Spanish Language and Sociolinguistic Analysis [1 ed.] 9789027267245, 9789027258076

Citation preview

Issues in Hispanic and Lusophone Linguistics

8

Spanish Language and Sociolinguistic Analysis

Edited by Sandro Sessarego and Fernando Tejedo-Herrero

John Benjamins Publishing Company

Spanish Language and Sociolinguistic Analysis

Issues in Hispanic and Lusophone Linguistics (IHLL) issn 2213-3887

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

Editors Jason Rothman

University of Reading

Jennifer Cabrelli Amaro

University of Illinois at Chicago

Editorial Board Patrícia Amaral

Kimberly L. Geeslin

Pilar Prieto

Sonia Colina

Michael Iverson

Liliana Sánchez

João Costa

Matthew Kanwit

Ana Lúcia Santos

Inês Duarte

Paula Kempchinsky

Scott A. Schwenter

Daniel Erker

Naomi Lapidus Shin

Carmen Silva-Corvalán

Timothy L. Face

Juana M. Liceras

Sónia Frota

John M. Lipski

University of Arizona

Ángel J. Gallego

Gillian Lord

State University of New York

María del Pilar García Mayo

Jairo Nunes

University of Maryland

Anna Gavarró

Acrisio Pires

University of Ottawa

Indiana University University of Arizona Universidade Nova de Lisboa Universidade de Lisboa Boston University University of Minnesota Universidade de Lisboa Universitat Autònoma de Barcelona Universidad del País Vasco

Universitat Autònoma de Barcelona

Indiana University

Macquarie University University of Pittsburgh University of Iowa

University of New Mexico University of Ottawa Pennsylvania State University University of Florida Universidade de São Paulo University of Michigan, Ann Arbor

Volume 8 Spanish Language and Sociolinguistic Analysis Edited by Sandro Sessarego and Fernando Tejedo-Herrero

Universitat Pompeu Fabra Rutgers University Universidade de Lisboa Ohio State University University of Southern California

Miquel Simonet Megan Solon

Juan Uriagereka

Elena Valenzuela Bill VanPatten

Michigan State University

Spanish Language and Sociolinguistic Analysis A descriptive and prescriptive analysis

Edited by

Sandro Sessarego University of Texas at Austin

Fernando Tejedo-Herrero University of Wisconsin-Madison

John Benjamins Publishing Company Amsterdam / Philadelphia

8

TM

The paper used in this publication meets the minimum requirements of the American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.

doi 10.1075/ihll.8 Cataloging-in-Publication Data available from Library of Congress: lccn 2016001811 (print) / 2016009684 (e-book) isbn 978 90 272 5807 6 (Hb) isbn 978 90 272 6724 5 (e-book)

© 2016 – 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 · https://benjamins.com

To Grace (Graziella)

Table of contents Acknowledgments

ix

Introduction

xi

Part I.  Cutting-edge Methodologies in Sociolinguistics Quantitative analysis in language variation and change Sali A. Tagliamonte Combining population genetics (DNA) with historical linguistics: On the ­African origins of Latin America’s black and mulatto populations Armin Schwegler

3

33

Part II.  Bilingualism Los Angeles Vernacular Spanish: An analytical approach to its indicators, ­markers, and stereotypes Claudia Parodi & Armando Guerrero Jr. On the tenacity of Andean Spanish: Intra-community recycling John M. Lipski Spanish and Valencian in contact: A study on the linguistic landscape of Elche Francisco Martínez Ibarra

91 109

135

Part III.  Language Acquisition Children’s Spanish subject pronoun expression: A developmental change in tú? Naomi Lapidus Shin

157

The role of social networks in the acquisition of a dialectal features ­during study abroad Joshua Pope

177

Lexical frequency and subject expression in native and non-native Spanish: A closer look at independent and mediating effects Bret Linford, Avizia Y. Long, Megan Solon, Melissa Whatley, & ­Kimberly ­Geeslin

197

.

 Spanish Language and Sociolinguistic Analysis

Part IV.  Phonological Variation On glottal stops in Yucatan Spanish: Language contact and dialect ­standardization Jim Michnowicz & Laura Kagan Vowel raising and social networks in Michoacán: A sociophonetic analysis Jennifer Barajas Bilingualism and aspiration: Coda /s/ reduction on the Atlantic Coast of ­Nicaragua Whitney Chappell

219 241

261

Part V.  Morpho-Syntactic Variation Spanish and Portuguese parallels: Impoverished number agreement as a vernacular feature of two rural dialects Sandro Sessarego & Letânia Ferreira The tuteo of Rocha, Uruguay: Is it as stable as it seems? Joseph R. Weyers A corpus-based sociolinguistic study of contact-induced changes in subject placement in the Spanish of New York City bilinguals Carolina Barrera-Tobón & Rocío Raña Risso

285 305

323

Part VI.  Lexical Variation Social factors in semantic change: A corpus-based case study of the verb ­afeitar ‘to adorn, to apply cosmetics, to shave’ David Korfhagen Attitudes towards lexical Arabisms in sixteenth-century Spanish texts Patricia Giménez-Eguíbar

345 363

“Trabajar es en español, en ladino es lavorar”: Lexical Accommodation in ­Judeo-Spanish Rey Romero

381

Index

401

Acknowledgments We owe the completion of this book to a number of people who supported us in a variety of ways during its preparation. We would like to thank the University of ­Wisconsin-Madison and in particular the Spanish and Portuguese Department and the Latin American, Caribbean and Iberian Studies Program for having provided us with a great academic environment to research Spanish sociolinguistics from a variety of disciplines and points of view. This book would not have been possible without the input of many University of Wisconsin-Madison students and colleagues. In particular, we would like to thank Francisco Scarano, Grant Armstrong, Diana Frantzen, Rajiv Rao, Catherine Stafford, Marisa Carpenter, Tammi Simpson, and Grace Bloodgood. We are indebted to all the colleagues who accepted to offer their professional opinion in the process of reviewing the manuscripts: Patrícia Amaral, Mark ­Amengual, Sonia Barnes, Ryan Bessett, José Luis Blas Arroyo, Travis G. Bradley, ­ Whitney Chappell, Clancy Clements, Manuel Delicado, Ana María Díaz Collazos, S­ tephen N. Dworkin, Daniel Erker, Francisco Gago-Jover, Kimberly L. Geesling, Patricia ­Giménez-Eguibar, Melvin González Rivera, Nicholas Henriksen, Jonathan Holmquist, Lewis C. Howe,  Mary Johnson, Felix  Julca Guerrero, Jeremy King, Bryan Kirschen, Delano Lamy, Michael J. Leeser, Natalia Mazzaro, Sean McKinnon, Antonio MedinaRivera, Jim Michnowicz, María Irene Moyna, Carmen Muñoz, Francisco Ocampo, Pablo Pastrana-Pérez, Rosa M. Piqueres, Kim Potowski, Ana de Prada Pérez, Aldina Quintana Rodríguez, Michelle Ramos Pellicia, Kathryn Ringer-Hilfinger, María José Serrano, Natalya Stolova, Catherine Travis, and Julio Villa García. Finally, we wish to thank Jason Rothman, Jennifer Cabrelli, and the publishing team of John Benjamins for their professionalism and help with the publication of this study. Thank you!

Introduction Sandro Sessarego & Fernando Tejedo-Herrero

University of Texas at Austin / University of Wisconsin-Madison

Spanish Language and Sociolinguistic Analysis explores the current state of Spanish sociolinguistics and its contribution to theories of language variation and change, from both synchronic and diachronic perspectives. It offers original analyses on a variety of topics across a wide spectrum of linguistic subfields from different formal, experimental, and corpus-based standpoints. During the last few decades the development of statistical models and their application to the study of language variation and change has found a strong presence in sociolinguistic research, and, as reflected in this volume, Spanish sociolinguistics has not been an exception to this trend. Indeed, several studies in this collection use a ­variety of statistical tools to analyze how linguistic and social factors pattern language variation. For instance, the Rbrul statistical analyses of glottalization in Yucatan Spanish discovered a hierarchy of social variables that constrain its use among younger speakers (Michnowitz & Kagan), while a variationist approach based on different regression analyses cast light on the factors affecting /-s/ weakening in costal Nicaraguan Spanish (Chappell). Mixed-effects logistic regression analyses were also employed to demonstrate the constraint hierarchy in the use of subject pronoun expression among first language learners (Shin), and a similar set of tests helped understand the relationship between lexical frequency and subject expression in the speech of native and non-native speakers (Linford et al.). Statistical analyses also proved to be essential in demonstrating the incipient stages of a change in progress concerning the use of tuteo in Rocha, Uruguay (Weyers), while the ANOVA tests and multivariate regression analyses applied to subject placement in the speech of New York City Spanish-English bilinguals offered important insights into the external conditions that need to be factored in to understand how this phenomenon operates (Raña-Risso & Barrera Tobón). Nevertheless, as important as it is to take advantage of existing statistical toolkits to analyze language variation and change, it is also crucial not to lose sight of the qualitative aspects of the analysis when it comes to asking the right questions and explaining the data. With that in mind, this volume begins with a contribution that offers not only an authoritative assessment of the advantages and disadvantages of widely-used statistical tools, but also useful advice for not losing focus on analyzing and ­interpreting the data in principled and informed ways (Tagliamonte).

doi 10.1075/ihll.8.002int © 2016 John Benjamins Publishing Company

 Sandro Sessarego & Fernando Tejedo-Herrero

Along with variationist analyses, the present volume covers a range of other analytic approaches to studying sociolinguistics. Thus, the present volume includes interdisciplinary methodologies that combine population genetics and historical linguistics to explain the origins of Afro-Hispanic populations in the Americas (Schwegler), models that draw data from historical and present day sources to study the retention of linguistic features in bilingual varieties (Lipski), and analyses that combine variationist and formal approaches within the Minimalist Program to study vernacular features in rural speech communities (Sessarego & Ferreira). This collection of studies also presents original research within other well-­ established frameworks, such as social networks theory applied to second language acquisition (Pope) and sociophonetic variation (Barajas); accommodation theory in the context of lexical loss in Judeo-Spanish (Romero); language attitudes toward the history of Arabisms (Giménez-Eguíbar) and non-standard features across generations in Los Angeles Spanish (Parodi & Guerrero). Additionally, this work includes some of the current research conducted in promising areas of study such as socio-cognitive semantics (Korfhagen), and language policy and linguistic landscape (Martínez Ibarra). As a whole, this book reflects an array of approaches and analyses that show how in its variation across speakers, speech communities, linguistic contexts, communicative situations, dialects, and time, the Spanish language provides an immense wealth of data to challenge accepted linguistic views and shape new theoretical proposals in the field of language variation and change. The present volume contains seventeen peer-reviewed articles, which were originally presented at the seventh meeting of the Workshop on Spanish Sociolinguistics, April 3–5, 2014, hosted at the University of Wisconsin-Madison. It is organized around six thematic sections: (i) Cutting-edge Methodologies in Sociolinguistics; (ii) Bilingualism; (iii) Language Acquisition; (iv) Phonological Variation; (v) Morpho-Syntactic Variation; and (vi) Lexical Variation. Section I includes two chapters that discuss and analyze some of the most advanced methodologies related to sociolinguistic research. In Chapter 1, Sali Tagliamonte provides an overview of traditional methods (e.g., logistic regression) as well as a presentation of more innovative tools, such as condition inference trees, that can be applied to the study of language variation and change. In her essay, Tagliamonte reviews the pros and cons of the available statistical models and reflects on the importance of the qualitative research that underlies any quantitative analysis of sociolinguistic data. In Chapter 2, Armin Schwegler offers an account of how population genetics (DNA) can be combined with historical linguistic information to shed light on the origins of Latin America’s black populations. In particular, he focuses on the ancestry of two AfroHispanic communities in the Americas: Palenque (Colombia) and Palo Monte (Cuba). Data point to a single origin for these populations, whose ancestry appears to descend from a relatively small African region, Mayombe, in the Republic of Congo.

Introduction 

Section II concentrates on the analysis of contexts in which Spanish is in contact with other languages: English, Quechua, and Valencian. In Chapter 3, Claudia Parodi and Armando Guerrero Jr. study a set of non-standard Spanish features and ­Salvadoran  lexical items found among speakers of Los Angeles Vernacular Spanish (LAVS). The authors demonstrate how Spanish-speaking immigrant communities in the United States negotiate the acceptability ratings of these features outside of the pressures of standard Spanish. The authors’ work focuses on acceptability ratings between first- and second-generation LAVS speakers and, unlike most current research in this area of study, they also examine interactions between each of these generations with monolingual speakers of Spanish. Given the lack of focus on the latter and the promising results of their research, the authors suggest the importance of this type of study with regard to other immigrant communities in the United States. In Chapter 4, John Lipski takes us to the Andes. In his study, the author proposes a model of trans-generational language change to account for the systematic appearance of Quechua-induced L2 Spanish features in Andean communities of Quechuadominant late bilinguals. The data collected by means of ethnographic interviews, sociolinguistic inquiries and interactive tasks helps build a framework that attempts to explain the nature and persistence of Andean-Spanish traits in a region characterized by prolonged Spanish/Quechua contact. In Chapter 5, Francisco Martínez-Ibarra studies an area of increasing interest among specialists in language policy issues: the linguistic landscape, or the use of one or more languages on public signs. The author examines the language attitudes toward Spanish and Valencian based on his analysis of over three thousand data items found in a wide variety of signs in the city of Elche. Martínez-Ibarra analyzes a number of factors related to the linguistic landscape in Elche (e.g., the choice of language, the type of business or author of the sign, and the location of the sign in the city) and offers an assessment of linguistic attitudes toward Valencian, the success of the language policies implemented by the local and regional governments, and what this might mean for spoken Valencian in Elche vis-à-vis the use of Spanish as the dominant language. Section III brings together three articles that focus on various aspects of Spanish language acquisition either as the L1 variety (Chapters 6 and 8) or as L2 (Chapters 7 and 8). In Chapter 6, Naomi Lapidus Shin offers an analysis of L1 acquisition by exploring the evolution of structured morpho-syntactic variation in the use of Spanish subject pronouns by Mexican children aged 6 to 16. Lapidus Shin sheds light on the constraints regulating variable pronominal use by showing that, while certain linguistic factors appear to have a relatively constant effect across different ages, the second-person singular variable tends to favor tú expression among younger children and tú omission among older ones. A combination of semantic and pragmatic factors seem to be responsible for such a developmental path.

 Sandro Sessarego & Fernando Tejedo-Herrero

In Chapter 7, Joshua Pope further analyzes an instance of language acquisition. This time, however, the focus is not on L1, but rather on L2. Pope relies on the examination of the social networks developed by American students studying abroad in Madrid to explain the different mastery of use of Spanish /θ/. Among other findings, he indicates that American students who developed close-knit social networks with other Americans presented lower levels of /θ/ than the learners who branched out and formed bonds with native Spanish speakers. In line with Milroy’s (1987) proposal, Pope suggests that the new social relationships can be seen as responsible for the introduction and adoption of new linguistic variants and that input from such social ties may lead to increased change in the American students’ spoken Spanish as their bonds with the native speakers grow stronger. In Chapter 8, Bret Linford, Avizia Long, Megan Solon, Melissa Whatley, and ­Kimberly Geeslin analyze the effects of lexical frequency on subject expression in native and advanced non-native varieties of Spanish. They find that lexical frequency does not have any significant independent effect on the data distribution of subject expression but that it can mediate the effect of other linguistic factors. Moreover, their analysis is not limited to overt and null pronominal forms. Rather, they also analyze full noun phrases (lexical NPs). Their results appear to be, for the most part, in line with Erker and Guy (2012) and Linford and Shin (2013), but are in contrast with the conclusions of Beyley et al. (2013). The authors’ findings provide intriguing similarities as well as differences between their study and the previous literature, suggesting future lines of research. Section IV includes chapters dealing with phonological/phonetic variation in Mexican and Nicaraguan speech communities. In Chapter 9, Jim Michnowitz and Laura Kagan focus on analyzing the internal and external factors regulating the production of glottal stops in Yucatan Spanish. The authors conclude that glottal insertion before vowel-initial words in this dialect can be understood as the byproduct of a contact-induced change driven by Mayan and mediated by internal development. This phenomenon, as well as many other dialectal features of Yucatan Spanish, are affected by standardization forces. Thus, younger, more educated speakers tend to drop the traditional Yucatan features, in favor of more prestigious standard Spanish forms. In Chapter 10, Jennifer Barajas presents an interesting study on vowel raising in rural Michoacán, México. Based on an acoustic analysis of /e/ and /o/ in unstressed post-tonic position, her study shows the relationship between vowel raising and particular social groups in the Colongo speech community. Specifically, her study demonstrates that vowel raising is a phonetic feature associated with members of close-knit networks in the community. The next chapter, by Whitney Chappell, looks at the late acquisition of Spanish by L1 Miskitu speakers in Bilwi, a town in northern Nicaragua. The author examines the production of /s/ in coda position; and, contrary to most research on this issue, she

Introduction 

concludes that, at least in the community under investigation, late L2 Spanish speakers reduce /s/ more frequently where they hear the /s/ reduction the best (i.e. in cue-rich, prevocalic environments). This differs from the linguistic behavior of monolingual Spanish speakers, who tend to reduce /s/ the most in cue-impoverished environments (i.e. in preconsonantal position). Section V has morpho-syntactic variation as its primary focus. In Chapter  12, Sandro Sessarego and Letânia Ferreira compare variable number agreement phenomena within the DP in Afro-Bolivian Spanish and Popular Brazilian Portuguese. They adopt an eclectic methodology that combines sociolinguistic techniques of data collection with generative models of linguistic analysis (Sessarego 2014). The nature of an impoverished agreement system in these two vernacular varieties is explained in light of a feature geometry account (Harley and Ritter 2002) that favors the emergence and selection of singular values as default values. In Chapter 13, Joseph Weyers analyzes the morpho-syntactic variation between voseo and tuteo in Rocha, Uruguay. He finds that, while tuteo represents a distinctive feature of this dialect and is connected to a strong sense of local pride and identity, younger speakers, primarily males, are more likely to prefer the vos (sos) form over tú (eres) with the verb ‘to be’ (ser). The reasons for such a preference appear to be related to the more informal status of vos and the linguistic influence exerted by the nearby capital, Montevideo, over Rocha. Results indicate that Rocha’s tuteo might not be as stable as traditionally assumed and that a potential shift from tuteo to voseo might take place in the speech of future generations. In Chapter 14, Carolina Barrera Tobón and Rocío Raña-Risso offer a corpusbased study of subject placement in the Spanish of New York City bilinguals. They compare the speech of first- and second-generation bilinguals to show that the latter group uses a more rigid word order, which the authors attribute to the higher level of contact with English. In addition, after combining these data with the rate of use of overt and null pronouns between the Caribbean and Mainland varieties of Spanish spoken in New York City, Barrera Tobón and Raña-Risso conclude that for changes in pronoun placement to occur, there must first be a change in the rate of overt pronoun use. For speakers to begin placing more pronouns preverbally, they first have to use more overt pronouns. Finally, Section VI is devoted to studies on lexical variation in which the authors employ socio-cognitive linguistics, standardization, and accommodation approaches to study various aspects of lexical variation and change. In Chapter  15, David ­Korfhagen offers a socio-cognitive approach to study historical shifts in the prototypical meanings documented for the word afeitar ‘to adorn’ > ‘to apply cosmetics’ > ‘to shave’. The word’s first semantic shift demonstrates an evolving negative attitude from the fifteenth through the seventeenth centuries, as afeitar acquired the meaning ‘to apply cosmetics [excessively]’. Korfhagen posits that the later semantic shift may stem from

 Sandro Sessarego & Fernando Tejedo-Herrero

the prototypical positive meaning that the word historically enjoyed, leading to a more socially accepted meaning of ‘to shave.’ Working within the standardization framework, in Chapter 16 Patricia GiménezEguibar discusses textual evidence that demonstrates the existing linguistic ideologies toward Arabisms in the sixteenth century. The author analyzes these ideologies through the prescriptive attitudes expressed in Guadix’s works (La Primera parte de una recopilación de algunos nombres arábigos), which may have led to the development of lexical stereotypes and eventual suppression of Arabisms over time, as evidenced in Herrera’s use of Arabisms in the early versions of his work, Obra de Agricultura, only later to excise these items in later revisions and editions of the same work. In Chapter 17, Rey Romero focuses on lexical accommodation in Judeo-Spanish in three communities: Istanbul, the Prince Islands, and New York City. Romero shows that the three communities present similar patterns of accommodation to model Peninsular or Latin American Spanish varieties: the adaptation is primarily lexically driven, although phonological adaptation is also found. The author discusses other factors of accommodation in these Judeo-Spanish communities such as group age, degree of contact with other varieties of Spanish, and language attitudes. We would like to conclude by saying that the articles in Spanish Language and Sociolinguistic Analysis represent a significant contribution to the growing field of Spanish sociolinguistics. The well-known scholars, as well as young and promising researchers, whose work makes up this volume, demonstrate the most advanced theoretical and technical approaches. We are honored to have had the opportunity to work with all of them and we hope this volume will provide students and scholars with new insights into the field of sociolinguistics.

References Bayley, R., Greer, K., & Holland, C. (2013). Lexical frequency and syntactic variation: A test of a linguistic hypothesis. University of Pennsylvania Working Papers in Linguistics, 19, 21–30. Erker, D., & Guy, G. R. (2012). The role of lexical frequency in syntactic variability: Variable subject personal pronoun expression in Spanish. Language, 88, 526–557. doi: 10.1353/lan.2012.0050 Harley, H., & Ritter, E. (2002). Person and number in pronouns: A feature-geometric analysis. Language, 78, 482–526.  doi: 10.1353/lan.2002.0158 Linford, B., & Shin, N. L. (2013). Lexical frequency effects on L2 Spanish subject pronoun expression. In J. Cabrelli et al. (Eds.), Selected proceedings of the 16th Hispanic Linguistics Symposium (pp. 175–189). Somerville, MA: Cascadilla Proceedings Project. Milroy, L. (1987). Language and social networks (2nd ed.). Oxford: Basil Blackwell. Sessarego, S. (2014). The Afro-Bolivian Spanish determiner phrase: A microparametric account. Columbus, OH: The Ohio State University Press.

part i

Cutting-edge Methodologies in Sociolinguistics

Quantitative analysis in language variation and change* Sali A. Tagliamonte University of Toronto

The essential goal of variation analysis is to understand the behavior of the dependent variable according to a series of factors, either external (social) or internal (grammatical), (Sankoff, 1988, p. 985). I begin with a brief review of the standard tool, logistic regression using the Varb family of programs. I will also introduce new tools, mixed effects models and condition inference trees relying on the discussions in Baayen (2008), Gries (2009), Johnson (2009). Keywords:  language variation; quantitative analysis; Varb

1.  Distributional analysis The first step in a quantitative analysis is to know how often the variants of a variable occur in a body of data. The second step is to assess the distribution of variants across the range of factors that are thought to condition them – a comparison of marginals (Rand & Sankoff, 1990, p. 4). This provides the frequencies and counts of each variant of the dependent variable in the data, either alone (an overall distribution of forms), or according to independent variables (a factor by factor analysis). A key component of a distributional analysis is to examine the cross-tabulation of factors in order to assess how different factors intersect with one another. Only after the analyst has examined the frequency and distribution of the linguistic variable and has explored and assessed the impact of their intersection, is it appropriate to turn to statistical modeling.

* The editors wish to thank the publisher, Wiley-Blackwell, and the author, Sali A. ­Tagliamonte, for allowing us to use an abbreviated version of Tagliamonte’s Chapter  5 entitled: ­“Quantitative analysis,” included in Variationist Sociolinguistics: Change, Observation, Interpretation, First Edition. Sali Tagliamonte © 2012 Sali A. Tagliamonte. Published 2012 by Blackwell ­Publishing Ltd.

doi 10.1075/ihll.8.01tag © 2016 John Benjamins Publishing Company



Sali A. Tagliamonte

1.1  Statistical modeling Statistical modeling provides a mathematical assessment of the relationship between the dependent variable and the independent variables. Variationists tend to use the variable rule program, which is in essence logistic regression. It is referred to as Varbrul (Cedergren & Sankoff, 1974), Goldvarb 2.0 (Rand & Sankoff, 1990), or Goldvarb X (Sankoff, 1988, p. 6; Sankoff, Tagliamonte & Smith, 2005). The variable rule program is ideal for data that have the following characteristics: (i) alternation between two or more specified sounds, words or structures; (ii) the alternation may seem chaotic or unsystematic when examined cursorily; and (iii) the alternation must recur repeatedly in discourse. Given these conditions statistical inference can be invoked (Sankoff, 1988, p. 2). Logistic regression is ideal for sociolinguistic data where a myriad of factors influence the choice of one variant or another.1

1.2  The three lines of evidence Three types of evidence are relevant for interpreting variation: (i) statistical significance (at the p.05 level), (ii) effect magnitude (strength of factors), and (iii) direction of effect (constraint ranking) (Poplack & Tagliamonte, 2001, p. 92; Tagliamonte, 2002, p. 731). This evidence permits comparisons between analyses, contexts, and groups, depending on how the data is partitioned.

1.2.1  Statistical significance The variable rule program assesses which factors are statistically significant (at the .05 level) and which are not. Both significant factors as well as non-significant factors are important for interpreting the results. 1.2.2  Constraints The variable rule program assigns factor weights or probabilities to each category of the factor groups included in the analysis. The factor groups are tests of independent variables thought to influence the choice of one variant over another. This provides as measure of each of the factors’ contribution to the dependent variable as well as an

.  There is an extensive list of work done in Spanish sociolinguistics using the Varb family of programs. Illustrative examples in the study of phonetic and phonological variation can be found in Brown & Torres Cacoullos (2003); Carvalho (2004); Díaz-Campos, Fafulas, & Gradoville (2010). Aspects of language variation and change in the areas of morpho-syntax and pragmatics can be represented, for instance, by Cameron (1993); Schwenter & Torres Cacoullos (2008); Schwenter (2011); Blas Arroyo & González Martínez (2014), and many of the contributions included in the book edited by Carvalho, Orozco, & Shin (2015) such as Alfaraz (2015), Lastra & Martín Butragueño (2015), and Carvalho & Bessett (2015).



Quantitative analysis in language variation and change

assessment of the factors’ relationships to each other. Constraint ranking is the hierarchy from more to less of the factor weights of categories within a factor group. This is the variable grammar (Poplack & Tagliamonte, 2001, p. 94).

1.2.3  Strength The strength of factors can be assessed using the variable rule program by c­ onsidering: (i) the range in each factor group and (ii) the order of selection of factors in the ­regression analysis. The range is determined by subtracting the lowest factor weight from the highest factor weight in the factor group. When these numbers are compared for each of the factor groups in an analysis, the one with highest number (i.e. range) typically identifies the strongest constraint. The lowest number identifies the weakest constraint, and so forth. The order of selection of factors can be found by looking at the progress of the multiple regression. Which factor group is selected first, second and so forth. The order of selection of factor groups normally mirrors the order of strength of factors as assessed by the range. One line of evidence is not sufficient to make a strong argument although it may provide partial confirmation of a hypothesis. A good argument arises after a course of many different analyses and comparisons performed on the same data set from which an argument is built using the total amount of consistent supportive evidence in the analysis to interpret the findings. In the next section, I will demonstrate variationist analysis by conducting a study of a commonly occurring linguistic variable. The data come from a small city in northeast England, York.

2.  The case study – variable (that) Variable (that) is the alternation between a overt complementizer, that, and its omission, as in (1). (1) a. In the- about 1926 or so, my mother decided that she’d have a … new house. (YRK/r) b. My mother, at the end of the meal, suddenly decided Ø she’d go to … town. (YRK/r)

There are two prevailing explanations for this variation. The zero variant is the result of: (i) grammaticalization of an epistemic phrase; and (ii) an over-riding complexity principle predicting the overt form. Thompson and Mulac (1991) hypothesize more zero with epistemic verbs, e.g. think, 1st person subjects, present tense, simple matrix verb phrases (i.e. no additional elements) and when the subject of the complement clause is a pronoun. This is due to the fact that constructions as in





Sali A. Tagliamonte

(2a–b), are developing their own pragmatic functions and have stopped functioning as complementizers. (2) a. That was the Rock and Roll era that I mean we’d never heard of Rock and Roll. (YRK/e) b. It’s just a sign of the times I think you could call it evolution. (YRK/e) c. We thought we were- you know we’d got the earth with thirty bob. (YRK/e)

Rohdenburg (1998) hypothesizes more zero with pronominal complement subjects and when there is no intervening material between matrix and complement clause. In his explanation, this is because the overt complementizer only appears when the phrase structure is complex. While both theories predict more overt that in certain conditions (e.g. pronominal subjects in the complement clause), only the grammaticalization hypothesis predicts more zero with epistemic verbs. These competing theories can be put to the test using statistical modelling.2

3.  Goldvarb logistic regression Table 1 displays the results of a logistic regression analysis using Goldvarb X. All tokens with tense/aspect morphology other than simple past or present were removed from the data set to facilitate direct comparison across platforms. The characters in square brackets reflect the mnemonic codes used for data analysis in Goldvarb and will reappear later in the chapter in other analyses. The information represented in Table 1 is assembled from different places in the Goldvarb output. The total number of tokens considered, proportions and Ns/cell are found in the comparison of marginals output. The input value and the factor weights come from the output of either a one-step regression or a step-up/step-down regression. The factor weights for the significant factors from the best stepping down iteration of the regression. When reporting non-significant factor weights, these come from the first iteration of the step down. –– –– –– ––

The following information is used to interpret the results: Title of table, with the application value noted All the factor groups in the model, those selected as significant and those not selected as significant with this indicated Total Number of tokens in the analysis (Total N) Corrected mean/input

.  For an analysis of variable que ‘that’ in Spanish see Rodríguez Riccelli (forthcoming).



Quantitative analysis in language variation and change

Table 1.  Logistic regression of the linguistic factors ­conditioning zero ­complementizers in York English Input

.877

Total N

1932 FW

%

N

LEXICAL VERB IN MATRIX CLAUSE Think [t]

.72

95.5

873

Say [S]

.50

75.4

240

Know [K]

.29

67.4

141

Other [O]

.26

61.7

678

Range

46

MATRIX SUBJECT 1st person singular [I]

.62

90.1

1248

NP [O]

.33

58.7

424

Other pronoun [P]

.29

59.2

260

Range

33

ADDITIONAL ELEMENTS IN MATRIX VERB PHRASE Nothing [N]

.56

83.4

1454

Something [A]

.37

65.9

478

Range

19

COMPLEMENT CLAUSE SUBJECT Personal pronoun [P]

.54

74.6

493

Other [O]

.40

80.6

1439

Range

14

INTERVENING MATERIAL None [N]

.54

83.2

1527

Some [Y]

.36

63.7

405

Range

17

VERB TENSE Present [R]

.56

80.4

573

Past [A]

.37

78.1

1359

Range

19

–– –– ––

Listing of each factor, with factor weight (probability) to at least two decimal points Listing of each factor’s proportion and number in cell (N’s) Range for each significant factor group





Sali A. Tagliamonte

The ‘application value’ is the variant the analysis is focused on, in this case zero. The ‘input’ is the overall tendency of the dependent variable. The ‘total N’ is the total number of contexts in the analysis. The N records the number of tokens per cell (the denominator, not the numerator). Each of the factor groups that have been considered in the analysis is listed with the results for each factor. Point-form numbers are ‘factor weights’. These indicate the probability of the dependent variable to occur in that context. The closer these numbers are to 1, the more favouring the effect is; the closer they are to zero the more disfavouring the effect is. The range provides a non-statistical measure of the relative strength of the factor. It is simply a number and should not be reported in point form. The higher this number is, the greater the contribution of that factor to the probability of the form. Good practice also requires the analyst to report the factor groups selected a significant and those that were in the modal but not selected as statistically significant. This is because non-significant factors are an important part of the interpretation. Three lines of evidence are used to interpret the results (Poplack & Tagliamonte, 2001, p. 92; Tagliamonte, 2002, p. 731): (i) statistical significance, i.e. which factors are statistically significant at the .05 level and which are not; (ii) relative strength, i.e. which factor group is most significant (largest range) or least? (smallest range), (iii) What is the order (from more to less) of factors within a linguistic feature? (constraint hierarchy). Finally, bringing in the interpretative component of variation analysis, (iv) Does this order reflect the direction predicted by one or the other of the hypotheses being tested? Each of these bits of information can, and should, be used to build the argumentation about the linguistic variable. Similarities and differences in the significance, ordering of constraints and strength of contextual factors provide a microscopic view of the grammar of the data under investigation, from which the analyst may infer the structure of the grammar. The results are interpreted using all the information in the table taken together (see discussionTagliamonte, 2007, p. 204). The results show that that the overall probability of a zero complementizer is .877 (the Input or Corrected Mean), a high value. Given that this is a binary variable (i.e. that vs. Ø) a disfavouring effect can be interpreted as favouring the alternate variant, i.e. overt that. The total number of tokens considered in the analysis is 1932 (Total N). This is quite a lot of data for a syntactic linguistic feature and sufficient to be relatively confident in the results even though six factor groups are in the model. The Ns/cell are robust in all cases (the lowest is N = 141 for the verb know). The logistic regression has determined that six factor groups are statistically significant at the p  that1.lmer How is the analyst to interpret this output? The list in (7) provides a basic guide (see Hay, 2006a): (7) – Intercept – This is the baseline of the model – The log odds of the dependent variable being one factor rather than the other – Coefficients (log odds) – Comparable to Goldvarb’s factor weights – A positive value is a favouring effect of zero complementrize; a negative value is a disfavouring effect – 0 is neutral (In Goldvarb .5 is neutral)

.  R interprets numbers in a factor group as continuous. There are ways around this. Consult the R literature.



Quantitative analysis in language variation and change 

Table 6.  R, Mixed Effects Model, Individual Random, Age as continuous Data: that AIC BIC logLik deviance 1389 1472 -679.3 1359 Random effects: Groups Name Variance Std.Dev. Indiv (Intercept) 0.41134 0.64136 Number of obs: 1932, groups: Indiv, 33 Fixed effects: Estimate Std. Error z value Pr(>|z|) (Intercept) 2.041509 0.559568 3.648 0.000264 *** Matrix.subjO -1.391238 0.192261 -7.236 4.62e-13 *** Matrix.subjP -1.519005 0.171171 -8.874 < 2e-16 *** Verbs.1OTHER -0.072345 0.236323 -0.306 0.759508 Verbs.1Say 0.935660 0.276523 3.384 0.000715 *** Verbs.1Think 1.894484 0.277216 6.834 8.26e-12 *** Add.elmN 0.844323 0.159095 5.307 1.11e-07 *** Sub.subjP 0.534426 0.159133 3.358 0.000784 *** Int.matY -0.927123 0.168948 -5.488 4.07e-08 *** TenseR 0.739611 0.152387 4.854 1.21e-06 *** Age -0.021039 0.006807 -3.091 0.001997 ** SexM 0.233471 0.300102 0.778 0.436586 Edu+ -0.414057 0.340109 -1.217 0.223443 OccW -0.650085 0.375800 -1.730 0.083653 . –-Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

– Deviance – Comparable to Goldvarb’s log likelihood – A measure of how well the model fits the data – The larger the deviance the worse the fit – The more factor groups the higher the number – Degrees of freedom – A measure of the number of parameters in the model or model complexity – The greater the number, the more complex the model This model shows the relative influence of the social and linguistic factors. The linguistic factors are all highly significant and at relatively high levels, as indicated by the three stars. The p values are high. The social are mostly not significant. Only Age and, at a much lower level, Education exert statistically significant effects.

6.5.2  Checking for interaction – R R (like Rbrul) also provides the analyst a simple and straightforward means to check for interactions. This is a relatively onerous process with Goldvarb; however, in Rbrul

 Sali A. Tagliamonte

and R it is simple. In R simply add a star (or variations thereof (see Baayen, 2008, p. 185) to the same formula, as in (8). For example, let us check for a potential interaction between education and occupation, as in Table 7.

(8) a. > that1.lmer = lmer(Dep.var ~ Verbs.1 + Matrix.subj + Add.elm + Sub.subj + Int.mat + Tense + Age + Sex + Edu * Occ + (1|Indiv), data = that, family = binomial) b. that1.lmer Table 7.  Test of interaction between Education and Occupation Data: that   AIC BIC logLik deviance  1388 1477 -678.1 1356 Random effects: Groups Name Variance Std.Dev. Indiv (Intercept) 0.39844 0.63122 Number of obs: 1932, groups: Indiv, 33 Fixed effects: Estimate Std. Error z value Pr(>|z|) (Intercept) -1.876441 0.568064 -3.303 0.000956 *** Verbs.1OTHER 0.068893 0.236555 0.291 0.770871 Verbs.1Say -0.940434 0.276738 -3.398 0.000678 *** Verbs.1Think -1.901034 0.277359 -6.854 7.18e-12 *** Matrix.subjO 1.389089 0.192070 7.232 4.75e-13 *** Matrix.subjP 1.523593 0.171280 8.895 < 2e-16 *** Add.elmN -0.836869 0.159105 -5.260 1.44e-07 *** Sub.subjP -0.530651 0.158990 -3.338 0.000845 *** Int.matY 0.917904 0.168802 5.438 5.40e-08 *** TenseR -0.737690 0.152363 -4.842 1.29e-06 *** Age 0.020228 0.006753 2.996 0.002738 ** SexM -0.313190 0.302868 -1.034 0.301099 Edu+ -13.390279 811.021183 -0.017 0.986827 OccW 0.502504 0.389378 1.291 0.196867 Edu+:OccW 13.911891 811.021267 0.017 0.986314

The assessment of interaction is shown by the colon, i.e. ‘Edu+:OccW’, bolded. The results show straightforwardly that there is no interaction between sex and education, p = < .986. The process of fitting different models, collapsing factor groups, testing for interactions, re-fitting etc. should continue until the analyst has found the model or models that best account for the variation. Then, the analyst is left to make use of all the bits of evidence to interpret and explain the results in the context of the speech community and in time and space.



Quantitative analysis in language variation and change 

6.5.3  Comparison across tools Table 8 directly compares the results across the analyses for the key factor group Lexical Verb in the Matrix Clause where think, say, tell, know and ‘other’ were tested as a categorical factor group. Table 8.  Comparison of calculations for matrix verb as a 4-way categorical factor group Goldvarb

Rbrul

Rbrul

R

R

FW

FW

LogOdds (sum coding)

Log Odds ­(treatment coding)

P value

Think [t]

.72

.768

1.196

1.89448

8.26e-12 ***

873

Say [S]

.50

.563

0.252

0.93566

0.000715 ***

240

Know [K]

.29

.335

-0.687

default

default

141

Other [O]

.26

.318

-0.762

-0.07235

0.759493

678

Count

This overview shows the different ways of calculating the results and how they were displayed across models. The results are consistent in that think and say highly favour the zero complementizer while the others are much more likely to appear with the overt complementizer. Further the distinct pattern of think over say is clear across the board. The treatment coding in R clearly shows that the category ‘other’ which is a mixed bag of over 60 additional, low frequency, verbs is not significantly different from know (the default). This suggests that know could be grouped with ‘other’. Otherwise, there is not a lot of difference among these analyses.

6.5.4  Conditional inference trees I now turn to a techniques embodied in R that is ideal for exploring the combined effect of multiple factors in complex data sets – conditional inference trees.9 For a more detailed discussion of the utility of these tools for analyzing variation see (Tagliamonte & Baayen, 2012). To use this model, install the Party package and dependencies in R, as in (9). (9) library(party) In a classification tree each factor group is assigned a p-value indicating the level of significance as well as its relationship to other factor groups in the model. Including every factor in the analysis would be too complex and unhelpful. Instead, let us focus on the factors that are most contentious in the literature on the borderline of significance

.  See Johnson and Barnes (2013) for an application of conditional inference trees to the study of variable haya vs. haiga in Mexican Spanish.

 Sali A. Tagliamonte

in the statistical models – sex, age (as continuous), occupation and education. The R code to produce this model is shown in command line in (10). The tree is shown in Figure 1. (10) a. > that14.ctree = ctree(Dep.var ~ Sex + Age + Occ + Edu, data = that) b. > plot(that14.ctree) 1 Occ p < 0.001 {B, S}

W 3 Age p = 0.011 > 68

≤ 68

5 Edu p < 0.001 –

+

6 Age p = 0.002 ≤ 76 Node 2 (n = 480) 1 T

Z

Node 4 (n = 1101) 1 T

Node 7 (n = 106) 1 T

> 76 Node 8 (n = 114) 1 T

Node 9 (n = 131) 1 T

0.8

0.8

0.8

0.8

0.8

0.6

0.6

0.6

0.6

0.6

0.4

0.4

0.4

0.4

0.4

0.2

0.2

0.2

0.2

0.2

0

Z

0

Z

0

Z

0

Z

0

Figure 1.  R, Conditional Inference Tree, Social Factors

Isolating the social factors in this way provides a visually dramatic portrait of the social nature of variable (that) in York. What does this view tell us? First, occupation at the top of the tree is the most important social factor conditioning complementizer choice. Recall that Occupation was selected as significant in the Goldvarb logistic regression (white collar workers, +edu, favoured the zero variant) and was marginally significant in the R mixed effects model in Table 6, p = < 0.1. In the random forest it is much lower down in importance than Age. Second, students, ‘S’



Quantitative analysis in language variation and change 

behave as blue collar workers ‘B’, a result that was also visible in one way or another in previous results. However, after this the results get interesting, at least from a sociolinguistic perspective. The York speech community is highly differentiated based on speaker age. Recall that age has been selected as significant by all the analyses regardless of whether age was run as a continuous or a categorical factor group. The conditional inference tree shows us how age intersects with the other social factors. A white collar worker aged ≤ 68 uses the most zero complementizers. Individuals over 68 use zero complementizers as well, but more so if they are less educated and especially if they are less than 76. In other words, education is a greater predictor of complementizer use among the older people in the community. In contrast, among the younger individuals occupation is the greater predictor of usage. Why would there be a split between the generation over 76 and those under 68? Recall that these data were collected in 1997. Anyone aged 68 in 1997 was 16 in 1945. Therefore the people under 68 are precisely the generations of the post World War II era. The split in the population at age 68, suggests that the zero complementizer was a social indicator for this generation. How does it contribute the explanation for variable (that) in York? The results in Figure 1 suggest that it has undergone a social re-analysis in the last 50 years or so. Where once it marked social groups (as reflected in education level), it has come to be a stylistic option (as reflected in the use of the standard language in one’s job type). It is still not clear how this plays in to the strong linguistic factors influencing this variable or why there has been a steady change towards more zero complementation in the history of English. Since this chapter is not about interpreting or explaining results, I will leave that part of the story of variable (that) for another time and place. The main purpose of the chapter has been to provide an introduction to quantitative approaches to language variation and change, to introduce new statistical tools and to instill confidence and curiosity for the analysis of language variation and change.

7.  Practical advice Let me end this chapter by offering some practical advice. First, set up your data files so that you can analyze it with an expansive toolkit and a set of hand-picked tools. Construct an EXCEL file with all the different types of factor groups you will need to conduct an analyses in Goldvarb, Rbrul or R. In order to make this data file usable across platforms, duplication of factors groups is necessary. For example, if you coded for individual in Goldvarb, you would have been restricted to single character codes. This requires using most of the characters in the courier keyboard. The weirder ones are not readable by Rbrul/R, so this factor group must be re-coded

 Sali A. Tagliamonte

or removed from analysis when the data is transported into R. It is simple enough to code each speaker as their actual name (see listing in 5b); however, then this factor group cannot be read in Goldvarb. Similarly, a Goldvarb token file may contain many tokens with a slash somewhere in the string (due to one factor or another being unknown or missing). These tokens can be analysed in Goldvarb, but not in Rbrul or R. Using EXCEL they can be sorted and deleted for analysis in Rbrul and R but the problem is, this means you will have fewer tokens to analyse. In sum, in order to retain maximum flexibility, you need multiple configurations for the same factor groups in order to use all the toolkits. Second, conduct a thorough distributional analysis of your data and copious crosstabulation of factors. From my perspective at the point of writing, this is simplest in Goldvarb. Third, conduct statistical modelling. Godvarb’s basic fixed effect runs seem to hold up to against the powerful mixed models I have tested here (see comparison in Table 8) at least when the community and individual grammars are parallel. However, the ability to provide statistical validation of this fact using mixed models is satisfying. Rbrul provides factor weights for all factors in a factor group (sum contrasts). This makes comparative studies simpler because it is possible to compare each factor’s level and effect across data sets. R is ideal for complex models and assessing the levels of significance of factor groups. Interactions can be tested easily either in Rbrul or R but I still like to see the data Goldvarb cross-tabulations first. Conditional inference trees are particularly fitting for data exploration and can be used as another prequel for informing logistic regression modelling, whether fixed or mixed. Figure 1 revealed unforeseen patterns of variable (that) in the York speech community. This is compelling testimony for its potential for other variables, contexts and situations. The last step is to interpret and explain the trends and patterns in the data.

8.  Summary The horizon for the quantitative study of language variation and change is bright and exciting. The difficult part will be to find the best way to use all these new tools effectively and to marshall their full potential for interpreting and explaining data. As a new body of findings builds over the next generations, analysts will be ideally armed to tackle the big questions of the field. In my explorations of different tools and practices for quantitative analysis, I have found that the most important thing of all is to understand your data, to be accountable to your data and to make every effort to explain the data. The statistical models are tools. They enable the analyst to accomplish all these goals more effectively. It is up to the analyst to ask the most telling questions, to wisely interpret the patterns, and to craft the best explanations.



Quantitative analysis in language variation and change 

References Alfaraz, G. 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. 5–18). Washington, DC: Georgetown University Press. Baayen, H. (2008). Analyzing linguistic data: A practical introduction to statistics. Cambridge: Cambridge University Press.  doi: 10.1017/CBO9780511801686 Baayen, H., Davidson, D., & Bates, D. (2008). Mixed-effect modeling with crossed random effects for subjects and items. Journal of Memory and Language, 59, 390–412.

doi: 10.1016/j.jml.2007.12.005

Bates, D. (2005a). Fitting linear mixed models in R. R News, 5(1), 27–30. Bates, D. (2005b). lme4: Linear mixed effects models using S4 classes, R package version 9.9975–7. Blas Arroyo, J. L., & González Martínez, J. (2014). ‘¿Qué tengo que/de hazer?’: Variación y cambio lingüístico en el seno de las perífrasis de infinitivo a partir de textos escritos de impronta oral en el español clásico. Studies in Hispanic & Lusophone Linguistics, 7(2), 241–274. Bresnan, J., & Ford, M. (2010). Predicting syntax: Processing dative constructions in American and Australian varieties of English Language, 86(1), 168–213.  doi: 10.1353/lan.0.0189 Brown, E. L., & Cortés-Torres, M. (2012). Syntactic and pragmatic usage of the [estar + Adjective] construction in Puerto Rican Spanish: ¡Está brutal! In K. Geeslin & M. Díaz-Campos (Eds.), Selected Proceedings of the 14th Hispanic Linguistics Symposium (pp. 61–74). Somerville, MA: Cascadilla Proceedings Project. Brown, E. L., & Torres Cacoullos, R. (2003). Spanish /s/: A different story from beginning (initial) to end (final). In R. Núñez-Cedeño, L. López, & R. Cameron (Eds.), A Romance perspective in language knowledge and use. Selected papers from the 31st Linguistic Symposium on Romance Languages (LSRL) (pp. 22–38). Amsterdam: John Benjamins.

doi: 10.1075/cilt.238.05bro

Bullock, B. E., & Toribio, A. J. (2015). An experimental approach to hypercorrection in Dominican Spanish. In R. Klassen, J. M. Liceras & E. Valenzuela (Eds.), Hispanic linguistics at the crossroads. Theoretical linguistics, language acquisition and language contact (pp. ­251–268). Amsterdam: John Benjamins.  doi: 10.1075/ihll.4.13bul Cameron, R. (1993). Ambiguous agreement, functional compensation, and nonspecific tú in the Spanish of San Juan, Puerto Rico and Madrid, Spain. Language Variation and Change, 5(3), 305–334.  doi: 10.1017/S0954394500001526 Carvalho, A. (2004).  I speak like the guys on TV: Palatalization and the urbanization of Uruguayan Portuguese. Language Variation and Change, 16(2), 127–151. Carvalho, A. M., & Bessett, R. M. (2015). Subject pronoun expression in Spanish in contact with Portuguese. In A. M. Carvalho, R. Orozco, & N. L. Shin (Eds.), Subject pronoun expression in Spanish:  A cross-dialectal perspective  (pp. 145–168). Washington, DC:  Georgetown University Press. Cedergren, H. J., & Sankoff, D. (1974). Variable rules: Performance as a statistical reflection of competence. Language, 50(2), 333–355.  doi: 10.2307/412441 Chambers, J. K. (2003). Sociolinguistic theory: Linguistic variation and its social significance. Malden, MA: Blackwell. Cofer, T. (1975). Performance constraints on relative pronoun deletion. Linguistics, 157, 13–32.

 Sali A. Tagliamonte Díaz-Campos, M., Fafulas, S., & Gradoville, M. (2010). Going retro: An analysis of the interplay between socioeconomic class and age in Caracas Spanish. In J. Michnowicz & R. Dodsworth (Eds.), Selected Proceedings of the Fifth Workshop on Spanish Sociolinguistics (pp. 65–78). Somerville, MA: Cascadilla Proceedings Project. Fries, C. C. (1940). American English grammar. New York, NY: Appleton, Century, Crofts. Geeslin, K., & Guijarro-Fuentes, P. (2008). Variation in contemporary Spanish: Linguistic predictors of estar in four cases of language contact. Bilingualism, Language and Cognition, 11, 365–380. 〈http://repository.upenn.edu/pwpl/vol17/iss2/12/〉

doi: 10.1017/S1366728908003593

Gries, S. T. (2009). Statistics for linguistics with R. Berlin: Mouton de Gruyter.

doi: 10.1515/9783110216042

Guy, G. R. (1980). Variation in the group and the individual: The case of final stop deletion. In W. Labov (Ed.), Locating language in time and space (pp. 1–36). New York, NY: Academic Press. Guy, G. R. (1988). Advanced VARBRUL analysis. In K. Ferrara, B. Brown, K. Walters, & J. Baugh (Eds.), Linguistic change and contact (pp.124–136). Austin, TX: Department of Linguistics, University of Texas at Austin. Guy, G. R. (1993). The quantitative analysis of linguistic variation. In D. Preston (Ed.), American dialect research (pp. 223–249). Amsterdam: John Benjamins.  doi: 10.1075/z.68.11guy Hay, J. (2006a). Modelling sociophonetic variation. Journal of Phonetics, 34(4), 405–408.

doi: 10.1016/j.wocn.2005.10.001

Hay, J. (2006b). Spoken Hay, J. (2006b). Spoken syntax: The phonetics of giving a hand in New Zealand English. The Linguistic Review, 23(3), 321–349. Huddleston, R., & Pullum, Geoffrey K. (2002). The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press. Jaeger, F. (2005). Optional that indicates production difficulty: Evidence from disfluencies. Proceedings of DiSS’05, Disfluency in Spontaneous Speech Workshop (pp. 103–109). Aixen‑Provence, France, 10–12 September 2005. 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.

doi: 10.1111/j.1749-818X.2008.00108.x

Johnson, D. E. (2010). Rbrul manual. Retrieved from 〈http://www.danielezrajohnson.com/ Rbrul.R〉 Johnson, M., & Barnes, S. (2013). Haya vs. haiga: An analysis of the variation observed in Mexican Spanish using a mixed effects model. In A. M. Carvalho & S. Beaudrie (Eds.), Selected Proceedings of the Sixth Workshop on Spanish Sociolinguistics (pp. 32–40). Somerville, MA: Cascadilla Proceedings Project. Johnson, M., & Grinstead, J. (2012). Variation in the voseo and tuteo Negative Imperatives in Argentine Spanish. In M. Tamminga (Ed.), University of Pennsylvania Working Papers in Linguistics, 17(2), Article 12. Available at 〈http://repository.upenn.edu/cgi/viewcontent. cgi?article=1198&context=pwpl〉 Labov, W. (1972). Language in the inner city. Philadelphia, PA: University of Pennsylvania Press. Labov, W. (1990). The intersection of sex and social class in the course of linguistic change. Language Variation and Change, 2(2), 205–254.  doi: 10.1017/S0954394500000338 Lamy, D. S. (2015). A sociophonetic analysis of trill production in Panamanian Spanish. In R. Klassen, J. M. Liceras, & E. Valenzuela (Eds.), Hispanic linguistics at the crossroads.



Quantitative analysis in language variation and change 

Theoretical linguistics, language acquisition and language contact (pp. 313–336). Amsterdam: John Benjamins.  doi: 10.1075/ihll.4.16lam Lastra, Y., & Martín Butragueño, P. (2015). Subject pronoun expression in oral Mexican Spanish. In A. M. Carvalho, R. Orozco, & N. L. Shin (Eds.), Subject pronoun expression in Spanish: A cross-dialectal perspective (pp. 41–60). Washington, DC: Georgetown University Press. Medina-Rivera, Antonio. (2011). Variationist Approaches: External factors conditioning variation in Spanish phonology. In M. Díaz-Campos (Ed.), The handbook of Hispanic sociolinguistics (pp. 36–53). Oxford: Wiley-Blackwell.  doi: 10.1002/9781444393446.ch2 Michnowitz, J. (2015). Subject pronoun expression in contact with Maya in Yucatan Spanish. In A. M. Carvalho, R. Orozco, & N. L. Shin (Eds.), Subject pronoun expression in Spanish: A cross-dialectal perspective (pp. 103–122). Washington, DC: Georgetown University Press. Paolillo, J. (2009). Model vs. software in variation analysis. Presented at a Workshop on Using Statistical Tools to Explain Linguistic Variation. Convenor: Sali A. Tagliamonte. New Ways of Analyzing Variation (NWAV) 38 Ottawa, Ontario, Canada. October 22–25, 2009. Pinheiro, J. C., & Bates, D. (2000). Mixed-effects models in S and S-PLUS. New York, NY: Springer.  doi: 10.1007/978-1-4419-0318-1 Poplack, S., & Tagliamonte, S. A. (2001). African American English in the diaspora: Tense and aspect. Malden, MA: Blackwell. Rand, D., & Sankoff, D. (1990). GoldVarb: A variable rule application for the Macintosh. Montreal, Canada: Centre de recherches mathématiques, Université de Montréal. Rodríguez Riccelli, A. (forthcoming). Espero estén todos:The distribution of the null complementizer in two varieties of Spanish. In J. King & S. Sessarego (Eds.), The dynamics of language variation and change: Varieties of Spanish across space and time. Amsterdam: John Benjamins. Rohdenburg, G. (1998). Clausal complementation and cognitive complexity in English. In F.-W.  Neumann & S. Schülting (Eds.), Anglistentag Erfurt (pp. 101–112). Trier: Wissenschaftlicher Verlag. Sankoff, D. (1988). Variable rules. In U. Ammon, N. Dittmar, & K. J. Mattheier (Eds.), Sociolinguistics: An international handbook of the science of language and society Vol. 2 (pp. 984–997). Berlin: Walter de Gruyter. Sankoff, D., Tagliamonte, S. A., & Smith, E. (2005). Goldvarb X. Department of Linguistics, University of Toronto, Toronto, Canada. Retrieved from 〈http://individual.utoronto.ca/ tagliamonte/Goldvarb/GV_index.htm〉 Sankoff, G. (1974). A quantitative paradigm for the study of communicative competence. In R. Bauman & J. Sherzer (Eds.), Explorations in the ethnography of speaking (pp. 18–49). Cambridge: Cambridge University Press. Sankoff, G., Blondeau, H., & Charity, A. (2001). Individual roles in a real-time change: Montreal (r->R) 1947–1995. In R. van Hout & H. vande Velde (Eds.), R-atics: Sociolinguistic, phonetic and phonological characteristics of /r/ (pp.141–157). Brussels: Université Libre de Bruxelles, Institute des Langues Vivantes et de Phonétique. Schwenter, S. A. (2011). Variationist approaches to Spanish morphosyntax. In M. Díaz-Campos (Ed.), The handbook of Hispanic sociolinguistics (pp. 123–147). Oxford: Wiley-Blackwell. Schwenter, S. A., & Torres Cacoullos, R. (2008). Defaults and indeterminacy in temporal grammaticalization: The ‘perfect’ road to perfective. Language Variation and Change, 20(1), 1–39.  doi: 10.1017/S0954394508000057 Storms, G. (1966). That-clauses in Modern English. English Studies, 47, 249–270.

doi: 10.1080/00138386608597258

 Sali A. Tagliamonte Tagliamonte, S. A. (2001). Come/came variation in English dialects. American Speech, 76(1), 42–61.  doi: 10.1215/00031283-76-1-42 Tagliamonte, S. A. (2002). Comparative sociolinguistics. In J. K. Chambers, P. Trudgill, & N.  Schilling-Estes (Eds.), Handbook of language variation and change (pp. 729–763). Malden, MA: Blackwell. Tagliamonte, S. A. (2006). Analysing sociolinguistic variation. Cambridge: Cambridge University Press.  doi: 10.1017/CBO9780511801624 Tagliamonte, S. A. (2007). Quantitative analysis. In R. Bayley & L. Ceil (Eds.), Sociolinguistic variation: Theory, methods, and applications, Dedicated to Walt Wolfram (pp. 190–214). Cambridge: Cambridge University Press.  doi: 10.1017/CBO9780511619496.011 Tagliamonte, S. A. (2012). Variationist Sociolinguistics: Change, observation, interpretation. Malden, MA: Wiley-Blackwell. Tagliamonte, S. A., & Baayen, R. H. (2012). Models, forests and trees of York English: Was/were variation as a case study for statistical practice. Language Variation and Change, 24(2), 135–178.  doi: 10.1017/S0954394512000129 Tagliamonte, S. A., & Smith, J. (2005). No momentary fancy! The zero ‘complementizer’ in English dialects. English Language and Linguistics 9(2), 1–21.  doi: 10.1017/S1360674305001644 Team, R Development Core (2007). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Retrieved from 〈http://www.Rproject.org〉 Thompson, S., & Mulac, A. (1991). The discourse conditions for the use of the complementizer that in conversational English. Journal of Pragmatics, 15, 237–251.

doi: 10.1016/0378-2166(91)90012-M

Torres-Cacoullos, R., & Walker, J. A. (2009). On the persistence of grammar in discourse formulas: A variaionist study of that. Linguistics, 47(1), 1–43.  doi: 10.1515/LING.2009.001 Wolfram, W. (1993). Identifying and interpreting variables. In D. R. Preston (Ed.), American dialect research (pp. 193–221). Amsterdam: John Benjamins.  doi: 10.1075/z.68.10wol Young, R., & Bayley, R. (1996). VARBRUL analysis for second language acquisition research. In R. Bayley & D. R. Preston (Eds.), Second language acquisition and linguistic variation (pp. 253–306). Amsterdam: John Benjamins.  doi: 10.1075/sibil.10.11you

Combining population genetics (DNA) with historical linguistics On the African origins of Latin America’s black and mulatto populations* Armin Schwegler

University of California, Irvine The purpose of this chapter is to explore the African provenience of some of Latin America’s Black inhabitants and to demonstrate how recent advances in research have made it possible to determine their ancestors’ origin with a high degree of specificity. To that end, the ancestry of two specific (and admittedly unusual) communities will be examined from a linguistic perspective: (1) the maroon village of Palenque (Colombia), and (2) ritual “families” of Palo Monte (Cuba). Part 2 of this chapter then sets these findings in dialogue with the latest investigations regarding the population genetics (DNA) of Palenque and over 40 sub-Saharan ethnolinguistic communities. The combined results of this interdisciplinary DNA and linguistic research point to a monogenetic hypothesis that places Bakongo slaves from a small region (Mayombe, also known as Yombe) in Western Central Africa at the center of Palenque’s and Palo Monte’s foundational story. Keywords:  African origins; population genetics; Chewa (Malawi); Colombia; Cuba; Loango; Kongo; Mayombe; Yombe; Vili; Palenque; Palenquero; Palo Monte; slave trade

*  Keynote address delivered at the 7th International Workshop on Spanish Sociolinguistics (WSS7), University of Wisconsin-Madison, April 4, 2014. Anna Hundt Golden kindly made expert editorial comments to an earlier version of this paper. I thank my colleagues and/or friends Naser Ansari-Pour, Miguel Gutiérrez Mate, John McWhorter, Yves Moñino, Danae M. Pérez, Brigitte Pakendorf, Steven Paas, Constanza RojasPrimus, Hiram Smith, and two anonymous reviewers for insightful comments to an earlier version of this paper. This study is dedicated to the pioneers of Palenquero studies: Aquiles Escalante, Germán de Granda, and Derek Bickerton.

doi 10.1075/ihll.8.02sch © 2016 John Benjamins Publishing Company

 Armin Schwegler

1.  Introduction This chapter explores the African provenience of some of Latin America’s Black (and mulatto) inhabitants, and how recent advances in DNA and linguistic research have made it possible to determine their ancestors’ origin with a high degree of specificity. Due in part to the complexity and long duration of the transatlantic slave trade, it had always been assumed that the provenance of Afro-Hispanic communities must have originated in multiple and very distant sub-Saharan regions. Because of the supposed heavy mixing of diverse ethnolinguistic groups in the Americas, scholars also presupposed that research surrounding the origins of an ethnic group based on linguistic (predominantly etymological) work was inherently speculative and, therefore, insufficiently conclusive to prove highly specific African ancestry for any New World population.

Palenque Department of Bolivar

Cartagena

Bogota

COLOMBIA

Map 1.  Palenque, Colombia (Department of Bolivar)

These long-held assumptions were as prudent as they were logical (I myself espoused them for over a decade), but in at least two instances – those of Palenque, Colombia (Maps 1 and 3) and Cuba – they were also thoroughly erroneous. As this study will demonstrate, Palenqueros along with the early adherents to Cuba’s Palo Monte ritual practice (still in existence today)1 formed surprisingly homogeneous .  See Palo Mayombe: The Garden of Blood and Bones by Nicholaj de Mattos Frisvold (2011) for a recent initiate’s account of this much maligned religion whose central nigromantic



Combining population genetics (DNA) with historical linguistics 

AFRICA REPUBLIC OF CONGO

Mayombe

Map 2.  Mayombe, Republic of Congo, where a Kikongo variety known as “Kiyombe” is s­ poken

e­ thnolinguistic communities whose roots lie in a small tropical region just north of the Old Kongo Kingdom, more specifically in an area known as “Mayombe” (Republic of the Congo, see Map 2; more detailed regional maps will be given in subsequent sections of this study).2 To properly contextualize the information offered in the following sections and to provide readers with a partial history of the discipline, this paper will revisit the early work by Escalante (1979), Granda (1971), Bickerton and Escalante (1970), and Del Castillo (1982, 1984, 1985) and how it sought to uncover Palenqueros’ African roots. Additionally, this paper will explore how subsequent linguistic investigations and triangulation with data from the secret ritual language of Cuba (Palo Monte) has gradually changed scholars’ thinking, eventually favoring the monogenetic origins hypothesis I espouse in this paper. Due to limitations of space, some of this scholastic

mystery is the prenda, a cauldron containing (inter alia) human skull or bones. Because of practitioners’ presumed supernatural powers or use of magic for evil purposes, the practice is sometimes referred to by outsiders as “Black Magic”. .  For the first extensive description of the Yombe (or Kiyombe) language (spoken in the Mayombe region), see Bittremieux (1922). Given the historical importance of ritual languages in the context of the present chapter, an equally relevant source is Bittremieux’s La Société secrète des Bakhimba au Mayombe (1936).

 Armin Schwegler

excursion into the “African” history of Palenque and the Palo Monte tradition will be superficial at best. However, the use of pertinent references to additional linguistic, historical, and/or anthropological sources will orient readers interested in learning more about the complex histories of two communities that have begun to fascinate a growing number of scholars. Part 2 of this chapter then relates these findings with the latest investigations into populations genetics (DNA), carried out over the last five years in Palenque and in over 40 sub-Saharan ethnolinguistic communities. The combined results of this genetic research and the aforementioned linguistic studies confirm the single origin hypothesis: ethnolinguistically cohesive groups of Kikongo-speaking slaves (from the Old Kongo and the Mayombe in particular) must have founded the communities of Palenque (Colombia) as well as Palo Monte (Cuba).3

2.  Palenque: Location and brief description Described extensively in recent scholarly literature (see Maglia & Schwegler [2012] and relevant sources cited therein), Palenque is a small rural village with approximately 4,000 inhabitants. It is located about 60 kilometers outside of Cartagena de Indias (Map  3). The village can be reached with relative ease by public bus in 2–3 hours (the final 6 kilometers are best traveled on hired motorbike taxis, generally driven by Palenquero youth). The community commands an extraordinary position within Latin American social and linguistic history: it harbors Latin America’s genotypically blackest population with sub-Saharan ancestry (Noguera, Schwegler, & Gusmão et al., 2014), and it is the only place in the entire South American mainland to feature a Spanish-based creole (locally known as Lengua lit. ‘tongue’, otherwise referred to in the literature as “Palenquero”). Due in part to its prolonged isolation from the rest of Colombian society until the 1980s or 1990s, Palenque has maintained a set of unique cultural traditions with deep African roots (Schwegler, 1992, 2006a, 2007).1 For example, Palenqueros have been singing Lumbalú funeral chants (of which more below) for centuries, a tradition passed down the generations until practitioners eventually forgot the literal meaning of some of their celebrated “African” liturgical texts. During the first half of the 17th century (and especially from 1595 to 1640), Cartagena was Latin America’s principal slave trade center. Around 1650, Silva Solís ­(a resident of Seville) estimated that the population of Spanish America included 329,000 .  For an introductory (albeit somewhat dated) examination of the relationship between human genetic variation, geographic distribution or “structuring,” and “race,” refer to Jorde and Wooding (2004).



Combining population genetics (DNA) with historical linguistics 

Atlántico

Barranquilla Cartagena

á

m na Pa PALENQUE

Pacífico Bogotá Cali

COLOMBIA

Brasil

Perú 300 km

Yves Moñino & Jeanne Zemer

Ecuador

Map adopted from Moñino & Schwegler (2002: vii). Photos by author.

Map 3.  Location of Palenque and Cartagena

African slaves. Two thirds (or about 200,000) of these slaves were located in South America, and most had been imported via Cartagena (Newson & Minchin,  2007b, p. 6). Cartagena was thus “not only the region’s most active slave market but also the main legal entry point for slaves for all destinations in South America” (Newson & Minchin, 2007b, pp. 136–137). For an extensive list of relevant works surrounding the transatlantic slave trade, consult Schwegler (2014b). Within that list, Newson (2004)

 Armin Schwegler

and Newson & Minchin (2007a, 2007b) are especially relevant regarding the provenance of Cartagena’s African slaves. The ancestors of the Palenqueros managed to escape from Cartagena to the tropical and (at the time) fairly densely forested hinterland of the Costa Caribeña. As revealed recently (Navarrete, 2003, and especially 2011, 2012; Schwegler, 2014b), there they formed multiple palenques,4 and eventually regrouped to form the community of San Basilio de Palenque (first known as San Miguel Arcángel) around the middle of the 17th century. By January of 1713, the population of the village consisted of 234 maroons, which was split into nearly equal numbers of Creoles and Bozales5 (for details see Navarrete [2008, p. 70] and Schwegler [2012]). It remains debatable whether the creole language of Palenque formed in situ, or was brought there by escaped slaves from other heavily multilingual localities where the creole may have been utilized for interethnic communication (as explained in section “6. Conclusions” below, the findings of this paper favor the second scenario). It is similarly unclear whether early Palenqueros used one or perhaps even several African languages in their daily discourse (but regarding this point, see “6. Conclusions” below). Heavy and widespread stigmatization of local traditions (language included) from both outside and within the community produced the progressive abandonment of most local traditions as early as the middle of the 20th century. As a result, Lengua, the Lumbalú, and other “exotic” manifestations of local culture (including dances with a patently “African” flavor) came to be viewed as antiquated and a principal cause of Palenque’s supposed “backwardness” and disconnect from modern life. Consequently, generational behavior shifted in an attempt to assimilate to cultural and linguistic patterns of the “more advanced” surrounding communities, Cartagena included. By 1990, especially among Palenquero youth, abandonment of the local creole had become a predictable accompaniment of this frenzied attempt at modernization. Creole language death, it seemed, was both imminent and unavoidable.

.  Palenques were [wooden] palisades or stockades used for defensive purposes. The term is related to Span. palo ‘stick, branch,’ and may have entered the Spanish language as a borrowing from Occitan (perhaps via Catalan). For details, consult Corominas & Pascual (1985, s.v. palo). .  Bozales are unacculturated first-generation Africans in the Americas. Creoles are their American-born descendants, and anyone else born to non-Amerindian parents during ­colonial times. Gómez is correct in pointing out that “in the seventeenth century Spanish Caribbean, blackness was not a proxy for a specific social stratum, no[r] did it carry the same kinds of racialised connotations it would assume decades later in most of the Atlantic world” (2014, p. 386). It follows, therefore, that then, as now, Palenque was characterized by at least some social stratification.



Combining population genetics (DNA) with historical linguistics 

However, around the turn of the millennium, a rather dramatic and unexpected shift redefined this situation of inevitable cultural attrition, in turn giving Lengua and local “African” traditions a new lifeline (Schwegler, 2011b, 2011c). Growing self-awareness and outward sociocultural and political projection, coupled with a national movement of negritud (Black awareness), raised public awareness about Palenque both in Colombia and on an international scale. A community that just a decade earlier was widely ignored by Cartagena’s academic and political establishment now garnered special attention from international entities such as the UNESCO (which in 2005 proclaimed Palenque a Masterpiece of the Oral and Intangible Heritage of Humanity) and global leaders that include African American President Barack Obama, who met with Palenque community leaders in Cartagena in April of 2012. The net result of the recently-attained fame is that, within Palenque, “African” roots and “African history” are now held in high esteem, so much so that heroic foundational and other myths about “African” maroon ancestors are routinely told to academic and other tourists who visit Palenque on an almost daily basis (on this point, see Schwegler, 2012, pp. 115–118). One byproduct of this new societal behavior is a rapidly growing local, national, and international interest in the true African origins of the Palenque community.

3.  Palenque and its African origins 3.1  Background I first arrived in Palenque in 1985 when it was still a rural village with no running water and only very intermittent electricity in some homes. Despite an early booklength work by Escalante (1979) and the subsequent publications by Granda (1968), Bickerton & Escalante (1970), Lewis (1970), Friedemann & Patiño Rosselli (1983), and Megenney (1986), prior to the mid-1990s the outside world almost completely ignored Palenque, and with it the unusual “African” traditions and unique situation of Palenquero creole/Spanish bilingualism, the product of almost four hundred years of geographic and social isolation. As the first white person to live in Palenque for extended periods of time, I learned to speak the local creole with fluency, and I also began to engage in linguistic and anthropological research with the ultimate goal of uncovering Palenqueros’ distant “African roots” (see, for instance Schwegler, 1989, 1990, 1992, 2002; Schwegler, 2012 lists these and many of my subsequent works on Palenque). As explained in Schwegler (2011a), my multiple extended stays in Palenque (1985 to 1996, and 2008, 2009, 2011) confirmed the supposition that prior to the

 Armin Schwegler

early arrival of scholars in the 1950s (and then again in the 1970s and 1980s)6, Palenqueros had no collective memory of the slave trade, nor of any historical events predating the late 1800s. As a result, none of Palenque’s oral transmission of local history proved helpful in narrowing down the ethnic or linguistic provenience of the community’s founding fathers. When asked explicitly about their ancestors’ geographic roots, most interviewees either offered no answer at all, or explicitly stated their ignorance on the matter. However, at the same time, virtually all elder Palenqueros were adamant about their supposedly “pure African genetic background.” Adopting an outwardly self-righteous posture, a Palenquero octogenarian, for instance, proudly offered me the following representative opinion on the matter: “¡ma hende ri Palengue nunka a hundá ku uto nu! Suto é ri Afrika memo!” (‘we Palenqueros have never mixed with anyone else! We are [straight] from Africa’ – fieldwork data 1988). As seen from their perspective, this and similar opinions regarding the community’s ethnic and cultural purity seemed entirely logical: then, as now, Palenqueros were – with very few exceptions – all Black, and their unusually dark skin color made them stand out even against the darkest of citizens of nearby Cartagena. However, as my recent collaboration with population geneticists has revealed (Noguera, Schwegler, & Gusmão et al., 2014), the social and genetic history of Palenqueros must have been considerably more heterogeneous than Palenqueros have assumed; our results indicate that some Palenquero DNA is almost one third ­Caucasian, while Amerindian genetic contributions are almost non-existent (less than 2%).7 Presently it remains unclear whether this sub-Saharan and European

.  The first extended scholarly visit to Palenque was that of Colombian anthropologist Aquiles Escalante in the early 1950s (see Escalante, 1954 [1979]). Starting around 1970, other scholars also visited Palenque, but their fieldwork in the community was typically limited to day trips. An exception to this general rule were Nina de Friedemann, Richard Cross and Carlos Patiño Rosselli, who repeatedly stayed in Palenque during the mid- to late 1970s (see Friedemann & Cross [1979] for a photographic record of these visits). I followed in their footsteps in the mid-1980s, and became the first outsider to speak their Lengua with fluency. In the 1990s, linguists Thomas Morton and Yves Moñino also gained fluency in the creole (for publications resulting from their fieldwork, see the reference list at the end of this chapter). .  From a historical perspective, the low Amerindian contribution is not unexpected. To date, not a single word of local Amerindian origin has been identified in the Palenquero creole. Local oral history dating back to approximately 1900 also suggests that intermarriage with nearby Amerindian or mestizo populations (e.g., San Cayetano) was practically non-­existent, in spite of the fact that on certain specific holidays, drumming festivals briefly brought the two communities into close social contact on a regular basis (fieldwork data, 1988). Rodas, Gelvez, and Keyeux (2003) had reported a higher Amerindian contribution for Palenque but subsequent studies did not confirm these results. Regardless, even these authors



Combining population genetics (DNA) with historical linguistics 

admixture occurred in colonial Africa, during the Middle Passage, in early C ­ artagena (prior to the escape of the maroons into the hinterland), or after Palenque was established. Taking into consideration a number of diverse factors too complicated to detail here, my current thinking suggests that much if not all of this interracial contact took place before the founding of Palenque. Recent and ongoing advances in DNA research (including the analysis of helicobacter pylori, a stomach bacterium whose DNA sequencing is very useful for population genetic research)8 promise to shed light on this question in the future.

3.2  S cholarship into Palenque’s African past: Early assumptions and a priori limitations Due in part to the profound lack of historical information surrounding the formative period of Palenque, early scholars of Palenquero language and culture faced an unusually daunting task: to reconstruct the African ancestry of slaves that successfully escaped 17th-century Cartagena de Indias, an exceptionally multilingual city where dozens of different sub-Saharan languages circulated on any given day (cp. Sandoval’s 1627 eyewitness testimony, as well as recent detailed work by historians like Gómez Zuluaga [2010], Newson & Minchin [2007a, 2007b], and Wheat [2009, 2011, forthcoming] about the social life of colonial Cartagena). Faced with a colonial history of extreme racial and linguistic admixture, and potentially confronted with African remnants (especially vocabulary) that could have originated from dozens of poorly or altogether undocumented sub-Saharan languages, early linguists and anthropologists were naturally reticent to postulate narrowly defined hypotheses about Palenqueros’ putative origin. At the time, etymological work on Palenque’s remaining lexical Africanisms seemed conjectural at best, and there was no reason to assume that its population, much like that of Cartagena, was anything other than the result of extreme convergence. Furthermore, there was early and consistent consensus among specialists on Afro-American languages and cultures that sub-Saharan words and/or linguistic features were likely to have multiple origins, and therefore attempting to isolate a specific ethnolinguistic group as the sole potential donor to an Afro-American community was by definition ill advised. Consequently, it was argued more than once that it is generally misguided to dwell on a specific African language when a feature is areal or shared by many of the African languages (for research regarding this ­perspective, concluded that “Palenque de San Basilio is one of the populations with the lowest frequencies of Amerindian mtDNA lineages (Table 3)” (2003, p. 23). .  For a brief description of this bacterium and its usefulness in genetic research, visit 〈http://en.wikipedia.org/wiki/Helicobacter_pylori〉.

 Armin Schwegler

see Huttar [1993], a creolist and scholar intimately familiar with African languages and Afro-American language history). Etymological vagueness thus seemed not only justifiable and advisable but also inescapable. Yet, basing my position on phonetic and etymological data obtained during my fieldwork in Colombia and in other regions, by around the year 2000 I became convinced that conventional thinking needed revision, and that “it is logical to infer that Western Kikongo (with its northern and southern branches) was especially instrumental in the formation of Palenquero” (­Schwegler, 2000a, p. 281). Since scholars also widely felt that sound research on Africanisms could progress successfully only if combined with careful anthropological fieldwork and historical research (Mufwene, 1993, p. 6, Introduction to Africanisms in Afro-American Language Varieties), I engaged further in such work and also shifted my main attention to Cuba in 1996 (see Section 3.4 below).

3.3  Palenque’s African past: First attempts at narrowing down origins The first in-depth study of Palenque was by the Colombian anthropologist Escalante  (1979), an astute and early observer of Caribbean coastal language and culture. Upon visiting Palenque in January of 1953, the village had 1,500 residents, all of whom were Black (Escalante, 1979, p. 28). The number of non-Palenqueros living in the village at that time was “extraordinarily low (10 or less)” (p. 29). This fact, together with Palenque’s observed geographic and social isolation, led Escalante to the conclusion that exogamy (intermarriage with outsiders) had historically been extremely rare. In addition to describing Palenqueros’ quotidian lives and local culture, Escalante took special interest in local traditions that seemed to have African roots. Among these were religious beliefs and practices, medicinal remedies, and an unusual type of funeral dirges called “Lumbalú”. These “strange” and (at the time) mostly impenetrable texts seemed “of great importance because they preserve the language originally used by the founding fathers of the rebel community” (Escalante, 1979, p. 78, my translation). As shown by Schwegler (1996), this assumption was proved erroneous, as Lumbalú chants are mostly a composite of contemporary creole and Spanish (often code-switched), routinely peppered with isolated African ritual jargon. Nonetheless, as I will demonstrate shortly, Lumbalú texts did play a major role in scholars’ search for Palenque’s African roots. Curiously, in his groundbreaking work, Escalante (1979) did not express an opinion about Palenqueros’ putative African ancestry. However, in a chapter of El negro en Colombia (1964) published a decade later, Escalante briefly discussed the role of Bantu peoples in the transatlantic slave trade. In this context, on page 100 he reproduced a Lumbalú song from his 1954 work, adding that “its lyrics (notably the words Congo,



Combining population genetics (DNA) with historical linguistics 

Luango, and Angola) may shed some light on the putative origins of its inhabitants” (my translation).9,10 Chi man congo Chi man luango Chi man ri luango de Angole. Eee calunga lunga manquisé Gombe manciale Yansú melacó Arió negro congo, chimbumbe.

Belief in a possible connection between Palenque and Congo/Angola further developed when Spanish linguist Germán de Granda, a creolist intimately familiar with central West Africa and the slave trade, published the first study dedicated solely to the African origins of Palenque’s creole language (see Granda, 1971). Granda accepted Escalante and Bickerton’s premise (based on external rather than internal linguistic data) that Bantu-speaking slaves from the Congo/Angola region constituted Palenque’s main substrate. He further solidified the hypothesis in two important ways: first, he offered pertinent historical information about the Portuguese slave trade by connecting the Congo/Angola region with Cartagena (in this context, Granda’s reference to Sandoval’s eyewitness account of slavery in Colombian port city was particularly relevant). Second, he specified (Granda, 1971, p. 88) that Kikongo (Map 4) must have been the principal source for Palenquero words (such as moná ‘child, kid, offspring’, ngombe ‘cow/ox’ nguba ‘groundnut’, etc.) rather than other Bantu languages. Although Granda’s corpus of less than a dozen Palenquero Africanisms was minute and his proposed etymologies not always correct (Schwegler, 2000a, 2002), his overall analysis appeared sound, thereby providing scholars with a working hypothesis on which to build.

.  The original wording reads: “Un canto funerario recogido por nosotros en la comunidad negra del Palenque de San Basilio […] contribuye a esclarecer en parte el origen de sus moradores” (Escalante, 1964, p. 100). Escalante did not attempt a translation of this song. Two pages later (p. 102), Escalante further expands on the usefulness of Lumbalú texts to investigate Palenqueros’ origins. In doing so, he refers to the Ganguelas of Angola, a group/community whose belief in certain spirits (according to Escalante) resembles that of the Palenqueros. The author refers to the mention of Calunga in one of the Lumbalúes he recorded, and he then correctly identifies it as a supreme deity among Angolan tribes. .  In examining the Lumbalú reproduced here, readers should keep in mind that the text seems “African” (and, therefore, “exotic” and unintelligible) even to native speakers of Palenquero and/or Spanish. Section 3.3 below will offer a translation as well as an improved transcription.

 Armin Schwegler

AFRICA

Kikongo

Bantu area Map 4.  Bantu area and approximate location of Kikongo

3.4  Palenque’s African past: The 1980s to mid-1990s The decade of the 1980s marked the beginning of creolistics as a separate and wellorganized sub-discipline of linguistics. The number of self-proclaimed creolists was still relatively small (50–100), but evidence for growth spurts of the emerging field could readily be found. For example, the period marked the founding of the Society for Pidgin and Creole Linguistics (SPCL) in 1987, as well as the launching of the Journal of Pidgin and Creole Languages (published bi-annually, now in its 30th year). At the same time, dialectologists and general linguists with an interest in Latin America (both the Spanish- and the Portuguese-speaking territories) essentially ignored pidgin or creole languages (for example, Bozal Spanish, Palenquero, Papiamento, Angolar, Principense, Sãotomense, and so on), and along with it the burgeoning field of contact linguistics that accompanied it. However, there were a few exceptions to this general lack of interest among Hispanists. Most notable among these were the Colombian scholars Nina de Friedemann (an anthropologist) and Carlos Patiño Rosselli (a linguist), who in 1983 coauthored the first book-length monograph on the language and culture of Palenque. Initially, within Colombia, their work did not attract the attention it deserved,



Combining population genetics (DNA) with historical linguistics 

but over time it nonetheless proved a game changer. It was groundbreaking for at least two reasons: (a) first, in addition to featuring a useful grammatical sketch of Palenquero, it also offered linguists worldwide the first corpus of genuine spoken Lengua in the form of an Appendix;11 and second, (b) even though Patiño Rosselli essentially ignored Palenque’s linguistic sub-Saharan legacy (for instance, he did not address the question of African roots nor engage in a systematic search for (archaic) African vocabulary), he nevertheless profoundly contributed in clarifying the true nature of Lengua. Because of his scholarship, we now know that Palenquero – in lexical terms – is far less “African” than Bickerton, Granda, and others pioneers had originally assumed it to be. This is in part because everyday Lengua makes use of less than two or three dozen words of African stock; more importantly perhaps, many of the presumed Africanisms in the Lumbalúes proved to be deceptive transcriptions of creole or Spanish words. Megenney’s El palenquero: un lenguaje post-criollo de Colombia (1986) differs greatly from that of Patiño Rosselli’s research in that it focuses on Palenque’s putative African origins (for example, an entire chapter [chap. 7] is devoted to lexemes with possible African stock). As explained in Schwegler (1999, 2000a, 2002), of the 43 s­ upposed Africanisms, only a dozen or so actually have African roots, and even in these cases, the author’s etymological arguments are often questionable at best.12 Markedly different from earlier publications on the Palenquero lexicon is Del Castillo’s “El léxico negro-africano de San Basilio de Palenque” (1984; see also 1985).13 This in-depth study unites for the first time a substantial corpus of over 100 AfroPalenquero words. For many of these, Del Castillo offers plausible etymologies, the majority of which are traced to Bantu languages (especially Kikongo and Kimbundu). By the time the Palenqueros welcomed me into their community in 1985, the previously mentioned gradual shift to more western models of behavior had become more accelerated, but had not yet completely erased their memory of distant African roots. Words such as “Africa,” “Kongo,” or “Angola” routinely elicited emotional responses with reference to their ancestors’ heroic past, and along with them, reinforced local cultural practices associated with the celebration of the Lumbalú, then Palenque’s most outwardly “African tradition.” However, by the mid-1980s, Palenqueros could no longer attach concrete semantic reference to such terminology. To them, the Lumbalúes

.  This corpus attracted the attention of linguists like me or, a few years later, Yves Moñino (CNRS, Paris), who then became fascinated by Palenque’s legacy. .  The same is true of Megenney (1976, 1980, 1996), and of Escalante’s (1989) examination of a small number of Palenquero ritual words. .  Also of interest in this context is Del Castillo’s “El aporte negro-africano al léxico de Colombia” (1992).

 Armin Schwegler

had essentially become a distant, vague, and yet culturally still very meaningful echo of a sub-Saharan legacy, anchored above all by its “African” rhythms (drums), accompanying dances, and occasional moments of trance. In a race to document Palenque’s rapidly disappearing African legacy, and convinced that still existent Lumbalúes may hold further clues regarding the community’s sub-Saharan roots, I embarked on fieldwork in the late 1980s and early 1990s that culminated in the publication of “Chi ma nkongo”: lengua y rito ancestrales en El Palenque de San Basilio (Schwegler, 1996). In this monograph I collected, transcribed, and analyzed dozens of ritual chants, and in doing so, continued the etymological reexamination of previously published Afro-Palenquerisms, adding numerous previously undocumented Palenquero ritual glossalia to the collection of putative Africanisms. Four main findings emerged from this work. First, on the whole, isolated ritual expressions – often of demonstrable Kikongo origin – are xenoglossic to the Palenquero public (generally speaking, the more mysterious and incomprehensible these ecstatic vocalizations are, the greater their perceived power and authenticity [i.e., Africanicity] is thought to be). Second, the ritual language – essentially modern Palenquero and Spanish – is a vernacular that is infused with obsolete or archaic words that have been preserved for years by a recognized group of musicians. Perhaps most important in addressing the Palenque’s origins question is the fact that none of the songs contain genuine, extended segments of African language. This is to say that contrary to the suppositions of Bickerton, Escalante and other pioneers of Palenquero studies, the ritual chants in question do not, in sensu strictu, proffer more than fragmentary bits and pieces of African language, be that of Kikongo or any other sub-Saharan vernacular that circulated in Cartagena during the height of the slave trade. As such, Lumbalúes cannot be taken to represent the early speech of Palenqueros. Finally, and importantly, careful research into contemporary Lumbalúes reveals that they can be deciphered and translated, and undoubtedly are remnants of ancestral Palenque; for this and other reasons, they are especially well suited for the recovery of language, orature, and origins. There is another point regarding the centrality of Lumbalúes to the “roots question” that deserves mention here. Of all the ritual songs collected in Palenque to date, only two (or possibly four) make explicit mention of African toponyms or ethnic groups (Map 5 shows the location of the territories or ethnic groups mentioned). It is important to note that these Palenquero chants as a whole point to areas where Kikongo was (and continues to be) spoken. Of course, a lack of contradictory evidence (in the form of additional geolinguistic data in Lumbalúes) may constitute no evidence at all, as we must suppose that Palenquero ritual chants with reference to non-Kikongo regions could simply have been lost over time. However, in this particular case, the exclusive mention of terminology referencing the Old Kongo alone may not be an accident of history, and therefore likely holds special significance. This I argue because additional recent linguistic and especially DNA evidence (see Section 5) s­ upports the



Combining population genetics (DNA) with historical linguistics 

L

O

Loango

A N

Kinshasa

Pointe Noire

Loango

G Malemba O

Cabinda

Africa

REPUBLIC OF THE CONGO

Boma

Suku

KONGO

Kingdom ANGOLA Luanda

Congo River

DEMOCRATIC REPUBLIC OF THE CONGO São Salvador/ Mbanza Kongo (Kongo capital)

(ANGOLA)

Map 5.  Ethnic groups and toponyms mentioned in Lumbalúes recorded to date As the travel account by Degranpré (1801, pp. 72, 82) shows, during the slave trade, “Angola” had a wider geographic meaning than it does today, and included the Loango Coast. Old ritual expressions like Chi ma ri luango di angola are thus perhaps best translated as ‘From the Loango [people] of the West Central African Coast [we are]”. For a Lumbalú referencing to the Suku group, see Schwegler (1996, p. 604). For Malemba, see Schwegler (1996, p. 626), where I did not yet identify the term as an ethnonym. At present, it is not entirely clear whether this is indeed its proper etymology.

hypothesis that Palenqueros’ ancestors originated exclusively from the Old Kongo and/or from an immediately adjacent region, which is perhaps implied by Palenqueros’ most cherished Lumbalú, repeated hereafter with a translation:14 chi ma nkongo

From the Kongo

Chi ma nkongo, From the Kongo people [I am], Chi ma luango, From the Loango people [I am], Chi ma ri Luango di Angola e; From the Luango people of Angola [I am], eh; Huan Gungú me ñamo yo; Juan Gungú is my name; Huan Gungú me a de nyamá, ee. Juan Gungú I shall be called, eeh. (Schwegler, 1996, pp. 524–525; slightly modified)

.  The version given here is accompanied by a translation (first proposed and analyzed in extenso in Schwegler (1996, pp. 524–537). It has since been confirmed by Bunseki Fu Kiau, native Congolese expert of old Kongo ritual traditions. Readers interested in viewing a recent 2013 performance of Chi ma nKongo (recorded in its natural setting) can visit 〈http://www.youtube.com/watch?v=ZS57wkyqbM0〉. The lead singer is Graciela Salgado Valdés, a member of the Batata clan, Palenque’s principal family of musicians. A photo of the “old Batata” and his large Lumbalú drum can be found in preface to Escalante (1954 [1979]).

 Armin Schwegler

With regards to research into Palenque’s African past, the information obtained by the mid-1990s may be summarized as follows: although it seemed increasingly obvious to scholars that Palenque was rooted geographically in the Old Kongo, the persisting question remained as to what extent slaves and languages from other regions of sub-Saharan Africa contributed to its formation. As we recall, the Cartagena slave trade had brought together ethnicities from many distant regions of Africa, and at the time, it seemed simultaneously counterintuitive to assume that the palenques in Cartagena’s hinterland would have been anything but a reflection of this profoundly interethnic mix of slaves. Furthermore, in Palenque there remained a substantial corpus of lexical Africanisms that had not been studied systematically, and some of these words defied ready identification with Kikongo sources. Also, Kimbundu – a language closely related to Kikongo – remained as a putative source, as Del Castillo in particular recognized, that some of the Afro-Palenquero vocabulary could be traced to Kikongo as well as Kimbundu and possibly other Bantu languages that once circulated in Cartagena de Indias. While these and similar limitations made etymological work challenging, the findings were often plausible, but still not (always) entirely convincing. Outside of the immediate arena of linguistics, cultural studies with a focus on origins (cp. Schwegler, 1992, 2006a, 2007) did suggest further Bantu connections. For instance, old Palenquero burial techniques pointed to coastal regions of C ­ entral West Africa (see Schwegler, 1992, and the recent detailed comparative study by ­Davidson, 2012). Closer to my own area of research, rare Afro-Portuguese remnants in the Palenquero language proved its erstwhile connection with the Afro-­ Portuguese slave trade, a connection whose importance to Palenque was first highlighted by Escalante and then de Granda. There was also a further limitation in attempts to significantly advance research into Palenque’s past: the internet had yet to be invented, and as a result, scholars from related disciplines (history, anthropology, etc.) could not communicate with today’s accustomed ease. Even the field of linguistics itself was still deeply separated in terms of geographic area of research: Africanists with native knowledge of key languages like Kikongo and Kimbundu could rarely – if ever – collaborate with their colleagues in the West. As a result, virtually all scholars studying Palenquero language and culture were trained in the West, and, for the most part, had no or only minimal access to scholars with expertise in relevant sub-Saharan languages. It is important to mention that around the mid-1990s, the limited group of ­Palenquero scholars had the distinctive sense that the uncovered evidence had shed some light on the origins question by strongly pointing in the direction of western Central Bantu languages (especially Kikongo and Kimbundu). However, more definitive answers seemed still well beyond reach, and for the most part, the academic world at large continued to be unaware of the existence (and historical importance) of Palenque. Within the next decade or so, this situation would change.



Combining population genetics (DNA) with historical linguistics 

3.5  Th  e growing centrality of Kikongo and connections to Afro-Cuban ritual language While working in Palenque in the early 1990s, a publisher in France contacted me regarding the translation of enigmatic and hitherto (for scholars) unintelligible AfroCuban ritual texts into English, texts that appear in abundance in Lydia Cabrera’s famous work El Monte (1954 [1971]). The book in question was the first major anthropological study of Afro-Cuban traditions, and like most of Cabrera’s prodigious work, had never been translated into English. Having learned of my expertise in Palenquero, the publisher invited me to meet in Paris, which I gladly accepted in the hopes of contributing to the resolution of a linguistic puzzle that had perplexed several generations of scholars. An example of such Afro-Cuban texts is reproduced hereafter. In this instance, it is a mambo (ritual song) that, like most of these liturgical texts, involves code switching with Spanish (for easier identification, “African” segments are given in caps): sála mi nganga sálaló nsunga da vuelta l’ ingenio arriba munda tó moana, súnga, vamo sunga …

(Cabrera, 1954 [1971], p. 128)

Similar and linguistically even more puzzling mambos (ritual songs) had also been collected by Cabrera’s compatriot and friend Fernando Ortiz. The following is a representative sample (“African” segments are once again in caps): Lube lube lubea lube la kueba nganga Si muanan doki ta sere sere palo kindiambo se so

Ortiz (1965 [1950], p. 313)

Despite my fluency in Palenquero and general familiarity with Afro-Hispanic registers, my journey to Paris resulted in utter disappointment: The Lengua of Palenque was of no help in deciphering these Afro-Cuban texts, and the mystery of their origin and true nature (to this day, El Monte and its ritual texts have yet to be translated into English). However, my examining these Afro-Cuban texts did lead to a preliminary conclusion: they clearly were not rooted in an Afro-Hispanic creole, and their linguistic typology was visibly far more “African” than any text – ritual or otherwise – ever collected in Palenque. However, my “Parisian defeat” did have a beneficial effect that would eventually prove decisive in my search for Palenque’s and Cuba’s African roots: it convinced me of the necessity (and urgency) to acquire a deeper knowledge of African languages, and Kikongo in particular. As a result, starting around 1992, I availed myself of diverse resources on Kikongo (including Laman, 1964 [1936]; Swartenbroeckx, 1973, and

 Armin Schwegler

later Nsondé, 1999, and Loëmbe, 2005),15 while also pursuing further investigations into the history and grammar of the Palenquero language, the results of which can be found (inter alia) in Schwegler (1998) and (2002). As my familiarity with the language of the Bakongo (speakers of Kikongo) people grew, so did the corpus of Palenquero data pointing to an ever-greater connection between Palenque and the Old Kongo and the adjacent Loango region to the north (Maps 6 and 7, pp. 50–51 show the location of modern Kikongo-speaking territories and that of the Old Kongo Kingdom). Around the same time (ca. 1992–1995) I had become integrated into the Palenquero community to a point where Palenqueros freely divulged overlooked lexical Africanisms, most of which no longer circulate in everyday parlance (due to the previously mentioned heavy stigmatization of Palenquero language and culture, such archaisms had become relegated to the private sphere). Many of these Africanisms could confidently be linked to Kikongo (for a sampling, see Schwegler, 1999); others were suggestive of Kimbundu and/or Kikongo origins. However, a relatively small proportion of such words did resist etymologization. The earlier lexical status thus persisted, albeit now with a larger database to be considered: broadly speaking, Palenque’s substrate appeared to be Bantu – to the complete exclusion of other African languages. The highlighted rectangular area is inhabited mainly by first-language speakers of the various dialects/languages of Kikongo (some of these language/dialect names are written on the map; for the genetic filiation of Kikongo within Bantu, see Guthrie (1971), H10 as well as the sub-branch H16; further information on the often confusing and complex classification of Kikongo and its sub-varieties can be found in Maho (2002, 2003, 2008a, 2008b, 2009); a summary list of Kikongo dialects is given in Fuentes and Schwegler (2014a), where references to pertinent works – including the Atlas linguistique de l’Afrique Centrale (ALAC): Atlas linguistique du Zaïre (inventaire préliminaire – are also provided). The light areas to the east and north of the rectangular box are inhabited by firstand second-language speakers of Kituba (a.k.a. Kikongo ya Leta ‘Kikongo of the state administration’, cp. French Kikongo de l’État – its etymological source), which is the lingua franca in the region. Most speakers of Kikongo readily understand Kituba. The vehicular Kikongo varieties of Kituba and Kikoongo ya leta are used in the Democratic Republic of the Congo (DRC; capital = Kinshasa) (the writing of the first – Kituba – was codified by missionaries; the second by the colonial Belgian state, and

.  These and similar works can now be complemented by Mavoungou & Ndinga-KoumbaBinz’s Civili, langue des Baloango: Esquisse historique et linguistique (2010). As shown in Maps 6 and 7, the Civili language is spoken in among the Vili (or Bavili) people who inhabit Loango (the people of this coastal region are the “Baloango”).



Combining population genetics (DNA) with historical linguistics 

Map 6.  Dialect map of Kikongo and Kituba Map adapted from: 〈http://upload.wikimedia.org/wikipedia/commons/c/c3/LanguageMap-Kikongo-Kituba.png〉 Original map drawn by Risto Kupsala. Details are based on the language maps of Ethnologue and several other sources. An easier-to-interpret color version of the map can be found at the link given above.

later by the government of DRC which formalized it as one of the four regional languages of the country. In the Republic of the Congo (capital = Brazzaville), the same vehicular language is named Munukutuba by its speakers; it has neither written codification nor official recognition. For the purpose of this chapter, readers should take special note of points 1–3 indicated on the map: #1 and #2 point to the Vili and Yombe (also Mayombe) regions located inside the Loango area just to the north of the mouth of the Kongo/Zaire River (point #3). As this study hopes to show, it is from either of these Kikongo-speaking regions that the vast majority of the ancestors of the Palenqueros (Colombia) and Paleros (Cuba) must have originated. For a reconstruction of the history of these forest lands, see Vansina (1990), Paths in the Rainforests: Toward a History of Political Tradition in Equatorial Africa. Within the larger historical context and especially long-term population DNA research, it is instructive to know that, as pointed out in Soret’s Les Kongo nord-occidentaux (1959), prior to the beginning of the 15th century, Kongo peoples inhabited only the left (lower) regions of the Kongo/Zaire River. According to the same source, northbound migrations to the Loango area (Map 7) by Kongolese groups started to

 Armin Schwegler

take place around the time when the Portuguese explorer Diego Cão first reached this part of Africa in 1482. According to Frank Hagenbucher-Sacripanti’s insightful work, the Loango kingdom was probably founded by Bavili clans coming from Mbanza Kongo (São Salvador, capital of the former Kongo Kingdom; see Map 5): Un important parti de Bawoyo fit son apparition sur la côte de Loango au xive siècle, comportant des forgerons groupés en une puissante confrérie, celle des Buuüdji, qui, s’appuyant sur un corps de guerriers entreprenants, s’imposa aux populations locales. Un état s’érigea, qui reçut le nom de Loango, terme désignant le pouvoir (Lwangu: le commandement politique) …  (1973, p. 23)

Martin similarly suggests that the Bavili had an original Kongo connection: The first Vili, according to some traditions, were farmers who broke away from other Kongo groups on account of a land shortage in the interior. They passed through the Mayombe region and settled in the coastal plain between the Kwilu and Chiloango rivers. This almost certainly took place in the thirteenth or fourteenth centuries.

Map 7.  A possible map of the Kingdom of Kongo, and its northerly neighbor, the Kingdom of Loango (circa 1711) Map and text adapted from: 〈http://www.africafederation.net/Kongo_History.htm〉

The Kingdom of Kongo was composed of six provinces: Mpemba, Mbata, Nsundi, Mpangu, Mbemba and Soyo, plus four vassal Kingdoms: Loango, Cacongo and Ngoye, at the North of the N’Zari river, and Ndongo, at the South of the Congo river.



Combining population genetics (DNA) with historical linguistics 

Even before the height of the slave trade in the late eighteenth century, European travelers were attracted to the region. As Sommerdyk (2012, p. 4) points out, Italian missionaries to the Kongo and Kakongo between 1666 and 1667, offered the following pertinent description: The first and most famous of these [ports] is that of Loango, the second of Capinda [sic], and the third and last that of Cacongo [Malemba] itself, but which is not very secure. … The Inhabitants of this country, considering they have been born pagans, are more courteous and humane than ordinary, and though they are Infidels, they cherish and respect our priests. (Guattini & Carli, “Account of the Congo,” 1674, p. 736)

From these and neighboring regions to the east, thousands of slaves were shipped to the Americas, many becoming pidgin-speaking Bozales in Colombia (Cartagena in particular) and Cuba during the height of their slave trade. For pertinent references to the history and sociolinguistic dynamics of this region, see Schwegler (2012, pp. 164–165, and 2014b) and references therein. These studies include a map of slaving routes leading from the interior to the ports of Loango, Malemba, Cabinda, Ambrizette, Ambriz, Mossula, and Luanda. Due to the presence of the Livingstone Falls – a succession of enormous rapids on the lower course of the Congo/Zaire River in west equatorial Africa, just below Malebo Pool in the Democratic Republic of the Congo – slave traders had to march their captives overland rather than ship them by river to the Atlantic coast.

3.6  A revelation: It is all Kikongo! In the middle of a summer night in 1996 and four years into my study of Kikongo, I recalled my ill-fated attempt in Paris at solving the linguistic puzzle posed by Cabrera’s Afro-Cuban texts. I wondered: could it be that their African elements were derived exclusively from Kikongo, thereby perhaps having the same monogenetic African origin as Palenque’s sub-Saharan lexicon? To my amazement, a glance of Cabrera’s El Monte (1979 [1954]) and its companion volume Reglas de congo: Palo Monte Mayombe (1979) proved my hypothesis to be true: for the most part, understanding individual words now seemed effortless, and their Kikongo etymologies and literal meanings straightforward. Contrary to what all scholars and practitioners of Palo Monte (the ritual practice to which these texts belong) had always believed, the sub-Saharan component of Afro-Cuban Lengua was not a mixture of dozens of African languages, but rather restructured Kikongo, brought to Cuba most massively after the first quarter of the 19th century (Fuentes & Schwegler, 2005, 2012, 2014; Schwegler & Rojas-Primus, 2010; Schwegler, 2006b, 2014a). The discovery of the true nature of Palo Monte’s ritual speech has had three notable consequences: first, it has facilitated the decipherment and translation of f­ ormerly

 Armin Schwegler

perplexing ritual texts, thereby highlighting the extraordinary role that Bakongo slaves and their creole descendants appear to have played during the transatlantic slave trade, and in the Spanish Caribbean in particular. Included among these texts is the above-cited mambo, whose literal meaning is as follows (cited from Schwegler & Rojas-Primus, 2010, p. 199, where Kikongo etymologies are given; the mambo has been reconstructed to render it more faithful to its original version – the English translation is mine): Reconstructed mambo Lumb’ e, lumb’ e, lumb’ e! Lubela cueva enganga (Sí), Sarabanda ta seresere! ¿Palo quindiambo sese agüé?

Literal translation Beat (the drum) eh! (3x) Creek … in the cave/hole of the Nganga16 Yes, Sarabanda, do mention/invoke the the magic potion! Into what kind of things is the magic potion today?

The second major consequence has been to solidify Kikongo etymologies of Palenquero words whose origin previously seemed plausible but not entirely proven. The increased confidence with which Palenquero Africanisms can now be traced rests in part on the fact that Palo Monte’s ritual Kikongo-based jargon is extraordinarily rich (hundreds of Palo Monte words have indisputable etymologies, and, as discussed recently in Fuentes  & Schwegler [2014] and Schwegler [2014a], some of these Africanisms are attributable only to Kikongo and no other Bantu language). Furthermore, the abundance of verifiable Palero etymologies has facilitated the establishment of rigorous Kikongo/Palo Monte sound correspondences, most of which are also applicable to Palenquero. As historical linguists have known for over a century, solid hypotheses about sound change are fundamental to the reconstruction of protoforms for each cognate set. The meanings of modern reflexes can then be compared (and hopefully linked) to the earlier meaning of the base form. The establishment of these sound and meaning correspondences for the Afro-Cuban case has reduced the formerly speculative nature of etymological work with Palenquero. A clear example thereof is seen in former ritual expressions mankisé (Lumbalú), reconstructed in Schwegler (1996, p. 462) as ¡ma nkisi, e!, with a former meaning of ‘ah, the fetish, eh!’. The ritual phrase mankisé formed part of a Lumbalú collected by Escalante in 1954, but by the time I arrived in Palenque in 1985, the members of the community could not recall its literal meaning, though elders did recognize it as former ritual parlance. In 1996, my etymological reconstruction suggested that ma nkis’ e, along with expressions such as Kalunga (in the same Lumbalú), was an unusually precious relic,

.  The Nganga is Palo Monte’s most sacred receptacle; in it live the spirits that the

Palero priest awakens with during his ritual.



Combining population genetics (DNA) with historical linguistics 

pointing back to a time when Palenqueros’ ritual beliefs used to adhere more closely to Old Kongo traditions. Subsequent fieldwork in Cuba (Fuentes & Schwegler, 2005; Schwegler & ­Rojas-Primus, 2010) has not only helped solidify the proposed Kikongo etymology for Palenquero manquisé → ¡ma nkis’ e!, but it has also highlighted the absolute centrality of the term and concept of nkisi to Afro-Hispanic religious practices derived from Bakongo territories. As explained in Fuentes and Schwegler (2005, pp. 28–29), a nkisi (also articulated nkise, enkise, enkisi, inkisi etc.) is Paleros’ most sacred object (usually a pot, also called nganga, enganga, or prenda) inhabited by venerated gods or ancestral “spirits” of Palo Monte. Similar to a practice in the former Kongo, Paleros consider the ritual of evoking a nkisi to conjure up profound religious and other sentiments associated with the use of magic herbs (“palos” in Spanish),17 the supernatural, trance, fraternities (known as “familias”), medicinal healing, the importance of death, proper burial, all-important connections to ancestral spirits, and so forth. Witnessing firsthand (fieldwork in Cienfuegos, Cuba) the profoundly symbolic significance of such nkisi and their central role in the lives of Cuban Paleros, I obtained a deeper understanding of why Palenqueros, Paleros, and other New World descendants of Bantuspeaking slaves held on to Kikongo-laden ritual language for generations: following in the tradition of the Old Kongo,18 at the most fundamental level their intra-muro communities were religious medicinal-based societies in which ritual activities played a central role. Ritual language was central to the proper functioning of their community and was most vividly and powerfully expressed in the parlance of their ancestors whose spirits reside in the nkisi. As long as this socio-religious mode of thinking was in place, the sacred tongue of their ancestors – Kikongo – was transmitted from generation to generation. During my fieldwork, information gathered from elders suggests that in Palenque prior to 1900, the imposition of Western models of thought and behavior (a Catholic church was erected in Palenque almost a hundred years ago, and a priest more or less permanently lived in the village in years past) gradually led to the

.  Hence “Palo Monte” = Palo del Monte = lit. ‘herb(s)/plant(s) from the hills (or from nature in general), used for magical or medicinal purposes’. .  For information on these old Kongo traditions, see the works by Bittremieux, Bouquet, Buganza, Dupré, Fu-Kiau, Hagenbucher-Sacripanti Kimbpianga, Laman, MacGaffey, Nsondé, Palmié, Thompson, Thornton, Wing, Warner-Lewis, cited in the References to Fuentes and Schwegler (2005). Martin’s “Power, cloth and currency on the Loango Coast” (1986) provides a most insightful account of how landmark events such as initiation and burial ceremonies play a central role among the Loango people in daily indigenous life (see especially page 6; Loango here refers to the coastal regions between southern Gabon and the Zaire river). In my fieldwork, I observed similar preoccupations among the Palenqueros and Cuba’s Paleros.

 Armin Schwegler

stigmatization of outwardly African spiritual traditions. As a result, the practice and memory of nkisi and everything it embodied began to fade along with Palenqueros’ recollection of Kongo-based chants and invocations. The third major consequence of the discovery of the true nature of Palo Monte was the realization that the African mambos collected by Lydia Cabrera, Fernando Ortiz, and their cohorts in the first decades of the 20th century contained such phonetically “unadulterated” Kikongo that they suggested an inescapable conclusion: there had to be (elderly) practitioners on the island who still possessed at least some fluency in this ritual language. Even though first deemed preposterous by the vast majority of my Cuban colleagues, this conclusion subsequently received confirmation during fieldwork carried out in Cienfuegos, Havana, Holguin, and several other localities (Schwegler, 2000b; Fuentes Guerra & Schwegler, 2005; Rojas-Primus, 2009; Schwegler & Rojas Primus, 2010). The fresh data revealed that adept practitioners of Palo Monte not only speak Lengua with fluency, but they also possess complete literal understanding of Kikongo phrases (this circumstance thus sharply contrasts with that of Palenque, where some of the old Lumbalú expressions are no longer understood by anyone). Similar to the early 20th-century mambos reproduced above, Paleros’ contemporary Lengua routinely code-switches between Spanish, Bozal Spanish, and restructured Kikongo. It is therefore logical that the unanticipated access to traditional “live” Palo Monte rituals is giving scholars of Afro-Hispanic speech new and exceptionally useful insights. This has made the fine-tuning of the sound laws governing Kikongo/ Palero articulatory changes a reality. By the same token, it has helped us better grasp socio-religious dynamics that have honed Paleros’ enduring verbal skills, and has led us to better appreciate the importance of relative chronology in the establishment of Afro-Hispanic enclaves. This difference in relative chronology explains why bozal speech and Kikongo ritual lexicon has been preserved far better in Cuba than in any other region in Latin America. On the other hand, in the Cartagena area (Palenque included), almost four long centuries of exposure to constant stigmatization and ridicule of “anything African” have forcefully eroded cultural and linguistic practices that must once have been ubiquitous.

4.  The state of the discipline from the mid-1990s to 2010 In the mid-1990s and in the years that followed, creolistics outgrew its infancy to become a burgeoning field of research in Europe and the United States. This interest in the study of contact languages transferred to other continents as well, eventually also reaching some parts of Asia and sub-Saharan Africa. The roots-oriented study of Palenquero profited considerably from this trend, as it garnered attention from ­well-known figures



Combining population genetics (DNA) with historical linguistics 

in the field. For instance, in 1999, John Holm (Portugal), Angela Bartens (­Finland), Derek Bickerton (USA), John Lipski (USA), Philippe Maurer (Switzerland), and others participated in the first international conference on Palenquero held in Cartagena (Moñino & Schwegler, 2002). Importantly, by then, Africanists like Yves Moñino (France) had joined in the research and carried out extensive fieldwork in Palenque. The general interest in pidgin and creole languages expanded in tandem with the growing recognition outside of academia of the sociocultural significance of these formerly ignored speech varieties. In part due to this societal shift, junior linguists from within Colombia (e.g., Marianne Dieck and later Alejandro Correa) began to develop a strong interest in the Palenquero language. Doctoral dissertations (e.g., Morton, 2005; Swearingen, 1997; Smith, 2014) and constructive collaboration among involved participants further strengthened the field and amplified the interest in questions related to African roots. Notable, widely read works like Lipski’s Latin American Spanish (1994) introduced Bozal Spanish, Palenquero, Papiamento, and other Black Spanishderived speech varieties to a broad audience of Hispanists. With the onset of the new millennium, a select group of junior scholars from Palenque and/or Cartagena also began to join in the study of the Palenquero language and culture (see, for instance, Cásseres Estrada, 2005; Pérez Miranda, 2011; Pérez Miranda & Miranda Reyes, 2012; Simarra Reyes & Triviño, 2012a [2007], 2012b, Simarra Obeso et al., 2008). A steady stream of monographs and articles flowed from these activities (see, for instance, Moñino & Schwegler, 2002; and Maglia & Schwegler, 2012, and relevant works cited therein), eventually leading to more theoretical, cross-Atlantic comparisons like that of Landa Buil (2014). As already mentioned, special attention from international entities such as UNESCO in 2005 and global leaders like North American President Barack Obama in 2012 led Palenque and its “African” history taking center stage within Colombia and beyond. Meanwhile, the world became a smaller place as ever-growing interconnectivity and ease of communication (internet, e-mail, skype, etc.) among scholars in different fields of study became a reality. This especially augmented interdisciplinary collaboration with historians (e.g., Navarrete, 2008, 2010, 2011), whose archive-based research into early Palenquero history has proven indispensable. During this period, Palenque’s “African” history also came into sharper focus thanks to Lipski’s extraordinarily rich and comprehensive A History of Afro-Hispanic Language: Five Centuries, Five Continents (2005). Together with Klee & Lynch’s El español en contacto con otras lenguas (2009), this and similar publications highlight the historical influence of African languages on Spanish by situating creoles, Bozal Spanish, and especially Palenquero squarely at the center of discussion. Limitations of space preclude me from examining the numerous works (e.g.,  Hualde  & Schwegler, 2008; Bartens, 1995; Lipski, 2005, 2009; Moñino, 2002, 2007a, 2007b; Moñino & Ortiz, 1999) that furthered our understanding of Palenque’s

 Armin Schwegler

African origins in the period under review here (approximately 1995 and 2010). However, it deserves clarification in that despite significant advances, the general state of the discipline at the time did not experience a fundamental shift, as definitive answers to the origins question still seemed well beyond reach. It is certain that at the time, scholars had been able to trace numerous additional Kikongo correspondences in both the lexis and Palenquero grammar. For example, this includes the pronoun system where first-person clitic Pal. i is a reflex of Kik. i ‘ibid’ (cp. i ta ablá ‘I am speaking/I speak’; for details see Schwegler, 2002). Other traces of Kikongo grammatical influence are found in the creole’s aspectual functions (cp. Moñino, 1999) or genitive constructions (cp. Moñino, 2002). Possible carryovers of Kikongo tone into Palenquero also garnered special attention (Hualde & Schwegler, 2008). Furthermore, investigations into Palenquero culture (rather than language) have provided a more nuanced assessment of substratal Bantu input (see Schwegler, 2006a). But while knowledge about connections between Palenque and the Old Kongo grew exponentially, so did the nagging sense that the reigning monogenetic hypothesis of Bakongo slaves as the sole donors of sub-Saharan input could still not be entirely proven. This led to a situation that contrasted sharply with that of Palo Monte, whose lexical derivation from Kikongo could be stated in categorical terms. The lingering doubts about Palenque’s true African substrate(s) persisted for several reasons. The following section briefly examines the rationale for some of these doubts.

4.1  The monogenetic Kikongo hypothesis around 2010: Lingering doubts Dozens of lexical Palenquero Africanisms – some recently discovered – had yet to be etymologized in print, and thus it remained to be seen whether Kikongo was their ­ultimate source. This naturally weakened the monogenetic claim, and raised the suspicion that scholars (myself included) may have erroneously focused their attention on a single language to the detriment of other and perhaps more scarcely documented substrates. As mentioned above, although it is true that Palenquero studies had attracted a growing number of scholars, it is also the case that by 2010, the group of specialists focusing on Palenquero African roots was still undersized and arguably too interconnected to be adequately self-critical. At the same time, very few participants in the debate were trained Africanists, and native speakers of pertinent African languages had yet to join the ranks of Palenqueristas. These two factors – which persist today – by necessity made us at least somewhat hesitant to embrace what the available linguistic evidence seemed to suggest (i.e., Kikongo monogenetic input). This relative reticence was further accentuated by the fact that contemporary anthropology had effectively ignored Palenque in spite of Escalante and Friedemann’s early lead. It is logical that anthropologists could have significantly reshaped the origins debate and potentially could have helped us sharpen our analytical reasoning.



Combining population genetics (DNA) with historical linguistics 

My self-doubts about my own monogenetic hypothesis persisted for yet another and more pressing reason: as Franz Boas’s admonished us over a century ago (and Wirtz [2014] has astutely reminded us recently), genealogy, culture, ethnicity, and language “have not necessarily the same fates” when we trace their historical movements (Boas, 1911, p. 11). Therefore, it is important to continue considering the possibility that, contrary to what the available linguistic data suggest, Bakongo slaves were only one – albeit the most dominant – of several ethnicities in early Palenque. In this scenario, multiple sub-Saharan languages could have potentially co-existed for several generations. Especially in the domain of religion and rituals, Kikongo conceivably dominated so that over time this led to a cultural and linguistic “ratchet effect.” As the value of (ritual) Kikongo rose or ratcheted upward in the generations that followed, the absolute supremacy of Kikongo could eventually have been established. This line of reasoning is consistent with what Sheller presents in her book, Consuming the Caribbean, when she argues that creolization is not simply about moving and mixing elements, but is more precisely about processes of cultural “regrounding” following experiences of violent uprooting from one’s culture of origin. It is deeply embedded in situations of coerced transport, racial terror, and subaltern survival […] Creolization is a process of contention.(Sheller, 2003, p. 189)

Viewed this way, the uniformity of Palenque’s African legacy may simply have been the result of multi-generational negotiation and linguistic power brokering, positioning one ethnolinguistic group against the other while leaving space for shifting alliances that eventually homogenized African languages and ethnicities into the Palenque we know today.19 The dynamics of such a transformation and readjustment would almost certainly have differed from generation to generation, but this most likely would have involved clans and subgroups that already held leadership roles in primitive Palenque. These have been the dynamics in the documentation of recent epochs of Palenque (approximately 1900 to present), and there is no reason to assume that the earlier social matrix was any different.

5.  The game changer: Population genetic (DNA) research 5.1  Contributors to the research Against this background of lingering doubts and origins questions, Yves Moñino and I independently sought collaboration with two separate teams of population g­ eneticists

.  For a different account of how and why Kongo influence may have become dominant in Palenque, see Moñino (2012, p. 234, including Note 18).

 Armin Schwegler

in the hope that an interdisciplinary approach may help resolve the theoretical stalemate (monogenesis vs. polygenesis) that had resulted from 50 years of humanistic research into Palenque’s African past. The first of these teams (Moñino’s affiliates) involves a group of researchers at The Centre for Genetic Anthropology of University College London, headed by renowned geneticist Dr. Neil Bradman. A junior member of this team includes Iranian scholar Nasar Ansari-Pour, whose 2010 dissertation20 will be the main source of DNA research examined in this study (see Section 5.2 below). My affiliation with geneticists comprises a joint Colombia-/Portugal-based group headed by Dr. Maria Claudia Noguera (Instituto de Genética Humana, ­Pontificia U ­ niversidad Javeriana, Bogota) and Dr. Leonor Gusmão (Institute of Molecular Pathology and Immunology of the University of Porto, Portugal). Our primary objective was to determine the African, European, and Native-American paternal contributions in genetic samples from the Department of Bolivar (Colombia; Map 1) with the aims of establishing (a) possible population substructures and (b) the proportion of biological African heritage in admixed populations of European, Amerindian, and African descent. One of the communities sampled in this study was Palenque. As expected, its genetic pool of African origin is highly preserved (with about 60%), contrasting sharply with the negligible Amerindian contribution (about 2%) (see F ­ igure  1). S­ urprisingly, Palenque’s proportion of male European DNA (36%) is far more significant than anticipated (Noguera, Schwegler, & Gusmão et al., 2014).

5.2  A  nsari-Pour’s comparative DNA study: 42 African ethnolinguistic communities vs. Palenque Judging by its title Human genetic variation, relationships of peoples of sub-Saharan Africa and implications for healthcare, Ansari-Pour’s doctoral dissertation (henceforth A.-P.) initially does not appear to have any direct bearing on the question of Palenque’s African genetic heritage. As will become apparent here, his work is, however, indeed highly relevant since Chapter 2 “The origins of the Palenque […]” explicitly compares the DNA of Palenque’s population to that of over 40 different and widely dispersed sub-Saharan ethnolinguistic communities (see Map 8 for their approximate location; a complete list is given in Figure 2). To that end, A.-P. compares and calculates the frequencies of NRY haplotypes and NRY-based genetic distances, and analyzes the distribution of mtDNA variation across the datasets obtained. The following ­sections on A.-P.’s goals and methods, results, and interpretation of data will explain how

.  The dissertation was resubmitted in 2011 with minor corrections. My study will draw from this revised version, but I will continue to cite it with its original 2010 filing date.



Combining population genetics (DNA) with historical linguistics 

Source: Noguera, Schwegler & Gusmão et al. (2014, Figure 1)

Figure 1.  Sub-Saharan genetic legacy in Palenque and five surrounding communities (Department of Bolivar) Localities sampled in this study are shown with the proportion of Sub-Saharan African, Native American and European chromosome contribution. The sample size is indicated for each population. As shown here, Palenqueros have the highest proportion of African Y chromosomes.

Map 8.  Approximate location of DNA sampling Except for South Africa, DNA was collected among multiple population groups. For a full list of these, see Figure 2 For a recent DNA study of the historical diffusion of these and other Niger-Congo groups, see Filippo (2011)

 Armin Schwegler

Geographic location

Population Group (Code) Bankim (BA) Foumban (FO) Cameroon Wum (WU) Bembe (BE) Kuni (KU) Lari (LA) Mboshi (MB) Sundi (SU) Teke (TE) Vili (VI) Congo Yombe (YO) Asante (AS) Ewe (EW) Ghana Fante (EW) Chewa (CH) Tumbuka (TU) Malawi Yao (YA) Mozambique Sena (SE) Nigeria Abak (AB) Afaha Eket (AE) Afaha Okpo (AO) Afaha Ukwong (AU)

South Africa

Awa-Onna (AW) Calabar (CA) Ediene lkono (ED) Efut Akpabuyo (EF) Efut Odukpani (EO) Ejagham-Akamkpa (EA) Ejagham-Calabar (EC) Eziagu Nenwe (EZ) lkono (IK) ltam (IT) Nike Enugu (NE) Nnung Ndem-Onna (NN) Nsit (NS) Ntan lbiono (NT) Obong ltam (OB) Oku-ltu (OI) Oku-Uyo (OU) Ukpom Ete (UE) Uwanse(UW) Bantu-Pretoria (BN)

Ethnic Identity Tikar Bamoun Aghem Bembe Kuni Lari Mboshi Sundi Teke Vili Yombe Asante Ewe Fante Chewa Tumbuka Yao Multiple Annang Ibibio Oron Oron Ibibio Igbo Ibibio Efik Efik

Language Latitude Longitude 6.083 Tikar 11.483 5.729 10.902 Bamoun 6.389 10.073 Aghem –4.795 11.846 Bembe –4.795 11.846 Kuni –4.259 15.285 Lari –4.259 15.285 Mboshi –4.259 15.285 Sundi –4.259 15.285 Teke –4.795 11.846 Vili –4.432 12.108 Yombe 6.106 –1.878 Akan 6.6 0.467 Ewe 5.817 –2.817 Akan 33.918 Nyanja –13.607 34.79 Tumbuka –14.27 –12.77 33.874 Yao –17.442 35.027 Sena 7.717 Annang 5.05 7.867 4.717 Ibibio 8.233 Oron 4.833 8.25 Oron 4.75 7.815 4.69 Ibibio Igbo 8.317 4.95 7.883 Ibibio 4.783 8.442 Efik 4.908 7.983 Efik 5.167

Ekoi Ekoi Igbo Annang Ibibio Igbo

Ejagham Ejagham Igbo Annang Ibibio Igbo

Ibibio Ibibio Ibibio Ibibio Ibibio Ibibio Ibibio Efik Bantu

Ibibio Ibibio Ibibio Ibibio Ibibio Ibibio Ibibio Efik Bantu

N 33 117 116 109 68 62 91 25 63 108 65 94 88 60 92 61 56 62 56 48 50 49 28 99 48 48 49

5.35 4.95 6.117 4.992 5.042 6.433

8.35 8.317 7.517 7.758 7.842 7.483

47 83 49 42 50 54

4.633 4.833 5.233 5.133 5.133 5.1 4.62 4.95 –25.746

7.85 7.9 7.933 7.967 7.933 7.967 7.65 8.317 28.187

48 36 50 50 49 48 50 50 98

Figure 2.  Details of population groups included in Ansari-Pour’s DNA study (2010, pp. ­32–33)



Combining population genetics (DNA) with historical linguistics 

this genetic, sex-specific DNA research complements and clarifies previous mostly ­language-based hypothesis of Palenqueros’ origins.

5.2.1  Goals and methods The majority of studies on sub-Saharan African populations have historically relied on either low sample numbers or a limited number of poorly defined populations (for references, see Ansari-Pour, 2010, p. 23). A.-P.’s investigation overcomes this limitation by using the African DNA collection of The Cancer Genome Atlas (TCGA),21 which consists of over 20,000 DNA samples. These DNA samples are available for many sub-Saharan populations that are defined by their ethnicities across the African continent. Precautions were taken to overcome polymorphism ascertainment bias as well as other limitations that have traditionally reduced the usefulness of large scale population genetics research. Overall, A.-P.’s database thus provides reliable fine-scale comparative DNA analyses that in toto offer the largest and most densely assembled sub-Saharan African data set yet reported (cp. AnsariPour, 2010, p. 23).22 A.-P.’s stated principal objective in Chapter 2 is “to investigate whether [Palenquero] genetic data are in accordance with the hypothesis of a Yombe founding group” (2010, p. 31). Readers will recall that the Yombe group (Map 6) is historically located in the Republic of the Congo and is the same ethnolinguistic (Kikongo-speaking) community that is referenced to in Cuban Palo Monte as Ma–Yombe > Mayombe (ma- is a class prefix indicating plurality or group membership; this same ma prefix is the universal pluralizer in Palenquero: cp. ma kasa ‘the homes’, ma ngombe ‘(the) cows/ steer = cattle’, ma kusa ‘(the) things’).

.  TCGA is a project that was started in 2005 with the goal of cataloguing genetic mutations responsible for cancer by using genome sequencing and bioinformatics. TCGA is supervised by The Centre for Genetic Anthropology and the National Human Genome Research Institute funded by the US government. For more information on TCGA project organization, visit

(Source: Wikipedia, April 29, 2014) .  However, readers should keep in mind that, as noted by Pakendorf, Bostoen, Filippo, “[o]verall the African continent is still severely understudied from a molecular a­ nthropological perspective. […] large areas, especially in central and southern Africa, still remain blank …” (2011, p. 58; see also the map in Figure 3, p. 59).

 Armin Schwegler

5.2.2  DNA data collection in Palenque, the Republic of the Congo, and beyond To render A.-P.’s comparative study maximally detailed, a n ­ umber of sex-specific NRY markers23 and mtDNA makers were24 analyzed in d ­ atasets from two continents (Africa and Americas) and from three different types of communities: (1) Palenque, (2) multiple groups from the Republic of the Congo (the Yombe were thus only one of several Congolese groups studied),25 and (3) peoples throughout sub-Saharan Africa (Map 8). It should be noted, however, that although DNA information was obtained from non-Bantu areas such as Nigeria and Ghana, the database does not include population groups from Senegal and other West African territories (see Table 1). This is a limitation that future studies may wish to address, since slaves from West Africa are known to have been shipped to Cartagena in fairly large numbers. In Palenque, buccal swabs were obtained from over 150 males over eighteen years of age (no two of them shared a common paternal grandfather). The fact that Y-chromosomal DNA is paternally inherited enables genealogical researchers to trace male lineages far back in time. The comparison of such DNA sequences is a pillar of phylogenetics in that it allows biologists to elucidate the evolutionary relationships among population groups (and/or species) while also permitting an examination of the genetic relatedness of populations. In the Republic of the Congo, samples were collected in different areas of ­Brazzaville, Pointe Noire, and the villages of Kakamoeka and Lovoulou, about 80 km inland from Point Noire (Ansari-Pour, 2010, p. 32). Historically, the Yombe people

.  NRY markers are sex-specific inherited markers; that is, those from the nonrecombining part of the human Y chromosome (NRY) transferred through the male germline (females do not normally carry Y chromosomes) and markers from the human mitochondrial genome transferred through the female germline (human sperm cells do not normally transmit mitochondrial DNA). Both parts of our genome are inherited in the absence of homologous recombination and thus are unchanged from one male (NRY-DNA) or female (mtDNA) generation to the next, as illustrated in (Figure 1), unless mutations occur. (Source: 〈http://www.biotechniques.com/BiotechniquesJournal/supplements/2007/December/Uni-parental-markers-inhuman-identity-testing-including-forensic-DNA-analysis/biotechniques-43536.html〉) .  In humans, mitochondrial DNA – just like male NRY – is sex-specific in that it is inherited (almost) exclusively from the mother. Since mitochondrial genome mutations are passed on 100% of the time from mother to all her offspring, such mutations are especially useful for reconstructing the population histories. For a tutorial providing an in depth lesson about mtDNA and how it allows us to trace our ancestry, visit 〈http://www.genebase.com/learning/article/17〉. .  All Congo groups were initially chosen by Yves Moñino.



Combining population genetics (DNA) with historical linguistics 

(­speakers of Kiyombe) have lived mainly in the Mayombe mountains, a hilly region about 100 kilometers inland from the Loango coast (see Maps 6–7). Recent wars between rival ethnic groups in the Congo have caused many to migrate to the city of Pointe Noire (Republic of the Congo), which is why some of the Yombe data samples were collected there. Additional buccal swab samples were obtained from another 34 population groups from distant areas of sub-Saharan Africa (Map 8 and Table 1). The total number of samples thus gathered was 2,649 and the total of sub-Saharan population groups compared was 42 (excluding Palenque). A graphic representation of the haplogroup26 distributions in sub-Saharan African and Palenquero population is given on page 41 of A.-P.’s dissertation (the interpretive, color-based nature of the graphed data prevents me from reproducing it here).

5.2.3  Results: Frequencies of NRY haplogroups and NRY-based genetic distances Of the sub-Saharan African groups studied, only two – the Yombe and the Chewa, an East African group from a remote region of Malawi – displayed a very close genetic proximity (Fixation Index = FST)27 to Palenque. The genetic margin between ­Palenqueros (0.027) and Chewa (0.035) and Yombe peoples was negligible to the point of being statistically insignificant; in contrast, the genetic margin in relation to all other African groups was significant at the 5% level (see Ansari-Pour’s Tables 2.3–2.4 and Figure 2.4 in Ansari-Pour, 2010, pp. 39–42, along with his explanatory text on pp. 43–44; his Table 2.3 is reproduced here as Table 1). Expressed differently, A.-P.’s data point to the following: (1) an unambiguously close genetic link between the male Palenqueros and the Yombe people; (2) a more or less deep (but always statistically significant) genetic distance between male and female Palenqueros to all other ethnolinguistic groups of sub-Saharan Africa included in the study; and (3) an “exception to the rule” in that Malawi’s Chewa people are also a virtual genetic match with the Palenqueros. The seemingly puzzling genetic correspondence between Palenqueros and the Chewa is addressed in Section 5.2.4 below.

.  In molecular evolution, a haplogroup is a group of similar haplotypes that share a common ancestor having the same single nucleotide poly-morphism mutation in all haplotypes. In human genetics, the haplogroups most commonly studied are Y-chromosome (Y-DNA) haplogroups and mitochondrial DNA (mtDNA) haplogroups. For this and additional information on haplogroups, visit 〈http://en.wikipedia.org/wiki/Haplogroup〉. .  For an explanation of “FST”, visit 〈http://en.wikipedia.org/wiki/Fixation_index〉.

 Armin Schwegler

Regarding haplogroup profiles, A.-P. makes the following general observation to offer a proper understanding of Palenqueros’ relationship to sub-Saharan groups: Examination of the haplogroup profiles of the three groups: Palenque, Chewa and Yombe reveals that of the twelve NRY haplogroups present in at least one of the groups six were present in all three. Of the other six, common presence or absence matched Palenque and Yombe for four haplogroups and Palenque and Chewa once while one haplogroup was only present in the Palenque (see Figure. 3). Notably (a) all the haplogroups present in the Yombe were observed in the Palenque and (b) the proposed signature haplogroup of the eastern expansion of the Bantu speaking peoples (Ansari-Pour et al., 2013) was not observed in the Palenque and the Yombe. (Ansari-Pour et al., 2014, p.10)

To contextualize the aforementioned citation, it is worth noting that “modern-day Bantu-speaking groups are characterized by low Y-chromosomal diversity” and only “two haplogroups are found in the majority of ethnolinguistic groups: E1b1a* and its subgroup E1b1a7” (Pakendorf, Bostoen, & Filippo, 2011, p. 64). Concerning the haplogroups studied by A.-P., two general and decidedly significant conclusions emerge from his work: the analysis of E1b1a haplogroups – i.e. the modal haplogroups that are found across sub-Saharan peoples studied (see his Table 3.3. in Ansari-Pour, 2010, p. 70) – reveals that (1)  the genetic distance between Palenque and Yombe peoples is smaller than that between Palenque and any of the other 41 sub-Saharan groups from West, West Central, East and South Africa, and, in addition, (2) of the eight groups studied from the Republic of the Congo, the Yombe were the only one for which there was no statistically significant genetic distinction to Palenque. A.-P.’s comparative DNA study of NRY and mtDNA sex-specific data is thus entirely consistent with our earlier humanistic (predominantly linguistic) hypothesis that Kikongospeaking Bakongo groups, and more specifically the Yombe people, are Palenqueros’ principal forefathers. A.-P.’s study demonstrates that large-scale population genetics research can illuminate past demographic events, even if these involve cross-Atlantic displacement of population groups whose history has been poorly documented. However, it is worth cautioning that the DNA studies presently available should not lead to the conclusion that the male Bayombe (Yombe) were the only Bakongo people among Palenque’s founders. As A.-P. astutely recognizes, his analysis “does not contradict the possibility that slaves from other parts of Congo joined the Palenque community either at its foundation or subsequently but that they were not in such number or authority so as to introduce the widespread use of their language” (Ansari-Pour, 2010, p. 57). His study does, however, highlight the overarching conclusion that Bakongo people can be identified with a reasonable level of confidence as the candidate group who heroically



Combining population genetics (DNA) with historical linguistics 

E1b1a*

E1b1a7

E1b1a8*

E1b1a8a1*

E1b1a8a1a

P*(xR1a)

BT*(xDE,KT)

E*(xE1b1a)

K*(xL,N1c,O 2b,P)

R1a1

Y*(xBT,A3b2)

DE*(xE)

A3b2

Table 1.  Haplogroup frequencies in Palenque and 42 sub-Saharan African groups from seven broad geographic regions (Ansari-Pour, 2010, pp. 39–40)

Palenque-Abajo (PAB) Palenque-Amiba (PAR) Palenque total (PA)

0.012 0.019 0.013

0.129 0.096 0.120

0.341 0.173 0.267

0.071 0.231 0.120

0.000 0.000 0.000

0.153 0.192 0.180

0.012 0.096 0.040

0.212 0.154 0.200

0.000 0.019 0.007

0.047 0.000 0.027

0.024 0.019 0.027

0.000 0.000 0.000

0.000 0.000 0.000

Bembe (BE) Kuni (KU) Lari (LA) Mboshi (MB) Sundi (SU) Teke (TE)

0.037 0.015 0.000 0.000 0.040 0.000

0.321 0.515 0.484 0.440 0.600 0.429

0.211 0.162 0.177 0.286 0.200 0.222

0.284 0.176 0.194 0.187 0.040 0.159

0.000 0.000 0.000 0.000 0.000 0.000

0.018 0.000 0.000 0.011 0.000 0.032

0.000 0.044 0.032 0.044 0.040 0.063

0.110 0.074 0.113 0.000 0.080 0.079

0.000 0.000 0.000 0.000 0.000 0.000

0.000 0.000 0.000 0.000 0.000 0.000

0.000 0.015 0.000 0.033 0.000 0.016

0.000 0.000 0.000 0.000 0.000 0.000

0.018 0.000 0.000 0.000 0.000 0.000

Vili (VI) Yombe (YO) Congo Total

0.056 0.031 0.024

0.500 0.277 0.430

0.250 0.369 0.239

0.139 0.138 0.181

0.000 0.000 0.000

0.009 0.015 0.012

0.028 0.046 0.034

0.019 0.092 0.066

0.000 0.000 0.000

0.000 0.000 0.000

0.000 0.031 0.012

0.000 0.000 0.000

0.000 0.000 0.003

Bankim (BA) Foumban (FO) Wum (WU) Cameroon Total

0.364 0.068 0.026 0.086

0.182 0.667 0.621 0.586

0.152 0.068 0.000 0.049

0.152 0.151 0.310 0.177

0.000 0.000 0.000 0.000

0.030 0.000 0.000 0.004

0.061 0.085 0.043 0.064

0.061 0.043 0.000 0.026

0.000 0.000 0.000 0.000

0.000 0.000 0.000 0.000

0.000 0.009 0.000 0.004

0.000 0.000 0.000 0.000

0.000 0.009 0.000 0.004

Asante (AS) Ewe (EW) Fante (FA) Ghana Total

0.245 0.114 0.300 0.211

0.309 0.420 0.250 0.335

0.106 0.102 0.167 0.120

0.245 0.273 0.217 0.248

0.000 0.000 0.000 0.000

0.000 0.023 0.033 0.017

0.000 0.023 0.000 0.008

0.043 0.045 0.017 0.037

0.000 0.000 0.000 0.000

0.000 0.000 0.000 0.000

0.053 0.000 0.017 0.025

0.000 0.000 0.000 0.000

0.000 0.000 0.000 0.000

Chewa (CH) Tumbuka (TU) Yao (YA) Malawi Total

0.043 0.033 0.071 0.048

0.239 0.410 0.321 0.311

0.293 0.230 0.250 0.263

0.130 0.082 0.054 0.096

0.054 0.033 0.036 0.043

0.000 0.000 0.000 0.000

0.076 0.082 0.196 0.110

0.130 0.098 0.071 0.105

0.011 0.016 0.000 0.010

0.000 0.000 0.000 0.000

0.000 0.000 0.000 0.000

0.000 0.000 0.000 0.000

0.022 0.016 0.000 0.014

Abak (AB) Afaha Eket (AE) Afaha Okpo (AO) Afaha Ukwong (AU) Awa-Onna (AW) Calabar (CA) Ediene Ikono (ED) Efut Akpabuyo (EF) Efut Odukpani (EO) Ejagham-Akamkpa (EA) Ejagham-Calabar (EC) Eziagu Nenwe (EZ) Ikono (IK) Itam (IT) Nike Enugu (NE) Nnung Ndem-Onna (NN) Nsit (NS) Ntan Ibiono (NT) Obong Itam (OB) Oku-Itu (OI) Oku-Uyo (OU) Ukpom Ete (UE) Uwanse (UW) Nigeria Total

0.054 0.104 0.020 0.041 0.000 0.061 0.083 0.125 0.020 0.021 0.024 0.286 0.000 0.080 0.019 0.042 0.028 0.040 0.000 0.000 0.042 0.060 0.060 0.053

0.357 0.354 0.460 0.408 0.500 0.475 0.479 0.438 0.449 0.426 0.518 0.469 0.429 0.520 0.685 0.563 0.361 0.380 0.480 0.510 0.396 0.480 0.460 0.464

0.107 0.042 0.060 0.102 0.000 0.121 0.042 0.042 0.143 0.085 0.048 0.020 0.167 0.060 0.000 0.063 0.111 0.160 0.060 0.102 0.083 0.140 0.060 0.080

0.161 0.125 0.080 0.143 0.179 0.101 0.083 0.042 0.082 0.277 0.193 0.102 0.143 0.120 0.074 0.250 0.250 0.140 0.140 0.102 0.208 0.100 0.060 0.135

0.232 0.208 0.280 0.163 0.107 0.141 0.125 0.125 0.204 0.043 0.108 0.082 0.071 0.160 0.130 0.021 0.167 0.180 0.120 0.163 0.167 0.080 0.240 0.145

0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.021 0.000 0.021 0.024 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.020 0.000 0.000 0.000 0.000 0.004

0.054 0.125 0.020 0.102 0.179 0.071 0.125 0.188 0.061 0.085 0.060 0.041 0.167 0.040 0.056 0.021 0.056 0.080 0.100 0.122 0.042 0.080 0.100 0.082

0.036 0.042 0.060 0.020 0.036 0.010 0.063 0.021 0.041 0.043 0.024 0.000 0.024 0.000 0.019 0.021 0.028 0.020 0.040 0.000 0.063 0.040 0.000 0.027

0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.020 0.000 0.000 0.000 0.000 0.001

0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

0.000 0.000 0.000 0.020 0.000 0.010 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.020 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.020 0.003

0.000 0.000 0.020 0.000 0.000 0.010 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.019 0.021 0.000 0.000 0.020 0.000 0.000 0.020 0.000 0.005

0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

Bantu-speakers pretoria (BN)

0.061

0.255

0.286

0.071

0.082

0.000

0.153

0.051

0.010

0.000

0.031

0.000

0.000

Sena-Mozambique (SE)

0.161

0.306

0.274

0.032

0.016

0.000

0.081

0.129

0.000

0.000

0.000

0.000

0.000

Grand Total

0.064

0.417

0.149

0.150

0.068

0.016

0.066

0.054

0.002

0.001

0.009

0.002

0.002

NRY UEP Haplogroup (according to the nomenclature proposed by Karafet et al. (2008))

 Armin Schwegler

and successfully established their little “Africa in the Americas” in the hinterlands of Cartagena de Indias. The fact that anthropological and linguistic data lead to the same conclusion naturally strengthens the DNA-based argument.

5.2.4  Resolving the Chewa paradox At first glance, the close Palenquero/Chewa kinship is bewildering, and has the potential for calling into question the overall validity of A.-P.’s research findings. The DNA correspondence with this Malawi group is troubling considering that during the critical period of Palenque’s formation (first half of 17th century), the Chewa were not embroiled in the slave trade to any significant extent.28 Therefore, a direct ancestral relationship can be confidently ruled out.29 The Chewa are Malawi’s largest ethnic group, the majority inhabiting the central area (Lilongwe, Dowa, and Ntchisi). The Chewa (Chichewa) tongue – also known as Nyanja30 or Chinyanja – is closely related to other Bantu languages. The Chewa also inhabit several regions along the bordering countries of Zambia, Tanzania, and Mozambique. Altogether, the Chewa number about 1.5 million people. However, as Gough explains, “the Chewa are not considered people of Malawi, nor people of Zambia, but people from the Nyanja group of Bantu” (2014 [2004]).31

.  During the 16th and especially 17th centuries, the Portuguese were already fairly active merchants on the east coast of Africa (Swahili Coast, including Mozambique). Vernet’s (2009) examination of historical sources (1500–1750) reveals, however, that the trade was predominantly material rather than human (slaves) in nature, and very modest in terms of shipment of number of slaves to the New World and Cartagena in particular. As noted most recently by Wheat (2015, p. 267), “Mozambique’s influence on the Americas is also generally limited, chronologically speaking, to the ninetheenth century; ninety-five percent of all known transatlantic slaving voyages [over 900 of them] originating in Southeastern Africa and the Indian Ocan islands took place after 1780” (2015, p. 267). .  The Portuguese had made some early contacts with the Chewa. Although the Portuguese did not reach the heartland of the Chewa chiefdom, there are well-documented records of contacts between 1608 and 1667 (Gough, 2014 [2004]). For the early history of Malawi, see Pachai (1972). Significant slave trade in Malawi did not begin until the 19th century, when under Omani rule over 40,000 slaves were sold through the Zanzibar slave market 〈http:// www.malawiproject.org/about-malawi/history/〉. For a map of the Major African regions that contributed to the transatlantic slave trade in the seventeenth and eighteenth centuries, visit 〈http://www.learnnc.org/lp/editions/nchistcolonial/1972〉. .  Literally ‘lake’, and hence ‘language of the lake(s)’. Cp. Chewa nyanja ya Malawi = ‘Lake (of) Malawi’ (〈http://translate.chichewadictionary.org/index.php〉). .  For the filiation of the Nyanja language, see Guthrie (Comparative Bantu 1971. II: 169 – N31a). For a recent comparative genome study of Malawians, see Joubert et al. (2010).



Combining population genetics (DNA) with historical linguistics 

Bakongoland

CHEWA MALAWI

Map 9.  Approximate location of the Chewa people of Malawi and their possible migratory route along the Kongo/Zaire River. The Chewa may also have followed other tributaries of the river Many Chewa traditions mention that presumably after their journey up-river from Bakongoland, the Chewa and other Malawi people migrated from Uluwa (Luba country) in the Katanga Province of the DRC (see black arrows). Ntara offers an approximate date between 1200 and 1500, and claims (1973 [1944/1945], p. 5) that a date nearer to 1500 is more likely. See also Banda (2002, pp. 12, 14, 16).

A historical explanation based on pre-colonial (14th or 15th-century) eastward migrations resolves the apparent Chewa paradox: as explained in Ntara’s The History of the Chewa (1973 [1944/1945]) and noted by A.-P. (2010, p. 57), as abundant oral tradition32 suggests, the Chewa migrated from the Congo to Malawi between 500–600 years ago, which preceded Colombia’s entry into the African slave trade (Cartagena de Indias, for instance, was not founded until 1533).33 Although the exact history s­ urrounding .  “This pioneer work entailed a great deal of time and effort being spent traveling throughout parts of present-day Malawi interviewing many old men who remembered the traditional history of their people. […] One of the strengths of this work is that it does include information from the primary sources – old men – which in many instances is no longer available.” Quoted from “Introduction” (p. ix) to Ntara (1973 [1944/1945]). .  This is one of a number of different interpretations of the early oral records of the Chewa. See also 〈http://en.wikipedia.org/wiki/Chewa_people〉. Language samples (given below) will illustrate the linguistic proximity between Kikongo, Chichewa, and Palo Monte vocabulary. For a useful diachronic map showing pre-colonial cultures of Africa (spanning roughly 500 BCE to 1500 CE), consult 〈http://en.wikipedia.org/wiki/File:African-civilizations-map-pre-colonial.svg〉. Unfortunately, the very informative website African Origins: Portal to African Origins: Portal to Africans Liberated from Transatlantic Slave Vessels at

 Armin Schwegler

Chewa’s exodus from the Old Kongo to Malawi is unknown, Map 9 reveals a route that traverses aside the main affluents of the Zaire/Kongo River, and eventually leads to the highlands and mountains of the East African Rift, as well as to Lake Tanganyika and Lake Mweru which lie to the northeast of Malawi. According to some sources,34 “(Chi)nyanja has its origin in the Eastern Province of Zambia from the 15th century to the 18th century.” The same source also claims that “[a] strong historical link of the Nyanja, Bemba and Yao people to the Shona Empire, who can point their earlier origins to Mashonaland, proves linguistically evident today” and that “[t]he ancient Shonas who temporarily dwelt in Malambo, a place in the DRC [Democratic Republic of the Congo], eventually shifted into northern Zambia, and then south and east into the highlands of Malawi.” When comparing the Bakongo and Chewa people, there exist striking similarities in language and socioreligious traditions, evidence which lends further credence to centuries-old historical genetic (DNA) link. While I cannot examine these linguistic and cultural similarities in any detail here, suffice it to say that in researching Chewa’s history and language, my familiarity with Kikongo and the Lengua of Palo Monte (Cuba) allowed me to readily identify the cognate nature of a dozen randomly selected core items (for a sampling, see Figure 3). Future research will need to determine whether or not these cognates owe their similarity to an old(er) Bantu common history (cp. Guthrie, 1971, and the recent assessments by Güldemann 2010 and Pakendorf, Bostoen, & Filippo, 2011), or whether (some of) these correspondences perhaps are powerful reminders that the ancestors of the Chewa – as well as the Paleros – originated in Bakongo-land.35

〈http://african-origins.org/〉 does not identify the historical origins of African slaves transported prior to the 19th century, therefore it is not relevant for the purpose of this paper. Similar observations are valid for The T ­ rans-Atlantic Slave Trade Database 〈http://slavevoyages.org/tast/index.faces〉 .  Source: 〈http://en.wikipedia.org/wiki/Chewa_language〉 .  Pakendo, Bostoen & Filippo rightly remind us that “no single modern-day Bantu language can be equated with ancestral language” and that “[i]n Africa, most individuals are multilingual and so there is no one-to-one correlation between linguistic and genetic affiliation, which poses a problem for the investigation of language dispersals using current affiliations” (2011, p. 53). However, the fact still remains that the correspondences – cultural, linguistic, and otherwise – between the Chewa and the Bakongo are striking, and need to be explained. Molecular perspectives (DNA comparison between Bachewa and Bakongo populations) in particular promise to provide novel, perhaps even definitive insights.



Combining population genetics (DNA) with historical linguistics 

Chewa (Malawi)36

Kikongo37

Lengua Palo Monte (Cuba)38

1. ‘village chief, king’

mfumu

mfúmu (L. 557)

mfumu (cp. V. C. 88)

2. ‘breasts’ (female)

mawere

mabèle (L. 28)

mabele (VC 118)

3. ‘water’

madzi

mási (Vili) (L. 504)

masi, mesi (VC 16)

4. ‘eye’

diso (= di-so) s.

di-isu (L. 124)

mesu ~ meso (VC 112)

5. ‘eyes’

maso (= ma-so) pl.

méso (pl.)(L. 550)~ madíisu (L. 124)

mesu ~ meso (VC 112)

6. ‘head’

mutu (= mu-ntu)

(mu)`ntu (L. 799)

ntu (VC 36)

7. ‘food, meal’

Chakudya

dya (L. 137) dia (L. 113; see also ndia in L. 668)

ndía, udia (VC 51)

8. ‘night, evening’

usiku (= u-siku)

ma-síka (L. 504)

masika (VC 108)

9. ‘moon(light)’

mwezi/miyezi

mw-ézi (L. 649) my-ézi (L. 649)

miesi (VC 94, see “luz de luna”)

10. ‘teeth’

ma-no (cp. singular dzino)

mi-inu (pl.) (L. 568)

menu, meno (VC 62)

11. ‘children’39

ana

á-ana, ‘aana (L. 3)

muana < mu40+ana (s./pl.)

12. ‘ten’

khumi

kuumi (L. 333)

kumi (VC 110)

Figure 3.  Sample Chewa (Nyanja) core vocabulary and cognates in Kikongo and Lengua (Palo Monte, Cuba)3637383940

.  For these and additional Chewa source materials (compiled by Steven Paas), visit: 〈http://translate.chichewadictionary.org/〉 〈http://en.wikipedia.org/wiki/Chewa_language〉 〈http://en.wikipedia.org/wiki/Chewa_language〉 (see “Sample phrases”) The fist of these links provides up-to-date information on the latest hardcopy edition of the Chichewa/ Chinyanja Dictionary. For the origins of Chichewa and its lexicography, readers should also consult Paas’ Rebmann: A Servant of God in Africa before the Rise of Western Colonialism (2011). Its Appendix II (pp. 252–266) surveys the history of Chichewa lexicography. For reviews of this work, see Hill (2012), and/or visit 〈http://www.chichewadictionary.org/johannes-rebmann/ book-reviews〉. .  Etymologies are from Laman (1964 [1936), here abbreviated “L.”. .  Sources are from Cabrera’s Vocabulario Congo (1984), abbreviated “VC”; due to the at times problematic nature of this and similar Palo Monte sources, some adjustments have been made to approximate to words to genuine Palero speech. .  Cp. also Palenquero moná ‘child, offspring’ < Kik. mw-‘áana ‘idem’. .  Mu = Kikongo class prefix, +singular/+human (Laman, 1964, p. 593).

 Armin Schwegler

The foregoing considerations surrounding the Chewa and their social and linguistic history are sufficiently informative so as to warrant A.-P.’s conclusion that “[t]he most parsimonious explanation of the observed [DNA] data is therefore that while the Yombe are the most closely related group to [the] Palenque, the Chewa are in turn also closely related to both the Palenque and the Yombe, but did not contribute to the Palenque NRY gene pool since they had migrated out of the area before slaves were transported by the Spanish to Colombia”41 (p. 57). We are therefore left with a circumstance in which the Yombe people may well be the sole direct African blood relatives of the Palenqueros, though Bakongos from neighboring or nearby geographic regions may also have been among the founders of Palenque.

5.3  Th  e relevance of shared articulatory features: Further thoughts on the common ancestry of the Bakongo, Chewa, Palenqueros, and Paleros The previous section establishes historical connections between the Bakongo, Chewa, Palenqueros, and Paleros that one could argue are rather general. However, to the specialist, it is certain minute dialectal details in their languages that perhaps most powerfully cement the hypothesis of a common Bakongo, and more specifically, northern Kongo (Yombe) ancestry. I now will focus on one such important detail. Before doing so, we must first consider critical language-external (historical) information surrounding the period when Kikongo entered into contact with European merchants and slave traders in the Old Kongo Kingdom. If our account in Section 5.2.4 above is correct, then the ancestors of the Palenqueros, Chewa, and Yombe resided in the Kingdom of Loango, a pre-colonial African state from approximately the 15th to the 19th century in what is now the Republic of the Congo. Then, as is the current reality now, its inhabitants were a branch of the Bakongo, and spoke a northern dialect of the Kikongo language, sometimes referred to as Fiote (for reasons made clear below, today this term is most often avoided due to its derogatory nature, and Kivili, Kiyombe and other ethnolinguistic labels are used instead). At its height in the seventeenth century, the Kingdom of Loango stretched from Cape St. Catherine in the north to almost the mouth of the Congo River. Missionaries and colonial merchants who would eventually visit the Loango coast tended to call the locals of Loango “Bafiote,” a derogatory term meaning “black person, negro”,42 and, as

.  In actuality it was the Portuguese, not the Spanish, who transported slaves to Colombia. But this is a detail that does not affect the validity of A.-P.’s overall conclusion. .  Cp. Swartenbroeckx (1973, p. 75: fiòti (Western Kikongo, archaic) ‘a black person’; the word is a derivation of fiòta (télé): to become black, dark, whence “fiòti.”’ Also Bentley: fiota ‘to be, grow black’ (1895, p. 834). Moñino (personal communication, May 5, 2014) reports that his recent Vili fieldnotes have the following entry from Matoombi-Loango: “fiota ‘black (color, hue)’ (final -a in fiota is unvoiced and is barely perceptible)”.



Combining population genetics (DNA) with historical linguistics 

reported in Swartenbroeckx (1973, p. 75),43 their native tongue fiòti (­Swartenbroeckx, 1973, p. 75). As noted below, the Loango people – also called Vili, Bavili, and, especially in older sources, Fiot(t)e, Fioti and Bafioti – entered into their first commercial contacts with Portuguese and then Dutch and French traders. It is most likely this common Fiote (Kikongo) vernacular of Loango that eventually evolved into the principal substrate of Palenquero and ritual Palo Monte of Cuba.44 In 1550, the Kingdom of Loango had become independent of the Kingdom of Kongo yet still remained loosely connected to that larger kingdom (Clark & Decalo,  2012, p. xxviii). By 1665 it had grown so independent that it eventually emerged as the capital of the Kongo Kingdom. The Vili of Loango began their first regular trade with the Portuguese in the 1570s, but it was not until the late 17th century and beyond that Vili merchants also engaged in extensive slave trade. As Martin’s short but very insightful study states, “[a] striking feature of the trade of Loango until about 1670, especially when compared with the trade of Luanda in the same period and with that of Loango in the eighteenth century, was the small volume of the slave trade” (1970, p. 147). While some of these slaves were acquired locally, many more were captured in various regions in the interior of the region. Throughout the 17th century (key for determining the ancestry of the Palenqueros), Loango remained a powerful kingdom (cp. Martin, 1970, 1972; S­ ommerdyk, 2012; Soret’s Les Kongo nordoccidentaux [1959] and Thomas [1997] are also most useful sources). With these language-external details in mind, let us now turn to the dialectal detail alluded to above that will assist us to further connect Palenque and Cuba (Palo Monte) with northern – rather than any other – varieties of Kikongo, especially the areas of the Old Loango where Vili and Yombe people reside. To that end, we will return to the earlier-cited Lumbalú song “Chi ma nkongo” and examine with more profundity the first word of its initial stanzas, i.e., chi: chi ma nkongo Chi ma nkongo, Chi ma luango, Chi ma ri Luango di Angola e;

From the Kongo From the Kongo people [I am], From the Loango people [I am], From the Luango people of Angola [I am], eh;

.  Laman (1964 [1936]) does not list these terms in his dictionary of Kikongo. .  I here concur with Moñino, who was first to highlight the central role that Fiote may have played: … el casi extinto fiot, un vehicular nacido de la lengua vili, hablado aún en los finales del siglo xix en la costa, desde el sur de Gabón hasta el sur de Cabinda. Este último, que presenta muchos portuguesismos integrados a su morfología bantú, fue probablemente la lengua de trata que los esclavos capturados por las fuerzas militares del rey de Loango aprendían fácilmente por su proximidad lingüística con las lenguas locales. (2012, p. 230)

 Armin Schwegler

In 1996, my attempt at deciphering Palenque’s Lumbalú chants was so to derive chi (unintelligible to Palenqueros) from Afro-Portuguese di ‘of, from,’ by way of a palatalization that is well attested elsewhere in the Afro-Iberian world (Schwegler, 1996, p. 527). However, Bunseki Fu Kiau’s extensive personal communications and especially his ability to readily connect the lyrics and melody of this archetypal Lumbalú to old Kongo songs convinced me that an alternative derivation – i.e. Palenquero chi < Kik. nsi ‘country, land, homeland, region, state, kingdom territory’ – is simultaneously more convincing and straightforward. Although this reanalysis does not affect the overall interpretation of the ritual lines under consideration, it does in all likelihood lead to a more faithful literal translation of the original text: Chi ma nkongo, [From] the land of the Kongo [I am], Chi ma luango, [From] the land of the Loango [people] [I am], Chi ma ri Luango di Angola e; [From] the land of the Luangos of Angola [I am], eh;

In much of Bakongo territory, nsi is pronounced as written (Laman, 1964 [1936], p. 764; Swartenbroeckx, 1973, p. 473), i.e. with a nasal element followed by an alveolar fricative (i.e.,[ns]). Therefore, the expected articulatory outcome in Afro-Hispanic speech (Palenque’s included) would be nsi > *si, *nsi, or *ensi, but not *chi (cp. Cuban Palero nsi ‘land’, Cabrera, 1984, p. 152). Expressed differently, Kikongo nsi – articulated [nsi] – could not have plausibly yielded Palenquero chi. However, in telling fashion, among the Vili and their Yombe neighbors to the east, /si/ ‘land’ is generally articulated not with an alveolar [s] but with a palatal [ʃ], as in [ʃi], or as [ts], as in [tsi] ‘land’.45 In Afro-Hispanic language (Lipski, 2005), [ʃi] predictably becomes chi [t͡ʃi], while tsi ‘land’ can potentially have a dual result, i.e., [t͡ʃi] or [si], though the former is far more probable. The upshot of these evolutionary phonetic details is as follows: only the Vili/Yombe forms of Kikongo nsi ‘land’ – articulated [ʃi] or [tsi] – can be the source of the first element in ritual Palenquero chi ma nkongo ‘land of the Kongo’. Ansari-Pour’s DNA-based analysis and the diachronic phonetic considerations offered here are thus mutually supportive in that they point to the northern Bakongo region as the focal point from where Palenqueros’ substrate originated.

.  I here base my argument on personal information from my colleague, Africanist Yves Moñino, who has carried out fieldwork in the Mayombe and the Vili region. See also Laman’s Introduction “9. Dialecte Vili” (1964, p. lxxvii-lxxxi), where it is noted that the prenasal in words like nsi tends to be omitted (see #20b and #25 of p. lxxviii) but that dialectal varieties often pronounce [ns] as [ts]. Once introduced in the Spanish world, this in Spanish inexistent [ts] would naturally shift to [t͡ʃ], a phonetic element that is common in Palenquero as well as Spanish. The shift from Kikongo nsi to Palenquero chi would thus have involved dialectal [ntsi] > [t͡ʃi] or [ntsi] > [ʃi] > [t͡ʃi].



Combining population genetics (DNA) with historical linguistics 

One final point deserves mention regarding the Congolese fiote connection we have sought to establish here. It cannot be coincidental that the Paleros in Cuba employ fiota or its phonetic variant fiuta to denote ‘Black person’. In light of the fact that Kik. fiòti ‘Black person; Fioti [vehicular] language’ was historically limited to (northern) coastal areas (Swartenbroeckx, 1973, p. 75), the word in question provides yet another indication that Paleros’ African legacy is intimately linked to the Loango region.

5.4  External historical data: Slave trade in Loango and the Mayombe Thus far, this study has based the crux of its monogenetically-oriented arguments on comparative linguistic and DNA data, which has been to the detriment of external historical information related to the slave trade. As Bickerton (1994, p. 65) once famously commented on the alleged African substrate influence on Saramaccan (a creole language of Suriname), in researching origins “one must show that the right speakers were in the right place at the right time” (1994, p. 65). Historical sources related to the slave trade in the Mayombe (shaded area in Map 10) and its adjacent territories indeed prove that Paleros’ and Palenqueros’ ancestors were “in the right place at the right time,” and importantly, shared the same language.

Mayombe L O Loango A N Pointe Noire G Malemba O Cabinda Boma

Africa

CONGO Kinshasa

Congo River

DEMOCRATIC REPUBLIC OF THE CONGO

ANGOLA Ambrizete Ambriz

Map adapted from Fuentes & Schwegler (2005: 33)

Map 10.  Mayombe and general area of West Central Africa from where Cuba’s Palero ­ancestors and most of Palenque’s maroons may originated

More than two hundred years ago, the French explorer Degranpré reported that at the time of the slave trade, “all these peoples spoke the same language [i.e., Kikongo]”

 Armin Schwegler

and in the second half of the 17th century, the provinces of Malemba, Cabinda, and Loango may have constituted “some 600,000 people” (Degranpré, 1801, pp. 167 and 216, respectively; my translation). As Heywood and Thornton note, “there is no continuous record of slave exports from Central Africa for most of the 17th century, though the Portuguese government did keep detailed tax books that have not survived” (2007, p. 160). Piecing together records from both the New and the Old World, the authors are nonetheless able to demonstrate that slave exports from the area in question were substantial, numbering in the thousands per year. For instance, they note that for the mid-1630s, annual exports from Luanda were around 15,000 slaves, before falling to 10,000 annually. The impact of this trade was felt in the Spanish West Indies, as well as elsewhere in Latin America where the slave trade was reaching its apogee. As Martin (1970, p. 147) explains, the hinterland of Loango (the Mayombe included) was probably only thinly populated prior to the middle of the 17th century, when Kongo groups started to move into the area of the Niari valley on the eastern side of the Mayombe. The Mayombe nonetheless played an important role in the slave trade, in part because “the roads from Loango to the interior had to pass through the treacherous Mayombe region, where the good paths were few and the routes passed along the sides of mountain slopes, by deep precipices, and through dense tropical forest” (Martin, 1972, p. 153). Most of the slaves were brought in caravans to the coast from distant regions, as described by Degranpré (1801), a French slave trader in the area for thirty years. Because of their coastal location, the Vili people had early and extensive (and therefore often privileged) contacts with European merchants and consequently became principal actors in the slave trade. This privileged position may explain (1) why Loango did not capture most of the slaves from its own region, but preferred to do so in neighboring Mayombe and beyond; (2) why European shippers – rather than establishing factories like in other parts of sub-Saharan Africa – anchored off shore and purchased slaves and colonial goods directly from local Vili officials; and (3) why Ansari-Pour’s DNA-based study points to Yombe rather than Vili people as the principal ancestors of the Palenqueros. But regardless, as is made amply clear by Proyart’s early account (1776) and more recent treaties on the history of West Central Africa,46 the external trade of the Loango Coast was such that the Loango-Vili-Yombe diaspora would – and understandably so – have a significant impact on nascent creole societies in the Americas.

.  For instance, Heywood (2002), Martin (1970, 1972); and, for a wider context, Thornton (1992) and Heywood & Thornton (2007). See also Thornton (1981).



Combining population genetics (DNA) with historical linguistics 

6.  Conclusions Part 1 of this investigation has sought to retrace how over fifty years of humanistic research have gradually favored a highly specific African origins hypothesis for two geographically distant Afro-Hispanic communities: Palenque (Colombia) and Palo Monte “families” (Cuba). It is certain that triangulation with data from the Old Kongo (especially the Loango/ Mayombe region) and these two communities produced respectable insights. However by 2010, linguists in particular still harbored lingering doubts concerning the monogenetic Kongo hypothesis, in part because there continue to exist a substantial body of Palenquero words (of incontestable sub-Saharan origin) that has yet to be etymologized. As I hope to have shown in this study, collaboration between linguists and population geneticists has now dissolved this theoretical impasse, and has dramatically shifted the momentum in the direction of the single origin (Kongo) hypothesis. As Ansari- Pour’s DNA-based investigations make clear, the Palenqueros indeed seem to have a narrowly definable genetic ancestry, located in the northwestern corner of Bakongoland. By examining research studies surrounding Palo Monte ritual speech, this paper has shown – as of yet without any DNA evidence from its practitioners47 – that the founders of Cuba’s Palero tradition must also have originated in this same Congolese homeland. To bolster this claim, we have placed the results in the greater context of the history of the slave trade, thereby confirming that the right speakers were indeed in the right place at the right time (pace Bickerton, 1994, p. 65). As we have documented, during the critical periods (17th century for Palenque, and 19th century for Cuba) the slave trade on the Loango Coast was intense, enduring, and far-reaching, involving both Cartagena de Indias and Havana on a grand scale. In arguing that “the momentum has decidedly shifted in the direction of monogenesis” we must, however, also be prudent and adopt a stance that recognizes the limitations inherent to the currently available data. In my view, at present there is value in taking a categorical stance in that the balance of arguments has been reversed: until

.  As regards to DNA research with a focus on African origins, contemporary Palo Monte “families” pose a challenge, as they are essentially “color blind.” Contrary to Palenque where ethnolinguistic uniformity has been preserved to a significant degree, Cuba’s Paleros today are members of very racially diverse groups including Blacks, mulattos, Caucasians, and Asians (the Chinese in particular are known to be adept practitioners, in no small measure because of their extensive familiarity with herbs used in magic practices). Especially in the Matanzas area, it should be possible to retrace the geneology of traditional Palero families to then compare their DNA to African groups. Regardless, Cuba is without a doubt an instance where genealogy, culture, ethnicity, and language have not always had the same fates.

 Armin Schwegler

now, the default assumptions were such that the burden of proof for monogenetic claims had squarely fallen on the proponents of a single-origin hypothesis. In light of the data presented here, the burden of proof now falls on those who continue to favor multiple substrate origins. From a strictly linguistic perspective this suggests that, unless incontrovertible evidence emerges that links Palenquero to African substrates outside of the Bakongo area, the monogenesis shall henceforth be the default working hypothesis. At the same time, I recognize that we do not yet possess sufficient evidence to make absolute claims about the supposed ethnolinguistic uniformity of Palenque’s founding fathers. While it is true that Ansari-Pour’s genetic tests cast a fairly wide net in terms of numbers of individuals sampled, there also is no question that more extensive DNA testing could yield a more complex foundational history, especially for Palenque’s female population. Furthermore, as noted earlier, Ansari-Pour’s sub-­ Saharan database does not include samples from West African areas known to have been important slave enclaves; similarly, and perhaps more importantly for this study, DNA data from Angola and adjacent Central West African territories are similarly absent from his study. Therefore, there continues to exist a substantial body of Palenquero words of incontestable sub-Saharan origin that has yet to be etymologized. As I have learned in my almost 30 years of fieldwork, research into Afro-­Hispanic language and culture is full of surprises, and just when the academic community seems to agree on a given hypothesis, fresh evidence often reverses the course of our thinking. Regardless, there is excellent reason to be optimistic about further collaboration with our colleagues in the biological sciences: population genetics is a rapidly advancing science, and the combination of fresh collections of DNA samples, novel techniques, and additional primary data from the field48 may soon reveal the relative chronology of Palenquero generations, thereby shedding light, for instance, on when exactly the earlier-mentioned Caucasian DNA entered Palenque’s genetic pool. When and until such future collaboration takes place, we will be well advised to remain cautious about our overall interpretation, a point made (and duly noted) by geneticist Brigitte Pakendorf (CNRS, Lyon) in a recent personal communication. Such collaboration will want (1) a larger DNA coverage of African populations, (2) Y-chromosomal short tandem repeat data in addition to the Y-chromosomal SNPs that were typed up to now, (3) mtDNA data for the maternal perspectives, as well as (4) autosomal DNA data from all relevant populations. As a final cautionary note, we should also remember that Ansari-Pour and his collaborators are merely the first team of geneticists to embark on a comparison of Palenquero and sub-Saharan population groups. In the future, perhaps more senior geneticists may offer novel perspectives and new findings

.  See, for instance, the recent volume by Maglia & Moñino Kondalo pa bibí mejó/ Contarlo para vivir mejor. Oratura y oralitura de San Basilio de Palenque (Colombia), published in 2015.



Combining population genetics (DNA) with historical linguistics 

to shed light on the central topic – the Palenqueros’ African origins – that has fascinated us here. Regardless, a preliminary conclusion to be drawn from this study is that African slave societies in places like Cartagena must have been socially and ethnolinguistically far more cohesive than has been assumed, thereby effectively safeguarding ancestral speech and customs for multiple generations. It is impossible to determine today the extent to which mastery of African languages like Kikongo played a role in slaves’ successful escape from white domination. However, the fact that the early inhabitants of Palenque were so uniformly Bakongo does suggest that shared linguistic, religious, and social practices may have been key factors in their securement of long-term freedom. It is my hope that the findings of this study provide new and engaging considerations surrounding a variety of interrelated topics. The evolution of bozal Spanish – as found in Palo Monte or the earliest generations of Palenqueros – and/or Spanishbased New World creoles continues to be a matter of profound debate (McWhorter, 2000; L ­ ipski,  2005; Clements, 2009; Schwegler, 2010, Schwegler & Rojas-Primus, 2010; Sessarego, 2013, 2014, 2015, in press). The crux of the issue lies in whether, in predominantly Black Spanish-speaking territories (Cartagena included), this pidginized interlingua subsequently creolized in locally-born generations, thus creating the kind of extended Spanish/creole bilingualism found in Palenque to this day. To some (the author of this study included), Palenque thus represents a direct continuation of this hypothesized and more widespread former creole/Spanish multilingualism. The fact that the early Palenque can be shown to have had a substantial (if not uniform) Bakongo configuration in my mind tilts the debate in the direction of the more widespread Afro-Hispanic creole hypothesis. This, I conclude, is for a simple reason: if Kikongo was indeed the sole African language spoken by bozales in early Palenque, they presumably would have clung to their ancestral speech (Kikongo) rather than invent a New World creole.49 If, on the contrary, slaves in the region generally made in-home use of a creole contact vernacular, its adoption in an emerging maroon community would have been a logical consequence of reigning sociolinguistic practices. One final point: during the course of this study, I have repeatedly characterized Palenque and Cuba’s Palo Monte as exceptional cases of language preservation. As we learn more about the Afro-Hispanic diaspora, further such cases of linguistic and

.  I concur with Lipski in that “[t]he presence of numerous Kikongo lexical items in Palenquero suggests that Kikongo was once spoken in the emerging maroon community” (Lipski, 2009, p. 126). I should note here that, as far as we know, the Yombe region never harbored a Portuguese creole similar to Palenquero.

 Armin Schwegler

c­ ultural survivals with highly locale-specific links to sub-Saharan Africa may emerge. Just this past week I learned of Emma Christopher’s (Senior Lecturer of History, Sydney University) long-term investigation into the community of Ganga-Longoba of Perico (Matanzas, Cuba), a community that has been singing the same “African” chants for generations. Christopher, too, initially assumed that these ceremonies resulted from extensive ethnic and cultural admixture, but after a two-year pursuit in Africa, she discovered that the chants came from just one village in a remote region of Sierra Leone. Her research finally led her to Mokpangumba (southern Sierra Leone), “where villagers not only identified the Banta language but recognized songs and dances from the initiation ceremony for their own secret society, devoted to healing” (Rainsford, 2014).50

Palenque

Mayombe

Figure 4.  Palenquero and Mayombe children Their resemblance has a highly probable explanation: a shared Bakongo ancestry

References African origins: Portal to African origins: Portal to Africans liberated from transatlantic slave vessels at 〈http://african-origins.org/〉 Ansari-Pour, N. (2010). Human genetic variation, relationships of peoples of sub-Saharan Africa and implications for healthcare (Published doctoral dissertation). University College London. [Slight revisions made in 2011] Ansari-Pour, N., Plaster C. A., & Bradman, N. (2013). Evidence from Y-chromosome analysis for a late exclusively eastern expansion of Bantu-speaking people. European Journal

.  For a video sample of this Afro-Cuban tradition, see 〈http://www.bbc.com/news/worldlatin-america-25876023〉.



Combining population genetics (DNA) with historical linguistics 

of Human Genetics, 21(4), 423–429. Retrieved from 〈http://www.nature.com/ejhg/journal/ v21/n4/full/ejhg2012176a.html.  doi: 10.1038/ejhg.2012.176 Ansari-Pour, N., Moñino, Y., Bradman, N., et al. (2014). African genetic descent in a village in Colombia. Manuscript in preparation. Atlas linguistique de l“Afrique Centrale (ALAC): Atlas linguistique du Zaïre (inventaire préliminaire). (1983). Paris: Agence de coopération culturelle et technique. Yaoundé, Cameroon: CERDOTOLA. Banda, L. E. C. (2002). The Chewa kingdom. Lusaka: No publisher. Bartens, A. (1995). Die iberoromanischen Kreolsprachen. Frankfurt: Peter Lang. Bickerton, D., & Escalante, A. (1970). Palenquero: A Spanish-based creole of northern Colombia. Lingua, 24, 254–267. Bickerton, D. (1994). The origins of Saramaccan syntax: A reply to John McWhorter’s ­“Substratal influence in Saramaccan serial verb constructions”. Journal of Pidgin and Creole Languages, 9, 65–77.  doi: 10.1075/jpcl.9.1.06bic Bittremieux, L. (1922). Mayombsch idioticon. Gent: Drukkerij Erasmus. Bittremieux, L. (1936). La société secrète des Bakhimba au Mayombe. Mémoires de l’Institut Royal Colonial Belge, 5(3). Brussels: Georges van Campenhout. Boas, F. (1911). Race and language. In F. Boas (Ed.), Handbook of American Indian languages (pp. 5–14). Washington, DC: Smithsonian Institution. Cabrera, L. (1971 [1954]). El monte. Igbo finda, ewe orisha, vititinfinda (notas sobre las religiones, la magia, las supersticiones y el folklore de los negros criollos y del pueblo de Cuba). Miami: Colección del Chicherekú. Cabrera, L. (1979). Reglas de Congo: Palo Monte Mayombé. Miami: Ediciones CR. Cabrera, L. (1984). Vocabulario Congo (el bantú que se habla en Cuba). Miami: Colección del Chichereku. Cásseres Estrada, S. (2005). Diccionario de la lengua afro palenquera español [sic]. Mompox: Ediciones Pluma. Clark, J. F., & Decalo, S. (2012). Historical dictionary of the Republic of the Congo (4th edition). Plymouth: Scarecrow Press. Clements, J. C. (2009). The linguistic legacy of Spanish and Portuguese: Colonial expansion and language change. Cambridge: Cambridge University Press.  doi: 10.1017/CBO9780511576171 Corominas, J., & J. A. Pascual (1985). Diccionario crítico etimológico castellano e hispánico. Madrid: Gredos. Davidson, J. M. (2012). “They laid planks ‘crost the coffins”: The African origin of grave vaulting in the United States. International Journal of Historical Archeology, 16, 86–134. doi: 10.1007/s10761-012-0170-5 Degranpré, L. (1801). Voyage à la Côte Occidentale d’Afrique, fait dans les années 1786 et 1787. Vol. 1. Paris: Imprimeur-Libraire Palais de Tribunal. Del Castillo, N. (1982). Esclavos negros en Cartagena y sus aportes léxicos. Bogota: Instituto Caro y Cuervo. Del Castillo, N. (1984). El léxico negro-africano de San Basilio de Palenque. Thesaurus, 39, 80–169. Del Castillo, N. (1985). Bantuismos en Cartagena de Indias: Vegetales, alimentos y bebidas. Muntu, 2, 85–109. Del Castillo, N. (1992). El aporte negro-africano al léxico de Colombia. In Presencia y destino: el español de América hacia el siglo XXI (Vol. 2; pp. 42–99). Bogota: Instituto Caro y Cuervo. Escalante, A. (1964). El negro en Colombia (Monografías Sociológicas, 18). Bogota: Universidad Nacional de Colombia.

 Armin Schwegler Escalante, A. (1979 [1954]). El Palenque de San Basilio. Barranquilla: Editorial Mejoras. [First published in 1954 under the title “Notas sobre El Palenque de San Basilio, una comunidad negra de Colombia”. Divulgaciones Etnológicas, 3(5), 207–358.]. Escalante, A. (1989). Significado del lumbalú, ritual funerario de Palenque de San Basilio. Huellas, 26, 11–24. Filippo, C. de, et al. (2011). Y-chromosomal variation in sub-Saharan Africa: Insights into the history of Niger-Congo groups. Mol. Biol. Evol., 28(3), 1255–1269. doi: 10.1093/molbev/msq312 Friedemann, N. S. de, & Cross, R. (1979). Ma ngombe. Guerreros y ganaderos en Palenque. Bogota: Carlos Valencia Editores. Friedemann, N. S. de, & Patiño Rosselli, C. (1983). Lengua y sociedad en el Palenque de San Basilio. Bogota: Instituto Caro y Cuervo. Fuentes Guerra, J., & Schwegler, A. (2005). Lengua y ritos del Palo Monte Mayombe (Dioses cubanos y sus fuentes africanas). Madrid: Iberoamericana. Fuentes Guerra, J. & Schwegler, A. (2012). Prácticas rituales afrocubanas: Deidades Kimbisa (Palo Monte) y sus fuentes kikongo. In M. L. Machado Caicedo (Ed.), La diáspora africana. Un legado de resistencia y emancipación (pp. 211–253). Amsterdam: National Institute for the Study of Dutch Slavery/NINSEE – Cali: Fundación Universitaria Claretian/FUCLA y Universidad del Valle. Fuentes Guerra, J., & Schwegler, A. (2014). El origen kongo del Palo Monte: Una aproximación etnolingüística. UniverSOS, 11, 9–61. Retrieved from 〈http://www.uv.es/~calvo/amerin dias/4_0.htm〉 Gómez Zuluaga, F. P. (2014). The circulation of bodily knowledge in the seventeenth Black Spanish Caribbean. Social History of Medicine, 26(3), 383–402.  doi: 10.1093/shm/hkt014 Gómez Zuluaga, F. P. (2010). Bodies of encounter: Health, illness and death in the early modern African-Spanish Caribbean (Unpublished doctoral dissertation). Vanderbilt University, Nashville, TN. Gough, A. (2014 [2004]). The Chewa people. Retrieved from: 〈http://www.peoplesoftheworld. org/hosted/chewa/〉 (9 May 2014). Granda, G. de. (1968). La tipología “criolla” de dos hablas del área lingüística hispánica. ­Thesaurus, 23, 193–205. Granda, G. de. (1971). Sobre la procedencia africana del habla “criolla” de San Basilio de Palenque (Departamento de Bolívar, Colombia). Thesaurus, 26, 84–94. [Reprinted in Granda 1978, Estudios lingüísticos hispánicos, afrohispánicos y criollos. Madrid: Gredos, pp. 441–452]. Guattini, A. M., & Carli, D. de. (1674). Viaggio del Padre Michael Angelo de Guattini da Reggio, et del P[adre] Dionigi de Carli da Piacenza Capuccini, predicatori & missionari apostolici nel regno del Congo. Bologna: G. Longhi. [Surdich, Francesco. 1997. Viaggio nel Regno del Congo. Cinisello Balsamo: Edizioni San Paolo. (French translation): Guattini, Angelo Michele, & Carli, Dionigi de. 2006. La mission au Kongo des pères Michelangelo Guattini & Dionigi Carli (1668) (A. du C. d’Abzac, Trans.). Paris: Chandeigne.]. Güldemann, T. (2010). Proto-Bantu and Proto-Niger-Congo: Macro-areal typology and linguistic reconstruction. In C. König & O. Hieda (Eds.), International Symposium of the Center of Corpus-based Linguistics and Language Education (CbLLE) (pp. 109–142). Amsterdam: John Benjamins.  doi: 10.1075/tufs.1 Guthrie, M. (1971). Comparative Bantu (vols. 1–4). Farnborough Hants: Gregg International Publishers.



Combining population genetics (DNA) with historical linguistics 

Hagenbucher-Sacripanti, F. (1973). Les fondements spirituels du pouvoir au royaume de Loango. Paris: ORSTOM. Heywood, L. (Ed.). (2002). Central Africans and cultural transformations in the American diaspora. Cambridge: Cambridge University Press. Heywood, L., & Thornton, J. K. (2007). Central Africans, Atlantic Creoles and the foundation of the Americas, 1585–1660. Cambridge: Cambridge University Press. Hill, P. (2012). Review of Paas (2011), Johannes Rebmann: As a servant of God in Africa before the rise of western colonialism. Nazareth Evangelical Theological Seminary (Dec. 2012), (pp.  1–2). Retrieved from: 〈http://www.nazarethseminary.org/datadir/en-events/ev97/ files/Hill%20Review%20of%20Paas.pdf〉 Hualde, J. I., & Schwegler, A. (2008). Intonation in Palenquero. Journal of Pidgin and Creole Languages, 23(1), 1–31.  doi: 10.1075/jpcl.23.1.02hua Huttar, G. (1993). Identifying Africanisms in New World languages: How specific can we get? In S. Mufwene (Ed.), Africanisms in Afro-American language varieties (pp. 47–63). Athens, GA: The University of Georgia Press. Jorde, L. B., & Wooding, S. P. (2004). Genetic variation, classification, and “race”. Nature Genetics, 36(11, Suppl. S), pp. S28–S33 (November 2004).  doi: 10.1034/ng1435 Joubert, B. R. et al. (2010). Comparison of genome-wide variation between Malawians and African ancestry HapMap populations. Journal of Human Genetics, 55, 366–374. doi: 10.1038/jhg.2010.41 Klee, C., & Lynch, A. (2009). El español en contacto con otras lenguas. Washington, DC: Georgetown University Press. Laman, K. E. (1964 [1936]). Dictionnaire kikongo-français (Vols. 1–2). Ridgewood, NJ: The Gregg Press. Landa, B. M. (2014). A comparison between the development of Palenquero Caribbean Creole and Swahili-Spanish interlanguage. In M. Morgan & V. Youssef (Eds.), Reassembling the fragments. Voice and identity in Caribbean discourse (pp. 127–142). Kingston: University of the West Indies Press. Lewis, A. R. (1970). A descriptive analysis of the Palenquero dialect (Unpublished MA thesis). University of the West Indies, Mona, Jamaica. Lipski, J. M. (1994). Latin American Spanish. New York, NY: Longman. Lipski, J. M. (2005). A history of Afro-Hispanic language. Five centuries, five continents. Cambridge: Cambridge University Press.  doi: 10.1017/CBO9780511627811 Lipski, J. M. (2009). Pitch polarity in Palenquero: A possible locus of H tone. In S. Colina, A. Olarrea, & A. Carvalho (Eds.), Romance Linguistics 2009: Selected Papers from the 39th Linguistic Symposium on Romance Languages (pp. 111–128). Amsterdam: John Benjamins. doi: 10.1075/cilt.315.07lip Loëmbe, G. (2005). Parlons vili: Langue et culture de Loango (Collection Parlons, 215). Paris: L’Harmattan. Megenney, W. W. (1976). El elemento sub-sahárico en el léxico costeño de Colombia. Revista Española de Lingüística, 6, 405–451. Maglia, G., & Moñino, Y. (2015). Kondalo pa bibí mejó/Contarlo para vivir mejor. Oratura y oralitura de San Basilio de Palenque (Colombia). Co-edited by Editorial Javeriana and Instituto Caro y Cuervo (Bogota) and CNRS (Paris). Maglia, G., & Schwegler, A. (Eds.). (2012). Palenque (Colombia): Oralidad, identidad y resistencia. Un enfoque interdisciplinario. Bogota: Instituto Caro y Cuervo & Universidad Javeriana.

 Armin Schwegler Maho, J. F. (2002). Comparative overview of three Bantu classifications. Göteburgo: Göteborg University, Department of Oriental and African Languages. (Available on-line). Maho, J. F. (2003). A classification of the Bantu languages: An update of Guthrie’s referential system. In D. Nurse & G. Philippson (Eds.), The Bantu languages (pp. 639–651). London: Curzon Press. Maho, J. F. (2008a). Indices to Bantu languages (An accompanying volume to “The new updated Guthrie list”). Munich: Lincom. Maho, J. F. (2008b). A referential classification of Bantu. NUGL online. The online version of the New updated Guthrie list, a referential classification of the Bantu languages. Maho, J. F. (2009). NUGL online. The online version of the New updated Guthrie list, a referential classification of the Bantu languages. Retrieved from: 〈http://goto.glocalnet.net/maho/ papers.html〉 (4 June 2009). Martin, P. (1970). The trade of Loango in the seventeenth and eighteenth centuries. In R. Gray & D. Birmingham (Eds.), Pre-colonial African trade (Chapter 7; pp. 139–161). London: Oxford University Press. Martin, P. (1972). The external trade of the Loango Coast, 1576–1870: The effects of changing commercial relations on the Vili kingdom of Loango. Oxford: Oxford University Press. Martin, P. (1986). Power, cloth and currency on the Loango Coast. Retrieved from: 〈https:// scholarworks.iu.edu/dspace/bitstream/handle/2022/3413/power+cloth+and=+currency. pdf?sequence=1〉 Mattos Frisvold, N. de. (2011). Palo Mayombe: The garden of blood and bones. No place of publication: Scarlet Imprint. Mavoungou, P. A., & Ndinga-Koumba-Binz, H. S. (2010). Civili, langue des Baloango: Esquisse historique et linguistique.Munich: Lincom. McWhorter, J. (2000). The missing Spanish creoles: Recovering the birth of plantation contact languages. Berkeley, CA: University of California Press. Megenney, W. W. (1976). El elemento subsahárico en el léxico costeño de Colombia. Revista Española de Lingüística, 6, 405–451. Megenney, W. W. (1980). Sub-Saharan influences in Palenquero and Barloventero. Revista/ Review Interamericana, 10, 143–155. Megenney, W. W. (1986). El palenquero. Un lenguaje post-criollo de Colombia. Bogota: Instituto Caro y Cuervo. Megenney, W. W. (1996). Mezclas culturales y raciales en la América Latina: Sus reflejos en la literatura, el lenguaje, la religión, y la música. Diaspora (Southern Arkansas University), 5, 1–44. Moñino, Y. (1999). L’aspect en palenquero: Une sémantaxe africaine. Actances, 10, 177–190. Moñino, Y. (2002). Las construcciones de genitivo en palenquero: ¿una semantaxis africana? In Y. Moñino & A. Schwegler (Eds.), pp. 227–248.  doi: 10.1515/9783110960228 Moñino, Y. (2007a). Convergencias lingüísticas iberocongolesas en palenquero: ¿integrarse a la sociedad mayoritaria o distinguirse de ella? In W. Mihatsch & M. Sokol (Eds.), Language contact and language change in the Caribbean and beyond – Lenguas en contacto y cambio lingüístico en el Caribe y más allá (pp. 37–58). Frankfurt: Peter Lang. Moñino, Y. (2007b). Les rôles du substrat dans les créoles et les langues secrètes: Le cas du palenquero, créole espagnol de Colombie. In K. Gadelii & A. Zribi-Hertz (Eds.), Grammaires créoles et grammaire comparative (pp. 49–72). Saint-Denis: Presses Universitaires de Vincennes. Moñino, Y. (2012). Pasado, presente y futuro de la lengua de Palenque. In G. Maglia & A. Schwegler (Eds.), Palenque (Colombia): Oralidad, identidad y resistencia. Un enfoque interdisciplinario (pp. 221–255). Bogota: Instituto Caro y Cuervo & Universidad Javeriana.



Combining population genetics (DNA) with historical linguistics 

Moñino, Y., & Ortiz, C. (1999). Réinterprétation de deux procédés de morphologie évaluative dans un créole de Colombie. SiLeXicales, 2, 253–261. Moñino, Y., & Schwegler, A. (Eds.). (2002). Palenque, Cartagena y Afro-Caribe: historia y lengua. Tübingen: Max Niemeyer.  doi: 10.1515/9783110960228 Morton, T. (2005). Sociolinguistic variation and language change in El Palenque de San Basilio (Colombia) (Unpublished doctoral dissertation). University of Pennsylvania, Philadelphia. Mufwene, S. (Ed.). (1993). Africanisms in Afro-American language varieties. Athens, GA: The University of Georgia Press. Navarrete, M. C. (2003). Cimarrones y palenques en el siglo XVII. Cali: Editorial Facultad de Humanidades. Navarrete, M. C. (2008). San Basilio de Palenque: Memoria y tradición. Cali: Programa Editorial Universidad del Valle. Navarrete, M. C. (2010). Nuevos aspectos en la historia de los palenques y los cimarrones del Caribe negroandino, siglos xvi–xvii. In J. M. de la Serna (Ed.), Africanos y ­afrodescendientes en Iberoamérica (pp. 23–81). Mexico, DF: UNAM, Centro de Investigaciones sobre América Latina y el Caribe; Mexicanos y Centroamericanos. Navarrete, M. C. (2011). El palenque de Limón (Cartagena de Indias, siglo xvii): El imaginario del poder y sus jerarquías. In J. M. de la Serna (Ed.), Vicisitudes negro africanas en Iberoamérica: Experiencias de investigación (Volume 7; pp. 101–134). Mexico, DF: Universidad Autónoma de México. Newson, L. A., & Minchin, S. (2004). Slave mortality and African origins: A view from Cartagena, Colombia in the early seventeenth century. Slavery and Abolition, 25, 18–43. doi: 10.1080/0144039042000302224 Newson, L.A., & Minchin, S. (2007a). Cargazones de negros en Cartagena de Indias en el siglo XVII: Nutrición, salud y mortalidad. In A. Meisel & H. Calvo (Eds.), Cartagena de Indias en el siglo XVII (pp. 206–247). Bogota: Banco de la República. Newson, L. A., & Minchin, S. (2007b). From capture to sale: The Portuguese slave trade to Spanish South America in the early seventeenth century. Leiden: Brill. doi: 10.1163/ej.9789004156791.i-373 Noguera, M. C., Schwegler, A., & Gusmão, L., et al. (2014). Colombia’s population crucible: Y chromosome evidence from six admixed communities in the Department of Bolivar. Annals of Human Biology, 41(5), 453–459. Retrieved from: 〈http://informahealthcare.com/ eprint/tGM9Tr3eb239bP5R7Vzu/full〉 (11 November, 2013. doi: 10.3109/03014460.2013.852244 Nsondé, J. de D. (1999). Parlons kikôngo: le lâri de Brazzaville et sa culture. Paris: L’Harmattan. Ntara, S. J. (1973 [1944/1945]). The history of the Chewa. (W. S. K. Jere, Trans., with commentary by H. W. Langworthy). Wiesbaden: Franz Steiner. (First published as Mbiri ya Achewa). Ortiz, F. (1965 [1950]). La africanía de la música folklórica de Cuba. Havana: Editora Universitaria. Paas, S. (2011). Johannes Rebmann: A servant of God in Africa before the rise of Western colonialism. Nürnberg: Verlag für Theologie und Religionswissenschaft. Paas, S. (2013). Dictionary/ Mtanthauziramawu Chichewa/ Chinyanja – English and English – Chichewa/ Chinyanja (4th ed.). Blantyre, Malawi: Christian Literature Action in Malawi (CLAIM) and Foundation Heart for Malawi (FHFM). Retrieved from: 〈http://translate. chichewadictionary.org〉 (Printed in Nanjing, China, by Amity). Pachai, B. (Ed.). (1972). The early history of Malawi. London: Longmans. Pakendo, B., Bostoen, K., & F., Cesare de. (2011). Molecular perspectives on the Bantu expansion: A synthesis. Language Dynamics and Change, 1, 50–88.  doi: 10.1163/221058211X570349

 Armin Schwegler Pérez M., B. (2011). Hablemos palenquero a través de los cuentos. Chitieno lengua ku ma kuendo. Cartagena: Pluma de Mompox. Pérez M., B., & Miranda Reyes, R. (Eds.). (2012). Alepuela e ñá Bisenta. Katiya pa insiñá mu kombetsá ri Palenge. Cartilla para la enseñanza de la lengua palenquera. Cartagena: Corporación Ataole/Ediciones Pluma de Mompox. Proyart, L.-B. A. (1776). Histoire de Loango, Kakongo et de autres royaumes d’Afrique. Paris: C. P.  Berton Librairie & N. Crapart. Rainsford, S. (2014). Cubans trace roots to remote Sierra Leone village. BBC News, Latin America & Caribbean. Retrieved from: 〈http://www.bbc.com/news/world-latin-­america25876023〉 (9 May 2014). Rodas, C., Gelvez, N., & Keyeux, G. (2003). Mitochondrial DNA studies show asymmetrical Amerindian admixture in Afro-Colombian and mestizo populations. Human Biology, 75, 13–30.  doi: 10.1353/hub.2003.0026 Rojas-Primus, C. (2009). Lengua ritual y sincretismo: Dinámicas de hibridez en el discurso mágico-religioso Palo Monte. Saarbrücken: VDM Verlag. Sandoval, Alonso de. 1987 (1627). De instauranda aethiopum salute. Un tratado sobre la esclavitud. Introduction and transcription by Enriqueta Vila Vilar. Madrid: Alianza Editorial. Schwegler, A. (1989). Notas etimológicas palenqueras: Casariambe, túngananá, agüé, monicongo, maricongo, y otras voces africanas y pseudo-africanas. Thesaurus, 44, 1–28. Schwegler, A. (1990). Abrakabraka, suebbesuebbe, tando, kobbejó, lungá y otras voces palenqueras: Sus orígenes e importancia para el estudio de dialectos afrohispanocaribeños. Thesaurus, 45, 690–731. Schwegler, A. (1992). Hacia una arqueología afrocolombiana: Restos de tradiciones religiosas bantúes en una comunidad negroamericana. América Negra, 4, 35–82. Schwegler, A. (1996). Chi ma nkongo: Lengua y rito ancestrales en El Palenque de San Basilio (Colombia) (Vols. 1–2). Madrid: Iberoamericana. Schwegler, A. (1998). Palenquero. In M. Perl & A. Schwegler (Eds.), América negra: Panorámica actual de los estudios lingüísticos sobre variedades criollas y afrohispanas (pp. 220–291). Frankfurt: Vervuert. Schwegler, A. (1999). El vocabulario africano de Palenque (Colombia). Segunda Parte: compendio de palabras (con etimologías). In L. Ortiz (Ed.), El Caribe hispánico: Perspectivas lingüísticas actuales (Homenaje a Manuel Álvarez Nazario) (pp. 171–253). Madrid: Iberoamericana. (Revised version in Yves Moñino & Armin Schwegler (Eds.), Palenque, Cartagena y AfroCaribe: Historia y lingüística [2002] (pp. 171–227). Tübingen: Max Niemeyer.) Schwegler, A. (2000a). The African vocabulary of Palenque (Colombia). Part 1: Introduction and corpus of previously undocumented Afro-Palenquerisms. Journal of Pidgin and Creole Language, 15, 241–312.  doi: 10.1075/jpcl.15.2.02sch Schwegler, A. (2000b). On the (sensational) survival of Kikongo in 20th-century Cuba. Journal of Pidgin and Creole Languages, 15, 159–164.  doi: 10.1075/jpcl.15.1.10sch Schwegler, A. (2002). On the (African) origins of Palenquero subject pronouns. Diachronica, 19(2), 273–332.  doi: 10.1075/dia.19.2.03sch Schwegler, A. (2006a). Bantu elements in Palenque (Colombia): Anthropological, archeological and linguistic evidence. In B. Haviser & K. C. MacDonald (Eds.), African re-genesis: Confronting social issues in the diaspora (pp. 204–222). London: University College London Press. Schwegler, A. (2006b). Bozal Spanish: Captivating new evidence from a contemporary source (Afro-Cuban “Palo Monte”). In J. Fuller & L. L. Thornburg (Eds.), Studies in contact linguistics: Essays in honor of Glenn G. Gilbert (pp. 71–101). Bern: Peter Lang.



Combining population genetics (DNA) with historical linguistics 

Schwegler, A. (2007). Black ritual insulting in the Americas: On the art of “vociferar” (Colombia), “vacilar” (Ecuador) and “snapping”, “sounding” or “playing the dozens” (USA). Indiana, 24, 105–155. Retrieved from: 〈http://www.iai.spk-berlin.de/publikationen/indiana/ bisherige-ausgaben/indiana-24.html〉 Schwegler, A. (2010). State of the discipline. Pidgin and creole studies: Their interface with Hispanic and Lusophone linguistics. Studies in Hispanic and Lusophone Linguistics, 3(2), 431–481.  doi: 10.1515/shll-2010-1082 Schwegler, A. (2011a). Palenque(ro): The search for its African substrate. In C. Lefebvre (Ed.), Creoles, their substrates, and language typology (pp. 225–249). Amsterdam: John Benjamins. doi: 10.1075/tsl.95.13sch Schwegler, A. (2011b). Palenque, Colombia: Multilingualism in an extraordinary social and historical context. In M. Díaz-Campos (Ed.), The handbook of Hispanic sociolinguistics (pp. 446–472). Malden, MA: Wiley-Blackwell.  doi: 10.1002/9781444393446.ch21 Schwegler, A. (2011c). On the extraordinary revival of a creole: Palenquero (Colombia). In M. Haboud & N. Ostler (Eds.), Endangered languages – Voices and images (pp. 153–165). Bath: Foundation for Endangered Languages. Schwegler, A. (2012). Sobre el origen africano de la lengua criolla de Palenque (Colombia). [Much augmented Spanish version of “Palenque(ro): The search for its African substrate”, Schwegler 2011 above]. In Maglia & Schwegler (Eds.), pp. 107–179. [Translated English version to appear in A. Schwegler, B. Kirschen & G. Maglia (Eds.), Palenque (Colombia): Orality, identity, and resistance. An interdisciplinary approach. Forthcoming in Creole Language Library series. Amsterdam: John Benjamins.] Schwegler, A. (2014a). Reply to “Commentaries”. UniverSOS, 11, 97–106. Retrieved from: 〈http://www.uv.es/~calvo/amerindias/4_0.htm〉 Schwegler, A. (2014b). Portuguese remnants in the Afro-Hispanic diaspora. In A. M. ­Carvalho & P. Amaral (Eds.), Portuguese-Spanish interfaces: Diachrony, synchrony, and contact (pp. ­403–441). Amsterdam: John Benjamins.  doi: 10.1075/ihll.1.21sch Schwegler, A., & Rojas-Primus, C. (2010). La “lengua” ritual del Palo Monte (Cuba): estudio comparativo (Holguín & Cienfuegos). Revista Internacional de Lingüística Iberoamericana (RILI), 15, 187–244. Sessarego, S. (2013). Chota Valley Spanish. Madrid: Iberoamericana. Sessarego, S. (2014). The Afro-Bolivian Spanish determiner phrase: A microparametric account. Columbus OH: Ohio State University Press. Sessarego, S. (2015). Afro-Peruvian Spanish: Spanish slavery and the legacy of Spanish Creoles. Amsterdam: John Benjamins.  doi: 10.1075/cll.51 Sessarego, S. (In press). 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. Sheller, M. (2003). Consuming the Caribbean. From Arawaks to zombies. London: Routledge. Simarra Reyes, L., & Triviño Doval, A. E. (2012a [2007]). Gramática de la lengua palenquera: introducción para principiantes. Gramátika ri luénga [sic] Palénque [sic]: pa ma lo ke tan komensá. Cartagena: Ministerio de Cultura/Corpoataole/Ediciones Pluma de Mompox. Simarra Reyes, L., & Triviño Doval, A. E. (2012b). Diccionario de la lengua palenquera. Rissionario [sic] ri luenga [sic] ri Palenge. Cartagena: Ministerio de Cultura/Corpoataole/Ediciones Pluma de Mompox. Simarra Obeso, R., Miranda Reyes, R., & Pérez Tejedor, J. P. (2008). Lengua ri Palenge. Jende suto ta chitiá. Cartagena: Instituto Manuel Zapata Olivella, Secretaría de Educación y Cultura Educativa Benkos Bioho de San Basilio de Palenque, Ministerio de Cultura, Cartagena.

 Armin Schwegler Smith, H. L. (2014). Patterns of variable tense-aspect marking in Palenquero creole (Doctoral dissertation). The Pennsylvania State University, Pennsylvania. Sommerdyk, S. (2012). Negotiating cross-cultural trade in the eighteenth century: From the Atlantic Coast markets to the Congo River Basin. WISH SEMINAR at WISER, November 5, 2012. Available at: 〈http://wiser.wits.ac.za/system/files/seminar/Sommerdyk2012.pdf〉 Soret, M. (1959). Les kongo nord-occidentaux. Paris: Presses Universitaires de France. Swartenbroeckx, P. S. J. (1973). Dictionnaire kikongo et kituba–français (vocabulaire comparé des langages kongo traditionnels et véhiculaires). Bandundu: Ceeba Publications. Swearingen Davis, M. (1997). A syntactic, semantic, and diachronic analysis of Palenquero -ba (Doctoral dissertation). Stanford University. Thomas, H. (1997). The slave trade: The story of the Atlantic slave trade, 1440–1870. New York, NY: Simon & Schuster. Thornton, J. (1981). Early Kongo-Portuguese relations: A new interpretation. History in Africa, 8, 183–204.  doi: 10.2307/3171515 Thornton, J. (1992). Africa and Africans in the Making of the Atlantic World, 1400–1650. ­Cambridge: Cambridge University Press. Transatlantic Slave Database. 〈http://slavevoyages.org/tast/index.faces〉 Vansina, J. (1990). Paths in the rainforests: Toward a history of political tradition in Equatorial Africa. Madison, WI: University of Wisconsin Press. Vernet, T. (2009). Slave trade and slavery on the Swahili coast, 1500–1750. In P. Lovejoy, B. A. Mirzai & I. M. Montana (Eds.), Slavery, Islam and diaspora (pp. 37–76). Trenton, NJ: Africa World Press. Wheat, D. (2009). The Afro-Portuguese maritime world and the foundations of Spanish Caribbean society, 1570–1640 (Doctoral dissertation). Vanderbilt University, Nashville, TN. Wheat, D. (2011). The first great waves: African provenance zones for the transatlantic slave trade to Cartagena de Indias, 1570–1640. Journal of African History, 52, 1–22. doi: 10.1017/S0021853711000119 Wheat, D. (2015). Global transit points & travel in the Iberian maritime world, 1580–1640. In P. C. Mancall & C. Shammas (Eds.), Governing the sea in the early modern era. Essays in honor of Robert C. Ritchie (pp. 253–274). Berkeley: University of California Press/Huntington Library Press. Wheat, D. (Forthcoming). Atlantic Africa & the Spanish Caribbean, 1570–1640. Chapel Hill, NC: Omohundro Institute of Early American History & Culture/University of North Carolina Press. Wirtz, K. (2014). Commentary to Schwegler 2014. UniverSOS, 11, 89–95. Available at: 〈http:// www.uv.es/~calvo/amerindias/4_0.htm〉

part ii

Bilingualism

Los Angeles Vernacular Spanish An analytical approach to its indicators, markers, and stereotypes Claudia Parodi & Armando Guerrero Jr. University of California, Los Angeles

Our goal in this chapter is to test the perception toward a typology of non‑standard Spanish features and Salvadoran lexical items among speakers of Los Angeles Vernacular Spanish (LAVS), a variant of Mexican Spanish used by the predominant Spanish-speaking speech community in Los Angeles, CA. We demonstrate how Spanish-speaking immigrant communities in the United  States reclassify acceptability ratings in a country that does not impose Standard Spanish, unless the speaker willingly enrolls in courses of L2 Spanish. Keywords:  LAVS; indicators; markers; stereotypes; speech community; (non) standard Spanish; Mexican Spanish

1.  Introduction In this study we set out to characterize the Spanish spoken in Los Angeles by primarily using the Labovian classifications of indicators, markers, and stereotypes (Labov, 1972, 2001; Santa Ana & Parodi, 1998). The goal is to establish a typology of the various features unique to Los Angeles Vernacular Spanish (LAVS), the vernacular used by the predominant speech community,1 and how this vernacular contrasts with Standard Mexican Spanish, which is used in limited social spheres in Los Angeles and is the variety that is taught across all levels of education. LAVS is a koine of several Mexican rural dialects that was formed in the Los Angeles area beginning in the

.  As delineated in Labov (1972, 2001), we also consider that grouping speakers who have similar speech patterns and the same valuation with respect to indicators, markers, and stereotypes characterizes a speech community. However, following recent scholarship on the Labovian speech community, such as Eckert (2008), we add the possibility of stylistic variation in our analysis.

doi 10.1075/ihll.8.03par © 2016 John Benjamins Publishing Company

 Claudia Parodi & Armando Guerrero Jr.

early 20th century as a result of mass migrations of working-class Mexican speakers (Barrera, 1979; Parodi, 2014). The current analysis is crucial for studying Spanish in Los Angeles as it aids in distinguishing a linguistic variety with a base that is rooted in working-class/rural Spanish from Mexico. It is possible to see how Spanish-speaking immigrants in the United States reclassify acceptability ratings in a country that does not impose any Spanish Standard on the various levels of education unless the speaker willingly enrolls in courses taught in Spanish; which may be the case for L2 Spanish students and 2nd generation speakers who attend post-secondary education majoring in Spanish or other similar fields. In general, elementary and secondary public school classes in Los Angeles are taught in English and most bilingual programs are transitional where English is the target language. Thus, LAVS is spoken in colloquial, non-academic, environments as a result of this sociopolitical climate. In order to identify certain sociolinguistic parameters governing the speakers of this variety, we chose ten (n  =  10) linguistic items that are stigmatized in Standard Mexican Spanish; but are nonetheless used in the spoken Spanish of working-class communities in Mexico, as recorded by several Mexican linguists.2 Also note that these items were already common in the Spanish used in the Iberian Peninsula before they were transferred to the Americas during the Spanish colonial period (from the 16th to the 18th century). These can all be found in a variety of written texts of the period in both the Americas and in Europe (Lapesa, 1982). Further, with our second goal of determining the extent that the use of LAVS has expanded among the Spanish-speaking Latino/a communites in Los Angeles in comparison to other dialectal varieties, we also investigated the use and maintenance of twenty-five (n = 25) regional lexical items that are characteristic of the Salvadoran Spanish spoken in Los Angeles. The purpose of this query is to find out if speakers of LAVS know the vocabulary of dialects other than Mexican Spanish.3 In Los Angeles, second-generation .  Here we refer to the following lexemes, which are stigmatized in Standard Mexican Spanish: semos, muncho, ajuera, nadien, güerfana, dotor, comistes, polecía, haiga, and asina. These appear in transcripts of working-class speech in publications such as Juan M. Lope Blanch (1995, 1976). They have also been noted as rural/working-class, or habla popular, in works by José Moreno de Alba (2001), Lope Blanch (1972), and Parodi (2011), among others. Please refer to the questionnaire used in this study in appendix 2. It is important to note that almost all of these stigmatized items have been identified as working-class speech in other geographical areas throughout Latin America (Lapesa, 1982). .  The Salvadoran vocabulary that we utilized in the present study is the following: matate, chuco, guaro, piscucha, marañero, ginas, ayote, cipote/bicho, socado, fustán, chumpa, plátano, guineo, majoncho, haragán, chabacán, chele, chambroso, pacha, keike, corvo/colin, cuchumbo, bayunco, chacalines and tunco. We compiled these lexemes by soliciting lists of words used in El Salvador from various Salvadoran participants. Moreover, we also confirmed these in different dictionaries, such as El Diccionario de la Real Academia Española (2010) and other



Los Angeles Vernacular Spanish 

­ orking-class Latino/a ­immigrants speak LAVS even if their parents are not Mexican. w This is so because LAVS is the speech variety that the Latino/a working-class customarily speaks. Lavadenz  (2005) shows that even 1st generation Guatemalans and ­Salvadorans feel compelled to learn vocabulary characteristics of LAVS in Los Angeles to avoid discrimination and even possible deportation to their countries of origin. Our analysis is guided by the Labovian paradigm of indicators, markers, and stereotypes, as well as Silverstein’s indexical order. Labov illustrates that indicators are any linguistic feature that socially or regionally characterizes speakers within a speech community or social sphere, but which are not perceived as marked in any way by the speakers themselves. Nevertheless, these features are usually identified by language specialists and can be said to be the core (n) or have the lowest indexical value, as described by Silverstein (2003). Markers, on the other hand, are features that are socially recognized by the speaker; these features index a community or social group, but are not connections that bear stigma toward the speaker. These include the semiotic act of noticing the features but not bringing ideology to bear on the relationship noticed, and can be said to be first-order indexicality (or n+1) in Silverstein’s model. Finally, stereotypes are perceived by the interlocutor and are clearly stigmatized; it is possible for speakers to stigmatize a feature but also not be aware of their own use in certain contexts; in other words, stigmatization does not translate to non-use. These can be said to have second-order indexicality (or n+2), which implies weighing ideology on the evaluation being made. Given the fact that social representations are constantly negotiated throughout the development of speech, it suffices for a speaker to use a single stigmatized form once during a series of utterances to immediately be identified as a member of a specific social group.4

2.  Methodology In order to characterize the vernacular Spanish spoken in Los Angeles (LAVS), using two questionnaires (see appendices 1 and 2) we interviewed a total of twenty-five (n = 25)

­ ictionaries dedicated to Salvadoran-specific vocabulary; see Geoffroy Rivas (1978) and d Romero (2003). We also consulted various studies about the Spanish of El Salvador, such as Lipski (2003) and Quesada Pacheco (2000). Please refer to the questionnaire used in this study in appendix 1. .  In order for statistical analyses on stigmatized items to prove significant they must be linked by co-occurrence across the different items. However, due to the existence of language academies in the Spanish-speaking world, evaluative awareness exists among schooled Spanish speakers in Latin America and Spain. This awareness reinforces a Spanish Standard and automatically marks, and often stigmatizes, forms that are not approved by these various language academies.

 Claudia Parodi & Armando Guerrero Jr.

participants, sixteen (n = 16) 1st generation and nine (n = 9) 2nd generation speakers. The participants reside in Lynwood and congregate at the local church in the area at the time of data collection.5 Lynwood is located in Southeast Los Angeles County; the ethnic makeup of the city’s population is predominantly of Mexican origin, though there is a relatively high number of Salvadorans, as can be observed in Table 2. Lynwood was not only ideal for being primarily Latino/a, but also because Spanish is audible in many different contexts and situations, such as the supermarket, in and out of church, commercial centers and malls, restaurants, and also in government institutions such as the different departments in the City Hall. Moreover, the use of Spanish can be observed among people of all ages and presumably of different generations due to the effortless use of bilingual code switching, common of 2nd and later generation speakers (Zentella, 1997a, p. 336). During the interviews we first inquired about the regional vocabulary used by Salvadorans, in order to solicit their knowledge of these lexical items we had our participants indicate to us the meaning of each lexeme using synonyms or through paraphrasing; during the participants’ responses we also registered any reactions to the vocabulary being tested, all of which revealed a clear dislike or acceptance of the lexical items presented (see appendix 1). Next, we investigated a select list of stereotypes in Standard Spanish using another questionnaire that incorporated these items in sentences that were narrated in a colloquial speech style or neutral speech common in working-class communities (see appendix 2). Although this part of the study focused on vocabulary that clearly stereotypes a speaker as rural or popular, the purpose of using colloquial language and style was to avoid accentuating the selected list of the lexical items that we were soliciting by a change in register; the sentences, therefore, would sound natural or familiar to the participants. Some of the colloquial features we included in the questionnaire, but were not part of the lexemes analyzed in the present study, are the following: re-feo for ‘muy feo’ (very ugly); viejo for ‘esposo’ (husband); or patas for ‘pies’ (feet), which are very common in colloquial Mexican Spanish. The aforementioned items are not assumed in Standard Mexican Spanish to be exclusively rural or popular, but rather colloquial in everyday language. We also included vocabulary that is not specific to El Salvador but is more common there than it is in Mexico; this vocabulary includes ayotes for ‘calabazas’ (squash) and haragán for ‘perezoso/huevón’ (lazy). The aforementioned colloquial lexemes were distractors and complemented the words that we were researching for analysis. Finally, we concluded

.  We chose the atrium of a church because it is a common place where local Latinos/as gather to socialize and eat on weekends. Catholicism is the predominant religion practiced by Latino/a communities in the US. Nevertheless, it is not the only one, as can be illustrated by the fact that there is a relatively high number of protestants among this community. 〈http:// www.bestplaces.net/religion/city/california/lynwood〉.



Los Angeles Vernacular Spanish 

our investigation with a short sociolinguistic interview to better understand the participants, their relationship to the Spanish-speaking communities in Lynwood, and to have a sample of naturalistic speech. The participant sample we compiled in our study includes sixteen (n  =  16) 1st generation adult participants who were born in Mexico or in El Salvador and who came to Los Angeles, CA after their eighth birthday. The other nine (n = 9) participants are 2nd generation heritage bilinguals and were born in Los Angeles, CA or came to Lynwood, CA from Mexico or El Salvador6 before their eighth birthday.7 Table 1 summarizes our participant data. Table 1.  Participant sample

1st Generation 2nd Generation Total

Mexico

El Salvador

Total

12 (48%)

4 (16%)

16 (64%)

7 (28%)

2 (8%)

9 (36%)

19 (76%)

6 (24%)

25 (100%)

In Table 1 we can observe that our sample includes a total of sixteen (n  =  16) 1st generation participants; twelve (n  =  12) of these are from tierras altas (represented by Mexico) and four (n = 4) from tierras bajas (represented by El Salvador). We also interviewed nine (n  =  9) 2nd generation participants; these folks were born in Los Angeles, CA or moved to the city before their eighth birthday – the eighth birthday is critical as it roughly delineates when a speaker can acquire a language variety natively or as a second dialect (Parodi, 2004; Chambers, 1992). From the 2nd generation participant sample, seven (n = 7) have parents who were born in tierras altas (Mexico) and two (n = 2) from tierras bajas (El Salvador). The division between tierras altas and bajas in the Spanish spoken in Latin America is highly instructive for studying Spanish

.  We did not restrict our study to participants from these countries, but our sample mostly represents them given the fact that the ethnic makeup of Lynwood is primarily from Mexico and El Salvador. .  Note that we made this division in order to be systematic with respect to age and the process of language acquisition. However, it is important to highlight that speakers who typically arrive to Los Angeles between the ages of eight and eighteen tend to be bi-dialectal. This is to say that these participants know and speak their native dialect as well as the variety spoken in Los Angeles, LAVS – see Parodi (2004, 2011). In these cases, we can refer to an intermediate generation, between the 1st and 2nd. The unique linguistic situation of this group could be due to the fact that after arriving in Los Angeles, these speakers may go to elementary and middle school in English and also maintain their use of Latin American Spanish/LAVS at home and LAVS/English in their respective community.

 Claudia Parodi & Armando Guerrero Jr.

in Los Angeles, CA given the fact that the two Latino/a groups that are most represented in the city are from either of these linguistically and typologically contrasted geographical areas. Generally speaking, the Spanish spoken in tierras altas is primarily characterized by its maintenance of consonants and weakening of vowels; in contrast to the Spanish spoken in tierras bajas, which is mostly defined by its weakening of consonants and maintenance of vowels. Hispanic linguists that work on dialectal typology have accepted this division as the most reliable for distinguishing dialects of Spanish (see, among others, Moreno de Alba, 2001; López Morales, 2005, p.155; and López Morales, 2010, p. 284). Parodi (2004, 2011) has a detailed description of the linguistic features that distinguish tierras altas and tierras bajas, and how these are systematically maintained across speakers in Los Angeles. The composition of our sample greatly reflects the population in Lynwood, CA, which has a relatively high number of residents from Central America (10%) in comparison to the portion of this same group in all of Los Angeles County (3.3%). The focus of our investigation, however, was not to numerically contrast Salvadorans and Mexicans, but rather to determine the differences between the vernacular Spanish spoken in Los Angeles8 and Standard Mexican Spanish; the latter of which is often used in second-language education, and is not representative of the Spanish spoken by speakers in the area. It should be noted that only a minority of immigrants from Latin America speak Standard Mexican Spanish or any other Spanish Standard variety in Los Angeles. This highly select population typically immigrated to the United States as members of the middle or upper class in their respective countries, which is not representative of the majority of immigrants from Latin America to the United States who are primarily from working-class communities (Barrera, 1979; Jiménez, 2010; Parodi, 2011). Furthermore, a social characteristic that radically distinguishes this minority group from the majority of Latin American immigrants in the US is their spoken vernacular. All of our participants live or have lived in Lynwood, a city in Los Angeles County that is predominantly working-class, and where there is a significant population of people born outside of the United States (43.6%). The 1st generation participants mostly immigrated from Mexico and some from El Salvador to this area, and the 2nd generation natively acquired LAVS in Lynwood. The city of Lynwood is geographically located in Southeast Los Angeles County, between the cities of South Gate and Compton and Interstate 710. According to the

.  The current study is part of a larger ongoing project by Claudia Parodi and Armando Guerrero, Jr. studying the use of Los Angeles Vernacular Spanish by geographic areas. First, we will analyze the specific localities where there is a high concentration of working-class Latinos/as to later examine the middle and upper class. This investigation contrasts with other research projects on Spanish Speaking communities in the US, such as New York, where speakers maintain their national varieties of Spanish, as in A. C. Zentella (1990, 1997b).



Los Angeles Vernacular Spanish 

2010 Census, Lynwood had a population of 60,452 (86.64%) Latinos/as. These figures, however, exclude the large undocumented population in the city, as in much of Los Angeles County. The ethnic distribution of the Latino/a population in Lynwood can be seen in Table 2. Table 2.  Latino/a population in Lynwood, CA9 Total Latino/a population

60,452 (86.64%)

Mexican

51,021 (84.40%)

Central American

5,761 (9.53%)

South American

349 (0.58%)

Puerto Rican

192 (0.32%)

Cuban

110 (0.18%)

〈http://www.usa.com/lynwood-ca-population-and-races.htm〉

For the most part, all of the Latinos/as who live in Lynwood speak Spanish at home according to the Census, a substantial 84.28%, of which many are heritage bilinguals.10 All of the interviews were recorded using a Sony digital ICD-PX312 with an Olympus ME-52W Noise Cancelling Microphone. The information was collected in a growing corpus of Los Angeles Spanish, housed in the Centro de Estudios del Español de Estados Unidos (CEEEUS) at the University of California, Los Angeles (UCLA). As previously mentioned, we also recorded brief sociolinguistic interviews; these took place shortly after the questionnaires and focused on the speakers’ experiences living in Los Angeles.

3.  Data analysis 3.1  Salvadoran vocabulary among Mexicans We began the interviews asking our participants to identify vocabulary characteristic of Salvadoran Spanish (see appendix 1). Our twelve (n = 12) 1st generation Mexican

.  The US Census registers official data, however, there is a large population of undocumented immigrants that make up more than 30% of the population of Lynwood, as noted in a bulletin addressed to residents of the city. Though the Census makes an effort to count this population, it may not always be able to provide an accurate number due to different personal reasons among the members of this group that go beyond the scope of this study. .  English is spoken exclusively in only 15% of the households in Lynwood, CA 〈http:// www.usa.com/lynwood-ca-population-and-races.htm〉

 Claudia Parodi & Armando Guerrero Jr.

participants were able to classify many of these items as Salvadoran dialectal markers. This is largely possible because Mexicans from Lynwood are in frequent contact with other 1st generation Salvadorans, mostly as co-workers but also in public settings such as schools, markets, or the atrium of the church where our data collection took place, which is representative of the linguistic contact these two communities have with one another outside of their homes – recall that in Lynwood almost 10% of the Latino/a population is from El Salvador. Moreover, in contrast to the 1st generation, the seven (n = 7) 2nd generation speakers of Mexican-heritage simply accepted these Salvadoran lexical items as indicators, or words that are just part of Spanish, but not regionally marked. This was independent of whether the speakers knew the meaning of the lexeme, and shows a different pattern from what we see in our 1st generation data set. Let us consider Table 3. Table 3.  Salvadoran vocabulary among Mexicans by generation 70 60

1a Generacíón

2a Generacíón

%

50 40 30 20 10

m

at at ch e uc g o pí uaro sc m uch ar a añ er o gí na cíp ay s ot ote e/ bí so cho ca d fu o stá ch n um b pl a át an gu o m íne ajo o n ha cho r ch agá ab n ac án ch ch am ele br os pa o ch a co ke rv ík o/ e cu co ch lín um ba bo yu ch nc ac o alí ne tu s nc o

0

As observed in Table 3, some Mexican participants from both first and second generations comparatively recognized the lexemes cipote/bicho, plátano, and keike. However, the percentages do not show a unanimous consensus recognizing these lexemes throughout this group of speakers to assume perception throughout the first and second generation. There were nevertheless lexical items that did d ­ emonstrate noteworthy percentages among speakers. However, there are sociolinguistic explanations for these high numbers. Both bicho and plátano coincide with words used in Mexican Spanish, though they have a different meaning. We eliminated from our analysis the instances in which speakers understood or expressed the Mexican meaning of these lexical items, as they did not conform to the specific goal of our questionnaire. Finally, an interesting amount of speakers that we interviewed recognized keike but this was due to its similarity to the English word cake.11 The Mexican participants did not stigmatize the regional

.  In LAVS, both pastel and the lexeme [k’eki] for ‘cake’ are used interchangeably.



Los Angeles Vernacular Spanish 

Salvadoran vocabulary in our questionnaire, through rejection, negative reaction, or mocking the words. The data indicates that these speakers only recognized a few of these words, which were considered markers by 1st generation speakers or indicators by 2nd generation speakers. However, most of them had little semantic knowledge of those words. ­Nevertheless, this was not the case with voseo. We registered clear rejection toward this paradigm and it often became the topic of the sociolinguistic interview, especially when noting that Mexicans used voseo to imitate Salvadorans with the goal of teasing or mocking them or other Central American Latinos/as.

3.2  Salvadoran vocabulary among all participants In the following section we analyze the perception of the Salvadoran dialectal markers among all of the participants in the study, a total of twenty-five speakers (n = 25). The goal here is to determine whether the use of these lexical items is only limited to Salvadorans, and if so, how. Additionally, we intend to ascertain whether this lexicon has expanded into the Mexican-origin population in Lynwood. In Table 4 we demonstrate its distribution. Table 4.  Salvadoran vocabulary among Salvadorans and Mexicans

m

at at e ch uc o gu pí aro sc u m cha ar añ er gí o na cíp ay s ot ot e/ e bí ch so o ca d fu o stá ch n um pl ba át an gu o m íne ajo o nc ha ho ra ch gá ab n ac án ch che am le br os pa o ch a co ke rv íke o/ cu col ch ín um ba bo yu ch nc ac o alí ne tu s nc o

%

100 90 80 70 60 50 40 30 20 10 0

Salvadoreña/os

Mexicana/os

In Table 4 we can observe that all of the Salvadorans from Lynwood in our sample (100%), both first and second generations, recognize the regional-specific lexemes of El Salvador: cipote/bicho, plátano, haragán, and pacha,12 which are potentially used to help maintain the group’s ethnic identity.13 Nevertheless, the remainder of

.  For the meaning of this and the other items see appendix 1. .  Future studies should investigate if something similar happens among other Salvadoran groups in Lynwood or in other communities in Los Angeles, CA.

 Claudia Parodi & Armando Guerrero Jr.

the Salvadoran lexicon tested sees losses in Los Angeles, as far as perception goes; this can be attributed to the fact that this population is in stable contact with Mexicans in these communities and the Salvadoran vocabulary becomes symbolic or no longer useful. Most of the Salvadoran participants, or 81% (n = 5), recognized eight (n = 8) more Salvadoran words.14 The same occurs with the following ten (n = 10) words, of which only 65% (n = 4) of the Salvadorans we interviewed recognized.15 An even lower 50% (n  =  3) only knew two (n  =  2) words.16 Finally, 100% of the ­Salvadorans interviewed did not recognize the lexeme, marañero, which is indicative of the diminishing usage of the Salvadoran lexicon in Los Angeles. The 1st generation participants indicated that they only used Salvadoran-specific vocabulary with their family or other Central Americans, which clearly demonstrates the metalinguistic knowledge of the participants – this is no surprise since they are a minority group among the Latino/a population in Los Angeles and are constantly in contact with the majority group, Mexicans, making a Mexican-based dialect more favorable or prestigious in Los Angeles. This contact provides multiple opportunities for regulating the language use of the minority participants, which we see occurring in other contexts, such as with Black English or Chicano English in the US (Lippi Green, 1997; Guerrero, 2014). On the other hand, the 2nd generation participants did not specify exclusive use of these lexemes within the Salvadoran community; this is not to say that there is no word that they recognize as only being understood by their Salvadoran family or other Central Americans, but rather that this particular word was not exclusively Salvadoran to them. The 1st and 2nd generation participants of Mexican-origin, for the most part, ignored almost the entire lexicon of Salvadoran-origin, as seen in Table 4. The number of words that the Mexicans from Lynwood did recognize was equal to the number of those that Salvadorans acknowledged the most, which potentially illustrates their high frequency among this community; these include cipote/bicho, plátano, and haragán.17 As previously mentioned, Mexicans recognized these words, but they were naturally less perceptive of them when compared to the usage among Salvadoran speakers, with percentages between 30% and 60% of relative frequency. The aforementioned data is evidence that the Salvadoran lexicon is almost exclusive to Central Americans and it does not expand among the Mexican-origin popula-

.  Socado, fustán, guineo, chele, chambroso, corvo/colín, bayunco and tunco. .  Matate, chuco, guaro, ayote, chumpa, majoncho, chabacán, keike, cuchumbo, and ­chacalines. .  Piscucha and ginas. .  Note that a much lower percent of Mexican-origin participants recognized keike and pacha, as illustrated in Table 4.



Los Angeles Vernacular Spanish 

tion in Lynwood, where it actually declines, illustrating its lack of prestige. This data shows that it would be worthwhile to ­investigate how acquainted the Salvadoran population is with Mexican vocabulary; we expect the number to be much higher since LAVS, with its Mexican Spanish base, is the variant that is most widespread and most prestigious in Los Angeles. The prestige is due to the fact that in Lynwood and elsewhere (see Parodi, 2011 and ­Guerrero, 2013), LAVS has historically been the dominant variety of Spanish among the first and second-generation Latino/a population.

3.3  S tandard Mexican Spanish stereotypes in Lynwood: 1st generation speakers Table 5 includes the responses of the 1st generation participants toward the stigmatized forms, which were elicited in the second questionnaire (see appendix 2). As we have indicated above, all of these forms came to the Americas after the 16th ­century and have been maintained in active use by working-class and rural vernaculars throughout North and South America, which includes Mexico and El Salvador -tierras altas and tierras bajas. Nonetheless, these forms have become stigmatized in the Standard varieties of Spanish in Latin America, as dictated by the various language academies, specifically the Real Academia Española (RAE) and Academia Norteamericana de la Lengua Española (ANLE), limiting their use to geographically rural regions and working-class communities in larger cities. Consequently, these forms have made their way to the United States through immigration, which is predominantly representative of rural and working-class populations in Mexico and other Spanish-speaking countries that are highly represented in the Southwest (Jiménez, 2010; Parodi 2011, 2014). In the United States, these linguistic forms have acquired distinct and unique valuations from those present in Latin America because of the lack of an imposing language academy and the preferred status of English.18 Among the 1st generation speakers in Lynwood, some of the words are now generalized and accepted, while others continue to be stigmatized, though differently still from what occurs among Standard Spanish speakers, who stigmatize all non-standard forms not accepted by the academies. As mentioned above, this is mostly due to the fact that academic instruction is in English throughout the primary and secondary education of 2nd generation speakers in the United States. Some may learn academic or Standard Spanish as an L2 language in high school or college, though most will acquire Spanish within their working-class communities, such is the case in Lynwood. These communities, as we have stated, .  Zentella (1997b, p. 30) finds the same features in New York’s working-class and/or rural Spanish dialects “as markers of low status.”

 Claudia Parodi & Armando Guerrero Jr.

have a vast 1st generation immigrant population; most with low finishing levels of formal education, which is why they speak a non-standard variety of Spanish (rural/working-class). This explains why among the 1st generation, the majority of lexemes are indicators or not perceived to bear indexical value, with an average of 70%, as shown in Table 5. Table 5.  Judgments among 1st generation participants 100

stereotype indicator

80 %

60 40 20

a

a

in as

ig ha

ist es

to r

co m

do

na fa

ra

o

en di

er gü

na

ue aj

ch un

le

cía m

po

se

m

os

0

The 1st generation speakers evaluated as indicators the use of final -s in the preterit of the 2nd person singular, comistes (87.5%); the present subjunctive haiga (93.75%); and the adverb asina19 (93.65%). We registered very high percentages of indicators, which contrast with the evaluation that is found among speakers of Standard Spanish, who stigmatize these forms in their totality for their lack of inclusion in language academies. These results indicate that the majority of participants from Lynwood do not perceive any stigma and categorize these lexical items neutrally, as part of their day-to-day vocabulary, with little regard to the conventions of the Standard – a consequence that is expected of speakers with little to no formal education in Spanish. Nonetheless, though many items remained indicators among the sample, there were some words that were still stigmatized. The lexeme that was stigmatized the most amongst the speakers was semos (43.65%); followed by polecía (25%), ­muncho (25%), and ajuera (25%), which were stigmatized less. Finally, there were certain lexemes that had very high acceptability ratings, nadien, güerfana, and dotor, all of which only 20% of the participants stigmatized. Although there were specific exceptions, as will be discussed below, in general it can be said that the 1st

.  It is important to clarify that some lexemes have multiple variants: ansina and ansí alternate with asina; naiden alternates with nadien. We limited our questionnaire to the most frequent lexical items in working-class speech represented in our database of Spanish in Los Angeles.



Los Angeles Vernacular Spanish 

g­ eneration participants in Lynwood had very few stereotypes or stigma toward Mexican w ­ orking-class Spanish, which is highly accepted in this region (and is naturally acquired by 2nd generation speakers). The fact that there are common acceptability ratings across different speakers demonstrates how a stable speech community can be maintained in this area, one that conforms to its own patterns of evaluation. These results are very exciting and beg for a comparison between LAVS speakers and other working-class communities in Latin America and the United States to identify how this variety has distinguished itself in a political setting where ­Standard Spanish is not imposed.

3.4  S tandard Mexican Spanish stereotypes in Lynwood: 2nd generation speakers On the other hand, the 2nd generation speakers in Lynwood accepted a much larger number of lexemes that are stereotyped in Standard Spanish; these were mostly considered indicators or simply neutral forms. In other words, these participants stigmatized Prescriptive stereotypes much less than the 1st generation participants, accepting lexical items that not all 1st generation speakers accepted. These participants valued the stigmas present in the vernacular of Mexican Standard Spanish speakers as indicators or neutral at an average of close to 90%. In Table 6 we included the responses of the 2nd generation participants toward the stigmatized lexemes in Standard Spanish.

100 90 80 70 60 50 40 30 20 10 0

na

o ch

er fa gü

un

m

ist es

ig a

co m

ha

a

os se m

in as

en di na

ue ra aj

to r

stereotype indicator

do

po

le cía

%

Table 6.  Judgments among 2nd generation participants

On Table 6 we show that the 2nd generation speakers highly valued as indicators the use of final -s in the preterit of the 2nd person singular, comistes (100%); the present subjunctive haiga (100%); the epenthetic /g/ before the diphthong /we/ in güerfana (100%); and epenthetic /n/ in muncho (100%) – all a much higher percentage than

 Claudia Parodi & Armando Guerrero Jr.

the 1st generation speakers. These results indicate that not a single participant perceived stigma in their use and categorized these lexemes as neutral and part of everyday vocabulary. These speakers generalize the use of similar lexical items as indicators when compared to the 1st generation by producing some and not others. The lexical items that were most stigmatized among this participant sample were somewhat different than the 1st generation speakers. Similarly, they stigmatized the following: polecía (25%) and ajuera (25%). They also added dotor (25%) among the lexemes that were most stigmatized in the sample, which was slightly more accepted by the 1st generation speakers (20%), 5% more. The responses given by the 2nd generation is also different from the 1st generation with respect to semos (87.50%), asina (87.50%), and nadien (87.50%), whose acceptability figures increased; most notably semos, which increased 31.25%. These results indicate that the Latin American non-standard Spanish features found in Lynwood are maintained without conforming to the acceptability patterns found in Mexican Standard Spanish. In general, when compared to the 1st generation, 2nd generation speakers double the acceptability numbers of the workingclass lexicon studied here, while still understanding the meaning of all of them. These results indicate an expansion of acceptability of non-standard features in Los Angeles Vernacular Spanish, especially among the 2nd generation, which potentially demonstrates the trajectory of Spanish in Los Angeles – where non-standard forms can flourish as the variety with prestige.

4.  Conclusions For the most part 1st generation speakers, especially those with higher levels of education, maintain more conventional patterns of perceiving stereotypes and regional markers – an expected result. In contrast, 2nd generation speakers reject fewer lexical items compared to the first group; in general many of the stereotypes for the 1st generation become indicators for the 2nd generation speaker. Similarly, within the 2nd generation group, the regional Salvadoran vocabulary loses its acceptability numbers as a marker of national identity and its lexemes are simply considered indicators, part of the Spanish language in general, losing all regional characteristics that could potentially affect other indexed attitudes. Standard varieties of Spanish do not play a significant role in LAVS, though they are audible on most, but not all, Spanishlanguage television channels and radio stations, in academic environments, and in the minute middle- and upper-class speech communities of Spanish in Los Angeles. The aforementioned results are exciting because they illustrate important differences between 1st and 2nd generation speakers of LAVS – a leveling of non-­standard forms. According to our participants, as they indicated in the sociolinguistic



Los Angeles Vernacular Spanish 

interview,20 there are multiple settings where 1st and 2nd generation LAVS speech differs. The differences surface mainly in two different environments: (a) when they speak with bilingual speakers of English and Spanish, and (b) when they speak with monolingual speakers of Spanish. Most research has focused on the former (a). However, it is crucial to also learn what are the linguistic patterns used between the 1st and 2nd generation immigrants, native speakers of LAVS, in monolingual environments: family gatherings, at school with other monolingual speakers, and in contact with monolingual speakers outside of the United States. As suggested by our results, speech perception and production changes from 1st generation speakers to 2nd generation speakers, though these two groups are more similar to each other than they are to the group of Standard speakers of Spanish, whose linguistic habits they disregard. Lastly, our study suggests that Spanish in Los Angeles should be studied alongside rural and working-class Mexican Spanish for a truly rich analysis since Standard Spanish plays such an insignificant role among the general population in this geographic area.

References Barrera, M. (1979). Race and class in the Southwest: A theory of racial inequality. Notre Dame, IN: University of Notre Dame Press. Chambers, J. K. (1992). Dialect acquisition. Language, 68, 673–705.  doi: 10.1353/lan.1992.0060 Eckert, P. (2008). Variation and the indexical field. Journal of Sociolinguistics, 12(4), 453–476.  doi: 10.1111/j.1467-9841.2008.00374.x Geoffroy Rivas, P. (1978). La lengua salvadoreña. San Salvador: Ministerio de Educación. Guerrero, A., Jr. (2013). A Los Angeles flavor of Spanish: Local norm & ideology of a US variety. Los Angeles, CA: University of California. Guerrero, A., Jr. (2014). ‘You speak good English for being Mexican.’ East Los Angeles Chicano English: Language & identity. Voices, 2(1), 53–62. Jiménez, T. R. (2010). Replenished ethnicity: Mexican Americans, Immigration, and identity. Berkeley, CA: University of California Press. Labov, W. (1972a). Language in the inner City: Studies in the Black English vernacular. ­Philadelphia. Philadelphia PA: University of Pennsylvania Press. Labov, W. (1972b). Sociolinguistic patterns. Philadelphia PA: University of Pennsylvania Press.

.  Due to the fact that these were comments going beyond the scope of our study, we only mentioned them here briefly. Nevertheless, it is important to develop a detailed and systematic study about these occurrences among the Spanish-speaking populations in Los Angeles, and possibly the United States. We have not found any references focused on the interaction between 2nd generation speakers of LAVS and monolingual speakers of LAVS and other monolingual varieties of Spanish.

 Claudia Parodi & Armando Guerrero Jr. Labov, W. (2001). Principles of linguistic change: Social factors. Oxford: Blackwell. Lapesa, R. (1982). Historia de la lengua española. Madrid: Gredos. Lavadenz, M. (2005). Como hablar en silencio (like speaking in silence): Issues of language, culture and identity of Central Americans. In A. C. Zentella (Ed.), Building of strength. Language and literacy in Latino families and communities (pp. 93–109). New York, NY: Columbia University, Teachers College. Lippi-Green, R. (1997). English with an accent: Language, ideology, and discrimination in the United States. London: Routledge. Lipski, J. M. (2003). Latin American Spanish. London: Longman. Lope Blanch, J. M. (1972). Estudios del español de México. México: Universidad Nacional Autónoma de México. Lope Blanch, J. M. (1976). El habla popular de la ciudad de México. México: Universidad ­Nacional Autónoma de México. Lope Blanch, J. M. (1995). El habla popular de la república mexicana. México: Universidad Nacional Autónoma de México. López Morales, H. (2005). La aventura del español en América. Madrid: Espasa Calpe. López Morales, H. (2010). La andadura del español por el mundo. México: Taurus. Moreno de Alba, J. G. (2001). El español de América. Mexico: Fondo de Cultura Económica. Parodi, C. (2004). Contacto de dialectos en Los Angeles: Español chicano y español salvadoreños. In Z. Estrada Fernández, I. Barreras Aguilar & M. Castro Llamas (Eds.), Séptimo Encuentro Internacional de Lingüística en el Noroeste (pp. 277–293). Hermosillo, Sonora, México: UniSon. Parodi, C. (2011). El otro México: Español chicano, koineización y diglosia en Los ­Ángeles, California. In P. Martín Butragueño (Ed.), Realismo en el análisis de corpus orales ­ (pp. 217–243). México: El Colegio de México. Parodi, C. (2014). El español de Los Ángeles: Koineización y Diglosia. In T. C. Smith-Stark, R.  Barriga Villanueva, & E. Herrera (Eds.), Lenguas estructuras y habantes. Estudios en homenaje a Thomas Smith Stark (pp. 1099–1121). México: El Colegio de México. Quesada Pacheco, M. Á. (2000). El español de América Central. In M. Alvar (Dir.), Manual de Dialectología. El Español de América (pp. 101–115). Barcelona: Ariel. Real Academia Española. (2010). Diccionario de la Real Academia Española. Madrid: Espasa Calpe. Romero, M. (2003). Diccionario de salvadoreñismos. La Libertad, El Salvador: Editorial Delgado. Santa Ana, O., & Parodi, C. (1998). Modeling the speech community: Configuration and variable types in the Mexican Spanish setting. Language in Society, 27(1), 23–51. doi: 10.1017/S0047404500019710 Silverstein, M. (2003). Indexical order and the dialectics of sociolinguistic life. Language & Communication, 23, 193–229.  doi: 10.1016/S0271-5309(03)00013-2 Zentella, A. C. (1990). Lexical leveling in four New York City Spanish dialects: Linguistic and social factors. Hispania, 73, 1094–1105.  doi: 10.2307/344311 Zentella, A. C. (1997a). Growing up bilingual: Puerto Rican children in New York. Malden, MA: Wiley-Blackwell. Zentella, A. C. (1997b). ‘Dime con quién andas y te diré quién eres’. Linguistic (in)security and Latino/a unity. In J. Flores & R. Rosaldo (Eds.), The Blackwell companion to Latino studies (pp. 25–39). Malden, MA: Blackwell.



Los Angeles Vernacular Spanish 

Appendix 1 Questionnaire on perception of Salvadoran-specific lexemes found in the speech of Salvadoran immigrants in Los Angeles. ¿Conoce usted alguna de las siguientes palabras? ¿Qué significa? ¿A quién ha escuchado usarlas? El Salvador 1. matate 2. chuco 3. guaro 4. piscucha 5. marañero 6. ginas 7. ayote 8. cipote/bicho 9. socado 10. fustán 11. chumpa 12. plátano 13. guineo 14. majoncho 15. haragán 16. chabacán 17. chele 18. chambroso 19. pacha 20. keike 21. corvo (colín) 22. cumbo (cuchumbo) 23. bayunco 24. chacalines 25. tunco

LAVS morral, bolsa sucio licor papalote, cometa tramposo sandalias calabaza, calabacita niño, muchacho, chamaco apretado medio fondo chamarra, saco plátano macho (grande) plátano tabasco (mediano) plátano dominico (chiquito) flojo, perezoso, huevón vulgar rubio, güero sucio botella, mamila, biberón pastel machete bote (de basura) bromista, simple, que dice tonterías camaroncitos cerdo

Appendix 2 Questionnaire on perception of non-standard linguistic items (in bold). They are used in nonstandard speech (habla popular) throughout the Spanish-speaking world, in Los Angeles Vernacular Spanish (LAVS), and in other working class/ rural variants of Spanish in the US. The sentences that contain these items reflect colloquial Mexican speech. ¿Le suenan mal o no le gusta alguna o algunas palabras en lo que le voy a decir?. Le pueden gustar todas o solo algunas palabras.

 Claudia Parodi & Armando Guerrero Jr. 1. Vimos una jirafa muy grande en la tele 2. Ayer comimos tamales 3. Ninguno quiso meterse a nadar 4. Me gustan las flores rojas 5. Pepe le dio de gritos a su hija re feo 6. Esas manchas se quitan con cloro 7. Los narcos son unos asesinos 8. Compramos pan y café ayer 9. Ana esta muy guapa 10. Rosa ayuda a su vecinito enfermo 11. Ojalá haiga comida en mi casa, ayer se me olvidó comprar ayotes 12. Ayer comistes pupusas con curtido 13. Los lunes nadien contesta porque son unos haraganes 14. Ana se disgustó con su viejo y llamó a la polecía 15. Los chavos se golpiaron y la escuela llamó al dotor 16. Caminé muncho y me duelen las patas 17. Me dijo que está achicopalada y asina no me quiere hablar 18. ‘Bia bastante gente ajuera de la casa 19. Me dio tristeza que mi vecinita se quedara güerfana 20. Semos cinco hermanas y un hermano

On the tenacity of Andean Spanish Intra-community recycling John M. Lipski

The Pennsylvania State University The present study draws on field data from communities of Quechua-dominant late bilinguals in northern Ecuador, a configuration typical of literary and folkloric portrayals of Andean Spanish. A comparison of contemporary data with earlier literary imitations as well as more trustworthy transcriptions reveals a combination of (possibly Quechua-induced) L2 Spanish features and a group of traits that cannot be reasonably analyzed as arising spontaneously and consistently generation after generation. Among the latter are the enclitic particles -ca and -tan, gerunds instead of finite verbs, and some specific lexical variants. Since it has generally been assumed that Quechua-Spanish interlanguage is an idiolectally variable and transitory register used primarily with monolingual interlocutors, the trans-generational survival of a cluster of Andean Spanish traits requires further explanation. This study presents a model for intra-community recycling, using data from ethnographic interviews, sociolinguistic inquiries, and interactive tasks performed by a broad cross-section of Spanish-Quechua bilinguals. The approach is offered as an addition to the toolkit for the analysis of long-standing language contact environments. Keywords:  language contact; Spanish; Quechua; bilingualism

1.  Introduction The Spanish of Quechua-dominant bilinguals in the Andean region is characterized by several recurrent morphosyntactic patterns, in addition to the range of idiolectal variation found among unbalanced bilinguals and second-language learners in general. Popular accounts of Quechua-influenced Spanish depict a picturesque jumble of mismatched vowels and erratic morphological agreement, while linguistic descriptions have concentrated on word order, clitic doubling, and double possessives. The present study draws on field data from communities of Quechua-dominant late bilinguals in northern Ecuador, a configuration typical of

doi 10.1075/ihll.8.04lip © 2016 John Benjamins Publishing Company

 John M. Lipski

literary and folkloric portrayals. Henceforth the language as used in Ecuador will be referred to as Kichwa, as it is known in that country.1 A comparison of contemporary data with earlier literary imitations as well as more trustworthy transcriptions of Andean Spanish reveals a combination of (possibly Kichwa-induced) L2 Spanish features and a group of traits that cannot be reasonably analyzed as arising spontaneously and consistently generation after generation. Since it has frequently been assumed that Kichwa-Spanish interlanguage is a transitory register used primarily with monolingual Spanish-speaking interlocutors, the trans-generational survival of a cluster of Andean Spanish traits requires further explanation. The present study focuses on three configurations that appear to be quasi-stable components of Ecuadoran Andean Spanish: use of the Spanish gerund instead of finite verbs and use of the particles -ca and -tan. The following fundamental q ­ uestions are addressed: –– –– ––

What are the linguistic sources of the non-canonical elements (Spanish, Kichwa, both, or neither)? How cohesive and systematic is the use of these items? How are these items propagated trans-generationally?

More generally this study attempts to depict the sort of sociolinguistic environment that has been conducive to the consolidation and retention of Andean Spanish traits, by combining contemporary field data, the results of interactive tasks, and a survey of  Quechua language planning – deliberate as well as unintentional – during the ­colonial period. Recent changes to the linguistic ecology of the speech communities under study have largely dismantled the mechanisms of intra-community recycling responsible for the retention and propagation of Andean Spanish interlanguage traits in previous generations.

2.  Data collection In order to provide an empirical basis for the examination of Andean Spanish ­morphosyntax, research was conducted in communities representative of the Kichwa-Spanish interface. The data were collected in Imbabura province in northern ­Ecuador, with one of the highest proportions of Kichwa speakers in the entire

.  In the present study the unified spelling of Kichwa Unificado now officialized by the ­Ecuadoran government (〈www.educacion.gob.ec〉; Franco, 2007; Ministerio de Educación, 2010) will be used for Quichua/Kichwa items.



On the tenacity of Andean Spanish 

Andean region (e.g. Andronis, 2004; Cachimuel, 2001; King & Haboud, 2002; ­Rindstedt & ­Aronsson,  2002). The data for the present study come from nine small rural communities in San Pablo del Lago parish and two communities in the neighboring San Rafael parish, in the cantón of Otavalo: Angla, Casco Valenzuela, El Topo, Ugsha, Gualacata, Huaycupungu, Pijal, Cochaloma, San Miguel Alto, Cachimuel, and Zuleta. The communities range from 20 to around 120 families (e.g. Cachimuel, 2001, pp. 52–53) and consist of clumps of dwellings interspersed with garden plots and pastures. All members of these communities are ethnically Native American and all but some of the youngest residents are native speakers of Imbabura Kichwa (Cole, 1982). The descriptions and analyses of the speech communities under discussion are the result of extensive field work conducted by the present author during the time period 2010–2014, and include observation and interaction with several hundred Kichwa-Spanish bilinguals of all ages, in schools, homes, fields, buses, and in neighboring towns. In order to obtain a representative corpus of Ecuadoran Andean ­Spanish, data were extracted from sociolinguistic interviews conducted between 2010 and 2013 with twelve speakers (ten women and two men) whose ages ranged from 40+ to 87. None had received any schooling nor lived for extended periods of time away from their communities. All participants were interviewed by the author after being introduced by close friends or family members, at least one of whom was present during most interviews. The format was free conversation, on topics dealing with rural life and agricultural practices. All of the individuals are Kichwadominant and acquired Spanish informally during adolescence or early adulthood through work on neighboring estates; all exhibit the full range of Andean Spanish traits, including partial n ­ eutralization of high and mid vowels /i/-/e/ and /u//o/ (Lipski, 2015), predominantly object-verb word order, elimination of definite articles, and lapses of verb-subject and adjective-noun agreement. Additional selection criteria included lifetime residence in their respective communities with no extended residence outside of the area, and little ongoing contact with canonical varieties of Spanish. The examples of gerunds, -ca, and -tan represent a generation and a sociolinguistic environment of the past, one that included effectively mandatory agricultural labor and no access to formal education. Younger bilinguals, including children now attending school, have greater exposure to Spanish and do not exhibit most of the interlanguage traits found among older Kichwa-dominant speakers. The present study focuses on the partially fossilized speech of older late bilinguals – the sociolinguistic profile forming the basis for p ­ opular stereotypes – in order to model the transition from an environment ­conducive to the retention of non-canonical Andean Spanish variants to one in which contemporary Spanish plays an increasingly prevalent role.

 John M. Lipski

3.  Extension of the Spanish gerund The first case involves the extension of the Spanish gerund in contexts where the g­ erund would be very infrequent or ungrammatical in monolingual varieties of ­Spanish. As early as 1615 the bilingual Peruvian Guamán Poma de Ayala noted the non-canonical use of the gerund in Andean Spanish:2 (1)

“[…] algunos yndios se hacían ladinos, los yanaconas dezían obeja chincando, pacat tuta buscando, mana tarinchos, uira cocha [Some Indians become Europeanized; the yanaconas {dark-skinned} would say “sheep losing, until dawn searching, we did not find, noble sir]

ya señor, sara paruayando, capón asando, todo comiendo […] yo agora ­mirando chapín de la mula” [yes sir, corn shucking, capon roasting, all eating … now I looking at mule’s hoof] Outside of the Andean region the use of the Spanish gerund as invariant exponent of finite verbal paradigms is almost never found, a fact for which at least two logical possibilities suggest themselves. Use of the gerund may be the direct result of transfer from Quechua, or the socio-historical circumstances surrounding the introduction of the Spanish language into the Andean region may hold the key to the unexpectedly broad use of the gerund. Initial research reported in Lipski (2013) suggests that both of the aforementioned factors are implicated, intertwined over a period of several ­centuries and operative even today. In the Andean Spanish of Ecuador there are commands of the sort dame cerrando la puerta ‘close the door [for me]’ that many scholars have attributed to a calque from Kichwa (e.g. Hurley, 1995; Kany, 1945, p. 158; Niño-Murcia, 1995, p. 90; Toscano Mateus, 1953, pp. 286–287; Vázquez, 1940, p. 127). However, as noted, e.g. by Haboud (1998, p. 219), Haboud & de la Vega (2008, p. 178–179), and Bruil (2008b), and ­confirmed by the present writer, Kichwa constructions based on ku-na ‘to give’ are proffered as equivalents to Andean Spanish dar + gerund primarily by balanced bilinguals with considerable proficiency in Spanish, while Kichwa-dominant speakers prefer simple honorific imperatives. This suggests that the constructions – of as yet undetermined origin – stabilized in local varieties of Spanish and then made their way into Kichwa (Albor, 1973; Bruil, 2008a; Cisneros Estupiñán, 1999; Haboud, 1998; Olbertz, 2002, 2008).

.  Taken from the on-line facsimile edition and transcription of Guamán Poma de Ayala’s El primer nueva corónica y buen gobierno (1615/1616), on the Royal Danish Library’s permanent website 〈www.kb.dk/permalink/2006/poma/397/en/text〉 (1 October 2014).



On the tenacity of Andean Spanish 

More relevant to the intra-community recycling are instances where gerunds appear in place of finite verbs only in Kichwa-dominant Spanish interlanguage. Some instances are quasi-lexicalized, e.g. the use of qué haciendo and qué diciendo ‘why?,’ representing direct calques of Kichwa ima rura-shpa-ta and ima ni-shpa-ta, respectively. Examples from Imbabura include: (2) ¿qué haciendo pegaste a la mujer? what do-ger hit-2s-pret prep art woman ‘why did you hit the woman?’ ¿qué diciendo faltaste la clase? what say-ger lack-2s-pret art class ‘Why did you miss class?’ These constructions occur in other Quechua-Spanish bilingual speech communities (Calvo Pérez, 2008, p. 200; Granda, 2001, p. 115; Merma Molina, 2004, p. 202; Merma Molina, 2005). Also found among Quechua-dominant speakers throughout the Andean region is phrase-final diciendo ‘saying,’apparently translating Quechua ­ni-s(h)pa ‘saying’ and used to report indirect speech or evidentiality (Calvo Pérez, 2001, p. 108; Gifford, 1969, p. 171); an example from Imbabura is: (3) mucha plata ha de tener diciendo much money aux of have-inf say-ger ‘[he] must have a lot of money, it seems’ Found frequently among older Kichwa-dominant bilinguals in Imbabura are gerunds used instead of finite verbs in combinations that would be completely ungrammatical in monolingual varieties of Spanish, for example: (4) yo-ca no siguiendo clase-ca ni firmar no pudiendo mismo I-top neg follow-ger class-top nor sign neg able-ger same ‘I didn’t take any classes, [I] can’t even sign [my name]’

sí allá en la asa chiquitica tenendo yes there in art house small-dim have-ger ‘yes there in that little house [I] have’ yo pobre aquí hacendo arar; yo hacendo chagra I poor here do-ger plow I do-ger farm ‘poor me plowing, I [have] a farm’ comprando comemos no teniendo papa buy-ger eat-1pl neg have-ger potato ‘when we have no potatoes, we eat by buying them’  ño patrón Galo, cuando vivendo ca, sembramos papa sir boss Galo when live-ger-top plant-1pl potato ‘When Galo [Plaza Lasso] was alive, we planted potatoes’

 John M. Lipski

cuando mama muriendo-tak, papá nomas quedando-tan when mother die-ger-affirm father only remain-ger-tan ‘When mother died, only father was left’ no sé cuánto año teniría; marido trayendo  neg know-1st how many year have-cond husband bring-ger para acá for here ‘I don’t know how old I was, my husband brought [me] here’

3.1  The Kichwa “gerund” -shpa Most accounts of non-canonical gerund usage in Andean Spanish implicate the ­Quechua verbal suffix -s(h)pa. Beginning in the 16th century traditional grammars of Quechua refer to this suffix as an “ablative gerund. In Kichwa all verbs in subordinate clauses are non-finite, and -shpa is in reality a subordinating complementizer (Muysken, 1997, p. 385; Cole, 1982, pp. 61–62). Most contemporary descriptive grammars and pedagogical materials (for Kichwa speakers as well as beginners) continue to treat -shpa as equivalent to Spanish forms in -ando and -iendo, leading several researchers to implicate direct transfer from Quechua in the expanded set of uses of the Andean Spanish gerund (e.g. Calvo Pérez, 2001, pp. 119–121; Gifford, 1969, pp. 171–172; Granda, 1995; Muysken, 1985, pp. 389–391; 2005; Niño-Murcia, 1995, pp. 94–96; Toscano Mateus, 1953, pp. 121, 273). In reality Kichwa -shpa and its congeners in other Quechua varieties correspond to only a subset of the manifestations of the Spanish gerund. In particular -shpa is never used in progressive constructions, which account for the most frequent appearance of the gerund in contemporary (non-Andean) Spanish. A comprehensive analysis can be found in Cole (1982) while correspondences with Ecuadoran Andean Spanish are summarized in Table 1 (from Lipski, 2013).

3.2  Spanish gerunds in the Ecuadoran Andean Spanish corpus In the Ecuadoran Andean Spanish corpus 1130 tokens of the Spanish gerund occurred, of which 208 (18.4%) correspond to normal Spanish progressive constructions. A total of 438 (38.8%) tokens of the Spanish gerund occurred in what can be construed as subordinate clauses in contexts where monolingual Spanish would require a finite verb but for which a form in -shpa would be appropriate in Kichwa, for example: (5) comprando comemos no teniendo papa ‘when we have no potatoes, we eat by buying them’ ese es que insiñen, poniendo escuela-ca ‘that is what they teach, having put a school there’ animalitos teniendo-ca no hay tiempo para ir a trabajar, ‘[since we have] animals, there is no time to go to work’



On the tenacity of Andean Spanish 

Table 1.  Comparison of Quichua -shpa and the Spanish gerund Use of -shpa in Quichua

example

gerund in Spanish?

time clauses

Quito-man chaya-shpa mi rijsi-ta riku‑rka-ni

possible; less common3

‘Arriving in Quito I saw my friend’ Sp. ‘Llegando en Quito vi a mi amigo.’ closely related (simultaneous) actions

kanda-shpa mi shamu-rka-ni

yes

‘I came singing’ Sp. ‘Vine cantando’ conditional clauses

Utavalo-man ri-shpa-ca ruwana-ta randi-sha

no

‘Upon going to Otavalo I will buy a poncho’ Sp. ‘Cuando [yo] vaya a Otavalo compraré un poncho’ successive actions

machiti-ta aisa-shpa ri-rka

yes

‘Grabbing a machete he/she went’ Sp. ‘Agarrando un machete se fue’ perfective

micu-shpa shamu-ngui

no

‘[after] eating you will come’ Sp. ‘Después de comer vendrás’ concessive clauses

mana Utavalu-man ri-shpa-pash ruwana-ta randi-sha

no

‘even not going to Otavalo [even if I don’t go] I will buy a poncho’ Sp. ‘Aunque [yo] no vaya a Otavalo compraré un poncho.’ “without doing”

mana jatu-shpa shamu-rka-nchij

no

‘not selling [without selling anything] we came’ Sp. ‘Vinimos sin vender.’ (Continued)

3  Spanish does allow the gerund in time clauses, both with same-subject (corresponding to Quichua -shpa) and even switch-subject (corresponding to Quichua -jpi) reference, e.g. Estando (yo) en casa, pude recibir la llamada; Estando el presidente de viaje los militares le dieron golpe de estado. Such combinations are comparatively rare and appear more often in literary texts and oratory than in everyday speech; they were documented for the medieval and early Spanish colonial period (e.g. Muñío Valverde, 1995, Chapter 1).

 John M. Lipski

Table 1. (Continued)  Comparison of Quichua -shpa and the Spanish gerund Use of gerund in Spanish

example

-shpa in Quichua?

progressive phrases

está lloviendo

no

‘it is raining’ manner clauses

manejando lentamente llegarás en dos horas

yes

‘[by] driving slowly you will arrive in two hours’ closely related simultaneous actions

el hombre salió corriendo

yes

‘the man left running’

A total of 34 tokens (3%) occur in main clauses, corresponding neither to Kichwa nor to Spanish patterns, for example: (6) sí allá en la casa chiquitica tenendo ‘there in that little house [I] have’ el ternero-tan hacendo enseñar mayores hacendo enseñar con huagra mayor ‘the older ones teach the steer, teaching by older cattle’ Assuming that the majority of the 436 tokens (38.6%) classified as fragments can also be construed as belonging to subordinate clauses, the prominence of gerunds in subordinate clauses rises even higher. This suggests that Kichwa-dominant speakers are not simply employing the Spanish gerund as the exponent for all verbs, but are distinguishing between main clauses – containing finite verbs – and subordinate clauses – which in Andean Spanish interlanguage frequently contain the nonfinite gerund. This distinction corresponds to Kichwa, in which verbs in subordinate clauses are given the suffix -shpa (with no change of subject) or -jpi (for change of subject), as per the examples in Table 1. Simple transfer from Kichwa does not, however, account for the choice of the Spanish gerund as the non-finite exponent of verbs in subordinate clauses; the scarcity of such combinations in monolingual Spanish – ranging from marginally possible to completely ungrammatical – rules out simple imitation and overgeneralization as the sole source of the free-standing gerund in Ecuadoran Andean Spanish.

3.3.  A translation task In an attempt to determine whether Spanish gerunds in place of finite verbs are the result of on-line interference from Kichwa, Lipski (2013) presented Imbabura Kichwa utterances containing subordinate clauses marked with -shpa to fourteen bilingual speakers from the same communities: eight adults with varying degrees of formal



On the tenacity of Andean Spanish 

e­ ducation but with no formal training in Kichwa and six adolescents who had received school classes in Kichwa. Overall the young bilinguals translated -shpa by the Spanish gerund 73% of the time – sometimes resulting in questionable Spanish sentences – as opposed to only 32% for adult bilinguals. Older bilinguals’ translations of -shpa were closest to those of school-trained bilinguals in those instances where a gerund is not unlikely in non-Andean Spanish, in particular simultaneous or immediately successive actions: picking up a machete and leaving, singing while walking, and running downhill. The largest discrepancies between children’s and adults’ Spanish translations of -shpa constructions involve conditional and concessive clauses, time clauses, and absence of actions (“without doing”). In those instances where -shpa corresponds to Spanish subordinate clauses, adult bilinguals did not turn to the Spanish gerund in spontaneous translation, while young bilinguals with formal training in Kichwa applied the equation-shpa ≡ Spanish gerund as a first gambit. Informal observations indicate that Kichwa-speaking children who acquire Spanish at home or in early pre-school do not pass through a stage in which gerunds systematically replace finite verbs. This suggests that the interpretation of Kichwa -shpa as equivalent to the Spanish gerund in subordinate clauses may be prompted by school instruction, reinforced by the lingering presence of the Spanish gerund in the speech of elderly Kichwa-dominant speakers. For preceding generations of nonliterate speakers, it is not likely that the infrequently occurring Spanish gerund would be spontaneously and independently proffered as the principal exponent of Imbabura Kichwa -shpa, which raises the question of why descriptive grammars have consistently described -shpa as a gerund translated by Spanish forms in -ando/-iendo.

3.4.  Historical reinforcement of the Spanish gerund The 1615 quote in (1) suggests that by the beginning of the 17th century Quechua speakers were turning to the Spanish gerund as invariant substitute for finite verbs. Given the tenuous nature of the correspondence between the Spanish gerund and the full range of Quechua structures based on -s(h)pa, and in view of the fact that the principal uses of the Spanish gerund have not changed substantially over the past few centuries (Muñío Valverde, 1995; Torres Cacoullos, 2000), additional sources must have been in play. By the mid-17th century several grammars of Quechua had been written and distributed by Spanish missionaries, beginning with the grammar of Domingo de Santo Tomás (1560), followed by Antonio Ricardo (1586), Diego de Torres Rubio (1603), Diego González Holguín (1608), Alonso de Huerta (1616), Juan de Aguilar (1690), an anonymous 16th or 17th century grammar (Dedenbach-Salazar Sáenz, 1993), and many others. Not only did all describe -s(h)pa as gerundio ablativo ‘ablative gerund’ but each patently copied the format and organization of previous works. In Ecuador the grammar of González Holguín was widely used as the source

 John M. Lipski

for Kichwa catechisms and other religious translations even past the 17th century (Hartmann, 1999, pp. 63–65). The colonial missionaries who wrote grammars of indigenous languages forced all structures into the classifications found in Latin grammars. Uniquely Quechua constructions were often overlooked, glossed as optional adornments, or misconstrued and mistranslated (Durston, 2007, p. 182). Once a Quechua-­Spanish grammatical correspondence had been asserted, it remained unchallenged. The incorporation of the Latinate grammars into liturgical and pastoral language resulted in grammatically distorted Quechua being presented by Spaniards to native Quechua speakers, who apparently employed these same constructions in oral confessions, songs and recitations, and in conversations with religious personnel. Suárez Roca (1992, pp. 30–31) notes that Spanish priests often preferred to give grammatical instruction to indigenous subjects in their own languages, learning from grammars written by missionaries. The priests evidently hoped that by urging indigenous speakers to address the clergy “by the book” (even if this meant departing from normally acceptable usage), the priests would more easily understand their speech. When speaking to indigenous subjects in Spanish, the priests may also have made their own use of Spanish conform to what was understood of Quechua grammar, a form of foreigner talk based not only on attitudes of cultural superiority but also on a desire to facilitate the comprehension of important concepts, both religious and secular. Originally used by religious personnel influenced by Quechua grammar treatises, the use of the Spanish gerund in speech addressed to indigenous people would eventually become part of the de facto foreigner talk employed by Spanish colonists and their descendents, a practice that – while no longer socially acceptable – continues to this day (Muysken, 1980, p. 71; 1984, p. 64). The mutual reinforcement between Quechua speakers’ Spanish interlanguage and the enhanced use of the gerund by native Spanish speakers (perhaps condescendingly) addressing Quechua speakers could contribute to the stabilization of the gerund as exponent of finite verbs in Ecuadoran Andean Spanish. Figure 1 summarizes the postulated relationships between the Ecuadoran Andean Spanish gerund and Imbabura Kichwa -shpa.

4.  Use of -ca and -tan as discourse markers Kichwa-dominant Spanish speakers in the Ecuadoran highlands lace their speech with two affixes not found in other varieties of Spanish. The first is -ca (the usual Spanish spelling), homophonous with and evidently derived from the Kichwa topicalizer -ka (the “Kichwa unificado” spelling).



On the tenacity of Andean Spanish 

foreigner-talk stereotypes

community-wide usage

Andean Spanish gerund

canonical Spanish usage

at the moment of speaking colonial Kichwa grammars

Kichwa -shpa

Figure 1.  Sources of the gerund in Andean Spanish

(7) nieto-ca en escuela de Quito ha de ser grandchild-top in school of Quito have-3s of be ‘[my] grandson is in school probably in Quito’ antes recién nacido-ca bautizamos ahora-ca ya  before recent born-top baptize-1pl now-top already maltoncito bautiza adolescent baptize-3s ‘In the past we baptized newly born [babies], now we baptize adolescents’ gavilán viniendo-ca se muere hawk come-ger-top die-refl ‘[when] the hawk comes [the chicken] dies’ The other discourse marker found in Ecuadoran Andean Spanish is -tan, which previous scholarship has derived from Spanish también ‘also.’ (8) mi mamita-tan todavía papá-tan my mother-tan still young die-pret-3s father-tan joven murió young die- pret-3s ‘My mother died [when she was] still young, my father also died young’ antes conejo-tan teniamos ahora ya no tiene before rabbit-tan have-1pl-imp now already neg have-3s ‘In the past we also had rabbits now [we] no longer have’

 John M. Lipski

Casco Valenzuela cómo-tan sería Casco Valenzuela how-tan be-cond-3s ‘I wonder what Casco Valenzuela is like’ me sufrí con diez hijos hijos-tan como  refl suffer-pret-1s with ten children children-tan like animalitos son animals-dim be-3pl ‘I suffered with ten children, children are like little animals’ Recent research reported in Lipski (2014) and summarized below suggests that -tan continues to be recycled, as do many instances of -ca.

4.1.  Ecuadoran Andean Spanish -ca and Kichwa -ka The only likely source for is Ecuadoran Andean Spanish -ca is the homophonous Kichwa suffix -ka (modern spelling), which marks topic or theme (old information) (Cole, 1982, pp. 166–167; Lefebvre & Muysken, 1988, p. 180; Muysken, 1995, p. 385; Sánchez, 2010, p. 45). In Imbabura Kichwa -ka can be attached to most parts of speech, including nouns occupying any argument position, can occur more than once per clause, cannot usually be suffixed to finite verbs, and is limited to matrix clauses. Conjoined nouns can each be marked with -ka and -ka often occurs with left-fronted constituents. In situ constituents can also be marked with -ka; this often but not always occurs when another token of -ka is attached to the clause-initial constituent: When fronted adverbials are marked with -ka a conditional ‘if ’ meaning obtains; in questions -ka can be construed as roughly ‘what about?’ The Ecuadoran Andean Spanish corpus yielded 440 tokens of -ca, with an overall distribution compatible with Kichwa -ka. A total of 174 tokens of -ca (39.5%) appear in clause-initial position, appropriate for topicalization. This includes not only subjects but also temporal and locative adverbs (which often begin narrations) as well as direct objects (reflecting Kichwa-influenced O–V word order). The corpus contains 52 full-clause utterances with more than one token of -ca; of these 45 (86.5%) contain an instance of -ca attached to the first constituent, for example: (9) ese tiempo-ca no eramos mucho-ca that time-top neg be-imp-1pl much-top ‘[at] that time there weren’t many of us’ antes-ca no sabía haber carro-ca before-top neg know-imp-3s exist car-top ‘before there were no buses’ Considering only clauses with a single token of -ca the rate of appearance of all tokens of Ecuadoran Andean Spanish -ca in clause-initial position rises to 44.3%; there are no



On the tenacity of Andean Spanish 

tokens in which -ca can be construed as appearing in a non-matrix clause. If -ca were simply an intensifier or stylistic ornament as some have suggested, at least some occurrences in non-matrix clauses would be expected. Also consistent with the behavior of Kichwa -ka is the fact that Andean Spanish -ca never attaches to finite verbs. Finally, no instance of -ca appears to provide new information or focus (rheme), coinciding with Kichwa usage.

4.2.  Ecuadoran Andean Spanish -tan and possible Kichwa sources The few descriptions of -tan in Ecuadoran Andean Spanish derive this item from a shortened version of Spanish también ‘also’, although many uses of -tan cannot be translated as ‘also’ and have been classified as intensifiers (Kany, 1945, p. 329), assertions of doubt (Toscano Mateus, 1953, p. 333), or negative emphatics (Muysken, 1982, p. 110). Toscano Mateus (1953, p. 333) offers a comparison with Kichwa -pish ‘also, even’ (­-pash in Imbabura). Imbabura Kichwa -pash is in principle not limited to matrix clauses and can occur several times within a single utterance, especially in coordinate constructions (Cole, 1982, p. 79), where -pash … -pash means ‘… as well as …’ Given the frequent interplay of topic- and comment-marking in Kichwa, Lipski (2014) has suggested that the Kichwa validator -mi may also have contributed to the range of uses represented by ­-tan. Kichwa -mi marks the main focus of an utterance, giving first-hand (evidential) and new (rheme) information; as such -mi only occurs in affirmative utterances (Cole, 1982, pp. 164–167). Like the topicalizer -ka this enclitic occurs only in matrix clauses; unlike -ka, -mi can occur only once per utterance. Imbabura Kichwa -mi is not restricted to left-fronted constituents and can occur after any main-clause element in situ. Evidential/focus markers like -mi occur either clauseinitially or attached to the rhematized element; in addition -mi does not occur on constituents to the right of the main verb (Muysken, 1995, pp. 380–383). The topicalizer -ka is usually placed before the focus/evidential marker -mi (Muysken, 1995, p. 385; Sánchez, 2010, p. 45). In Imbabura Kichwa -mi (focus, new information, rheme) and -ka (topic, previous information, theme) co-exist symbiotically, but in most circumstances neither is obligatory. The Ecuadoran Andean Spanish corpus yielded a total of 212 analyzable tokens of -tan, of which all but one occurred in a matrix clause; -­ tan never occurred more than once per sentence. A total of 96% of the tokens of -tan occurred in affirmative utterances, and as interpreted by the present author and confirmed by two bilingual consultants from the research area, 99 (47%) cannot be reasonably construed as meaning ‘also’ or ‘even’ or embodying interrogative or indefinite reference, i.e. in circumstances that would correspond to Kichwa -pash. All of these tokens are consistent with the introduction of new (theme) and first-hand information and there are no tokens of

 John M. Lipski

-tan followed by -ca. This general distribution is similar to that of the Kichwa validator/focus marker -mii and does not closely resemble that of any other Kichwa affix, including -pash, which is not limited to single occurrences, to matrix clauses, or to specific clause-internal positions. Such circumstantial similarity is not sufficient to postulate Kichwa -mi as a contributing source for -tan, but does leave open the possibility that -tan has grammaticalized into a syncretic discourse marker.

4.3.  Probing for -ca and -tan: A translation experiment An analysis of the Andean Spanish corpus reveals a distribution of -ca that appears to differ in no qualitative fashion from Kichwa -ka, while -tan has expanded its original status as an apocopated form of Spanish también ‘also.’ All of the late bilinguals who provided data for the present study employed both elements, as did most of the other speakers of similar sociolinguistic profiles interviewed by the author. In order to explore the intra-community viability of Ecuadoran Andean S­ panish, a Spanish-to-Kichwa translation task was administered (described in detail in ­Lipski, 2014). It was hypothesized that if -ca and -tan are consistent elements in the Ecuadoran Andean Spanish speech communities then individuals from these same communities should have no difficulty in parsing and translating into Kichwa utterances containing -ca and -tan. It was further hypothesized that participants would frequently translate Andean Spanish -ca with Kichwa -ka attached to the equivalent elements and with similar word order. Given the possibility of more than one Kichwa source for -tan, the translation task serves as a probe for other possible Kichwa sources. From the corpus twenty-five representative utterances containing one or more instance of -ca and thirty-nine utterances containing -tan were extracted. The stimuli were presented to eleven men and twenty-seven women from the same communities; all are Kichwa-Spanish bilinguals, with ages ranging from the mid-twenties to the late seventies and formal education ranging from no schooling to a completed university education. Observed proficiency in Spanish covered a similarly broad spectrum. Participants were told that they would hear samples of Spanish from their own and neighboring communities and were asked to quickly translate each utterance into Kichwa as completely as possible, with no embellishment or abbreviation.

4.4.  Results: Ecuadoran Andean Spanish -ca and Kichwa -ka Kichwa ­-ka is closely tied to speakers’ assumptions about shared knowledge and focus. Since the translation task did not presuppose participants’ knowledge of specific events, not all instances of Andean Spanish -ca were rendered with -ka in Kichwa. At the same time there were instances of -ka being added in the translation of stimuli lacking -ca.



On the tenacity of Andean Spanish 

The most consistently translated cases of -ca involved temporal and ­locative adverbs (e.  g. dealing with previous conditions in the communities under study), possibly because participants felt a personal involvement with the utterances being translated. Table 2 gives the breakdown. In the aggregate the results of this translation task are consistent with the convergence of Kichwa -ka and Ecuadoran Andean Spanish -ca not only in form but also in general function. Table 2.  Translations of Ecuadoran Andean Spanish -ca as Kichwa -ka by ­Kichwa-Spanish bilinguals (N = 38) subject (N = 11)

DO (N = 6)

obj. prep. (N = 3)

temporal adverb (N = 12)

locative adverb (N = 3)

predicate nom. (N = 3)

gerund (N = 3)

#

306

137

80

380

84

76

53

%

73%

60%

70%

83%

74%

67%

46%

4.5.  Why -ca? Although all residents of the communities in which the data were collected acknowledge the frequent presence of -ca in local Spanish, only Kichwa-dominant bilinguals actually produce this form – and do so with considerable inter-speaker variability. That this variable usage has been attested for several generations suggests that -ca originates from on-line momentary interference rather than borrowing, only later becoming partially grammaticalized. This suggested trajectory is not unexpected given the linguistic ecology of the communities under study. The behavior of -ca is consistent with largely involuntary intrusion of Kichwa into L2 Spanish. There is a body of research indicating that even balanced bilinguals do not completely inhibit the non-target language (Kroll, Misra, Bobb, & Guo, 2008; Kroll, Dussias, Bogulski, & Valdés Kroft, 2011; van Heuven, Schriefers, Dijkstra, & Hagoort, 2008). The high frequency of occurrence of -ka in Kichwa combined with the absence of a suitably generic and invariant Spanish equivalent augments the possibility of eventual automatization of -ka insertion (e.g. Antón Méndez, 2011). Hatzidaki et al. (2011, p. 128) observe that activation of features such as agreement from the non-source language may be stronger in contexts where speakers have to use both languages and also when the non-source language is dominant. Both of these conditions obtain in the case of Kichwa-dominant bilinguals’ use of Spanish within their communities. Parallel activation of Kichwa and Spanish is consistent with the inter-speaker variability of -ca in local vernacular Spanish, which although frequent does not show signs of complete grammaticalization. Figure 2 graphically displays the proposed trajectory and status of -ca.

 John M. Lipski

community-wide usage

Andean Spanish -ca

incomplete inhibition of L1 at the moment of speaking

Kichwa -ka

Figure 2.  Proposed sources of -ca in Ecuadoran Andean Spanish

4.6.  Results: Ecuadoran Andean Spanish -tan into Kichwa The stimuli, derived from the Ecuadoran Andean Spanish corpus, contained thirtynine tokens of -tan, of which eighteen had been judged to be equivalent to ‘also/even’ (e.g. corresponding to Kichwa -pash) or in interrogative utterances (corresponding to Kichwa ima-shina ‘how,’ ima … ka-nga ‘what can it be,’ or ima-pash ‘how can it be’). Twenty-one tokens of -tan could not be clearly correlated with the semantic range associated with Spanish también ‘also’ or with the aforementioned distribution of Kichwa -pash. The results are shown in Table 3. Table 3.  Translations of Ecuadoran Andean Spanish -tan by Kichwa-Spanish ­bilinguals (N = 38)

-pash

“also”/“even” (N = 15)

interrogative (N = 3)

other (N = 21)

473/83%

3/3%

77/10%

-mi

0

0

106/13%

other particle

2

61/54%

3

95/17%

50/44 %

612 /77%

no particle

Of the tokens of -tan that a priori could correspond to ‘also, even,’ participants’ ­translations included -pash at the rate of more than 80%, suggesting that -tan is not a capricious embellishment but rather an element that in many contexts is clearly associated with Spanish también. In one interview a participant performed an apparent repair by saying ellos-tan ‘they-tan’ immediately followed by ellos también tienen … ‘they all have…’ When -tan could not be clearly construed as meaning ‘also, even’ or



On the tenacity of Andean Spanish 

expressing an interrogative speculation, in more than three quarters of the instances there was no direct translation. The Kichwa validator/focus marker -mi was sometimes added in an approximate location corresponding to -tan. However, the low rate of translation of -tan as -mi does not immediately suggest -mi as the primary motivating element, despite the fact that the instances of -tan not corresponding to Kichwa -pash coincide with the distributional properties of -mi as well as signaling new information. One possible contributing factor is the fact that -mi “indicates the authority for assertion, the degree of certainty of the speaker, and whether the matter is of importance to him” (Cole, 1982, p. 165). In a translation task based on stimuli presented without a full pragmatic context, the conditions that would facilitate assertions of authority frequently do not obtain. The results of the translation task indicate that Kichwa-Spanish bilinguals are not automatically translating all instances of -tan as Kichwa -pash, but are sensitive to a wider range of meanings of -tan in Kichwainfluenced Spanish interlanguage. Unlike -ca, ­­-tan is a syncretic element in both form and function, and cannot be attributed to a simple failure to suppress Kichwa when speaking Spanish. Figure 3 depicts the proposed development and current status of -tan.

Andean Spanish -tan at the moment of speaking

Kichwa -mi

at the moment of speaking

community-wide usage

Kichwa -pash

Spanish también

Figure 3.  Sources of -tan in Ecuadoran Andean Spanish

4.7.  F  actors influencing the presence of -ca and -tan in Ecuadoran Andean Spanish In Ecuadoran Andean Spanish -ca is the only direct transfer of a Kichwa affix, while -tan has expanded its semantic/pragmatic range among Kichwa-dominant bilinguals.

 John M. Lipski

Lexical exchanges between Quechua and Spanish are frequent and apparently unconstrained, but morphological crossover is at least partially determined by the properties of each language’s functional categories (e.g. Muysken, 2012a, 2012b). Sánchez (2003, p. 15) has proposed the Functional Convergence Hypothesis for bilingual contact, according to which “Convergence […] takes place when a set of features that is not activated in language A is frequently activated by input in language B.” Muysken (2012a, p. 204) expands this definition by adding that convergence “takes place in a contact situation when a set of features is unstable or a new functional category emerges due to input in one of the languages that is compatible with input from the other language.” The Kichwa topicalizer -ka and the evidential/focus marker -mi have been analyzed as heads of functional projections (e.g. Muysken, 1995, p. 386; Sánchez, 2003, pp. 31f.). At the same time Spanish phrase structure exhibits a (usually null) Topic Phrase (TopP) and a Focus Phrase (FocP) (e.g. Camacho, 2006; Méndez Vallejo, 2009; Ordóñez & Treviño, 1999; Toribio, 2002). The emergence of lexically overt heads (e.g. -ca and/or -tan) would not fundamentally alter Spanish phrase structure, and partial convergence would result from the “unstable” features available to the L2 Spanish speaker. Although the artificial nature of the Spanish-to-Kichwa translation task (utterances devoid of discourse/pragmatic context) is not conducive to the production of evidential validators, the lack of direct transfer of the Kichwa topicalizer -mi to Ecuadoran Andean Spanish cannot be accounted for by the Functional Convergence Hypothesis since this element shares the same basic syntactic and pragmatic status as -ka. Transfer of -mi may be impeded by the potentially confusing presence of the homophonous Spanish elements mí ‘me’ and mi ‘my’ as well as the first-person singular object clitic me (given the effective neutralization of Spanish mid and high vowels in Ecuadoran Andean Spanish). The equation Spanish también > tan ≡ Kichwa -pash provides the principal basis for convergence; the reason(s) for a less robust Ecuadoran Andean Spanish representation of the Kichwa FocP has yet to be determined.

5.  H  ow do speakers acquire and retain Ecuadoran Andean Spanish‑specific elements? Analyses based on a corpus of naturalistic speech data reveal recurring patterns in the Spanish of Kichwa-dominant bilinguals as well as possibly emergent traits that cannot be fully accounted for by postulating on-line transfer from Kichwa. In some instances the non-canonical uses of the gerund may result from spontaneous transfer of the Kichwa complementizer -shpa, but the translation of -shpa by the Spanish gerund is not fully predictable, apparently reflecting stereotypes formed during the colonial and early post-colonial periods and subsequently reinforced in foreigner-talk. Insertion of



On the tenacity of Andean Spanish 

-tan includes patterns not directly attributable either to Spanish también or to specific Kichwa particles. Insertion of -ca comes the closest to patterning like on-line Kichwa transfer, although the decision to mark topic with considerable consistency but rarely to mark comment/focus invites further analysis. What is noteworthy about the three aforementioned non-canonical features is that they co-occur among all of the Kichwadominant bilinguals who provided data for the present study as well as among speakers with similar linguistic profiles.

5.1.  Other non-canonical features in Ecuadoran Andean Spanish Additional non-canonical or archaic features also recur in the speech of older Kichwadominant bilinguals, although the same traits are not found in surrounding ­(Spanish monolingual) Euro-mestizo communities. One such feature is the use of singular 2nd person singular verb forms (with and without the locally prevalent subject pronoun vos) with the diphthong corresponding to Peninsular Spanish 2nd person plural vosotros (Toscano Mateus, 1953, pp. 230, 240). This usage has all but disappeared from contemporary Ecuador, but forms like vos sois ‘you are,’ and vos vais ‘you go’ can still be heard in the communities under study. Other non-canonical features include use of pasque < Sp. parece que ‘it seems that,’ archaic verb forms such as vide (modern Sp. vi ) ‘I saw,’ and truje (modern Sp. traje ‘I brought’), pronunciation of seis ‘six’ as [sajs], the use of andi for donde ‘where,’ and the use of castilla for castellano ­‘Spanish language.’ Fluent Spanish speaking bilinguals from the same communities use none of these features and have not done so in recent generations, but the translation experiment reveals that all members of these speech communities recognize the non-canonical traits as characteristic of older Kichwa-dominant speakers. Circumstantially the evidence suggests intra-community recycling of these configurations for an extended period, which implies circumstances in which Kichwa-dominant bilinguals communicated with one another in some approximation to Spanish or at least received considerable input from Kichwa-Spanish interlanguage. The data also point to a recent configurational shift following which these non-canonical features are no longer transmitted to successive generations.

5.2  The changing linguistic ecology of highland indigenous communities Until only a generation or two ago, the rural communities represented in the present study were effectively monolingual in Kichwa. Schools were rarely found in rural areas and attending schools outside of the communities entailed considerable personal and financial sacrifice. Most families did not send their girls to school, and boys’ attendance was sporadic at best, given the need to work in agriculture. The absence of formal education produced the dual effect of inhibiting the acquisition of canonical varieties of Spanish and providing an environment with few ­sociolinguistic s­ trictures

 John M. Lipski

against the use of Andean Spanish variants. For several generations the main vector for the introduction of Spanish was quasi-feudal labor on several large estates, including one owned by former Ecuadoran president Galo Plaza Lasso ­(Antamba ­Antamba  et  al.,  2011; Cachimuel, 2001; Cañarejo Quilumbaquí, 2010; López ­Paredes, 2002; Quemac Benalcázar & Ipiales Rosero, 2010). On these estates both men and women worked in the fields and many women worked indoors as cooks, cleaners, and personal servants. Kichwa speakers acquired at least some Spanish from overseers, service providers, and other servants, most of whom were mestizos with little or no formal education and whose use of Spanish may also have embodied some Kichwainfluenced configurations. According to several former peons and domestic servants interviewed by the author, on many estates use of Kichwa was discouraged, which led to use of Kichwa-based Spanish interlanguage. Incipient bilinguals not only overheard Kichwa-­influenced Spanish from fellow workers but in some instances used the same interlanguage when communicating with these workers. At the same time children in these communities were increasingly exposed to the Spanish interlanguage of older speakers, even before receiving any formal training in Spanish. In recent generations effectively monolingual Kichwa-speaking men began to leave the communities for military service and to work in economically more promising areas of the country, including Quito (in construction), Cayambe (commercial flower production) as well as the Pacific coast and later also the Amazon Basin oil-­producing region. These experiences put them into closer contact with canonical varieties of Spanish, although judging by the author’s conversations with several such individuals, many Andean Spanish traits persisted and were even reinforced upon return to the indigenous communities. During the same period some women began to migrate to urban areas such as Quito in search of employment, generally in domestic service; once more conversations with women who have returned from such employment reveal only partial acquisition of more canonical Spanish varieties. Younger bilinguals with increasing proficiency in Spanish were often raised by parents and other older relatives whose Spanish interlanguage – as overheard and as used with younger bilinguals – could persist into a generation of more balanced bilinguals alongside more monolingual-like configurations. Another aspect that has changed considerably in the last two generations is contact with monolingual Spanish speakers in surrounding towns. Three of the communities are located on or near the Pan-American Highway, but several miles away from the nearest town where significant numbers of Spanish speakers could be found, while another one is located on a major road leading to Ibarra, the provincial capital. Currently all of the communities are served by local buses that circulate several times per day, with most dwellings located no more than a few minutes’ walk from the nearest point for boarding a bus. Informal transportation is also provided by numerous pickup trucks and private automobiles. In previous



On the tenacity of Andean Spanish 

generations travel was by foot (or on horse- or mule-back for a few residents), with travel times ranging from less than an hour to half a day or more. Since most of the communities are located on mountainsides above the Spanish-speaking towns, return trips – which often involved carrying large bundles of merchandise – represented a considerable uphill sacrifice that limited such treks to the most essential. Of all of the communities only Pijal is located within easy walking distance of a Spanish-speaking town (González Suárez), and in Pijal comparatively fewer Andean Spanish traits remain in the speech of older residents. Zuleta, relatively close to Ibarra, also harbors fewer traditional Andean Spanish features. Conversely the hilltop community of Ugsha is located the furthest from any Spanish-speaking town, and older residents of this community show the least proficiency in Spanish and rarely use the language within the community.

Table 4.  Language recycling in the bilingual Imbabura communities SPEAKERS

hear Spanish from

speak Spanish to

children in the community; some adult bilinguals; monolingual Spanish speakers in other locations; overhear Kichwadominant speakers children and adult bilinguals in the community; monolingual Spanish speakers in other locations; overhear Kichwa-dominant speakers other children; adult bilinguals, Kichwa-dominant speakers

children in the community; some adult bilinguals; monolingual Spanish speakers in other locations

monolingual Spanish speakers in other locations; overhear some Kichwa-dominant Spanish and sometimes participate in conversations (during hacienda labor) some adult bilinguals in the community; Kichwa-dominant speakers during hacienda labor; monolingual Spanish speakers in other locations;

monolingual Spanish speakers on other locations; other bilingual & Kichwa-dominant speakers during hacienda labor

CONTEMPORARY TIMES older Kichwa-dominant

adult bilinguals

children

children in the community; some adult bilinguals; monolingual Spanish speakers in other locations all children and adults

PREVIOUS GENERATIONS older Kichwa- dominant

adult bilinguals

children

some adult bilinguals in the community; monolingual Spanish speakers in other locations; overhear some Kichwa-dominant speakers

some adult bilingual (mostly away from the community); Kichwa-dominant speakers during hacienda labor; monolingual Spanish speakers in other locations other children; some adult bilinguals in the community; monolingual Spanish speakers in other locations

 John M. Lipski

According to extensive informal observations by the present author, when Kichwa-dominant bilinguals use Spanish to communicate in towns, most of the speech acts take place in markets, health clinics, and buses, and are simple transactions that require little grammatical competence and evoke no strictures against interlanguage usage. The situation was similar for previous generations, so brief contacts with monolingual Spanish speakers in such settings had little effect on the Spanish of Kichwa-dominant speakers. The contemporary environment is considerably different as regards exposure to more canonical varieties of Spanish as well as opportunities to speak Spanish within the communities themselves. Although Kichwa is still being retained as the primary home language in many families and is taught in some schools, children learn Spanish beginning in pre-school and – increasingly – at home, and speak more Spanish than Kichwa with one another. All of the communities are served by elementary schools, and secondary schools are located in many nearby towns. Older Kichwa-dominant speakers are increasingly interacting in Spanish with children and young adults, and as the critical mass of speakers of canonical Spanish continues to grow at the same time that the population of older interlanguage speakers decreases, forms such as -ca, -tan, and gerunds instead of finite verbs are no longer being recycled and acquired by younger speakers. A comparison of the sociolinguistic environment of previous generations as well as the contemporary situation is given in Table 4, which summarizes the recycling of non-canonical Andean Spanish elements in previous generations and the attenuation of such recycling at the present time.

6.  Summary and conclusions The instances of -ca and -tan, and non-canonical gerunds in Ecuadoran Andean Spanish provide common threads linking speakers with widely varying degrees of ­proficiency in Spanish. By combining historical explorations, ethnographic fi ­ eldwork, and interactive tasks the complex nature of these speech communities has been brought into sharper focus. Although the impact of Kichwa is i­ndisputable in all of the phenomena described here, features such as over-use of the gerund and the particles -ca and -tan cannot be dismissed as spontaneous learners’ errors. The presence of other non-canonical items in these speech communities that are no longer found in neighboring areas provides evidence of intra-community recycling of Spanish linguistic features, even set against the background of predominantly Kichwa usage in previous generations. Although by their very nature preliminary and tentative, the observations presented in this study are offered as a potentially useful contribution to the study of a 500-year language ­contact environment.



On the tenacity of Andean Spanish 

References Albor, H. (1973). “Da” + gerundio, ¿un quechuismo? y otras maneras de atenuar los imperativos. Hispania, 56, 316–318.  doi: 10.2307/339024 Andronis, M. A. (2004). Iconization, fractal recursivity, and erasure: linguistic ideologies and standardization in Quichua-speaking Ecuador. Texas Linguistic Forum, 47, 263–269. Antamba Antamba, M. E., Cacuango Morán, L. C., Calapaqui Chicaiza, S. M., Pazmiño Sevillano, W. A., & Perugachi Peña, L. (2011). Revitalización de la memoria histórica sobre la “Toma de tierras en el Cantón Cotacachi parroquia: El Sagrario y en el Cantón Otavalo parroquias: Eugenia Espejo, San Rafael de la Laguna, González Suárez y San Pablo del Lago.” Universidad Politénica Salesiana Sede Quito, thesis, Licenciado en Ciencias de la Educación Mención en Docencia Básica Intercultural Bilingüe. Retrieved from: 〈http:// dspace.ups.edu.ec/handle/123456789/2063〉 (1 October 2014). Antón-Méndez, I. (2011). Whose? L2-English speakers’ possessive pronoun gender errors. Bilingualism: Language and Cognition, 14, 318–331.  doi: 10.1017/S1366728910000325 Bruil, M. (2008a). Give + gerund in Ecuadorian Spanish: A calque from Quichua or a large process of contact induced change? Leiden Working Papers in Linguistics, 5(1), 1–23. Bruil, Martine. (2008b). Innovations in the Ecuadorian Converb Systems: Grammatical change in language contact situations (Unpublished MA thesis). University of Leiden, Netherlands. Cachimuel, G. (2001). ‘Vuelta arriba acabamos la escuela. Vuelta abajo no podemos’: Opiniones de la comunidad educativa frente a la EIB en una comunidad del Ecuador (Unpublished master’s thesis). Universidad Mayor de San Simón, Cochabamba, Bolivia. Calvo Pérez, J. (2001). Caracterización general del verbo en el castellano andino y la influencia de la lengua quechua. In T. Fernández, A. Palacios, & E. Pato (Eds.), El indigenismo americano: Actas de las primeras jornadas sobre indigenismo (pp. 99–129). Madrid: Universidad Autónoma de Madrid. Calvo Pérez, J. (2008). Perú. In A. Palacios (Ed.), El español de América: contactos lingüísticos en Hispanoamérica (pp. 189–212). Barcelona: Ariel. Camacho, J. (2006). In situ focus in Caribbean Spanish: Towards a unified account of focus. In N. Sagarra & A. J. Toribio (Eds.), Selected proceedings of the 9th Hispanic Linguistics Symposium (pp. 13–23). Somerville: MA, Cascadilla Proceedings Project. 〈www.lingref.com〉, document #1362. Cañarejo Quilumbaquí, I. (2010). Particularidades del capital social en las comunidades indígenas: Casos de Angla, Gualaví y Cusipamba. (Unpublished master’s thesis). Facultad Latinoamericana de Ciencias Sociales (Programa de Políticas Públicas y Gestión, maestría en Desarrollo Local y Territorial), Ecuador. Retrieved from: 〈www.flascoandes.org/dspace17/ bitstream/10469/2467/4/TFLASCO-2010ICQ.pdf〉 (1 October 2014). Cisneros Estupiñán, M. (1999). Peculiaridades del gerundio en el suroccidente de Colombia. Thesaurus, 44, 1003–1035. Cole, P. (1982). Imbabura Quechua. Amsterdam: North-Holland. Dedenbach-Salazar Sáenz, S. (Ed.). (1993). Una gramática colonial del quichua del Ecuador. Bonn/St. Andrews: Bonner Amerikanistische Studien/Institute of Amerindian Studies, University of St. Andrews. Durston, A. (2007). Pastoral Quechua: The history of Christian translation in colonial Peru, ­1550–1650. Notre Dame, IN: Notre Dame University Press. Franco, W. P. (2007). Gramática kichwa: Kichwa unificado. Quito: n. p.

 John M. Lipski Gifford, D. (1969). Serrano speech: Notes on the mestizo dialect of SE Peru. Forum for Modern Language Studies, 5(2), 162–179.  doi: 10.1093/fmls/V.2.162 González de Holguín, D. (1901). Arte y diccionario qquechua-español correjido y aumentado por los RR. PP. Redentoristas al que en 1608 publicó el Rvdo. P. Diego González de Holguín, S.J. Lima: Imp. del Estado. Granda, G. de. (1995). Un quechuísmo morfosintáctico en dos áreas extremas del español andino: las perífrasis verbales de gerundio con valor perfectivo en el noroeste argentino y el sur colombiano. Anuario de Lingüística Hispánica, 11, 151–159. Granda, G. de. (2001). El español del noroeste argentino y su inserción en el área lingüística andina: implicaciones metodológicas. In K. Zimmermann & T. Stolz (Eds.), Lo propio y lo ajeno en la lenguas austronésicas y amerindias (pp. 107–125). Madrid: Iberoamericana. Haboud, M. (1998). Quichua y castellano en los Andes ecuatorianos: Los efectos de un contacto prolongado. Quito: Abya-Yala. Haboud, M., & de la Vega, E. (2008). Ecuador. In A. Palacios (Ed.), El español de América: Contactos lingüísticos en Hispanoamérica (pp. 161–187). Barcelona: Ariel. Hartmann, R. (1999). El obispo de Quito Luis Francisco Romero y el catecismo quichua de 1725. In L. Crickmay & S. Dedenbach-Salazar Saenz (Eds.), La lengua de la cristianización en Latinoamérica: Catequización e instrucción en lenguas amerindias. The language of Christianisation in Latin America: Catechisation and instruction in Amerindian languages (Bonner Amerikanistische Studien, Vol. 32; Centre for Indigenous American Studies and Exchange St. Andrews Occasional Papers, 29), (pp. 61–87). Markt Schwaben: Anton Saurwein. Hatzidaki, A., Branigan, H., & Pickering, M. (2011). Co-activation of syntax in bilingual language production. Cognitive Psychology, 6, 123–150.  doi: 10.1016/j.cogpsych.2010.10.002 Huerta, A. de. (1993[1616]). Arte breve de la lengua quechua. With a commentary by Ruth Moya and transcriptions by E. Villacís. Quito: Proyecto Educación Bilingüe Intercultural & ­Corporación Editora Nacional. Hurley, J. K. (1995). The impact of Quichua on verb forms used in Spanish requests in Otavalo, Ecuador. In C. Silva-Corvalán (Ed.), Spanish in four continents: Studies in language contact and bilingualism (pp. 39–51). Washington, DC: Georgetown University Press. Juan de Aguilar. (1690). Information can be found here: http://www.worldcat.org/title/ gramatica-quechua-de-juan-de-aguilar/oclc/18351293 Kany, C. E. (1945). American Spanish syntax. Chicago, IL: University of Chicago Press. King, K., & Haboud, M. (2002). Language planning and policy in Ecuador. Current Issues in Language Planning, 3, 359–424.  doi: 10.1080/14664200208668046 Kroll, J., Bobb, S., Misra, M., & Guo, T. (2008). Language selection in bilingual speech: evidence for inhibitory processes. Acta Psychologica, 128, 416–430. doi: 10.1016/j.actpsy.2008.02.001 Kroll, J., Dussias, P., Bogulski, C., & Valdes Kroft, J. (2011). Juggling two languages in one mind: What bilinguals tell us about language processing and its consequences for cognition. ­Psychology of Learning and Motivation, 56, 229–262. doi: 10.1016/B978-0-12-394393-4.00007-8 Lefebvre, C., & Muysken, P. (1988). Mixed categories: Nominalizations in Quechua. Dordrecht: Kluwer. Lipski, J. (2013). ¿Qué diciendo nomás? Tracing the sources of the Andean Spanish gerund. Spanish in Context, 10, 227–260.  doi: 10.1075/sic.10.2.03lip Lipski, J. (2014). Syncretic discourse markers in Kichwa-influenced Spanish: transfer vs. emergence. Lingua.  doi: 10.1016/j.lingua.2014.07.002



On the tenacity of Andean Spanish 

Lipski, J. (2015). Colliding vowel systems in Andean Spanish: Carryovers and emergent properties. Linguistic Approaches to Bilingualism, 5(1), 91–121.  doi: 10.1075/lab.5.1.04lip López Paredes, D. (2002). Aproximación histórica de los cambios de la seguridad y consumo alimentario entre los pueblos quichuas de la sierra ecuatoriana: Otavalos y Cayambis en la cuenca del Lato San Pablo, provincia de Imbabura (Unpublished MA thesis in anthropology). Facultad Latinoamericana de Ciencias Sociales, Quito, Ecuador. Retrieved from: 〈http://67.192.84.250/dspace/handle/10469/709〉 (1 October 2014). Méndez Vallejo, D. C. (2009). Focalizing ser (‘to be’) in Colombian Spanish (Unpublished Ph.D. dissertation). Indiana University, IN. Merma Molina, G. (2004). Lenguas en contacto: Peculiaridades del español andino peruano. Tres casos de interferencia morfosintáctica. Estudios de Lingüística Universidad de Alicante (ELUA), 18, 191–211. Merma Molina, G. (2005). El gerundio en el español andino peruano: Una perspectiva pragmática. UniverSOS: Revista de Lenguas Indígenas y Universos Culturales, 2, 117–128. Ministerio de Educación, Ecuador. (2010). Kichwa: kichwata yachaymkanta/gramática pedagógica. Quito: Ministerio de Educación, Colección Runakay. Retrieved from: 〈www.educacion.gob.ec/wp-content/plugins/download-monitor/download.php?id=3114〉 (1 October 2014). Muñío Valverde, J. L. (1995). El gerundio en el español medieval, s. XII–XIV. Málaga: Librería Ágora. Muysken, P. (1980). Sources for the study of Amerindian contact vernaculars in Ecuador. Amsterdam Creole Studies, 3, 66–82. Muysken, P. (1982). The Spanish that Quechua speakers learn: L2 learning as norm-governed behavior. In R. Anderson (Ed.), Second languages: A cross-linguistic perspective (pp. ­101–124). Rowley, MA: Newbury House. Muysken, P. (1985). Contactos entre quichua y castellano en el Ecuador. In S. Moreno Yáñez (Ed.), Memorias del primer simposio europeo sobre antropología del Ecuador (pp. 377– 452). Bonn: Instituto de Antropología Cultural de la Universidad de Bonn & Quito: Abya-Yala. Muysken, P. (1995). Focus in Quechua. In K. É Kiss (Ed.), Discourse configurational languages (pp.375–393). Oxford: Oxford University Press. Muysken, P. (1997). Media Lengua. In S. Thomason (Ed.), Contact languages: A wider perspective (pp. 365–426). Amsterdam: John Benjamins.  doi: 10.1075/cll.17.13muy Muysken, P. (2005). A modular approach to sociolinguistic variation in syntax: The gerund in Ecuadorian Spanish. In L. Cornips & K. Corrigan (Eds.), Syntax and variation: Reconciling the biological and the social (pp. 31–53). Amsterdam: John Benjamins. doi: 10.1075/cilt.265.03muy Muysken, P. (2012a). Spanish affixes in the Quechua languages: A multidimensional perspective. Lingua, 122, 484–493.  doi: 10.1016/j.lingua.2011.10.003 Muysken, P. (2012b). Root/affix asymmetries in contact and transfer: Case studies from the Andes. International Journal of Bilingualism, 16, 22–36.  doi: 10.1177/1367006911403211 Niño-Murcia, M. (1995). The gerund in the Spanish of the North Andean region. In C. SilvaCorvalán (Ed.), Spanish in four continents: Studies in language contact and bilingualism (pp. 83–100). Washington, DC: Georgetown University Press. Olbertz, H. (2002). Dar + gerundio en el español andino ecuatoriano. Sintaxis, semántica y origen. Círculo de Lingüística Aplicada a la Comunicación 12. Retrieved from: 〈http://www. ucm.es/info/circulo/no12/olbertz.htm〉 (1 October 2014).

 John M. Lipski Olbertz, H. (2008). Dar + gerund in Ecuadorian Highland Spanish: Contact-induced grammaticalization? Spanish in Context, 5, 89–109.  doi: 10.1075/sic.5.1.06olb Ordóñez, F., & Treviño, E. (1999). Left dislocated subjects and the pro-drop parameter: A case study of Spanish. Lingua, 107, 39–68.  doi: 10.1016/S0024-3841(98)00020-5 Quemac Benalcázar, D. C., & Ipiales Rosero, W. A. (2010). Propuesta de repoblación forestal en el Taita Imbabura del Cantón Otavalo (Unpublished doctoral dissertation). Universidad Técnica del Norte (Facultad de Ingeniería en Ciencias Agropecuarias y Ambientales, Escuela de Ingeniería Forestal), Ecuador. Retrieved from: 〈http://repositorio.utn.edu.ec/ handle/123456789/113〉 (1 October 2014). Rindstedt, C., & Aronsson, K. (2002). Growing up monolingual in a bilingual community: The Quichua revitalization paradox. Language in Society, 31, 721–742. doi: 10.1017/S0047404502315033 Sánchez, L. (2003). Quechua-Spanish bilingualism: interference and convergence in functional categories. Amsterdam: John Benjamins.  doi: 10.1075/lald.35 Sánchez, L. (2010). The morphology and syntax of topic and focus: Minimalist inquiries in Quechua. Amsterdam: John Benjamins.  doi: 10.1075/la.169 Santo Tomás, Fray D. (1951 [1560]). Grammatica o arte de la lengua general de los indios de los reynos del Perú. Lima: Edición del Instituto de Historia, Universidad Nacional Mayor de San Marcos (facsimile edition). Suárez Roca, J. L. (1992). Lingüística misionera española. Oviedo: Pentalfa Ediciones. Toribio, A. J. (2002). Focus on clefts in Dominican Spanish. In J. Lee, K. Geeslin, & J. C. ­Clements (Eds.), Structure, meaning, and acquisition in Spanish: papers from the 4th Hispanic Linguistics Symposium (pp. 30–146). Somerville, MA: Cascadilla Press. Torres Cacoullos, R. (2000). Grammaticization, synchronic variation, and language contact: A study of Spanish progressive -ndo constructions. Amsterdam: John Benjamins. doi: 10.1075/slcs.52 Torres Rubio, D. de. (1754[1603]). Arte y bocabulario de la lengua Quichua general de los indios del Perú añadido por el P. Juan de Figueredo. Ahora nuevamente corregido y comentado en muchos vocablos, por un Religioso de la misma compañía. Lima: En la imp. de la Plazuela de San Christoval. Toscano Mateus, H. (1953). El español del Ecuador. Madrid: Consejo Superior de Investigación Científica. van Heuven, W., Schriefers, H., Dijkstra, T., & Hagoort, P. (2008). Language conflict in the ­bilingual brain. Cerebral Cortex, 18, 2706–2716.  doi: 10.1093/cercor/bhn030 Vázquez, H. (1940). Reparos en nuestro lenguaje usual. Quito: Ecuatoriana.

Spanish and Valencian in contact A study on the linguistic landscape of Elche Francisco Martínez Ibarra Towson University

The purpose of my study is to analyze the contact between Spanish and Valencian in the linguistic landscape of Elche. What are the languages being used and on what types of signs? Does the linguistic landscape reflect the language use and attitudes of the population of Elche? What does the linguistic landscape say about the way language policies are being implemented by local and regional governments? Over 3000 data items were collected and examined based on a number of factors such as the languages being used, the author of the signs, and the location where each sign was observed. Results confirm the dominance of Spanish and suggest a weak implementation of language policies aimed at the normalization of Valencian. Keywords:  linguistic landscape; Catalan; Valencian; Elche

1.  Introduction The purpose of my study is to analyze the linguistic landscape (LL) of Elche, Spain, and describe the contact between Spanish and the western variety of Catalan (Valencian henceforth). Among other advantages, the study of the linguistic landscape helps us understand the sociolinguistic and cultural diversity of speech communities. In this case in particular, it may direct our attention to the presence of speakers of languages that we might otherwise not be aware of, such as those of foreign immigrants. Furthermore, it shows how official language policies are implemented in a region where the native language, Valencian, is in poor health, as supported by data on language use from academic and official sources. My objective with the current paper is to answer the following questions: (1) what languages can be found in the LL of Elche and what is the type of signage employed? (2) Does the linguistic landscape reflect the language use of the population of Elche? (3) What does the linguistic landscape say about the way language policies are being implemented by the local and regional governments?

doi 10.1075/ihll.8.05mar © 2016 John Benjamins Publishing Company

 Francisco Martínez Ibarra

In the following subsections to the introduction I underline the importance of studying the LL for the field of sociolinguistics by briefly looking at contemporary and past research. A description of both the speech community under study and the most significant traits that characterize its linguistic behavior is provided. Next, in Section 2, I explain the methodology employed for the collection of the data and how these data were analyzed. The third section describes the main findings, organized into three groups: signage, author, and location. Finally, the reader is encouraged to reflect upon the benefits that the study of LL may offer to the field of sociolinguistics as well as upon the role the LL may play in the process of linguistic normalization of a particular language.

1.1  Studying the linguistic landscape The study of the LL is a relatively new area of research in the field of sociolinguistics. Although the first attempts at analyzing public signage from a linguistic perspective date back to the end of 1970s (cf. Rosenbaum, Nadel, Cooper, & Fishman, 1977; Tulp, 1978), it was the last decade of the 20th century and the first decade of the 21st that produced the greatest number of studies. Research on the field has focused on issues such as the implementation of language policies (Sloboda, Szabó-Gilinger, Vigers, & Simicic, 2010), the consequences of immigration (Castillo Lluch & Sáez Rivera, 2011), or the impact of English as a lingua franca (Piller, 2003). The most quoted definition of LL is probably that of Landry and Bourhis: The language of public road signs, advertising billboards, street names, place names, commercial shop signs, and public signs on government buildings combines to form the linguistic landscape of a given territory, region, or urban agglomeration.(Landry & Bourhis, 1997, p. 25)

Certainly, current technological innovations such as neon lights, flat-screen panels, or electronic scrolling banners widen the possibilities for data collection. In any case, the study of the LL focuses on any and all written samples as they are found on the public streets, usually, within the city limits. However, there are cases in which the object of study extends to larger territories such as the country of Malta (cf. Sciriha & ­Vassallo, 2001) or the region of Brittany in France (cf. Vigers, 2013). In the case of Landry and Bourhis’s (1997) work in Canada, significance is derived from the fact that it was one of the first projects to show that the presence of a certain language (i.e. French) in the LL of any community can influence the perception that speakers in that community have about that language. These findings confirmed Tulp’s (1978) results two decades earlier when he studied how public signage in B ­ russels was being ‘Frenchified’. Not surprisingly, research on LL is most prolific in multilingual contexts. Israel and its capital city Jerusalem have proven to be a popular focus. Among the different projects that have analyzed its LL, we can find Spolsky



Spanish and Valencian in contact 

and ­Cooper’s (1991) description of how the LL of the city reveals its recent history of ­British, Jordanian, and Israeli dominance. Likewise, Ben Rafael, Shohamy, Amara, and Trumper-Hecht (2006) explain how the use of Arabic, Hebrew, and English on public signage changes considerably depending on whether the author of the sign is private or governmental (i.e. Bottom-up vs. Top-down). Such findings reveal dominance/­ subordination behaviors among different ethnolinguistic groups. Other researchers have preferred to focus on contexts with a priori lower indexes of linguistic diversity due to the presence of a strong, prominent language that controls all interaction. This is the case of Japanese in Tokyo (cf. Backhaus, 2007) or of ­Italian in Rome. In the case of Rome, Griffin (2004) analyzed the use of English within the LL and concluded that English was not being employed because of its communicative function as a lingua franca in an international context, but rather because it was perceived as an attractive and fashionable language. Also in Rome, Bagna, and Barni (2005) studied how the LL reflects the presence of new immigrant communities. The most fascinating part of their study is the fact that they developed an innovative methodology for the collection and analysis of the data that involved the use of software and electronic devices designed exclusively for the study of the LL. In Spain, research on the LL has produced a great variety of studies as well, although two main themes seem to prevail: first, research has been aimed at describing how the LL reflects the languages of new immigrant communities; second, substantial attention has been directed at examining how language policies are implemented in those areas in which Spanish shares co-officiality with other languages such as ­Catalan, Basque, or Galician. In the former group of research interests, we find both projects that analyze the presence in the LL of languages other than Spanish (e.g. Castillo Lluch & Sáex Rivera, 2011) and projects that analyze linguistic contact between different dialects of Spanish (e.g. Pons Rodríguez, 2011). Regarding the presence in the LL of languages that share co-officiality with Spanish, the situation seems to vary depending on the region. In Catalonia, Miró Vives (1998) looked at how the LL of Tarragona evolved during the 1990s and noticed that monolingual signage in Catalan had increased considerably, primarily in economically dynamic areas as well as in neighborhoods with the youngest populations. Similarly, Solé i Carmadons (1998) documented the predominance of monolingual Catalan signs in the LL of Barcelona, underscoring the fact that their presence decreased depending on the type of business being advertised (e.g. telecommunications and automobiles). Other projects have noticed the predominance of Catalan on monolingual signage was lower in the LL of coastal cities with high volumes of tourism because the LL is shared with Spanish as well as other international languages such as English and/or German (e.g. Lepretre Alemany & Romaní Olivé, 2000). In a study of the LL of Majorca, in the Balearic Islands, Bruyèl Olmedo and Juan Garau (2009) observed how the use of Catalan in such a touristic city is limited by the presence of

 Francisco Martínez Ibarra

Spanish, English, and even German. According to these authors, the use of Catalan in the LL is aimed at portraying authenticity and identity in connection with local culture. In the Basque Country, we find a different situation from that of Catalonia. According to Cenoz and Gorter (2008), Spanish seems to control the LL in general, particularly in monolingual signage produced by a private author (i.e. bottom-up). In contrast to Catalonia, where monolingual signage was predominant, bilingual and multilingual signs seem to be the most popular type of signage in the Basque Country. The presence of the minority language, Basque, is relatively high in this context. In a previous study, the same authors compared the LL of San Sebastian and Ljouwert in the Netherlands (Cenoz & Gorter, 2006). Their findings suggest that the implementation of a strong linguistic policy by the Basque Government seems to have had a positive effect on the use of Basque on bottom-up signs. Such results contrast with the situation in Galicia, as described by Regueira, López Docampo, and Wellings (2013), who explain how the use of Galician in the LL of Pontevedra and Santiago de Compostela is limited to top-down signs, that is, signs produced by the government. The same authors interpret these results as clear evidence of the failure of the Galician government and their language policies to influence the use of Galician on bottom-up signage. According to Dunlevy (2012), the presence of Galician on a bottom-up sign seems to increase only in more rural contexts. Thus, with the exception of Catalonia, Spanish usually controls the LL of those regions where co-officiality is shared with other language. This is mainly true for ­bottom-up, monolingual signs, which is also the situation we find in the Autonomous Community of Valencia (ACV). By comparing the LL of Gandía, where a majority of the population speaks Valencian daily, with the LL of the city of Valencia, where the use of Spanish for daily interaction prevails, Ladó (2011) noticed that past and current language policies aimed at the promotion of Valencian seem to have had very little impact concerning the use of the language on public signage, independent of the level of bilingualism in the cities. According to the same author, her findings are evidence of the ambivalence displayed by the Valencian government toward normalizing the minority language since the beginning of the most recent democracy. Unfortunately, to my knowledge, Ladó’s work is the only project to have examined the presence of Valencian in the LL.

1.2  Language use: Beyond a linguistic issue The first attempts at studying linguistic behavior and the contact between Spanish and Valencian are found in the work of Ninyoles (1969, 1971). His efforts have been continued by Blas Arroyo (1994, 2005), Casesnoves Ferrer (2001, 2005, 2010), Ciscar Ramírez, González Martínez & Pérez Ledo (2002), Nicolás Amorós (2004), and Pradilla (2005, 2008), among others. In the southern territories of the ACV, sociolinguistic research on



Spanish and Valencian in contact 

Valencian and Spanish is conducted by authors such as Mas i Miralles (1994), Montoya Abat (1996), Segura i Llopes (1996, 2003), Torres i Selva (2003), and Baldaquí (2004, 2005), among others. Their academic work, together with the information provided by sociolinguistic surveys on language use and attitudes conducted by the Valencian Department of Education, describe the linguistic situation of the region. Data from the last of such surveys (SIES, 2010) show that, for the whole ACV, 68.8% of the population can understand Valencian, 48.5% can speak it, 45.3% can read it, and 26.4% can write it. When compared to data from the 1992 sociolinguistic survey, data show a decline in oral skills: 14.4% in listening and 12.6% in speaking. In contrast, data on writing, a skill that is mainly learned at school, show a substantial improvement, from 15.8% in 1993 to 26.4% in 2010. However, when compared to results from 2005 it should also be noted that writing skills decreased from 29.8%. Either way, it seems evident that the population’s writing skills in Valencian are not thriving. The situation in the city of Elche is similar to the rest of the ACV as available data show a general decline in linguistic skills, particularly in communicative skills, such as speaking (Martínez Ibarra, 2013).1 According to findings by the same author, this decline has a direct impact on the social use of the language, as only 22% of the total population in Elche reports using Valencian quite/very frequently. In general, Elche seems to follow a similar pattern to that which characterizes the province of Alicante, an area defined by active bilingualism with a predominance of Spanish in the private domain, and passive bilingualism with a predominance of ­Spanish in the public domain (Gimeno-Menéndez & Gómez Molina, 2007, p. 10). This means that the area is characterized by a strong diglossia, as understood by ­Fishman (1967), in which the use of Spanish governs any type of interaction, private or public, while the use of Valencian is limited to private conversations, such as conversations with family members, close friends or neighbors in certain neighborhoods. In particular, the daily use of Valencian in Elche seems most frequent among older speakers (60+), women, speakers whose parents were born in Elche, and speakers who live in more traditional neighborhoods such as the city center, Pont Nou and El Plà (cf.. Martínez Ibarra, 2013). In order to understand the motivations behind these results, we need to consider a number of factors and historical events that have shaped the relationship between valencianos, which includes people in Elche, and in Valencia as a whole. Such factors relate to sociolinguistic issues such as the steady loss of social prestige in favor of ­Spanish; or political issues such as Valencian secessionism, that argue for a ­distinct

.  The most current official sociolinguistic survey available for the city of Elche was conducted in 1994. Martínez Ibarra (2013) provides a more contemporary account of linguistic practices in Elche.

 Francisco Martínez Ibarra

identity for Valencian territories, independent from the neighboring region of ­Catalonia (cf. Climent Ferrando, 2005; Montoya Abat & Mas i Miralles, 2011a, 2011b).

1.3  The city of Elche Elche is located in the province of Alicante, in the southern territories of the Autonomous Community of Valencia, on the Mediterranean coast. According to the 2013 census, the population of Elche was 230,224, which placed it among the 20 most populated cities in Spain. The largest population growth took place in the 1960s and 1970s, when the population of Elche grew from 73,320 in 1960 to 162,873 in 1981. The expansion was motivated by plentiful employment opportunities in the booming shoe industry, which remains the main economic sector to this day. As can be expected, such a dramatic shift in the demographics of the city caused a series of consequences that have shaped Elche structurally, socially and linguistically. The majority of the newly arrived immigrants came from neighboring regions in which Spanish was the only language spoken (e.g. Murcia, Jaen, Almeria, Albacete). This reinforced the already existing situation of language shift that had originated in the sixteenth century and still poses a threat to the survival of Valencian. Spanishspeaking immigrants who arrived in Elche during the 1960s and 1970s first settled in the neighborhoods of Pont Nou and el Plà, where they mixed with the native population. However, the construction of new neighborhoods such as Carrús and the expansion of others such as Altabix were soon necessary to address the large number of immigrant workers. Towards the end of the 20th century and beginning of the 21st century, we find a second significant wave of immigration motivated by employment opportunities, this time in the construction industry. The vast majority of this new group of foreign immigrants arrived from South America, Eastern Europe, and North Africa. Among the countries with significant migration to Elche, we find Colombia (1,773), Ecuador (1,610), Morocco (4,196), and Romania (5,416). The majority of these foreign immigrants settled in the neighborhood of Carrús. Politically, the Partido Socialista Obrero Español (PSOE), a left-wing party, was in control of the Valencian government from the beginning of the most recent period of democracy in Spain until 1995. During the party’s tenure, they supported a series of policies, such as the Llei d’Ús i Ensenyament del Valencià, aimed at promoting the use of Valencian in education and society in general. To a certain extent, these policies reduced the language shift from Valencian to Spanish. However, many agree that the way they were implemented by the PSOE was too weak and unconvincing (cf. ­Climent-Ferrando, 2005; Mas i Miralles & Montoya Abat, 2011b). Since 1995 the Partido Popular (PP), a right-wing party that supports the idea of a strong central government, has governed the ACV. Throughout the nearly twenty



Spanish and Valencian in contact 

years that they have been in power, their decisions concerning language policies have been characterized by a lack of support for Valencian, and a favoring of other languages such as Spanish and English. Pradilla (2008) refers to their actions as a counter policy for language planning. In the city of Elche, the local government has been mainly controlled by the PSOE, although support from Compromís,2 a coalition of left-wing parties that support a distinctive Valencian agenda, was occasionally needed in order to gain control of the municipal government. In exchange for their political support, PSOE always allowed Compromís to administer the local office for cultural affairs. During their tenure, they developed several programs and campaigns to promote the social use of Valencian in Elche, though their efforts were frequently restricted by the central Valencian government. For instance, one of their campaigns for the promotion of Valencian, Fes teu el Valencià or ‘Make Valencian yours,’ had to seek financial support from the Catalan government, after funding from the Valencian government, controlled by the PP, was denied. After the local elections of 2011, the PP took control of the municipal government and began implementing policies that were in line with the central Valencian government, and were aimed at favoring the use of Spanish over Valencian (cf.  Moltó,  2011). As our data will show, their lack of interest in the promotion of Valencian has, on occasion, been in violation of local and regional laws.

2.  Methodology and data The most challenging part of studying the LL of a certain community is defining the object of study. In order to answer the questions anticipated in the introduction, a total number of 1,832 texts were identified. A text is defined here as “any and all types of publicly displayed writing whose content can be associated to the business, institution, or private individual who displays it” (Franco Rodríguez, 2008). For instance, using a bar as an example, the content displayed on a small blackboard in the street with information about the menu, and the name of the bar as displayed on a large sign on the exterior are considered parts of the same text, that is, of the same unit of analysis. If the language used on the blackboard is Spanish but the language of the large sign on the wall is Valencian, we would consider the text an example of a bilingual text. It was decided not to include in the analysis any cases of moving signage (i.e. buses, cars, clothing); mainly because “their presence cannot be accurately contrasted in subsequent research, especially in diachronic research” (Franco Rodríguez, 2009). Sometimes it was not possible to determine the language being used either because .  Political alliance that groups together Izquierda Unida, BLOC Nacionalista Valencia, and Verds.

 Francisco Martínez Ibarra

all words belong both to Spanish and Valencian lexicons (e.g. Bar Capri) or because they are cases of lexical innovation (e.g. Nutrimed). Any such cases of ambiguity were not included in the final analysis. Similarly, last names (e.g. Tous, García) were also excluded. With these considerations, the total final number of texts analyzed for our study was reduced to 1,649. Fieldwork took place in June, July, and August of 2013 and consisted of the observation of the languages being used for signage in the streets of Elche. The following information was entered on a form: name of the business, number of languages being displayed, type of actor (i.e. top-down vs. bottom-up), and location by neighborhood and type of street. In those cases where two or more languages were used, it was also noted if any of the languages being used could be identified as predominant. This was determined by looking at the type of font employed for each language or by the spatial location of languages on the sign. For instance, a bilingual sign with Spanish written above Valencian was categorized as Spanish-predominant. Similarly, a bilingual sign with Valencian in bold or using a larger font size would be Valencian-predominant (cf. Scollon & Scollon, 2003 on ‘place semiotics’). Two main factors were taken into consideration in the selection of each location for observation: the type of street and the neighborhood. The city of Elche is organized into 29 neighborhoods, each of them subdivided into administrative sections. Based on Larrosa Rocamora (2000), all neighborhoods were grouped into four urban areas or neighborhoods (i.e. Carrús, PontNou/ElPla, Centro, Altabix). The decision took into consideration historical events and demographic data. A minimum of two main streets, two secondary streets, and two plazas were observed for each of the neighborhoods. When deciding whether streets were categorized as primary or secondary, the main investigator, who is a native of Elche and speaks Spanish, Valencian and English, used the 2013 Guide for Elche available at the Tourism Office and followed the following criteria: (1) main streets had to be longer than 15 blocks and have a minimum of five businesses per block; (2) secondary streets had to be shorter than 10 blocks and have a maximum of three businesses per block. In addition, the main investigator asked his family members and friends who live in Elche for a list of five main streets for each neighborhood. The top two names were chosen if they satisfied the ‘block criteria.’ Regarding secondary streets, any street could be selected for observation as long as those names had not been mentioned in the list for primary streets, and met the block criteria.

3.  Results 3.1  Types of signs Monolingual signage governs the LL of Elche, as it accounts for more than two thirds (71%) of the total number of texts analyzed. Only 28% of collected texts were



Spanish and Valencian in contact 

­ ilingual texts. In general, Spanish prevails as the main language in the LL of Elche, b being present in 69% of all texts.3 Valencian and English presence is much lower: 18% and 13%, respectively. Similarly, data show that most monolingual signage texts (85%) are only written in Spanish, while just 13% of monolingual texts are written only in Valencian. The use of English in monolingual texts is almost insignificant (2%). However, the use of Valencian and English is higher in the case of bilingual texts, as illustrated in Figure 1. 7% Spanish/Valencian Spanish/English Spanish/Other

48 %

45 %

Figure 1.  Distribution of bilingual texts

We can see how almost half of the bilingual texts in Elche display a combination of Spanish and English (48%), a percentage slightly higher than the Spanish-Valencian (45%). Other combinations of languages, having Spanish always present, account for 7% of bilingual texts and include languages such as Chinese, Arabic, Romanian, Italian or French. The predominance of Spanish is even greater when we examine the differences between languages considering the type of font and the spatial location on the signs. Spanish is the most prominent language in 86% of the cases, while Valencian takes prominence in only 5% of bilingual texts. Regarding the type of business, Figure 2 shows the percentages of use of Valencian and English for this type of text.4 The categorization is an adaptation from FrancoRodríguez (2008) and is not intended to be exclusive nor exhaustive but merely illustrative since there are cases in which a type of text could belong to two categories.

.  Data referring to the presence of a language in the LL is determined by combining all texts that display that language, either by itself or together with other language(s). For example, the presence of Valencian is determined by combining the number of monolingual texts in Valencian and the number of bilingual and multilingual texts that use Valencian. The final percentage is based on the total number of analyzed texts. .  Data displayed in Figure 2 come from the analysis of texts only belonging to businesses. Any other types of texts (e.g. urban infrastructure) were excluded from this analysis.

 Francisco Martínez Ibarra

For example: a driving school, which in my analysis is included in the category of ­‘education,’ could also be analyzed under ‘auto.’ Home 19

Clothing

32

Technology Legal and professional services

12

Healthcare

23

Dining Education

24

Personal Care Auto Food 0

18 20

40 English

%

60

80

100

Valencian

Figure 2.  Use of Valencian and English in the linguistic landscape according to the ­economic field

Language choice for written advertising has been shown to have significant consequences in relation to the way we associate languages with certain values (cf. Noriega & Blair, 2008). Previous research such as Wölck (2004) or Casesnoves Ferrer (2010) claim that a minority language (e.g. Valencian) is frequently associated with positive values such as identity, trust, and familiarity. These characteristics can be highly valued in fields such as education, dining, and legal and professional services, where the use of Valencian seems to be the greatest. English, however, appears to serve a different role. Ross (1997) explains how it is internationally perceived as a highly attractive language and any business using it is often identified as modern and up-to-date. Such qualities seem to be critical in constantly innovating fields such as technology (32%), healthcare (23%) and ­clothing (19%). Finally, if we analyze the data in more detail and we subdivide texts into main and secondary signs, we can identify two distinctive functions for Spanish and Valencian. Main signs are understood as “the part of a text that identifies the issuer – be it a particular individual, a business, or an official institution – and the type of activity or semantic field with which it is associated” (Franco-Rodríguez, 2008, p.15).5 A ­secondary sign .  “…la parte del texto por la que podemos identificar al emisor – ya sea un particular, una institución o un negocio – y el tipo de actividad que realiza o el ámbito semántico al que se circunscribe” (Franco-Rodríguez, 2008, p.15).



Spanish and Valencian in contact 

is the part of a text that is used to communicate information related to the particular individual, business, or official institution. Figure 3 provides a comparison of the presence of Valencian and Spanish on main and secondary signs. 100

Spanish Valencian

80

English

%

60 40 20 0

Main Sign

Secondary Sign

Figure 3.  Distribution of languages according to the type of sign

As illustrated, the general dominance of Spanish continues to be the norm. Its use is even greater in secondary signs (92%), which are mainly used to communicate information (e.g. discount offers, methods of payment, schedules). The use of Valencian on secondary signs is much lower (8%), while the use of English is non-existent. That is, Spanish is the preferred language for communication in Elche. In main signs, a type of sign that may suggest identity and symbolic values, the presence of Valencian is much higher (19%). When we take a closer look at that percentage, we find that 81% of main signs in Valencian consist of four or fewer words, which may suggest that the use of this language carries symbolic values. Griffin (2004) examined the presence of English in the LL of Rome. Among his conclusions, he claims that the fact that three quarters of public signs in English consisted of four or fewer words suggests that, in general, English is not a language chosen to communicate information but rather it is mostly used as a symbol. A qualitative study based on personal interviews with shop owners could help us confirm or contradict the claim that the use of Valencian on main signs is associated primarily with symbolic motivations.

3.2  Author: Top-down vs. bottom-up signage The Llei d’Ús i Ensenyament del Valencià (LUEV) groups all cities in the ACV into two main linguistic areas: a predominantly Spanish-speaking area, and a predominantly Valencian-speaking area. The way in which language policies are implemented, as in the case of bilingual education, depends on this division of the territories, and the same applies for any policies concerning public, governmental signage. In agreement with the LUEV, the city of Elche, located in a predominantly Valencian-speaking

 Francisco Martínez Ibarra

area, passed the Guidelines for the Linguistic Normalization for the City of Elche and its Administration (Reglament de Normalització Lingüistica per al Municipi d’Elx i el seu Ajuntament) in order to corroborate the law locally and support the recovery of ­Valencian in 1997. Article 15 of these guidelines explains the following:6 1. All signs, posters and notices found at buildings, offices and municipal services must be written in Valencian. 2. Registration and labeling that identifies municipal-owned property, such as vehicles, machinery, general equipment or buildings, must be written in Valencian. 3. Signs on public roads intended to provide information to passers-by will have to be written in Valencian. Ajuntament d’Elx (1997, p. 9; author’s translation). According to those guidelines, we might think that all signage in Elche produced by the local government would be monolingual in Valencian or, where not possible, it would be a Valencian/Spanish bilingual type of signage. We would expect the presence of Valencian on top-down signs to be the preferred option by large margins. However, data in Figure 4 show a different situation. 100

Spanish

80

%

60

Valencian

74

48

English

52

40 20 0

10

15

0 Top-down

Bottom-up

Figure 4.  Language distribution depending on the author

Two different trends can be observed. First, in an institutional or governmental context, the presence of Valencian is slightly higher than the presence of S­ panish;

.  (1) Els rètols, cartels i indicacions existents als immobles, dependències i serveis municipals, hauran d’estar escrits en valencià. (2) Les inscripcions i retolacions que identifiquen béns de propietat municipal, com ara vehicles, maquinària, immobles o utillatge en general, s’hauran d’escriure en valencià. (3) Els rètols de la via pública destinats a donar informació als transeünts hauran de ser redactats en valencià.



Spanish and Valencian in contact 

though Spanish is the dominant language in more than 80% of bilingual top-down texts if we look at the type of font utilized and the spatial organization on the ­different signs. Also, more than one third of top-down signage is monolingual in Spanish, despite local and regional regulations on language use that clearly s­ pecify that Valencian should be given preference over Spanish. Second, the presence of Spanish is the highest on bottom-up texts (74%). English is second (15%) and Valencian third (10%). The use of English and Valencian occurs mainly on bilingual texts, as their presence on bottom-up monolingual texts is almost non-existent at 2% each. Given such findings, the use of Valencian on top-down signage, although slightly higher than the use of Spanish, does not seem to reflect the language policies passed by local and regional administrations. This weak institutional enforcement seems to favor the predominance of Spanish, strengthening the already obvious presence of Spanish in the LL of Elche.

3.3  Location: A diglossic linguistic landscape? The LL of Elche was analyzed taking into consideration the location where each text was observed with the objective of determining if it was possible for the LL of a certain community to reflect the linguistic behavior of its speakers. Such an analysis was organized into two parts: the neighborhood in which the texts were observed and the type of street (i.e. main vs. secondary). As explained earlier, four main areas or neighborhoods were identified for the study: Carrús, Centro, PontNou/ElPla and Altabix. The neighborhood of Carrús is characterized as the area where most immigrants, foreign and national, have traditionally settled down. Altabix is the area in Elche that has seen the greatest growth in urban development, which is why many young families have chosen it as their current home. The area of PontNou/ElPla provided accommodation for the first waves of immigrants in the 1960s and 1970s, so we can find a mix of native Elche families and immigrant families. Finally, in the city center, the Centro area, we find a majority of native Elche families. Regarding the language use of people in Elche, Martínez Ibarra (2011) explains how people in PontNou/ElPla and Centro seem to speak Valencian more often. The lowest levels of use of Valencian were found in Carrús. Figure 5 helps us determine if that behavior is reflected at all on the LL. We can see the LL of Elche seems to reflect, to a certain extent, the linguistic behavior of its population since the highest presence of Valencian is identified in the neighborhoods of Centro and PontNou/ElPla. The lowest presence corresponds to Carrús, where the use of Valencian on public signage decreases to almost half of that observed in Centro. Apart from the fact that the city center is the area in town

 Francisco Martínez Ibarra 100

Spanish Valencian

80

English

%

60 40 20 0

12 Carrús

17

Altabix

23

Centro

18

PontNou/ElPla

Figure 5.  Language presence by neighborhood

where most governmental buildings are located, two other factors may motivate such results, particularly in regards to the type of signage created by a private author. Firstly, Centro is the most traditional area in Elche, where the highest percentage of elderly people live (Larrosa Rocamora, 2000). This type of population appears to also be the one that speaks Valencian most frequently (Martínez Ibarra, 2013) so the presence of Valencian in the LL could respond to the fact that either the author of the text speaks the language or the text is intended to be read by people who speak Valencian. Secondly, Centro is also the most touristic area in Elche and the use of the minority language, Valencian, may be motivated by an attempt to display ‘authenticity’ and attachment to local culture, as suggested in Kallen (2009) and Bruyèl-Olmedo and Juan-Garau (2009). Altabix (17%) and Centro (15%) show the strongest presence of English. In ­Altabix, the neighborhood is characterized by having the highest percentage of young people in Elche, a type of population that seems to highly value English as a language associated with economic development and a modern appeal, mostly in fields such as technology and clothing. In contrast, in Centro, an area with a high volume of tourism, English could be used as much for communication as it could to project an international and cosmopolitan sophistication, exploiting the prestige associated with English-speaking countries (cf. Griffin, 2004). In addition to examining the LL of Elche by neighborhood, we also analyzed it depending on the type of street where each sign was observed (i.e. main vs. secondary streets). The objective was to determine whether the situation of diglossia that characterizes the use of Valencian in the ACV and in Elche could be reflected somehow in the LL of the city. To that end, it was decided that if the presence of Valencian were to be higher in the LL of secondary streets, as compared to that of the main streets, we could say we have a diglossic LL. Figure 6 shows the findings for the city of Elche.



Spanish and Valencian in contact  100

Spanish Valencian

80

English

%

60 40 20 0

Main

Secondary

Figure 6.  Presence of Valencian depending on the type of street

Overall, it is not possible to establish any clear difference between the presence of Valencian and Spanish in the LL of main or secondary streets in Elche. Data show almost identical percentages and, in both contexts, Spanish is the dominant language and Valencian secondary. The presence of English is slightly higher in the LL of main streets. In evaluating bottom-up signage, although Valencian ranks third, both for main and secondary streets, we can appreciate a moderately higher percentage for the use of Valencian in the LL of secondary streets (11%) as opposed to main streets (9%). However, such findings are not strong enough to suggest the situation of diglossia present in Elche is reflected in the LL.

4.  Conclusions and discussion Data show the LL of Elche mainly consists of monolingual texts written in Spanish. The presence of Valencian on public signage seems to be limited to bilingual texts in certain neighborhoods such as PontNou/ElPla and Centro. Distinguishing between main and secondary signs allowed us determine that Valencian is very rarely used to communicate information. The use of Valencian seems to respond to symbolic motivations. English is, to a certain extent, another language with a significant presence in the LL of Elche, primarily due to being highly valued in the fields of clothing and technology. Finally, other languages such as Arabic and Chinese are mainly used in the neighborhood of Carrús, a neighborhood with significant foreign immigration. As a consequence, the LL of Elche seems to reflect, at first, the linguistic behavior of the population. Notably, Spanish, as an unmarked linguistic system, is the clear predominant language. Additionally, the presence of Valencian in the LL is higher in those neighborhoods where previous studies determined that the daily use of ­Valencian is most frequent.

 Francisco Martínez Ibarra

In regard to the implementation of language policies, the LL of Elche shows that both the local and regional governments do not seem to enforce existing regulations. Despite Valencian having the greatest presence in top-down texts, the number of monolingual texts in Spanish is still quite significant, accounting for more than a third of all top-down texts. Such findings are in line with Ladó’s (2011) conclusions for the city of Valencia, where monolingual Spanish texts accounted for 32% of all top-down texts. Because the analysis of linguistic landscapes is still a relatively new field in sociolinguistic research, more studies are needed to complement those already conducted. In Spain, such studies would improve our understanding of the similarities and differences among regions. For now, due to the scarce, but growing, number of studies on the LL of Spain, any generalization should be done cautiously. In language contact situations in which a certain language replaces another, as in the case of Spanish and Valencian, the vitality of the languages in contact is a fundamental factor to be considered for any process of linguistic normalization (Baldaquí, 2004). Studying the LL of a multilingual community provides information on written communication patterns among users. Also, it helps to depict ethnolinguistically such communities. In Elche’s case, a greater presence of Valencian in the LL could help reduce its seeming invisibility and increase its social and communicative values. The LL can help societal perceptions of a certain language by making its use seen as something normal within everyday life. As such, it appears necessary that all attempts at normalizing a minority language be accompanied by a rigorous linguistic analysis of public signage.

References Ajuntament d’Elx. (1997). Reglament de normalització lingüística per al municipi d’Elx i el seu ajuntament. Elche/Elx: Ajuntament d’Elx. Backhaus, P. (2007). Linguistic landscapes: A comparative study of urban multilingualism in Tokyo. Clevedon: Multilingual Matters. Bagna, C., & Barni, M. (2005). Dai dati statistici ai dati geolinguistici: Per una mappatura del nuovo plurilinguismo. Studi Italiani di Linguistica Teorica e Applicata, 34(2), 329–355. Baldaquí Escandell, J. M. (2004). La percepció de la vitalitat etnolingüística pels jóvens de l’Alacantí. Barcelona: Publicacions de L’Abadia de Montserrat. Baldaquí Escandell, J. M. (2005) A contribution to the study of Valencian linguistic secessionism: Relations between the perception of the supradialectal unity of the Catalan language and other sociolinguistic variables. Catalan Review: International Journal of Catalan Culture, 19(1–2), 47–58. Ben-Rafael, E., Shohamy, E., Amara, M. H., & Trumper-Hecht, N. (2006). Linguistic landscape as symbolic construction of the public space: The case of Israel. International Journal of Multilingualism, 3(1), 7–30.  doi: 10.1080/14790710608668383



Spanish and Valencian in contact 

Blas Arroyo, J. L. (1994). Valenciano y castellano. Actitudes lingüísticas en la sociedad valenciana. Hispania, 77(1), 143–155.  doi: 10.2307/344468 Blas Arroyo, J. L. (2005). Realidad sociolingüística y educación en la Comunidad Valenciana. In H. U. Cárdenas & T. F. å. Ulloa (Eds.), La educación plurilingüe en España y América (pp. 9–40). Madrid: Dykinson. Bruyèl-Olmedo, A., & Juan-Garau, M. (2009). English as a lingua franca in the linguistic landscape of the multilingual resort of S’Arenal in Mallorca. International Journal of Multilingualism, 6(4), 386–411.  doi: 10.1080/14790710903125010 Casesnoves Ferrer, R. (2001). Contexts d’utilització d’una llengua minoritzada en context minoritari. Documentos del Español Actual, 3(4), 25–37. Casesnoves Ferrer, R. (2005). L’evolució del coneixement del valencià 1991–2001: Una dècada de vanvis socials, polítics, demogràfics i lingüístics. Journal of Catalan Studies, 1(1), 32–51. Casesnoves Ferrer, R. (2010). Changing linguistic attitudes in Valencia: The effects of language planning measures. Journal of Sociolinguistics, 14(4), 477–500. doi: 10.1111/j.1467-9841.2010.00450.x Castillo Lluch, M., & Sáez Rivera, D. M. (2011). Introducción al paisaje lingüístico de Madrid. Lengua y Migración, 3(1), 73–88. Cenoz, J., & Gorter, D. (2006). Linguistic landscape and minority languages. In D. Gorter (Ed.), Linguistic landscape: A new approach to multilingualism (pp. 67–80). Clevedon [England]: Multilingual Matters. Cenoz, J., & Gorter, D. (2008). El estudio del paisaje lingüístico. Retrieved from: 〈http://www. euskara.euskadi.net/r59-bpeduki/es/contenidos/informacion/artik22_1_cenoz_08_03/ es_cenoz/artik22_1_cenoz_08_03.html〉 Ciscar Ramírez, L., González Martínez, D., & Pérez Ledo, P. (2002). Language attitudes and loyalties in the Valencian Country Noves SL. Revista de Sociolingüística, Primavera, 1–8. Climent Ferrando, V. (2005). The origins and evolution of language secessionism in Valencia. An analysis from the transition period until today. Working Papers 18. Retrieved from: 〈http://www.ciemen.org/mercator/Menu_nou/index.cfm?lg=ct〉 Dunlevy, D. (2012). Linguistic policy and linguistic choice: A study of the Galician linguistic landscape. In C. Hélot, M. Barni, R. Janssens, & C. Bagna (Eds.), Linguistic landscapes, multilingualism and social change (pp. 53–68). Frankfurt: Peter Lang. Fishman, J. A. (1967). Bilingualism with and without diglossia; diglossia with and without bilingualism. Journal of Social Issues, 23(2), 29–38.  doi: 10.1111/j.1540-4560.1967.tb00573.x Franco Rodríguez, J. M. (2008). El paisaje lingüístico del condado de Los Ángeles y del condado de Miami-Dade: Propuesta metodológica. Círculo de Lingüística Aplicada a la Comunicación, 35, 3–43. Gimeno Menéndez, F., & Gómez Molina, J. R. (2007). Spanish and Catalan in the Community of Valencia. International Journal of the Sociology of Language, 184, 95–107. Gorter, D. (Ed.). (2006). Linguistic landscape: A new approach to multilingualism. Clevedon: Multilingual Matters. Griffin, J. (2004). The presence of written English on the streets of Rome. English Today, 20(2), 3–7 & 47. Kallen, J. (2009). Tourism and representation in the Irish linguistic landscape. In E. Shohamy & D. Gorter (Eds.), Linguistic landscape: Expanding the scenery (pp. 270–283). London: Routledge.

 Francisco Martínez Ibarra Lado, B. (2011). Linguistic landscape as a reflection of the linguistic and ideological conflict in the Valencian Community. International Journal of Multilingualism, 8(2), 135–150. doi: 10.1080/14790718.2010.550296 Landry, R., & Bourhis, R. (1997). Linguistic landscape and ethnolinguistic vitality: An empirical study. Journal of Language and Social Psychology, 16(1), 23–49. doi: 10.1177/0261927X970161002 Larrosa Rocamora, J. A., & Universidad de Alicante (2000). Atlas demográfico y social de la ciudad de Elche. San Vicente del Raspeig: Universidad de Alicante. Leprêtre Alemany, M., & Romaní Olivé, J. M. (2000). L’ús de les llengües a la publicitat exterior a Barcelona i a sis altres ciutats l’any 1999. Llengua i Ús, 17, 55–59. Martínez Ibarra, F. (2011). Spanish-Valencian contact in Elche: Language use and attitudes (Unpublished doctoral dissertation). University at Buffalo – SUNY, Buffalo, NY. Martínez Ibarra, F. (2013). Why would I speak Valencian? Understanding language rejection in the southern territories of the Autonomous Community of Valencia. International Journal of the Linguistic Association of the Southwest, 32(1), 67–99. Mas i Miralles, A. (1994). La substitució lingüística del català. L’administració eclesiàstica d’Elx en l’edat moderna. Alcoi: Diputació d’Alacant. Miró Vives, C. (1998). Estudi sobre la llengua dels rètols a la ciutat de Tarragona. Llengua i Ús, 12, 54–58. Moltó, E. (2011, July 11). La alcaldesa del PP ‘renacionaliza’ Elche. El País. Retrieved from: 〈http://www.elpais.com〉 Montoya Abat, B. (1996). Alacant: La llengua interrompuda. Valencia: Denes. Montoya Abat, B., & Mas i Miralles, A. (2011a). La transmissió familiar del valencià. València Acadèmia Valenciana de la Llengua. Montoya Abat, B., & Mas i Miralles, A. (2011b). La situació social del català al País Valencià en el trànsit dels segles XX al XXI. Zeitschrift für Katalanistik, 24, 293–316. Nicolás Amorós, M. (2004). De la identitat del poder al poder de la identitat: algunes consideracions sobre la situació de la llengua catalana al País Valencià. Revista Catalana de Sociologia, 20, 63–83. Ninyoles, R. (1969). Conflicte lingüístic valencià. Valencia: Tres i Quatre. Ninyoles, R. (1971). Idioma i prejudici Palma de Mallorca: Editorial Moll. Noriega, J., & Blair, E. (2008). Advertising to bilinguals: Does the language of advertising influence the nature of thoughts? American Marketing Association, 72, 69–83. Piller, I. (2003). Advertising as a site of language contact. Annual Review of Applied Linguistics, 23, 170–183. Pons Rodríguez, L. (2011). Hispanoamérica en el paisaje lingüístico de Sevilla Itinerarios, 13, 97–127. Pradilla, M. A. (2005). La deriva estandarditzadora valenciana. Del secessionisme rupturista a l’aïllacionisme particularista. Zeitschrift für Katalanistik, 18, 141–170. Pradilla, M. À. (2008). La tribu valenciana. Reflexions sobre la desestructuració de la comunitat lingüistica. Benicarló: Onada Edicions. Regueira, X. L., López Docampo, M., & Wellings, M. (2013). El paisaje lingüístico en Galicia. Revista Internacional de Lingüística Iberoamericana, 1(21), 39–62. Rosenbaum, Y., Nadel, E., Cooper, R. L., & Fishman, J. A. (1977). English on Keren Kayemet Street. In J. A. Fishman, R. L. Cooper, & A. W. Conrad (Eds.), The spread of English (pp. 179–196). Rowley, MA: Newbury House. Ross, N. (1997). Signs of international English. English Today, 13(2), 29–33. doi: 10.1017/S0266078400009597



Spanish and Valencian in contact 

Sciriha, L., & Vassallo, M. (2001). Malta: A linguistic landscape. Malta: University of Malta. Scollon, R., & Scollon, S. W. (2003). Discourses in place: Language in the material world. London: Routledge.  doi: 10.4324/9780203422724 Segura i Llopes, J. C. (1996). Estudi lingüístic del parlar d’Alacant. Alicante: Institut de Cultura Juan Gil-Albert. Segura i Llopes, J. C. (2003). Una cruïlla lingüística. Caracterització del parlar del Baix Vinalopó. Alicante: Quinta Impresión. Shohamy, E., & Gorter, D. (Eds.). (2009). Linguistic landscape: Expanding the scenery. London: Routledge. SIES (2010). Encuesta de conocimiento y uso social del valenciano. Valencia: Consellería de ­Educación. Área de Política Lingüística. Sloboda, M., Szabó-Gilinger, E., Vigers, D., & Simicic, L. (2010). Carrying out a language policy change: Advocacy coalitions and the management of linguistic landscape. Current Issues in Language Planning, 11(2), 95–113.  doi: 10.1080/14664208.2010.505067 Solé Camardons, J. (1998). La llengua de la publicitat exterior a Barcelona. Llengua i Ús, 12, 49–53. Spolsky, B., & Cooper, R. L. (1991). The languages of Jerusalem. Oxford: Clarendon Press. Torres i Selva, C., & Montserrat i Buendia, S. (2003). La interrupció generacional del català a la ciutat d’Elx. In V. c. Martines (Ed.), Llengua, societat i ensenyament (Vol. III). Alacant: Symposia Philològica. Tulp, S. M. (1978). Reklame en tweetaligheid: Een onderzoek naar de geografische verspreiding van franstalige en nederlandstalige affiches in Brussel. Taal en Sociale Integratie, 1, 261–288. Vigers, D. (2013). Signs of absence: Language and memory in the linguistic landscape of ­Brittany. International Journal of the Sociology of Language, 223, 171–187. Wölck, W. (2004b). Universals of language maintenance, shift and change. Collegium ­Antropologicum, 28(1), 5–12.

part iii

Language Acquisition

Children’s Spanish subject pronoun expression A developmental change in tú?* Naomi Lapidus Shin

University of New Mexico This study explores children’s acquisition of structured morphosyntactic variation by examining Spanish subject pronoun expression. Analyses of 5,923 verbs produced by 154 Mexican children, ages 6 to 16, show that the variables that most strongly constrain the oldest children’s pronoun usage – Person, Reference, Priming – are acquired first during childhood. These variables exert similar effects across age, with the exception of second-person singular, which favors tú expression among younger children and tú omission among older children. The developmental trajectory from more to less tú expression is explained as the result of (a) increasing production of nonspecific reference, which in turn decreases rates of tú, and (b) abundant reported speech in the younger children’s data, which rendered tú expression pragmatically appropriate. Keywords:  Spanish subject pronouns; child language development; variation in child language; nonspecific reference; politeness

1.  Introduction Structured morphosyntactic variation is highly systematic among adults (e.g. Labov,  1994), yet we know very little about when and how patterned variation emerges during childhood. The current study investigates the development of variable morphosyntax by analyzing Spanish subject personal pronoun expression (e.g. tú ­bailas ~ ­bailas) in Shin’s Corpus of Mexican Children’s Spanish. The corpus consists of *  The author wishes to express gratitude to the children who participated in this research, as well as their families and the schools that made the project possible. Support was provided from the University of New Mexico and the University of Montana, and the research could not have been completed without assistance from: Dora LaCasse, Davin Poulin, Joshua Rodríguez, and Sandra Warren. Thanks also go to Danny Erker, Jackelyn Van Buren, three anonymous reviewers, and the editors of this volume for helpful feedback on this article. Any remaining errors are mine alone

doi 10.1075/ihll.8.06shi © 2016 John Benjamins Publishing Company

 Naomi Lapidus Shin

interviews with 154 children in two interior cities in Mexico, Oaxaca and Queretaro. The participants were divided into four age groups: 6/7-, 8/9-, 10/11-, 12+-year-olds. A total of 5,923 verbs that occurred either with or without a subject pronoun were analyzed with two goals in mind. One was to determine whether pronoun expression patterns found in previous studies of adult Spanish are also present in the developing grammars of ­Spanish-speaking children. The second goal was to explain an age-related difference found in second-person singular tú expression. Results from a mixed effects logistic regression performed in Rbrul (­Johnson, 2009) show constraints emerging in a predictable order: the strongest predictors of pronoun expression among older children – Person, Reference, and Priming – emerge earliest in childhood. The effects of these constraints were the same across age, with the exception of 2sg tú. Among the younger children 2sg verbs favored tú expression, but among 12+year-olds, 2sg verbs favored tú omission. Why would children express lower rates of tú as they grow older? This decrease in tú expression is surprising given the more general tendency to progress from fewer to more expressed pronouns with age (­Grinstead, 2010; Shin, 2015). Also, since the children were interviewed by an adult, we might expect them to omit specific tú to avoid sounding inappropriately intimate or dominant (Brown & Levinson, 1987). Two compatible explanations are suggested for the seemingly exceptional trajectory of tú. The first links cognitive development during childhood to increasing production of nonspecific 2sg reference, which favors tú omission. The second is that an abundance of reported speech contexts in the younger children’s data rendered tú expression pragmatically appropriate. That is, expressing tú may be too direct when addressing an adult interviewer, but it is not pragmatically problematic when addressing characters in a story. To summarize, the current article has two main goals. The first is to investigate how patterns of Spanish subject pronoun expression develop during childhood. The second is to explain the finding that tú expression appears to decline with age.

2.  Adult Spanish subject pronoun use: Rates and constraints Variationist research on Spanish subject pronouns has explored how often ­pronouns are expressed as well as the contexts that favor their expression. The proportion of verbs that appear with pronouns, often referred to as the pronoun rate, varies depending on dialect; Caribbean speakers tend to produce higher pronoun rates than M ­ ainland Latin Americans do (Otheguy & Zentella, 2012; Shin & Otheguy, 2013). Mexican adults’ pronoun rate tends to be around 20 percent (Lastra & Martín ­Butragueño, 2015; ­Michnowicz, 2015; Shin & Erker, 2015; Shin & Otheguy, 2013). While pronoun rates vary, linguistic predictors of pronoun use tend to be similar across varieties (Carvalho, Orozco, & Shin, 2015; Otheguy & Zentella, 2012). The



Children’s Spanish subject pronoun expression 

c­ urrent study focuses on six predictors briefly described here. (i) Person/­number: ­Singular pronouns are expressed significantly more often than plural pronouns. (ii)  Reference: Pronouns are expressed more often when switching reference than when maintaining it, as in (1) (1) Yo me enojé porque Ø le dije que si él creía que era eso… [920]1 ‘I got mad because (I) told him that if he believed that it was that…’ The second clause is a same-reference context, that is, the subject of porque le dije is the same as in the previous clause. Here the speaker omits yo ‘I’. In the third clause, reference is switched and él ‘he’ is expressed. (iii) Priming: Expressing a pronoun tends to ‘prime’ subsequent pronoun expression, resulting in a clustering of expressed pronouns, as in the repetition of yo ‘I’ in Yo no me sé palabras difíciles. S-, yo sé la-, yo sé las que yo me aprendo. [205] (‘I don’t know difficult words. I know the, I know those that I learn.’). (iv) Tense/mood/aspect (TMA): Forms with ambiguous person morphology, such as the imperfect, favor pronoun expression, while unambiguous forms, such as the preterit, disfavor it. (v) Clause: Main and dependent clauses favor pronoun expression, while coordinate clauses favor omission; (vi) Reflexive: Verbs without a reflexive pronoun, such as bailo ‘I dance’, favor pronoun expression more so than verbs with reflexive pronouns, such as me baño ‘I bathe myself ’. Longer descriptions of the six trends described above and an exhaustive list of studies illustrating these trends can be found in Carvalho et al. (2015). Not only do the same variables consistently impact adult subject pronoun expression, the hierarchy of these variables is similar across communities. Person/number, Priming, and Reference tend to be the strongest predictors, followed by TMA; other variables such as clause type and reflexive are ranked lower (e.g. Otheguy & Zentella, 2012; Torres Cacoullos & Travis, 2011).

3.  Previous research: Children’s Spanish subject pronoun use During the earliest stages of Spanish language production Spanish-speaking children rarely produce subjects (Grinstead, 2010; Villa García, 2013). Between two and five years old, pronoun expression increases, but is mostly restricted to the first person singular (Bel, 2003; Silva-Corvalán, 2014, p. 153). Given the scarcity of expressed ­pronouns in very young children’s speech, it would be difficult to conduct variationist-style multivariate analyses examining the contexts in which pronouns

.  All examples come from Shin’s Corpus of Mexican Children’s Spanish. The number at the end of each example refers to the participant identification number.

 Naomi Lapidus Shin

are expressed or omitted. To date, the few variationist studies of children’s subject pronouns have focused on school-age children (Bayley & Pease-Álvarez, 1997; Shin, 2015; Shin & Erker, 2015). Recently Shin (2015) and Shin and Erker (2015) found that six- to eight-year-olds expressed relatively few subject pronouns. Nevertheless, the most robust adult patterns, such as expressing more pronouns when switching reference, were also evident in the children’s data. Also, once a predictor variable significantly constrained usage, it generally exerted the same effect on children as it does among adults. A notable exception found in Shin (2015) was that second ­person singular (2sg) verbs favored pronoun expression among the younger children, but omission among the 12+­year-olds. This unexpected trajectory of 2sg tú was left unexplained. The current study builds on the previous analyses of Shin’s corpus by (a) analyzing subject pronoun expression by means of a mixed effects logistic regression run in Rbrul (Johnson, 2009), and (b) more closely examining expression of 2sg pronoun tú.

4.  Method 4.1  Participants & Interviews Data come from Shin’s Corpus of Mexican Children’s Spanish, which consists of interviews conducted with 154 children in Queretaro and Oaxaca, Mexico, all of whom were born in Mexico and reported speaking only Spanish in the home. The children, ages 6;4 to 16;4, were divided into four age groups: 6/7, 8/9, 10/11, and 12+.2 Sociolinguistic interviews were conducted in a quiet room in the children’s school. In addition, the children were prompted to retell familiar stories or to make up new ones, and to narrate a picture book in a no-shared knowledge context, i.e. the researcher told the children she was unfamiliar with the pictures, and she did not look at the book while the children narrated.

4.2  The envelope of variation Only contexts that permit variation between pronoun expression and omission were included in the current study. Consider example (2).

.  An anonymous reviewer asked why younger children were not included. Young Spanishspeaking children produce very few subject pronouns (e.g. Grinstead, 2010), which would make a variationist-style multivariate analysis challenging. Nevertheless, including younger children in future research would be worthwhile.



Children’s Spanish subject pronoun expression 

(2)  En Italia era una niña, y el fin de semana Ø se quedó a dormir en la casa de la abuela María. Y ella hurgaba en los…en los sillones y en los muebles. Y y Ø encontraba collares y guantes muy antiguas… [105] ‘In Italy there was a girl, and on the weekend (she) slept over in her grandmother Maria’s house. And she dug around in the … in the chairs and in the furniture. And and (she) found necklaces and very old gloves…’ Example (2) includes three variable contexts, ella hurgaba, se quedó a dormir, and encontraba. In all three cases expression or omission of ella is possible. Not all contexts are variable. For instance, expressed pronouns rarely occur in subject-headed relative clauses (Vi a la niña que *ella/ø estaba sentada allí ‘I saw the girl that *she/ø was sitting there) or with verbs that make reference to meteorological conditions (llueve, ‘it rains’). There are also contexts in which an expressed pronoun is required, as in Yo/*ø mismo hice la cena, ‘I myself made the dinner’, and yo qué sé/qué sé yo ‘what do I know’. These types of contexts in which there is little to no variation are excluded from the current study.3 Imperatives were also excluded because only 5/176 (3%) occurred with expressed subject pronouns. Also excluded were cases in which the intended referent was unidentifiable because (a) these cases could not be coded for relevant predictor variables, such as Reference, and (b) there was almost no variation, that is, there were almost no cases of expressed subject pronouns with unidentifiable referents. Another ­context in which pronouns are rarely expressed is third-person plural (3pl) nonspecific reference, as in (3). (3)  Una vez este mi tía, a su hijo se le- Ø le operaron porque tenía ­apendicitis. [908] ‘One time my aunt, (they) operated on her son because (he) had ­appendicitis.’ When the speaker says operaron ‘operated,’ he is not referring to a specific referent and, following the general trend in adult Spanish, ellos ‘they’ is omitted. There were 185 third person plural verbs with nonspecific subject reference; all occurred without a subject pronoun.4 Since there was no variation, all 185 cases of nonspecific 3pl verbs were excluded. The only remaining nonspecific referents were 2sg. After excluding contexts of little to no variation, the data extraction process yielded a total of 5,923 verb tokens.

.  For a more detailed discussion of the envelope of variation for Spanish subject pronoun expression, see Otheguy and Zentella (2012, pp. 45–67). .  The count is 209 if it includes cases that were excluded based on other criteria. After excluding ambiguous reference, imperatives, etc., there were 185 3pl verbs with nonspecific referents. Omission of ellos was categorical whether the count of nonspecific 3pl includes 209 or 185 cases.

 Naomi Lapidus Shin

4.3  Coding and statistical analyses Each verb token was coded for Person, Priming, Reference, TMA, Clause type, and Reflexive,5 described here: 1. Person/number included five factors: 1sg (yo, ‘I’), 2sg (tú), 3sg (él/ella ‘he/she’),6 1pl (nosotros, ‘we’), and 3pl (ellos/ellas, ‘they’). 2. Priming: Each verb included in the study was coded for priming only if there was a previous, coreferential subject that occurred (a) as either an expressed or omitted subject pronoun in a variable context, henceforth labeled Pronoun_ and Ø_, respectively, and (b) within the previous six clauses of the verb under study. The two contexts – previous mention = expressed or omitted subject pronoun – are henceforth labeled Pronoun_ and Ø_, respectively. Not all verbs could be coded for priming. Verb tokens were coded as ineligible, henceforth ‘other’, if (a) there was no previous mention of the referent or (b) there was a previous mention of the referent but it occurred as a lexical NP, in object position, or in any context in which there is little to no variation between expression and omission of pronouns. In summary, the priming variable included three factors: (i) previous mention = expressed pronoun (Pronoun_ ), (ii) previous mention = omitted pronoun (Ø_), or (iii) other. 3. Reference included two factors: same- and switch-reference. These contexts are operationalized here in terms of the relationship between two consecutive grammatical subjects.7 Co-reference across subjects is considered ‘same-reference,’ while a switch in reference is ‘switch-reference.’ 4. TMA included four factors: simple present indicative (e.g. bailo), preterit (e.g. bailé), imperfect (e.g. bailaba), and ‘other’.8 This last factor ‘other’ included future

.  The data were coded for other variables not included here, such as semantic class of the verb. The results reported in the current article are not altered by the inclusion of semantic class (Shin, 2015). .  Cases of usted, ustedes (both ‘you-formal’) and uno (‘one’) were excluded from the current study due to the scarcity of these forms in the children’s data. .  The restriction of Reference to consecutive clauses helps distinguish it from Priming. Only the previous clause is examined for Reference, while the previous six clauses are considered for Priming. Note, too, that Reference only compares coreference between subjects, which has been a common way to operationalize this variable. Some studies have examined coreference with a preceding object (e.g. Bayley & Pease-Álvarez, 1997; Michnowicz, 2015), but operationalizing Reference as a binary variable does not appear to affect its ranking in the adult variable hierarchy (compare Otheguy, Zentella, & Livert, 2010, p. 789 to Otheguy & Zentella, 2012, p. 160). .  Present indicative, preterit and imperfect verbs made up 91.3 percent of the entire data set. The heavy reliance on the three major tenses is not particular to child language. In the



Children’s Spanish subject pronoun expression 

(e.g. bailaré or voy a bailar), conditional (e.g. bailaría), subjunctives (e.g. baile, bailara), and perfect compounds (e.g. he bailado, había bailado, habré bailado). 5. Clause type included three factors: main, subordinate, and coordinate clauses. 6. Reflexive: included two factors: verb with a reflexive pronoun (e.g. me baño), verb without a reflexive pronoun (e.g. bailo). Mixed effects logistic regression analyses were performed in Rbrul (­Johnson, 2009). The model included presence/absence of the subject pronoun as a dependent variable; six fixed factors, Person/number of the verb, Priming, Reference, TMA, Clause type, and Reflexive; and two random factors: Speaker and Verb Lexeme. Reference*Clause was also included as an interaction term.

5.  Results: Pronoun rates Overall rates of pronoun expression for the Mexican children are presented in Table 1. For the sake of comparison, pronoun rates reported in studies of adult Mexican Spanish are also included in the table. Table 1.  Subject pronoun expression rates, Mexican children and adults Mexican Children-this study Age group

N speakers

N verbs

% pronoun

6/7

19

1551

9

8/9

47

1597

8

10/11

35

1230

10

12+

53

1545

12

N speakers

N verbs

% pronoun

Yucatan (Solomon, 1999)

12

1545

19

Yucatan (Michnowicz, 2015)

20

1940

20

NYC-1st generation (Shin & Erker, 2015)

19

8319

21

Mexico City (Lastra & Martín Butragueño, 2015)

18

2040

22

Chipilo, Mexico (Barnes, 2010)

16

1178

26

Mexican Adults Location (study)

Otheguy-Zentella corpus of adult Spanish in NYC, 87 percent of 63,492 finite verbs are in the present indicative, preterit and imperfect.

 Naomi Lapidus Shin

As shown in Table 1, Mexican adults express subject pronouns at a rate of about 20 percent.9 By comparison, the children produced very few pronouns.10 Even though pronouns are relatively scarce in the children’s data, it is still possible that the same linguistic factors previously found to constrain adult subject pronoun expression also constrain children’s pronoun use. To explore whether this is so, I now turn to multivariate analyses.

6.  Results: Linguistic constraints on pronoun use Table 2 indicates, for each age group, which variables were selected as significant predictors of pronoun expression, as indicated by asterisks under the column marked p. Table 2.  Variable hierarchies, Mexican children’s subject pronoun expression Age 6/7

Age 8/9

Age 10/11

Age 12+

Variables

p

Variables

p

Variables

p

Variables

p

Person

***

Person

***

Person

***

Person

***

Reference

***

Priming

***

Clause

***

Reference

***

Priming

***

TMA

*

TMA

*

Priming

***

[TMA]

Reference

*

Reference

**

Clause

**

[Clause]

[Clause]

Priming

*

Ref*Clause

***

[Ref*Clause]

[Ref*Clause]

[Ref*Clause]

TMA

*

[Reflex]

[Reflex]

[Reflex]

[Reflex]

Deviance

834.15

751.89

638.91

970.68

Intercept

-2.45

-2.10

-2.66

-2.87

Centered input ­probability

.08

.11

.07

.05

R2 fixed effects

.20

.17

.28

.29

R2 fixed & random

.30

.26

.44

.42

†

***p