas Arroyo, 2008, p. 86; Osborne,
2008, p. 17–29; or Orozco, 2015, p. 350).
At the same time there are very few sociolinguistic studies that examine this
issue in Peninsular Spanish. Based on what can be gleaned from the limited
research that is available Blas Arroyo (2008, p. 86) suggests that in Spain the MF
still enjoys a certain degree of vitality; nevertheless, as most studies focusing on
Peninsular Spanish are based almost completely on written sources there is no
strong empirical data to support this claim. At least one sociolinguistic study
(Osborne, 2008), shows that in Andalusian Spanish rates of use of PF are on a
par with the ones observed in Latin American communities, a finding that
contradicts Blas Arroyo’s proposal of a less developed change in peninsular
Spanish. At any rate it is fair to conclude that we do not know much about the
actual distribution of futures in Peninsular Spanish.
A related issue is the possible influence of language contact in the
distribution of futures. On the one hand it has been claimed that in the United
States Spanish contact with English has accelerated the advance of the
periphrastic future in the Southwest (Gutiérrez, 1995, 2002) and among
Colombian immigrants to New York (Orozco, 2015). In contrast, in the
peninsular areas where Spanish coexists with Catalan, it has been observed that
the decline of the MF is less advanced (Blas Arroyo, 2004, 2007, 2008; Enrique-
Arias, 2008, 2014a; Wesch, 1997), a phenomenon that seems to be due to the
structural asymmetry of the expression of future in the two languages involved
as Catalan does not have a periphrastic future with ir ‘go’.
Until Blas Arroyo’s (2007, 2008) study came about, the alleged robustness
of morphological future in the Spanish spoken in Catalan-speaking territories
had only been mentioned in a handful of studies (cf. Blas Arroyo, 2004, p. 1068;
Wesch, 1997) and there was only one preliminary empirical study (Ramírez &
Blas Arroyo, 2000). Blas Arroyo conducted the first systematic variationist study
on the influence of Catalan contact on the use of Spanish futures in the
community of Castellón. He compared Catalan-dominant versus Spanish-
dominant bilinguals to find that indeed the former retain MF more than the latter.
He did not consider, however, data from monolingual areas of Spain, which
leaves us with the question of whether the retention of MF is a peninsular trait or
something that only happens in Catalan-speaking areas. There is another
variationist study done on Peninsular Spanish, in the Andalusian community of
Puente Genil (Osborne, 2008) which exhibits much lower frequency of MF, but
Osborne’s study is not entirely comparable as she uses different methodology
and looks at different variables.
In view of the insufficient empirical evidence regarding the distribution of
futures in Peninsular Spanish, and more specifically the alleged effect of Catalan
contact in the inhibition of the spread of PF, this investigation aims to explore
further the influence of Catalan–Spanish bilingualism in the variable expression
of futurity in Spanish. To that effect we carry out a contrastive study that
compares a peninsular monolingual community (Alcalá de Henares, a city of
200,000 inhabitants that lies 20 miles northeast of Madrid) and a bilingual one
(Palma, a city of 400,000 inhabitants located in the Catalan-speaking island of
Majorca). To ensure the validity of our contrastive analysis we use two
comparable corpora compiled following PRESEEA guidelines, and apply the
same methodology, considering identical predictor variables. The study centers
around two internal factor groups explored in previous studies of the expression
of futurity (temporal proximity to the speech act, and sentence modality) as well
as gender, age, level of schooling and, in the case of the data collected in Palma,
language dominance. Beyond the empirical value of our contrastive approach in
testing the effect of Catalan contact in the inhibition of the spread of PF, our
study is relevant because it studies the distribution of futures in two communities
that have not been studied so far, thus contributing to a better knowledge of a
seldom studied variable in peninsular Spanish. Although Blas Arroyo and
Osborne present their results in terms of multivariate logical regression analysis
and we use tokens and percentages, we are confident that our results are valuable
in that they complement previous studies.
This chapter is structured as follows. First, we provide a brief overview of
the mechanisms that facilitate the inhibition or deceleration of a change in
language-contact situations. Then we explain the data sources and methodology
used in this study, in particular our intention of contrasting comparable data from
monolingual and bilingual varieties to examine the effect of Catalan language
contact in the evolution of the Spanish future forms. In the next section we
present the results to then conclude with some observations about the relevance
of our comparative approach in the empirical establishment of contact-induced
change.

2. Contact and inhibition of change

While change is widely considered to be a common result of language contact
(Thomason, 2001, p. 10), very little attention has been given to alternative
scenarios, such as when bilingualism favors the retention of traditional features
that are receding in monolingual varieties. As Enrique-Arias explains
(2010, pp. 100–102) the mechanism that facilitates this often overlooked
outcome is fairly straightforward: when there is a change in progress (i.e. an
innovative variant is increasing its frequency and encroaching on new contexts),
the traditional variant may be reinforced amongst bilingual speakers by (a) the
existence of a parallel structure in the contact language and/or (b) the absence of
a structural equivalent for the innovative variant in the contact language. In such
cases change could be delayed, and thus the spread of the innovative variant
would progress more slowly than in non-contact varieties of the same language.
Enrique-Arias (2010, pp. 101–102) illustrates this mechanism with the evolution
of three phonetic phenomena in the Spanish spoken in Mallorca. The
phenomenon known as yeísmo (the loss of the phoneme /ʎ/ and its merger into
the phoneme /ʝ/) is a trend that has become part of the dominant urban speech
pattern in Peninsular Spanish, but the existence of a more robust distinction
between /ʎ/ and /ʝ/ in Catalan apparently has the effect of slowing down this
change in Majorcan Spanish. In Romera’s study in Palma, speakers with Catalan
as L1 exhibited considerably higher levels of conservation of the distinction
compared to those classified as Spanish L1 (Romera, 2003, pp. 371–372). In the
same study, another widespread phenomenon in the Spanish speaking world, the
weakening of /s/ in syllable-final position, exhibits a similar distribution:
speakers classified as Catalan-dominant exhibit a higher level of maintenance of
/s/. Again, the reason is that final syllable [s], which tends to weaken in many
Spanish dialects, is rather robust in Catalan. The same is true of the weakening
and loss of [d] in words ending in -[ado], a tendency in general Spanish which in
the case of Catalan-dominant bilinguals is curbed by the opposite tendency in
Catalan, where the equivalent structure employs a sound articulated with higher
tension (Blas Arroyo, 2007, pp. 269–270; Romera, 2003, p. 373). In sum,
contact-induced change is not the only consequence of language contact:
alternate outcomes, including the retention of conservative variants or the
inhibition or slowing down of a change, are also possible.1
The mechanism that we have illustrated effectively predicts that, even
though the substitution of the morphological future by the periphrastic one is
virtually a pan-Hispanic trend, the structural asymmetry of the expression of
future in Catalan and Spanish will favor the retention of the morphological
future in those communities in which the two languages are in a contact
situation. Both Catalan and Spanish share the widespread Romance MF derived
from the Latin periphrasis infinitive + HABERE ‘to have’ (AMARE HABEO > amar-
é ‘I will love’). Nevertheless, unlike Spanish, which has developed a new PF
with ir ‘go’, the literal Catalan equivalent, that is, the periphrasis with anar ‘go’,
is used for the expression of preterite actions and states. This asymmetry is quite
evident if we look at dialectal data: in Lara Bermejo’s (2016, p. 545) map with
the distribution of future forms in Ibero-Romance varieties, the PF is non-
existent in Catalan-speaking areas including Majorca, while it is widespread in
the Portuguese, Galician, Leonese, Castilian and Aragonese areas. The result of
this asymmetry is that, while all varieties of Spanish have shown a historical
tendency to replace the MF with the periphrasis with ir, bilingual Catalan
speakers tend to retain it more. Blas Arroyo’s (2007, 2008) detailed study of the
distribution of future forms in the Spanish of the Catalan-speaking region of
Castellón provides empirical support for this hypothesis: speakers that are
Catalan-dominant exhibit systematically higher percentages of use of the
morphological future. In short, this is a case in which the outcome of language
contact is not the acceleration of a change but rather the inhibition or
deceleration of a change that is taking place in non-contact varieties (Enrique-
Arias, 2010, pp. 104–105, 2014a, pp. 290–291, 2019).
The effect of Catalan language contact on the retention of the
morphological future in Majorca has also been studied from a diachronic
perspective. Enrique-Arias (2014a, 2008) examined the historical evolution of
the distribution of the two future forms to see whether the Spanish-language
documents produced in Majorca exhibit a slower tendency towards the adoption
of the innovative form, that is, of the PF. To this end he looked at the relative
frequencies of future forms in Aaron (2010), which represents the most complete
work on the historical evolution of Spanish future forms in peninsular Spanish
and compared them to those of Majorcan documents produced by Catalan–
Spanish bilinguals. Aaron (2010) analyzed a number of literary texts from
several historical periods in the modern era (we leave aside Medieval Spanish as
there is an insignificant number of examples of the periphrastic future in that
period): (a) the beginning of the 17th century, (b) the end of the 18th century to
the beginning of the 19th century and (c) the end of the 20th century to the early
21st century. As for texts produced by Catalan–Spanish bilinguals, Enrique-
Arias uses a collection of over a hundred personal letters dated between 1739–
1788 from the Cecilia Zaforteza epistolary archive (for a description see
Enrique-Arias, 2014b) and court testimonies from some thirty legal cases in the
Royal Court of Palma dated between 1769 and 1841. In terms of results, Aaron’s
corpus registers 4.3% periphrastic future in the 17th and 18th centuries
combined and 13.1% in the 19th century (Aaron, 2010, p. 5). A comparison of
these results with the available data for the Spanish of bilingual Majorcans
effectively suggests that there is indeed a slower evolution of this phenomenon
among speakers of the contact variety. In the private 18th century letters,
Enrique-Arias registered 210 examples of the MF as opposed to only two
examples of the PF. Similarly, the presence of the PF in the court documents is
minimal: out of 132 occurrences of future forms found in the 18th-century texts,
none is the PF, and in the 19th-century documents there are only two examples
in comparison to 166 occurrences of the synthetic future (in other words, the
percentage of periphrastic future usage is less than 1%). Table 1 presents a
summary comparison of both corpora:

Table 1. Historical evolution of the expression of future in monolingual and

bilingual areas

Century Monolingual writers (Aaron, 2010) Bilingual writers (Enrique-Arias 2014a)

17–18 4.3% (59/1367) 0.07% (2/280)

19 13.1% (77/588) 1.2% (2/168)
TOTAL 6.9% (136/1955) 0.09% (4/448)

x2 = 24.425; p < .000

Enrique-Arias (2014a) acknowledged that there are comparability problems

between the two corpora regarding dating of the chosen texts, as well as the
distribution of textual genres. At any rate the comparison of the global data from
each corpus seems to confirm the hypothesis that contact with Catalan would
have had the effect of slowing down the development of the periphrastic future
in the Spanish of bilinguals: whereas by the 19th century the PF represents
13.1% of future forms in monolingual peninsular Spanish, the innovative form is
virtually absent in the Majorcan corpus (1.2%). Moreover, the temporal
reference in the four PF examples contained in the Majorcan corpus corresponds
to events very near the moment of enunciation such as when a reference is made
to an action that the speaker is about to undertake; that is, there is an inchoative
aspectual value that expresses a preparatory phase of the verbal action, a usage
not unknown to Catalan (anem a veure ‘let’s see’).

3. Corpus and methodology

The data used in this study are made of the future indicative (MF) and the
periphrastic future (PF) forms used to express future events in 108
sociolinguistic interviews belonging to two different speech communities in
Spain: Alcalá de Henares, a Spanish monolingual community near Madrid, and
Palma de Majorca, a Spanish-Catalan bilingual community in the Balearic
Islands. We are aware that present indicative is also used to express future
events, and this variable has been included in other studies, but we have decided
to limit our investigation to the future tense forms in order to follow the
methodology used in Blas Arroyo’s (2008) study as closely as possible.
The Alcalá interviews were collected within the PRESEEA project in the
1990s – some in 1991 and most of them in 1998 – while the Palma interviews,
also part of the PRESEEA project, were carried out at a later date, between 2007
and 2010. Both sets of informants were selected using identical sampling based
on quotas of sex, age, and level of education (for the PRESEEA project
methodological guidelines see [Link] In each community
the 54 participants are equally divided between men and women distinguishing
three age groups and three education levels, as summarized in Table 2.

Table 2. Distribution of participants according to social variables in Palma and

Alcalá

First age group (18–34 Second age group (35– Third age group (+55
years) 55 years) years)

Male Female Male Female Male Female

Education level 3 3 3 3 3 3
(primary)
Education level 3 3 3 3 3 3
(secondary)
Education level 3 3 3 3 3 3
(college)
In addition, the Palma participants were selected distinguishing three levels of
language dominance; the levels were determined according to a survey of
biographical data and language use. Therefore, the three participants in each cell
(see Table 2) corresponded to a Catalan-dominant speaker, a Spanish-dominant
one and a balanced bilingual (we included the latter category as we observed that
a large number of bilingual speakers in the city of Palma make use of both
languages on a daily basis).
Each interview lasted around 45 minutes. Following the methodology
designed for the PRESEEA project, a fixed number of thematic modules were
introduced during the course of all the interviews. Some of these topics facilitate
the appearance of future tense forms with reference to different degrees of
temporal distance (plans for the weekend, for vacations, career plans) and also
ensure reference to inanimate subjects (predictions about weather in the context
of global warming).
As it is well known, the verbal forms employed by the MF and the PF have
other uses that do not convey future temporal reference. The MF, for instance,
has developed values such as uncertainty, conjecture or probability (Rojo &
Veiga, 1999, p. 2895). Likewise, some constructions that use forms of ir a ‘go to’
with an infinitive and thus resemble the PF clearly retain the lexical meaning
‘move from a place to another’ rather than expressing a future action (e.g.
cuando voy a trabajar… ‘when I go to work’). All such cases, in which the
verbal variable did not bear a clear prospective temporal meaning, and
consequently the two alternatives were not interchangeable, were discarded in
the analysis. In total the corpus contained 1478 valid tokens (836 from Palma
and 642 from Alcalá).
We coded the resulting future forms according to two linguistic factors
(temporal distance to the speech event and sentence modality) as well as
extralinguistic criteria: gender, age, level of education, geographic locality
(considering Palma vs. Alcalá), and in the case of Palma, dominant language.
For data analysis we used the SPSS statistical package; we generated tables with
tokens and percentages and we performed statistical significance texts (chi-
squared and p values). In the following sections we justify the choice of
variables and analyze the results.

4. Results

4.1 Effects of bilingualism in the distribution of variants

Our main objective, as we have explained before, is to explore the influence of

Catalan-Spanish bilingualism in the variable expression of futurity in Spanish.
Table 3 exhibits the frequency distribution for future time forms in a Catalan-
Spanish bilingual setting such as Palma in contrast with the city of Alcalá de
Henares where Catalan is not spoken. This distribution provides evidence that
the MF is considerably stronger in Palma (53.6%) compared to a monolingual
Peninsular Spanish community like Alcalá (23.2%). The data in Table 3 also
suggest that, despite the report of the slower development of PF in Peninsular
Spanish, the frequency registered in Alcalá is very close to that of another
peninsular community, Puente Genil (24.8%; cf. Osborne, 2008), with figures
that in turn are comparable to what has been reported in Latin America (cf.
28.3% in Barranquilla, Colombia according to Orozco, 2015).2 At the same time,
the rates in Palma are very close to those found by Blas Arroyo (2008) in
Castellón, another Catalan–Spanish bilingual community (55.4%).3

Table 3. Distribution of future forms in Palma and Alcalá

Palma Alcalá
MF 53.6% (448) 23.2% (149)
PF 46.4% (388) 76.8% (493)
Total 100% (836) 100% (642)

X2: 137.94; <.000

The influence of Catalan in the retention of the MF is confirmed by the

distribution of futures in the Palma speakers sorted by language dominance (see
Table 4). Whereas Spanish-dominant bilinguals use MF the least (42.2%) the
opposite is true of the Catalan-dominant ones (61%) while balanced bilinguals
lie somewhere in the middle (55.6%).

Table 4. Distribution of future forms in Palma sorted by language dominance

Spanish Both Catalan TOTAL

MF 42.2% (106) 55.6% (154) 61.0% (188) 53.6% (448)

PF 57.8% (145) 44.4% (123) 39.0% (120) 46.4% (388)
TOTAL 100% (251) 100% (277) 100% (308) 100% (836)

x2 = 20.341; p < .000

Blas Arroyo (2008) also considered language dominance in his study. First, he
divided his participants into three groups according to place of origin: speakers
from rural districts in Castellón province, urban residents from the city of
Castellón, and immigrants born in Spanish monolingual areas that had moved to
the province only after adolescence. This classification was done under the
assumption that the diverse provenance of his informants reflects three different
levels of Catalan dominance: the presence and use of Catalan is more prevalent
in rural areas compared to the city of Castellón, while Spanish-speaking
immigrants use Catalan the least. The percentages of inflectional future in his
study (50%, 48% and 39%, respectively) confirm that greater levels of Catalan
dominance entail more use of the MF. Likewise, the probabilistic weights exhibit
a slight tendency toward the use of MF among rural speakers (0.53) while
immigrants from other regions of Spain where Catalan is not spoken clearly
disfavor this form (0.37). At the same time, speakers from the city of Castellón
lie somewhere in between, at a level that is just about neutral (0.49). These
results confirm that the higher density of Catalan speakers in rural areas, as
compared to the city of Castellón de la Plana, where Spanish is the more
commonly used language, is reflected in the probabilistic data.
Still, Blas Arroyo (2008) conducted a second experiment in which he
divided his speakers according to language dominance much in the way we do in
our study. Thus he considered two groups: speakers whose dominant language
was Spanish, including those who were only passive bilinguals, and speakers
whose dominant language was Catalan and who were capable of using Spanish
to some degree. The frequency analysis supports the hypothesis that Catalan
dominance influences the distribution of future forms: those speakers classified
as Catalan dominant exhibit higher rates of use of the MF (53%) vs. 45% for
Spanish dominant speakers. The logistic regression analysis, however, only
found this factor to be significant when taken in isolation and not when
combined with the other factors.
In conclusion, the data presented here support the hypothesis that Catalan
bilingualism favors the retention of the MF in Spanish. First, in the two bilingual
communities studied (Castellón and Palma) the frequency of use of the MF
doubles that of the communities where Catalan is not spoken (Puente Genil and
Alcalá); and second, in both Palma and Castellón, those individuals classified as
Catalan-dominant or that live in areas where there is a higher density of Catalan
speakers retain the MF with higher frequency compared to speakers who are
Spanish-dominant or live in areas where Spanish is used more often.
4.2 Internal factors

Once we have established the importance of Catalan-Spanish bilingualism in the

retention of the MF, we turn now to see whether the distribution of the futures is
influenced by the same internal factors in bilingual and monolingual
communities. Previous studies have looked at a great number of internal factors,
such as temporal distance, adverbial specification, grammatical person, subject
animacy, verb frequency, lexical type, among others. Because of limitations of
space, and for the sake of comparability with the only detailed variationist study
of peninsular Spanish in contact with Catalan, we will just look at the two
internal factors that are most significant in Blas Arroyo’s (2008) analysis,
namely Temporal proximity and Sentence modality. The first one refers to the
temporal distance between the speech act and the event expressed by the verb, a
factor that is often understood as the most relevant when it comes to the
selection of the MF vs. the PF forms. The PF is often linked to the expression of
close or imminent actions or processes, while the MF is supposedly the preferred
variant for distant time. In fact, temporal distance turned out to be the most
relevant factor according to range in the multivariate analysis in Blas Arroyo
(2008) and the third most relevant one in Osborne (2008).
In order to test the relevance of temporal distance in the distribution of
future forms in our data, and also with the intention of making our study
comparable to what has been observed in peninsular Spanish, both in the
bilingual setting of Castellón and in the monolingual one in Puente Genil, we
coded the temporal distance for each verb in the corpus according to four
categories: immediate, intermediate, maximum and indefinite. Immediate events
were those that were to take place right away or within the same day, such as (1)
below. The intermediate category comprised actions that were expected to take
place in the seven days (i.e. a week) following the moment of speech, as in (2).
An event that was to take place at a future moment more distant than a week was
coded as maximum distance, as in (3). Finally, any event for which the moment
of realization could not be determined, as in (4), was coded as indefinite.

(1)
1. en casa de mis padres, están comiendo allí; ahora iré a recogerlos
[speaker 27, Alcalá]
at my parents’ home, they’re eating there; I’ll go now to pick them up
2. ¿me vas a preguntar por qué? [speaker 25, Palma]
are you going to ask me why?

(2)
1. estoy estudiando pero bueno, este martes empezaré a trabajar [speaker
9, Alcalá]
I’m studying but, well, this Tuesday I’ll start working
2. bueno, los va a cumplir el día 1 [speaker 10, Palma]
well, his birthday will be on the first of next month

(3)
1. yo no sé si podré tener vacaciones en Semana Santa todavía [speaker 37,
Palma]
I still don’t know whether I’ll be able to have a vacation at Easter
2. me parece que no va a ser acabar la carrera y encontrar trabajo
inmediatamente [speaker 41, Alcalá]
it seems to me that it won’t be a matter of graduating and finding a job
right away

(4)
 1. y si sale la opción de poder ser bombero, me voy a a hacer bombero
[speaker 2, Palma]
and if the option to become a firefighter comes out I will become a
firefighter
2. me voy a morir antes de todo lo que tengo que hacer [speaker 52,
Alcalá]
I will die before everything I have to do

These categories coincide to a great extent with the ones used in Blas Arroyo
(2008) and Osborne (2008) but with some differences. Blas Arroyo (2008) adds
one more category that he terms attenuated, corresponding to situations that
refer to a distant point in time, but which psychologically display a certain
degree of closeness to the moment of speech due to the presence of elements
such as the demonstrative este ‘this’ in the immediate linguistic context (este
verano ‘this summer’). Due to the extremely low number – four cases in total –
of occurrences of this type in our corpus we have not considered a separate
category for attenuated distance. On the other hand, Osborne (2008, p. 34)
distinguishes between immediate distance, for actions that were to take place
right away or within a period of one to two hours after the moment of speech,
and within a day distance, for events that were to take place within 24 hours. As
these two categories fall within the period characterized as immediate distance
both in Blas Arroyo’s (2008) classification and in our own, in our comparisons
we have collapsed these two categories under the same one.

Table 5. Distribution of MF (as opposed to PF) in Alcalá and Palma sorted by

temporal distance

Immediate Intermediate Maximum Indefinite TOTAL

MF 15.0% 18.2% (4/22) 24.6% (14/57) 28.4% (99/349) 23.2%

(Alcalá) (32/214) (149/642)
MF (Palma) 42.7% 26.0% 52.5% 62.6% 53.6%
(91/213) (12/46) (72/137) (273/436) (448/836)

Alcalá x2 = 13.764; p < .003; Palma x2 = 12.810; p < .000

As seen in Table 5, the data in Alcalá and Palma exhibit similar tendencies: the
MF form is more frequent as the temporal distance increases and is most
frequent in the indefinite distance. Thus in Alcalá the percentages are: immediate
distance 15%, intermediate distance 18.2%, maximum distance 24.6%, and
indefinite distance 28.4%. In the Palma materials we find 42.7%, 26%, 52.5%
and 62.6% for the same categories, respectively. The only discordance to this
pattern involves immediate distance in Palma, with rates of MF (42.7%) that are
actually quite high compared to the intermediate one (26%). This result could be
the effect of factors not considered in the current investigation. One possibility is
that contents in the immediate distance often entail interactions with the
interviewer. For instance, in examples such as es una cosa delicada, te contaré
pues hasta donde yo puedo contar (‘this is a sensitive matter, I will only tell you
as much as I can’ [speaker 33, Palma]), the reference of the form te contaré ‘I
will tell you’ is the immediate future; the use of the MF may be a hedging device
in the part of the speaker as to not appear too aggressive or imposing in the
interaction with the interlocutor. As it is well known (cf. Albelda et al., 2014)
modifying verbal tenses is a common hedging mechanism in Spanish and other
languages (for instance, in placing requests quiero ‘I want’ is more direct than
quería ‘I wanted’, and that is why the latter is commonly used for polite
requests); as the MF is associated with greater distance to the moment of
enunciation, and also because it can be used to express conjecture, its use results
in a less direct and imposing interaction as opposed to using the PF.
The data from Puente Genil and Castellón exhibit a similar pattern (see

4
Table 6).4 First, just like what we have seen for Palma and Alcalá, the values for
MF are always much higher in the bilingual setting of Castellón as opposed to
Puente Genil.

Table 6. Distribution of MF (as opposed to PF) in Puente Genil and Castellón

sorted by temporal distance

Immediate Intermediate Maximum Indefinite

MF (Puente Genil) 16% 12.5% 46.4% 34.1%

MF (Castellón) 34% 51% 73% 40%

Furthermore, in the contexts that are temporally furthest away from the speech
event there are higher rates of MF: 16%, 12.5% and 46.4% respectively for
immediate, intermediate and maximum distance in Puente Genil, and 34%, 51%
and 73% in Castellón. But there are differences as well; in contrast with the
results registered in Puente Genil and Castellón, in the Palma and Alcalá data
indefinite distance has the highest percentage of MF.
In sum, maximum temporal distance is the context in which the MF is used
the most in all four corpora considered. This confirms that even though the
percentage of use of the MF will vary considerably between monolingual and
bilingual settings all four speech communities studied share the general value of
greater temporal distance to the moment of speech associated with the MF.
Next to temporal distance, modality is one of the categories that have been
most often considered in studies of the variable expression of future. One
dimension of this category involves speaker’s attitude, in particular the degree of
certainty that the speaker attributes to the content of the utterance; although we
acknowledge the relevance of speaker’s attitude we find it is a very difficult
factor to formalize in objective terms. For this reason, in our study we have
opted for analyzing sentence modality – whether the sentence that contains the
future form is a statement, interrogation or a command – which can be easily
coded using objective criteria. In considering sentence modality, and in
accordance with Blas Arroyo’s (2008) reasoning, we aim to determine to what
extent the speakers’ degree of involvement in their utterances as well as their
level of certainty as regards their content influences the choice of the future
form. As before, and for the sake of comparability, we have adopted the
categories in Blas Arroyo (2008, pp. 101–102): affirmative, negative,
exclamatory-exhortatory and interrogative, distinguishing between direct and
indirect interrogatives. The predictions are as follows: higher levels of emotional
involvement should favor the PF; therefore affirmative and interrogative
sentences will favor the MF, while exclamatory and exhortatory ones will favor
the PF forms. As for negative sentences, although some studies have related
them to MF in Canadian French (Poplack & Turpin 1999, p. 154), it has been
shown repeatedly that negation does not produce a significant effect favoring the
inflectional future in Spanish (cf. Aaron, 2006, p. 267; Durán Urrea &
Gradoville, 2006, p. 4; Osborne, 2008, p. 61).
As for the results, in Blas Arroyo’s study only affirmative sentences favor
the MF (0.56 probability, 51%) while exclamatory-exhortatory ones disfavor it
the most (0.27, 27%); the remaining categories lie somewhere in the middle,
which bears out the predictions we have established for this category.

Table 7. Distribution of MF (as opposed to PF) in Alcalá and Palma sorted by

modality

Affirmative Negative Interrogative Indirect Exclamative TOTAL

Interrogative

MF 26.6% 10.0% 4.2% 36.8% (7/19) 9.8% 23.2%

(Alcalá) (130/488) (7/70) (1/24) (4/41) (149/642)
MF 56.7% 39.8% 0% (0/7) 58.3% (7/12) 12.5% 53.6%
(404/713) (35/88) (2/16) (448/836)

Alcalá x2 = 21.103; p < .000; Palma x2 = 28.514; p < .000

The data in our study gives some support to the predictions set forth (see
Table 7). Affirmative and indirect interrogatives (which in the grammatical
tradition are considered declarative sentences) are the ones that exhibit the
highest percentages of MF in both corpora (26.6% and 36.8% respectively in
Alcalá, 56.7% and 58.3% in Palma) while exclamative ones exhibit much lower
percentages of MF (9.8% in Alcalá, 12.5% in Palma). As for negative polarity
items, they do not favor the MF, just like other studies have found. Finally,
interrogative sentences exhibit very low percentages of MF, with one case for the
entire corpus; this does not allow establishing comparisons with other studies.

4.3 Age, gender and level of instruction

As we have explained before, the decline of the MF and its replacement by the
PF is a historical trend that can be traced back to the 1600s. If this change still
continues in the two communities considered, then it stands to reason that the
distribution of the two variants should be sensitive to sociolinguistic variables
such as age, gender and level of instruction. The results show, however, that
while this is true for the Palma data, none of these factors reached statistical
significance in the Alcalá group (cf. Table 8).

Table 8. Distribution of MF (as opposed to PF) in Palma and Alcalá sorted

according to sociolinguistic variables

Palma Alcalá

Age
18–55 years 50.8% (286/563) 24.6% (103/418)
56 years and older 59.3% (162/273) 20.5% (46/224)
Palma (x2 = 5,050; p < .024); Alcalá (x2 = 1,380; p < .501)
Gender
Women 49.5% (199/402) 22.4% (58/258)
Men 57.4% (249/434) 23.6% (91/384)
Palma (x2 = 5,198; p < .023); Alcalá (x2 = 0,128; p < .720)
Education level
Primary 50.4% (136/270) 27.3% (50/183)
Secondary 49.6% (129/260) 24.8% (60/241)
Post-secondary 59.8% (183/306) 17.8% (39/218)
Palma (x2 = 7,527; p < .023); Alcalá (x2 = 5,583; p < .061)

In regards to age, we should expect that in consonance with the pan-Hispanic

tendency towards the encroachment of the PF, older speakers would retain the
MF more in their speech when compared to younger speakers. In our study,
participants are stratified in three age groups following the PRESEEA
methodological guidelines: 18 to 34 years, 35 to 55 years, and over 55 years.
Because the younger and middle group exhibited very similar frequencies of MF
(50.5% and 51.2% respectively), the analysis distinguishing the three groups was
not statistically significant; thus we collapsed these two groups and contrasted
the results with the older group, which renders statistically significant results in
Palma (in Alcalá the results were not significant regardless of distinguishing
three or two age groups). In Palma, speakers that are over 55 years of age use the
MF more (59.3%) compared to the younger groups (50.8%), which implies that
we should expect a tendency towards lower rates of MF in the future. The other
study conducted in a Catalan-Spanish bilingual setting (Blas Arroyo, 2008)
presents similar results: after restructuring the age groups from the original four
categories to two, speakers under 40 and speakers over 41, Blas Arroyo
(2008, p. 112) found that the older speakers favored the MF (55%, 0.55) whereas
it was disfavored by those in the younger age groups (44%, 0.41).
The results in peninsular areas where Catalan is not spoken are not as clear;
in the Puente Genil study (cf. Osborne, 2008, p. 54) the data indicated that the
MF was favored over the present tense by the older (over 50) group of speakers,
and in turn the present tense was favored over the MF by the younger (less than
45) group, but there were no statistically significant results indicating that
younger speakers favored the PF. And in the Alcalá sample we actually find a
tendency that is opposite to that of Palma and Castellón, as we have relatively
higher percentages of MF for the younger and middle groups (24.6%) compared
to the older one (20.5%), but again this result does not reach statistical
significance.
As for gender and level of instruction (our study did not consider social
class as a separate category), in Palma the MP is retained more by men (57.4%)
compared to women (49.5%), and by college-educated speakers (59.8%) versus
individuals with a secondary (49.6%) or primary education (50.4%). Again these
results differ from those in Alcalá: first there is hardly a difference between
women and men (22.5% and 23.7%, respectively); and second, the MF is
retained more by the lower education group (27.3%), followed by the speakers
with secondary education (24.9%), and was least used by the college-educated
participants (17.9%), exactly the opposite of what we found in Palma, but with
results that do not reach statistical significance. If we cross gender and education
it turns out that in Palma – we leave out Alcalá as none of these variables are
statistically significant – the MF is retained the most by the higher education
group both among women (55.6%) and men (64.1%) while the middle and lower
groups exhibit lower percentages: for men it is 57% in the lower group and
49.2% in the middle one; as for women, those with a secondary education retain
the MF 50% while participants with only primary education use the traditional
variant the least (40.4%).
In Blas Arroyo’s (2008) study gender and education level did not reach
statistical significance. As for social class, speakers belonging to the highest
social stratum (upper middle class) disfavored the MF; this is opposite to what
has been observed in other communities, where upper class individuals are the
ones that typically display a greater tendency to retain the MF. For instance, the
studies by Gutiérrez (1995, 2002) in several communities of Mexican origin
show that the rapid advance of the PF at the expense of the MF is driven mainly
by the lower classes, a fact that he interprets as a change from below. As we
have mentioned already, gender taken in isolation was not significant in Blas
Arroyo’s (2008) study, but when crossing gender and social class there were
some interesting results: the females in higher social positions produced the
highest percentages of the PF at the expense of the morphological variant (cf.
p. 114).
While it is difficult to extract a clear conclusion from all of the above, some
of the general traits presented so far could indicate a change in progress in Palma
whereby, as has been attested in many other Spanish speaking areas, the PF is
gaining ground at the expense of the MF. The fact that younger and middle-age
speakers are using the PF more often compared to the older ones is in agreement
with the general pan-Hispanic tendency towards limiting the use of the MF.
Likewise, the higher level of MF among educated speakers would confirm that
this is the conservative variant, as this form is the most traditional and
(presumably) educated form of the future tense in the Spanish verb system. And
if indeed there is a change in progress, the relatively higher use of the innovative
form by the Palma females would be evidence that, although educated speakers
favor the more traditional MF, there are no negative connotations attached to the
innovative variant, as this would be a process led by one of the groups that are
sensitive to the sociolinguistic prestige attached to the variants.
 At the same time, it appears that the use of the PF is quite stable in Alcalá,
as none of the sociolinguistic variables considered in our study reached
statistical significance. At any rate, the frequency of MF in Alcalá is
considerably low (23.2% overall), which makes it hard to become even lower
and thus it may be difficult to detect a tendency towards a further reduction of its
uses.

5. Conclusions
The possible influence of Catalan bilingualism in the retention of the MF had
been pointed out in a number of publications but there were only two empirical
studies on the issue: Blas Arroyo (2008) had compared different ethnolinguistic
groups in the community of Castellón and Enrique-Arias (2014a, 2019) had
looked at historical data contrasting texts produced by Castilian writers and
Majorcan bilinguals. In order to provide more robust empirical support for the
hypothesis that Catalan contact inhibits the spreading of the PF, in this article we
conduct for the first time a contrastive study of a bilingual community and a
monolingual one, applying the same methodology in regards to data collection
and analysis. Our results show that the frequency of use of the MF among Palma
bilinguals is double what it is in a non-contact setting such as Alcalá.
Additionally, we compare our results with available data from other peninsular
communities to find a similar contrast between bilingual Castellón and non-
contact Puente Genil.
The quantitative analysis in our study also reveals how certain linguistic
factors that had already been detected in previous variationist studies influence
the choice of future form. Temporal distance, which is possibly the variable that
has been considered most often, appears to work in the same way in the four
peninsular communities for which we have data (Palma, Castellón, Alcalá and
Puente Genil): the MF is favored in the contexts that are temporally furthest
away from the speech event. As for the other linguistic variable that we have
examined, sentence modality, Palma and Alcalá yield the same results:
affirmative and indirect interrogatives exhibit the highest percentages of MF
while exclamative sentences and negative polarity items exhibit lower
percentages, which in turn coincides at least partially with what has been found
in Castellón. We can thus conclude that, while the frequency of use of the MF
will vary considerably between monolingual and bilingual varieties, the different
speech communities studied here share the general value of greater temporal
distance to the moment of speech associated with the MF.
Finally, we have considered the possibility of a change in progress. In
Palma the supposedly innovative variant (i.e. the PF) is favored by younger and
middle-aged speakers, by women, and by individuals with primary and
secondary education, while the traditional variant is more prevalent with men,
older speakers, and speakers with a college degree. This distribution would
indicate that, just like in Castellón, there is a tendency of change in progress
where the MF is being replaced by the PF, although the sociolinguistic profile of
the speakers leading the change differs in the two communities: both in
Castellón and Palma the change is led by women but, in contrast to what
happens in Castellón, in Palma the conservative variant is more prevalent among
the higher social groups.
This tendency of change is consonant with other historical developments in
which Catalan contact phenomena in Palma have receded resulting in a process
of convergence with general Spanish; traits such as seseo (the lack of distinction
between the pair of sibilants /s/ and /θ/), or the directional use of the preposition
en as in voy en Palma (‘I’m going to Palma’) were once widespread in the
Spanish of Majorca but are now restricted to rural speakers (Enrique-Arias,
2008, 2019). In contrast with the Palma data, the distribution of the two future
forms appears to be quite stable in Alcalá, where none of the sociolinguistic
variables reached statistical significance.
We are well aware that, for a complete understanding of the variable use of
the future forms, we still need to perform a detailed multivariate analysis which
will include other factors such as adverbial specification, grammatical person,
subject animacy, verb frequency or lexical type. Likewise, we should include
other peninsular and American communities as well as historical data in our
analysis. For the time being, however, the contrastive study that we have
presented throughout this chapter provides evidence in support of the influence
of Catalan contact in the evolution of the Spanish future forms and serves to
illustrate the mechanisms through which language contact can result in the
retention of conservative variants and the inhibition of changes.

Acknowledgements
The research reported here was funded by two grants from the Ministry of
Economy and Competitiveness of Spain (AEI/FEDER, UE) references FFI2014-
59135-R and FFI2017-83899-P. We are thankful to Francisco Moreno, Isabel
Molina, Florentino Paredes and Ana Cestero for facilitating us the PRESEEA –
Alcalá corpus and to Laura Camargo for her assistance in compiling the
PRESEEA – Palma corpus. We also thank two anonymous reviewers for their
comments.

Notes
1. Enrique-Arias (2010, pp. 102–103) further suggests that these structural factors may be reinforced when
the language in question, as has been traditionally the case with Spanish in Majorca, is the socially
dominant one, which means that it is acquired and used in formal contexts. As Spanish textbooks and
grammars tend to only include the MF, and in general written sources make higher use of this form, Catalan
speakers that learn Spanish through formal registers (i.e. schooling, printed material, mass media) have
limited exposure to the innovative variant and are thus less likely to use it in their speech. ①
2. For comparability sake, when quoting figures from those studies, such as Osborne (2008) or Orozco
(2015) which consider three variants (MF, PF and present tense) we recalculate percentages discarding
present tense occurrences. ①

3. This percentage has been calculated considering Blas Arroyo’s (2008, pp. 91–93) figures in his first
count, in which he considered all valid occurrences of MF and PF, just like we do in our study. Blas Arroyo
undertook a second count in which he considered only those occurrences that appeared isolated in
discourse, and this brough down the percentage of MF to 46%, still quite high if compared to monolingual
variaties of Spanish. ①

4. The Puente Genil data has been adapted from Osborne’s (2008) Table 6 in p. 36. ①

References
Aaron, J. E.
(2006.) Me voy a tener que ir yendo: A corpus-based study of the grammaticization of the ir a + INF
construction in Spanish. In N. Sagarra & J. T. Almeida (Eds.), Selected proceedings of the 9th
Hispanic Linguistics Symposium (pp. 263–272). Somerville, MA: Cascadilla Proceedings Project. ①
Aaron, J. E.
(2010.) Pushing the envelope: Looking beyond the variable context. Language Variation and Change, 22,
1–36. ①②③④⑤
Albelda, M., Briz, A., Cestero, A. M., Kotwica, D., & Villalba, C.
(2014.) Ficha metodológica para el análisis pragmático de la atenuación en corpus discursivos del español
Oralia: Análisis del discurso oral, 17, 7–62. ①
Blas-Arroyo, J. L.
(2004.) El español actual en las comunidades del ámbito lingüístico catalán. In R. Cano (Ed.), Historia
la lengua española (Vol. 2. pp. 1065–1086). Barcelona: Ariel. ① ②
Blas-Arroyo, J. L.
(2007.) El contacto de lenguas como factor de retención en procesos de variación y cambio lingüístico.
Datos sobre el español en una comunidad bilingüe peninsular. Spanish in Context, 4(2), 263–29.
①②③④
Blas-Arroyo, J. L.
(2008.) The variable expression of future tense in Peninsular Spanish: The present (and future) of
inflectional forms in the Spanish spoken in a bilingual region. Language Variation and Change, 20(1),
85–126. ①②③④⑤⑥⑦⑧⑨⑩⑪⑫⑬⑭⑮⑯⑰⑱⑲⑳
Durán-Urrea, E. & Gradoville, M.
(2006.) Variation in the future tense of New Mexican Spanish. Texas Linguistics Forum, 50 (SALSA XIV
Proceedings). Retrieved from [Link] (26
December, 2017). ①
Enrique-Arias, A.
(2008.) Materiales para la historia del castellano hablado en Mallorca: El archivo epistolar de Pedro de
Santacilia. In C. Sinner & A. Wesch (Eds.), El castellano en las tierras de habla catalana (pp. 219–
232). Frankfurt & Madrid: Vervuert & Iberoamericana. ①②③
Enrique-Arias, A.
(2010.) On language contact as an inhibitor of language change: The Spanish of Catalan bilinguals in
Majorca. In A. Breitbarth, C. Lucas, S. Watts, & D. Willis (Eds.), Continuity and change in grammar
(pp. 97–118). Amsterdam: John Benjamins. ①②③④
Enrique-Arias, A.
(2014a.) Efectos del contacto de lenguas en el castellano de Mallorca: Una perspectiva histórica. In A.
Enrique-Arias, M. J. Gutiérrez, A. Landa, & F. Ocampo (Eds.), Perspectives in the study of Spanish
language variation (Anexos de Verba, 72), pp. 271–297. Santiago de Compostela: Universidad. ① ②
③④⑤⑥
Enrique-Arias, A.
(2014b.) Lengua y escritura en la Mallorca del XVIII: El archivo epistolar de Cecilia Zaforteza. In A.
Castillo-Gómez & V. Sierra-Blas (Eds.), Cinco siglos de cartas. Historia y prácticas epistolares en las
épocas moderna y contemporánea (pp. 313–328). Huelva: Universidad. ①
Enrique-Arias, A.
(2019.) Testing contact-induced change in the Spanish of Mallorca. Insights from a historical perspective
In G. Rei-Doval & F. Tejedo-Herrero (Eds.), Lusophone, Galician and Hispanic linguistics: Bridging
frames and traditions (pp. 168–183). London: Routledge. ①②③
Gutiérrez, M.
(1995.) On the future of the future tense in the Spanish of the Southwest. In C. Silva-Corvalán (ed.),
Spanish in four continents: Studies in language contact and bilingualism (pp. 214–226). Washington,
DC: Georgetown University Press. ① ②
Gutiérrez, M.
(2002.) Innovación en el español de los Estados Unidos: ¿Progreso o decadencia? Revista de Lingüística
Teórica y Aplicada, 40, 113–128. ① ②
Lara-Bermejo, V.
(2016.) La expresión del futuro en las lenguas romances de la Península Ibérica. Boletín de la Real
Academia Española, 96(314), 529–558. ①
Orozco, R.
(2015.) Castilian in New York City: What can we learn from the future? In S. Sessarego & M. González-
Rivera (Eds.), New perspectives on Hispanic contact linguistics in the Americas (pp. 347–372). Madrid
& Frankfurt: Iberoamericana & Vervuert. ①②③④
Osborne, S. D.
(2008.) Variable future tense expression in Andalusian Spanish (Unpublished MA thesis). University of
Georgia, Athens. ① ② ③ ④ ⑤ ⑥ ⑦ ⑧ ⑨ ⑩ ⑪
Poplack, S., & Turpin, D.
(1999.) Does the FUTUR have a future in (Canadian) French? Probus, 11(1), 133–164. ①
Ramírez, M., & Blas-Arroyo, J. L.
(2000.) La expresión variable del futuro verbal en el español castellonense. In V jornadas de fomento a la
investigación. Castellón: Servei de Publicacions de la Universitat Jaume I (CD-ROM edition). ①
Rojo, G., & Veiga, A.
(1999.) El tiempo verbal. Los tiempos simples. In I. Bosque & V. Demonte (Eds.), Gramática descriptiva
de la lengua española (pp. 2867–2935). Madrid: Espasa. ①
Romera, M.
(2003.) La variedad del castellano actual en Baleares. Moenia, 9, 359–381. ① ②
Sedano, M.
(2007.) Future tense expressions in several Spanish corpora. In G. Parodi (Ed.), Working with Spanish
Corpora. Research in Corpus and Discourse (pp. 132–144). London: Continuum. ①
Thomason, S.
(2001.) Language contact. An introduction. Edinburgh: Edinburgh University Press. ①
Wesch, A.
(1997.) El castellano hablado de Barcelona y el influjo del catalán. Esbozo de un programa de
investigación. Verba, 24, 287–312. ① ②
Index

A
accent 115–118, 131–132, 281, 292, 310
Afrogenesis Hypothesis 43–44, 47, 55
see also social ecology 11, 14, 16, 18

Afro-Hispanic 2, 8, 11–12, 18, 38, 43–45, 47–50, 55, 57–58

black slavery 43, 51
Decreolization Hypothesis 43–48, 55
Chocó Spanish 2, 43, 45–46, 48–49, 55

Andean migrants 5–6, 85, 91, 283–289, 292–293, 295, 297, 299, 301, 304–305,
307–308, 310–311
Andean Spanish 2, 5, 12, 18, 31, 110, 165, 283, 287, 290–291, 294, 302, 312
Aymara 1–2, 91, 287, 290, 292–293

B
bilingualism 1, 5–7, 14, 31, 36, 38, 56, 59, 61, 64, 66–68, 72, 75, 79–81, 139,
146, 154, 159–162, 179–180, 187, 189, 198, 209–211, 232–233, 235, 243,
254–259, 281–282, 292, 313, 315–317, 322–324, 331, 333
child bilingualism 61, 64, 75
codeswitching 7, 235, 255–258
C
Caribbean Spanish 3, 50, 57, 139–140, 142–144, 151, 156, 158, 161, 265, 276
Dominican Spanish 3–5, 89, 110, 161, 163, 166–169, 172–174, 176–177,
263, 281–282
Puerto Rico vii, 1, 7, 20, 47, 65, 111, 140, 142–143, 146, 152, 161, 173,
263–267, 269–272, 277–278, 280–282

Chile 85–87, 91–92, 99, 102, 105–108, 110–111, 224

Chilean Spanish 85, 87–90, 92, 96, 105–106, 108–109

G
gender agreement 2, 7, 20, 48–49, 55, 61–68, 72–80, 233, 297

H
heritage language 61–64, 67, 71, 75–78, 140, 159, 215, 220–221, 223, 230, 233
heritage speakers 1–4, 8, 51, 61, 63–64, 66, 69, 71, 78–80, 139, 161, 211,
215–216, 220–231, 233, 254–255, 258

I
imperfect 4, 141, 161, 216, 218–219, 221–223, 225–226, 230–233
intonation 57, 116, 118, 131–134, 136

L
language attitudes 111, 282–284, 286–289, 293–296, 298, 301, 308, 310,
312–314
language change 12, 14, 21, 36–37, 39, 109, 115, 135, 181, 210–211, 258, 313,
315, 333
language contact 1–2, 4, 6, 8, 11–14, 17–18, 28, 35, 37–38, 40–41, 58, 61, 63,
115, 140, 144, 146, 155–159, 161–162, 177, 179, 181, 189, 210–211, 222,
232, 237, 254, 256–257, 263, 287, 290, 316–319, 332–334
Chocó Spanish 2, 43, 45–46, 48–49, 55
dialect contact 6, 85–86, 109, 143, 161, 263, 282, 285, 287
dialectal contact 263, 265
Spanish–Catalan contact 315
Spanish-contact varieties 46
Spanish Creole debate 40, 43, 45–46
Spanish creoles 8, 19–20, 38, 43, 46–47, 57–58

language processing 11, 25–26, 28, 39–40, 57, 235, 314

linguistic ideologies 111, 283–284, 286, 288–289, 293, 301

M
Majorca 4, 6, 179–182, 186, 189, 191–195, 199–200, 204, 206, 208–210, 315,
317–320, 332–333
Maya 3, 115–117, 121–123, 126, 130–136
Yucatan Spanish 3, 115, 135

migration 2, 19, 90–91, 111, 263–264, 266–267, 281, 283, 285–286, 288, 291,
297, 306, 313

N
naturalistic SLA 11–15, 25, 27–28, 31, 35
P
perception 5, 110, 118, 132, 237, 263–264, 267–268, 271–275, 277–278, 280,
283–285, 292, 301, 303, 305, 310, 312–314
Peru 1, 47, 85–86, 90–92, 102, 110–111, 224, 239, 241, 283–286, 289, 292–293,
299, 309, 313
Lima 2–3, 85–87, 90–93, 102–107, 109–111, 113, 165, 283–289, 291–293,
295, 297–303, 305–307, 309–314
Limeño Spanish 87, 91, 104, 283–284, 291, 302
Quechua 1–2, 31, 91, 165, 177, 284–285, 287, 290–293, 298, 312

Portuguese 1, 4–5, 13–14, 19–20, 22–23, 27, 32, 36–37, 44, 46–47, 56–57, 211,
215–216, 218–219, 224, 233, 235, 237–240, 242–257, 318
proficiency 3, 28, 30–31, 63, 66, 71, 76, 78, 139–140, 144–148, 150–159, 171,
206, 222–224, 237, 239, 243, 253, 290, 292, 298
pronominal clitics 179–188, 190, 194, 197, 199–202, 204, 208–210
prosodic rhythm 115–116, 119–121, 123, 136
provinciano 283–284, 289, 292–296, 298, 300–306, 308–312

R
rhythm 115–121, 123, 126, 130–136

S
semantic verb type 3, 140, 142, 156, 158
sociolectal leveling 85, 90, 105–106, 108
sociolinguistic 1–3, 6–7, 11, 13–14, 17, 19, 21–22, 25, 35, 61–63, 65–74, 76–81,
87, 90, 92–93, 102, 105–107, 109, 111, 122, 135, 146–147, 160, 162, 168,
172, 222, 231, 238–239, 243, 252, 255–256, 263–265, 267, 277, 279–282,
 284, 289, 293, 295, 312–314, 316, 320, 328, 330–332
subject pronoun expression 3, 8, 66, 81, 139–140, 143, 153–155, 159–161
rates of overt pronominal subjects 3, 140, 144–145, 152, 155, 158
switch reference 3, 141–142, 145, 147–149, 153–160

T
tense, aspect, mood 3–4, 26–27, 80, 141, 163, 166, 171, 177–178, 215–221, 223,
225, 227, 230–233, 315, 320–322, 329–330, 333–334

U
U.S. Spanish 139, 143, 222
Spanish in the US 163

V
variation 2–3, 7–8, 18, 21, 38, 57–58, 80, 85–87, 90–91, 94, 96, 98–99,
102–103, 105–111, 119–121, 126, 130, 132–136, 139–140, 142, 152–153,
159–162, 165, 176–178, 188–190, 197, 199, 202, 206–209, 211, 221,
232–233, 255, 281–282, 284, 290, 292–293, 312–315, 332–333
future tense 315, 320–321, 330, 333–334
present perfect 3–4, 7, 163–166, 177–178, 233, 291, 297
preterite 161, 163–165, 177, 216, 233, 318
progressive verb form 215
sibilant variation 85–86, 106, 109
social variables 86, 100–101, 103, 105, 107, 111, 179, 181, 190, 192–193,
200, 204–205, 208–209, 321
 verb morphology 215
verb form 27, 139, 141–142, 144, 149, 153–154, 215, 222–223, 225, 227,
231
verb form ambiguity 139, 141–142, 144, 153–154

W
weakening 50, 85–87, 89–90, 98, 100, 102, 105–107, 109–110, 145, 158, 222,
318
/s/ 2, 3, 85–111, 160, 161, 318, 332