New perspectives on hispanic contact linguistics in the Americans 9783954874729, 3954874725, 9788484898771, 8484898776

742 118 7MB

English Pages 461 [463] Year 2015

Report DMCA / Copyright


Polecaj historie

New perspectives on hispanic contact linguistics in the Americans
 9783954874729, 3954874725, 9788484898771, 8484898776

Citation preview

Sessarego, Sandro; González-Rivera, Melvin (eds.): New Perspectives on Hispanic Contact Linguistics in the Americas

Lengua y Sociedad en el Mundo Hispánico Language and Society in the Hispanic World Consejo editorial / Editorial Board: Julio Calvo Pérez (Universitat de València) Anna María Escobar (University of Illinois) Luis Fernando Lara (El Colegio de México) Francisco Moreno Fernández (Universidad de Alcalá) Juan Sánchez Méndez (Université de Neuchâtel) Armin Schwegler (University of California, Irvine) José del Valle (The Graduate Center, CUNY) Klaus Zimmermann (Universität Bremen) Vol. 35

Sessarego, Sandro; González-Rivera, Melvin (eds.):

New Perspectives on Hispanic Contact Linguistics in the Americas

Iberoamericana - Vervuert - 2015

© Iberoamericana, 2015 Amor de Dios, 1 – E-28014 Madrid Tel.: +34 91 429 35 22 Fax: +34 91 429 53 97 [email protected] © Vervuert, 2015 Elisabethenstr. 3-9 – D-60594 Frankfurt am Main Tel.: +49 69 597 46 17 Fax: +49 69 597 87 43 [email protected] ISBN 978-84-8489-877-1 (Iberoamericana) ISBN 978-3-95487-472-9 (Vervuert) E-ISBN 978-3-95487-831-4 Diseño de la cubierta: Carlos Zamora

This book is dedicated to Fabrizio, Marino, Jara and Magda.


of contents

Acknowledgments ............................... ................................


Introduction ...................................................................... Sandro Sessarego and Melvin González-Rivera


Section 1. Spanish in contact with Indigenous languages Maya-Spanish Contact in Yucatan, Mexico: Context and Sociolinguistic Implications ...................................................................... Jim Michnowicz


Rapanui Features in the Morphosyntactic System of Easter Island Spanish ........................................................................... Verónica González López


The Formal Guaraní and Spanish of Paraguayan Bilinguals ................. Shaw Nicholas Gynan, Ernesto Luís López Almada, Carlos Marino Lugo Bracho and María Eva Mansfeld de Agüero


Continuity and Innovation in Peruvian Spanish: Pragmatics and Contact in (Differential) Object Marking ............................................... Elisabeth Mayer and Manuel Delicado Cantero


Borrowed Clause Combining Patterns in Two Arawakan Languages Baure and Paunaka .............................................................. Swintha Danielsen and Lena Terhart


Section 2. Spanish in contact with coerced-migration languages Codeswitching and Borrowing in Aruban Papiamentu: The Blurring of Categories .... .................................................................... Yolanda Rivera and Patrick-André Mather


Nominal Ellipses in an Afro-Hispanic Language of Ecuador: The Choteño Case .................................................................... Sandro Sessarego and Javier Gutiérrez-Rexach


Cimarroneras in Venezuela: The Role of Isolated Communities in the Potential Development of a Spanish Creole .. ................................. Avizia Yim Long and Manuel Díaz-Campos


The Individual as the Locus of Variation and Change in a Contact Situation in Panama ............................................................. Delano S. Lamy


Section 3. Spanish in contact with free-migration languages Romance Language Contact in Mexico: The Case of Veneto-Spanish Bilingualism ..................................................................... Hilary Barnes Portuguese/Portuñol in Misiones, Argentina: Another “Fronterizo”? ....... John M. Lipski Preposition Stranding in a Non-Preposition Stranding Language: Contact or Language Change? ........................................................... Melvin González-Rivera, Ramón Padilla-Reyes, and John Rueda-Chaves Definite and Indefinite Articles in Nikkei Spanish ............................ Ana María Díaz-Collazos

235 253

283 301

Section 4. Latin American Spanish outside of Latin America Doing Being Boricua on the Island and in the U.S. Midwest: Perceptions of National Identity and Lateralization of /ɾ/ in Puerto Rican Spanish ...... Wilfredo Valentín-Márquez


Castilian in New York City: What Can We Learn from the Future? ......... Rafael Orozco


Language Attitudes and Linguistic Identities in Miami ...................... Diego Pascual y Cabo


Heritage Speakers’ Spanish in California: How Unbalanced Bilingualism Affects Reverse Construction of the gustar-type .............................. Viola Miglio and Stefan Th. Gries


Spanish and English in Contact in the Cyber World .......................... Antonio Medina-Rivera


Contributors ......................................................................


A cknowledgments We would like to thank the panel of scholars who reviewed the chapters and provided valuable feedback for the authors: Michelle Ramos-Pellicia, Chad Howe, Jacqueline Toribio, Kimberly Geeslin, David Korfhagen, Pilar Chamorro, Edith Beltrán, Juliana de La Mora, Marisa Carpenter, Scott Schwenter, Emily Kuder, Rey Romero, Brenda Stelter, Nate Maddux, Jason Doroga, Bruno Estigarribia, Diego Pascual, Nick Faraclas, Ian Tippets, Fernando Tejedo, Ksenija Bilbija, Javier Gutiérrez-Rexach, Nina Longinovic, Bill Cudlipp, Rajiv Rao, Diana Frantzen, Cathy Stafford, Grant Armstrong, Johannes Kabatek, Rafael Orozco, Jim Michnowicz, Ana Carvalho, Tim Gupton, Iván Ortega Santos, Frankie Larson, Chelsea Pfaff, Francoise Rose, Rand Valentine, Haralambos Symeonidis, Cecilia Montes Alcalá, Miquel Simonet, Marcos Rohena Madrazo, Yayoi T. Aird, and Antonio Ruiz Tinoco. We also owe a debt of gratitude to all the authors whose works appear in this volume. Very special thanks go to the University of Texas at Austin, the Universidad de Puerto Rico, Mayagüez, and the Centro de Investigaciones Lingüística del Caribe (CILC) for their support, which made this publication possible. Last but not least, we wish to thank Klaus Vervuert, Rebecca Aschenberg and the publishing team of Lengua y Sociedad en el Mundo Hispánico for their professionalism and help with the publication of this study.

INTRODUCTION Sandro Sessarego1 and Melvin González-Rivera2 1 University of Texas at Austin and 2Universidad de Puerto Rico - Mayagüez

The present volume is an edited collection of original contributions, all of which focus on Hispanic contact linguistics in the Americas. The project is composed of four main sections, organized according to the type of socio-historical scenario that characterizes the nature of the contact situation: (i) Spanish in contact with indigenous languages; (ii) Spanish in contact with coerced-migration languages; (iii) Spanish in contact with free-migration languages; and (iv) Spanish in contact with languages outside of Latin America, but still within the Americas. In so doing, the present project covers a variety of languages distributed across Northern, Southern, Central America, and the Caribbean. In Chapter 1, Jim Michnowicz provides an account of Yucatan Spanish (Mexico). The study describes certain lexical, phonological and morphosyntactic traits of this dialect, which at various times by various researchers have been directly or indirectly attributed to Maya influence (cf. Lope Blanch 1987). The present chapter seeks to present an overview of this contact situation in Yucatan, while addressing areas of possible or likely contact-induced change and the sociolinguistic factors surrounding the use of (perceived) indigenous language forms. In Chapter 2, Verónica González López focuses on the extended patterns of language shift and diglossia between Rapa Nui and Spanish on Easter Island. Spanish, the prestigious language, is employed in all types of public contexts and official events, while Rapa Nui is relegated to the domestic and private spheres. After collecting the data by means of sociolinguistic interviews, the author provides an analysis of a wide range of morphosyntactic phenomena. Findings appear to be mostly in line with those of other studies on SLA, creolization and bilingualism (e.g., Clements 2009; Escobar 2000; Lipski 2007; Otheguy and Stern 2010), thus suggesting that apart from a set of specific Rapa Nui-driven constructions, the rest of these grammatical elements have to be seen as the result of universal processes that are at work in all cases of language contact.


Sandro Sessarego and Melvin González Rivera

In Chapter 3, Shaw Nicholas Gynan, Ernesto Luís López Almada, Carlos Marino Lugo Bracho, and María Eva Mansfeld de Agüero take us to Paraguay and present data on the mutual influence of Spanish and Guaraní, which reflect the differential status of urban bilinguals and rural Guaraní-speaking monolinguals (cf. Gynan 2007). This analysis of 100 Guaraní-Spanish guided oral interviews reveals that the phonology of even formal Paraguayan Spanish is influenced by the Guaraní substrate, but there is almost no lexical or morphological influence from Guaraní. This appears to be due to the fact that, for most Paraguayans, borrowing from Guaraní while speaking Spanish carries a stigma, while using Spanish words and phrases when speaking Guaraní is generally acceptable. This unequal acceptance of loanwords and other contact phenomena favors the Spanish norm over the Guaraní norm and has generated heated linguistic ideological conflicts in the past, especially between the Ateneo de Lengua y Cultura Guaraní and the Paraguayan Ministry of Education. Resolving the issue of standardization of Guaraní is now the responsibility of the Academia de la Lengua Guaraní, recently created by the Secretaría de Políticas Lingüísticas. In Chapter 4, Elisabeth Mayer and Manuel Delicado Cantero analyze the evolution of differential object marking (DOM) on primary object marking in certain Peruvian Spanish contact varieties. In particular, they analyze the cases of extended DOM, that is, the extension of the prepositional accusative to topical inanimate objects. The authors argue that this change is regulated by pragmatic strategies, continuing the diachronically well-attested struggle between the dative and the accusative for primary object status in monotransitive clauses (Company 2003). This constitutes continuity as well as innovation of differential object marking. The paper highlights the contact avenues which arguably favored such changes (between Quechua and Spanish, and between Andean Spanish and Standard Peruvian Spanish), thus illustrating the role of contact as an integral mechanism of change. In Chapter 5, Swintha Danielsen and Lena Terhart provide an account of contact phenomena in Baure and Paunaka, two Bolivian Arawakan languages, both seriously endangered and currently being documented by the authors (cf. Danielsen 2007; Danielsen & Terhart 2014). Arawakan languages are generally very verby and many concepts are therefore expressed by predicates and sometimes by lexicalized verb-like constructions. Clause coordination and subordination are done by specifically marked verbal serial constructions and less often by particles that act as conjunctions. Contemporary Baure and Paunaka show numerous constructions that do not appear to be the result of internal language change; they are better analyzed as the result of a prolonged language contact with Spanish. In particular, the authors find several Spanish conjunctions and adverbs functioning as



conjunctions in Paunaka and Baure. Moreover, a few lexicalized verbal constructions in both languages appear to be applied like Spanish conjunctions. In this article, Danielsen and Terhart explore the distribution of these borrowed elements and try to explain the motivations for these types of contact-induced change. In Chapter 6, Yolanda Rivera and Patrick-André Mather present results from a study carried out in Aruba (The Netherlands Antilles). They offer an analysis of borrowing, code-switching, and phonological adaptation phenomena in Papiamentu, a Spanish-based creole that is going through a significant process of Hispanization due to more recent contact with Spanish dialects from the Caribbean. Results suggest that code-switching, in the case under inspection, involves much phonological interpenetration of languages, thus showing that there is no clear-cut distinction between single-word codeswitches and integrated borrowings (Muysken et al 1996; Poplack 2004; Bullock and Toribio 2009). In Chapter 7, Sandro Sessarego and Javier Gutiérrez-Rexach develop an analysis of nominal ellipses in Choteño Spanish, an Afro-Hispanic vernacular from Ecuador, and compare these grammatical phenomena to their respective counterparts in standard Spanish. The authors build on the literature dealing with Spanish N-drop (Brucart 1987; Kester and Sleeman 2002; Ticio 2003, 2005; etc.) to provide a unified account of nominal ellipses in these two dialects. In Chapter 8, Avizia Yim Long and Manuel Díaz-Campos examine the potential for Creole language development in colonial Venezuela. They provide demographic, sociohistorical, and linguistic data to cast light on the ongoing debate concerning the (non)creole origin of Afro-Caribbean Spanish (Álvarez and Obediente 1998; Díaz-Campos and Clements 2008; McWhorter 2000; Schwegler 1996; etc.). They suggest that the conditions for creole development in Venezuela may have been in place in cimarroneras and cumbes, marooned communities where fleeing slaves hid during colonial times. In Chapter 9, Delano S. Lamy studies the variability of voice onset time (VOT) in bilingual Creole English-Spanish speakers in Panama. Data are statistically analyzed through the incorporation of a mixed-effect linear regression in which the individual speaker is included as a random effect factor. The idea is that the individual speaker’s results represent stylistic variation in bilingual speech, which includes both Spanish-like and Creole English-like VOT duration (Lisker and Abramson 1964; Poplack and Meechan 1998). By gleaning information from the sociolinguistic interviews and language background questionnaires, it is observed that factors such as language attitudes, language loyalty and maintenance, and cultural identity affect VOT variability, thus shedding light on a variety of social factors that characterize this particular bilingual speech community.


Sandro Sessarego and Melvin González Rivera

In Chapter 10, Hilary Barnes examines the language contact situation observed in the small town of Chipilo, a Veneto-Spanish bilingual community in Mexico founded in 1882 by immigrants from Northern Italy (Barnes 2009). Data collected from sociolinguistic interviews and reading tasks are discussed, showing that while there are features of bilingual Chipilo Spanish that are common among other varieties of rural Spanish, several characteristics appear to be due to sustained contact with Veneto. In Chapter 11, John Lipski offers data on a group of Portuguese/“Portuñol”-speaking enclaves within the northeastern Argentine province of Misiones. These communities were founded in the first half of the twentieth century, within living memory, and sufficient sociodemographic information is available to allow for a more accurate diachronic representation. This analysis of the factors responsible for the presence and characteristics of Portuguese in Misiones, supplemented by recently collected data from border areas of Paraguay, Bolivia, Peru, and Colombia (cf. Kaufmann 2009; Lipski 2010, 2011, among others), sheds additional light on the formation of border speech communities. The data also expand the possibilities for studying the contact and alternation—both voluntary and involuntary—between two closely related languages in ways that transcend commonly observed constraints on intrasentential language switching. In Chapter 12, Melvin González-Rivera, Ramón Padilla-Reyes and John RuedaChaves examine preposition stranding under sluicing in Puerto Rican Spanish and argue that speakers of this Caribbean dialect tend to judge this construction acceptable, even though it is not allowed in standard Spanish (Bosque and Gutiérrez Rexach 2009; Campos 1991; Zagona 2002). They suggest that the grammatical judgment reflected by Puerto Rican Spanish speakers may be due to the contact situation between Spanish and English on the island. In Chapter 13, Ana María Díaz-Collazos analyzes the linguistic outcomes of Spanish in contact with Japanese in Colombia, specifically in the Nikkei community (Befu 2002), and provides a variationist analysis of Spanish articles in the spontaneous speech of these Japanese/Spanish bilingual speakers. Since the Japanese language lacks of a system of articles, this feature is problematic for this population. Her results show that articles are linked to certain noun types or verbal complements in all types of speakers. Bilinguals show different levels of lexicalization according to their specific language situation. In Chapter 14, Wilfredo Valentín-Márquez examines the sociolinguistic distribution of syllable-final (r) (Canfield 1981; Guitart 1978; Lipski 1994) in two Puerto Rican (PR) communities with different situations of language contact. He compares a community in which Puerto Rican Spanish (PRS) is the only language



spoken by most of the population (Cabo Rojo, Puerto Rico) with a community where PRS is a minority language (Grand Rapids, Michigan). Besides the contribution of linguistic context, life stage, and gender, the author explores whether the degree of integration into the PR community of the informants on the mainland offers explanatory insight into differences between the communities in terms of the variable’s distribution. Also, he considers the speakers’ perceptions of national identity—based on the meanings and uses of the word “boricua,” typically associated with core Puerto Ricanness—and he explores whether those judgments are related to the use of [l], the stigmatized variant of (r), in the two communities. In Chapter 15, Rafael Orozco explores Spanish-English contact in New York City through the prism of the expression of futurity among speakers of Colombian and Puerto Rican origin. The distribution of the variants of futurity (simple present indicative, morphological future, and periphrastic future) reveals that the periphrastic future is the most frequent form while the morphological future occurs the least. Futurity is conditioned by an intricate combination of internal and external constraints. Internal factors largely condition the variants of futurity similarly in both speaker groups. However, apparently due to contact with English, several constraints no longer affect the Puerto Rican cohort. The impact of linguistic contact is also reflected in younger speakers’ lack of use of the morphological future. Furthermore, interesting similarities in the effects of external constraints reflect Colombians’ assimilation to their new sociolinguistic landscape as they follow the Puerto Rican lead. Orozco’s findings help explain other instances of morphosyntactic variation leading to change, especially those involving analytic and synthetic variants (cf. Silva-Corvalán 1994). They also provide important information that helps compare the sociolinguistic forces constraining variation in New York City to those in other Hispanic speech communities. In Chapter 16, Diego Pascual y Cabo examines the perceived attitudes towards language and language use of three distinct groups of Cuban and Cuban-American young adults in Miami, Florida (cf. Lynch 2000; Alfaraz 2002). In this analysis, he presents combined data from semi-structured oral interviews and surveys that examine these speakers’ attitudes towards the Spanish and English languages, their language use, and the extent to which these languages’ social realities manifest in their everyday lives. The data presented show that these groups share many of the core aspects that form their social, cultural, and linguistic makeup, but present differences in terms of the linguistic values they assign to each language. These differences, coupled with the existence of additional complex divisions within this community, seem to suggest that language choice is employed to establish social boundaries which may point to a cultural and linguistic shift towards mainstream American monoculturalism and monolingualism.


Sandro Sessarego and Melvin González Rivera

In Chapter 17, Viola Miglio and Stefan Th. Gries focus on the Heritage Mexican Spanish speakers and L2 speakers from central and southern California. The authors study the recognition of different forms of the reverse construction with the Spanish verb gustar, ‘to like’. Using a questionnaire, they manipulate the grammaticality of gustar constructions. Results indicate that, on the whole, heritage speakers (HS) achieved a higher rate of correct judgments. However, the superior performance of HS is not found across the board; rather, it is part of significant interactions with other factors. In fact, in some cases (such as sentences with negation, or where the position of the syntactic subject is before the verb) HS perform on a par or worse than advanced learners, thus showing, in line with other studies, that “[…] HS’ initial linguistic advantages over L2 learners seem to diminish when both groups are compared at the high end of the proficiency spectrum” (De Prada Pérez and Pascual y Cabo 2011: 111). Finally, in Chapter 18, Antonio Medina-Rivera builds upon Crystal’s work on cyber language (2001/2006) to explore language contact outcomes between English and Spanish speakers outside of the conventionalized national borders. He navigates the cyber space to focus on speech communities using different varieties of Spanish and different levels of Spanish/English proficiency. He analyzes a wide range of language contact situations, including, among other phenomena, cases of language innovation as well as the correspondences between oral and written language. The current collection of articles is a contribution to the growing field of Hispanic contact linguistics. It consists of studies carried out by researchers with a solid and well-established academic profile in the field, as well as articles by younger academics applying the latest theoretical tools to the study of language-contact phenomena. We are honored to have had the opportunity of working with all of them and we hope that this volume will provide students and professors with a forward-looking perspective on Hispanic contact linguistics in the Americas. REFERENCES Alfaraz, G. (2002). “Miami Cuban Perceptions of Varieties of Spanish”. In D. Long, and D. Preston (eds.), Handbook of Perceptual Dialectology, Vol. 2. Philadelphia: John Benjamins, pp. 1-12. Álvarez, A./Obediente, E. (1998). “El Español Caribeño: Antecedentes Sociohistóricos y Lingüísticos”. In P. Matthias, and A. Schwegler (eds.), América Negra: Panorámica Actual de los Estudios Lingüísticos sobre Variedades Hispanas, Portuguesas y Criollas. Madrid/Frankfurt: Iberoamericana/Vervuert, pp. 40-61. Barnes, H. (2009). A Sociolinguistic Study of Sustained Veneto-Spanish Bilingualism in Chipilo, Mexico. Doctoral dissertation, The Pennsylvania State University.



Befu, H. (2002). “Globalization as Human Dispersal: Nikkei in the World”. In L. R. Hirabayashi, A. Kikumura-Yano, and J. A. Hirabayashi (eds.), New Worlds, New Lives: Globalization and People of Japanese Descent in the Americas and from Latin America. Stanford: Stanford University Press, pp. 1-18. Bosque, I./Gutiérrez-Rexach, J. (2009). Fundamentos de Sintaxis Formal. Madrid: Akal. Brucart, J. M. (1987). La Elisión Sintática en Español. Doctoral dissertation. Universitat Autònoma de Barcelona. Bullock, B./Toribio, J. (2009). “Trying to Hit a Moving Target: On the Sociophonetics of Codeswitching”. In L. Isurin, D. Winford, and K. de Bot (eds.), Multidisciplinary Approaches to Codeswitching, pp. 189-206. Campos, H. (1991). “Preposition Stranding in Spanish?”. In Linguistic Inquiry 22, 4, pp. 741-750. Canfield, D. L. (1981). Spanish Pronunciation in the Americas. Chicago: The University of Chicago Press. Clements, J. C. (2009). The Linguistic Legacy of Spanish and Portuguese. Cambridge: Cambridge University Press. Company, C. (2003). “Transitivity and Grammaticalization of Object. The Struggle of Direct and Indirect Object in Spanish”. In G. Fiorentino (ed.), Romance Objects. Transitivity in Romance Languages. Berlin: Mouton de Gruyter, pp. 217-260. Crystal, D. (2001/2006). Language and the Internet. Cambridge: Cambridge University Press. Danielsen, S. (2007). Baure: An Arawak Language of Bolivia. Indigenous Languages of Latin America (ILLA) 6. Leiden: CNWS. Danielsen, S./Terhart, L. (2014). “Paunaka.” In: M. Crevels, and P. Muysken (eds.), Las lenguas de Bolivia, Vol. 3. La Paz: Plural Editors, 221-258. Díaz-Campos, M./Clements, C. J. (2008). “A Creole Origin for Barlovento Spanish? A Linguistic and Sociohistorical Inquiry.” In Language in Society, 37, 3, pp. 351-383. Escobar, A. M. (2000). Contacto Social y Lingüístico: El Español en Contacto con el Quechua en el Perú. Lima: Pontificia Universidad Católica del Perú. Guitart, J. M. (1978). “A Propósito del Español de Cuba y Puerto Rico: Hacia un Modelo no Sociolingüístico de lo Sociodialectal.” In H. López Morales (ed.), Corrientes Actuales en la Dialectología del Caribe Hispánico. San Juan, PR: Editorial Universitaria de la Universidad de Puerto Rico, pp. 77-92. Gynan, S. N. (2007). “The Language Situation in Paraguay: An Update”. In R. B. Baldauf, and B. Kaplan (eds.), Language Planning and Policy in Latin America, Vol. 1: Ecuador, Mexico and Paraguay. Multilingual Matters: Clevedon, England, pp. 284-301. Kaufmann, G. (2009). “Falar Espanhol or Hablar Portugués: Attitudes and Linguistic Behavior on the Brazilian-Uruguayan and Brazilian-Argentinian Borders”. In Romanistisches Jahrbuch 60, pp. 276-317. Kester, E. P./Sleeman, P. (2002). “N-Ellipsis in Spanish”. In Linguistics in the Netherlands 2002, 107-116. Amsterdam: John Benjamins. Lipski, J. (1994). Latin American Spanish. London: Longman. Lipski, J. (2007). “El Español de América en Contacto con Otras Lenguas”. In M. Lacorte (ed.), Lingüística Aplicada del Español. Madrid: Arco Libros, pp. 309-345.


Sandro Sessarego and Melvin González Rivera

Lipski, J. (2010). “Spanish and Portuguese in Contact”. In R. Hickey (ed.), Handbook of Language Contact. New York: Wiley-Blackwell, pp. 550-580. Lipski, J. (2011). “Encontros Lingüísticos Fronteiriços”. In Ideação 13, 2, pp. 83-100. Lisker, L./Abramson, A. (1964). “A Cross-Language Study of Voicing in Initial Stops: Acoustical Measurements”. In R. J. Baken, and R. G. Daniloff (eds.), Readings in the Clinical Spectrography of Speech. San Diego, CA: Singular Publishing Group, Inc, pp. 384-422. Lope Blanch, J. M. (1987). Estudios sobre el Español de Yucatán. México City: Universidad Nacional Autónoma de México. Lynch, A. (2000). “Spanish-Speaking Miami in Sociolinguistic Perspective: Bilingualism, Recontact, and Language Maintenance among Cuban-Origin Population”. In A. Roca (ed.), Research on Spanish in the United States. Somerville, MA: Cascadilla, pp. 271-283. McWhorter, J. (2000). The Missing Spanish Creoles: Recovering the Birth of Plantation Contact Languages. Berkeley, CA: University of California Press. Muysken, P./Kook, H./Vedder, P. (1996). “Papiamento/Dutch Codeswitching in Bilingual Parent–Child Reading”. In Applied Psycholinguistics 17, 4, pp. 485-505. Otheguy, R./Stern, N. (2010). “On So-Called Spanglish”. In International Journal of Bilingualism 15, 1, pp. 85-100. Poplack, S. (2004). “Code-Switching”. In U. Ammon, N. Dittmar, K.J. Mattheier, and P. Trudgill (eds.), Sociolinguistics. An International Handbook of the Science of Language, second edition, Berlin: Walter de Gruyter, pp. 589-596. Poplack, S./Meechan, M. (1998). “How Languages Fit Together in Code-Mixing”. In International Journal of Bilingualism 2, pp.127-138. Schwegler, A. (1996). “La Doble Negación Dominicana y la Génesis del Español Caribeño”. In Hispanic Linguistics 8, pp. 247-315. Silva-Corvalán, C. (1994). Language Contact and change: Spanish in Los Angeles. New York City: Oxford University Press. Ticio, M. E. (2003). On the Structure of DPs. Doctoral dissertation, UConn, Storrs. Ticio, M. E. (2005). “NP-Ellipsis in Spanish”. In Proceedings of the 7th Hispanic Linguistics Symposium 7, pp. 128-141. Zagona, K. (2002). The Syntax of Spanish. Cambridge: Cambridge University Press.



1. Introduction Throughout the southern Mexican states of Yucatan, Campeche and Quintana Roo, as well as in northern Belize, Spanish is in contact with an indigenous language, Yucatec Maya (Lewis 2009). Yucatec Maya is part of a larger family of Maya languages that stretches from the Central American countries of Honduras and Guatemala in the south, to Tabasco on the Gulf of Mexico in the north, and is the second largest Maya language, after K’iche’, spoken in the highland region of Guatemala (Lewis 2009). As of 2005 (the last date for which these data are available), the Mexican census reported 752,316 Yucatec Maya speakers across the Yucatan peninsula, with a majority in the state of Yucatan (527,107), and lesser, but still substantial amounts, in Campeche (69,249) and Quintana Roo (155,960) (INEGI 2005). This chapter will focus on the largest group of Yucatec Maya1 speakers in Yucatan State. First, I will contextualize the current situation of language contact by examining the historical background and demographics of Maya and Spanish speakers in Yucatan. I will then present the possible linguistic consequences of Maya-Spanish contact, synthesizing data from studies on phonetics/phonology, morpho-syntax and the lexicon. Finally, I will address the future of Maya-Spanish contact in Yucatan, outlining both efforts to revitalize the Maya language and the overall standardization of (possibly) Maya-influenced Spanish. 2. Historical background and context According to historical records, Maya and Spanish first came into contact around 1511, when survivors of a Spanish shipwreck washed ashore in the southern Yucatan peninsula. Several short expeditions were made in the following decades, until 1

Since “Maya” is the term used by native speakers of the language and for simplicity, this term will be used throughout the chapter.


Jim Michnowicz

Figure 1. Map showing the distribution of Maya languages. Font size reflects the number of speakers. Modified under the GNU Free Documentation License from source:

1527, when Francisco de Montejo, known as “El Adelantado”, landed at Cozumel to undertake the conquest of Yucatan for the Spanish Crown. This first attempt at conquest failed, and one year later the Spanish abandoned Yucatan to regroup. Montejo, accompanied by his son (also Francisco de Montejo, known as “El Mozo”), returned around 1530, and again was forced to temporarily suspend military actions due to Maya resistance, the climate, and lack of supplies (Quezada 2001: 34). It was not until a decade later, in 1542, that Montejo the Younger founded the city of Merida on the site of the Mayan city Tiho. The following year, in 1543, yet another Francisco de Montejo (“The Nephew”) founded Valladolid in eastern Yucatan. Still, the conquest would not be considered complete until 1687, 160 years after it began (see Quezada 2001 for a more detailed overview of the conquest).

Maya-Spanish Contact in Yucatan, Mexico


The result of this extended conquest and the difficulties the Spanish faced in colonizing Yucatan was that Spanish speakers were by and large isolated in the main Spanish cities of Campeche, Merida and Valladolid, with the countryside dominated by Maya speakers (Mosely 1980: 86; Lipski 2004: 99). Generally speaking, Yucatan (which in the colonial period referred to the modern states of Campeche, Yucatan and Quintana Roo) received less outside immigration than the rest of Mexico, with lower numbers of colonists from Europe and slaves from Africa than many other areas (Weber 1980: 173), and Spanish speakers comprised a very small percentage of the total population. In 1580, for example, the population of Yucatan was comprised of approximately 2000 Spaniards and 300 Africans, compared with 200,000 Mayans. In 1700, the Spanish population had risen to 20,000 people, while the Mayan population remained much higher, at 182,000. Later in that century, the Spanish population rose dramatically to 103,000, while the Mayan population also rebounded to its highest level since the conquest (254,000 people) (Mosely 1980: 102-104). Even well into the 20th century, Maya speakers made up almost half of the population of Yucatan State, as seen in Figure 2. The isolation that separated Maya speakers in rural areas and Spanish speakers in urban areas during the colonial period has diminished over the past century, and the demographic situation in present day Yucatan is such that Spanish speakers are in constant contact with both Maya and Maya-influenced Spanish (Lipski 2004: 99). This became especially true in the latter half of the 20th century, as increasing numbers of Maya speakers have moved to cities looking for work as manual laborers, vendors, domestic help, and as nannies for Spanish-speaking families (Michnowicz 2009, 2011; Lipski 2004: 99). Thus contact between Maya and Spanish is of two different types: the first-hand contact that occurs among Maya-Spanish bilinguals, and the second-hand contact that brings Spanish monolinguals into contact with a Maya influenced, L2 variety of Spanish.

Figure 2. Maya speakers as percentage of total population in Yucatan State. Source: INEGI.


Jim Michnowicz

Importantly, while Maya speakers have slowly abandoned their native language in favor of Spanish over the past centuries (Pfeiler and Zámišová 2006), the Maya language enjoys a level of prestige uncommon among indigenous languages in Latin America. Lope Blanch (1987: 9) observes that the prestige of Maya may be due to several factors, including its status as an adstrate, rather than substrate, language; the relatively large percentage of the population that still speaks Maya, including city dwellers; and the existence of a single indigenous language throughout the territory, in comparison to other areas with many small, mutually unintelligible languages. Additionally, the use of Maya phrases has a certain popularity among some, particularly young, Spanish speakers (see Michnowicz 2008: 298-299; Kolmer 2006). These factors have led numerous researchers to attribute a wide-variety of linguistic features of Yucatan Spanish to Maya language influence, with Lope Blanch (1987: 8-9) stating that while in many contact varieties features are erroneously attributed to the indigenous language, “[t]his is not the case with Yucatan Spanish. In it, the influence of the Maya language is patently clear, and cannot be argued” (my translation). Likewise, Klee (2009) and Lipski (2004) identify Yucatan as one of the three regions most likely to demonstrate indigenous influence, along with Paraguay and the Andean region. 3. Linguistic consequences of Maya-Spanish contact Before beginning the discussion of possible areas of Maya influence on Yucatan Spanish, it is important to distinguish between L2-interference features on the one hand, that for the most part only occur among the bilingual, Maya-dominant population, and on the other hand, possibly Maya-influenced features that have also permeated the Spanish of monolinguals, likely through processes of large-scale shift of Maya-speaking populations to Spanish over the last century or more (see Thomason and Kaufman 1988; Winford 2003 for an overview of these processes). This distinction is also important in other areas of indigenous language contact, such as the Andes (Escobar 2011: 328). Although there are some exceptions, generally speaking, phonetic/phonological variants are more likely to have been passed to the monolingual Spanish population than morphosyntactic variants, which are primarily restricted to bilingual speech (see Escobar 2011 for a comparison with Quechua influenced Spanish in the Andes, where similar patterns exist). 3.1 Phonetics/Phonology The phonetics/phonology of Yucatan Spanish, frequently attributed to Maya contact, represent the most studied area of the dialect. Lope Blanch (1987: 34) provides an overview of phenomena that have been attributed, rightly or wrongly, to contact with Maya. These include the substitution of [p] for /f/, as in [ˈpe.ʧas] for fechas

Maya-Spanish Contact in Yucatan, Mexico


(Michnowicz 2012; Suárez 1979); the elision of /x/, as in [tra.ˈba.o] for trabajo (Michnowicz 2012; Barrera Vásquez 1937); and the depalatalization of /ɲ/, as in [ˈni.njo] for niño (Barrera Vásquez 1937; Yager 1982). Most of these features, however, have not been systematically analyzed in the literature. Therefore, the discussion below will focus on the main phonetic variables analyzed in previous work. 3.1.1 Final -m One of the most studied features of Yucatan Spanish is the possibility of labializing word final nasals in absolute final position (i.e. before a pause), where most other varieties allow only [n] or [ŋ], for example pan [ˈpam] or camión [ka.ˈmjom] (Michnowicz 2006, 2007, 2008; Alvar 1969; Cassano 1977; Yager 1982, 1989; García Fajardo 1984; Lope Blanch 1987, 1990; Pfeiler 1992). Michnowicz (2008: 289) found 25% [m] in absolute final position, although that frequency may be much higher for particular lexical items, such as place names (Michnowicz 2006). Importantly, this same alternation appears in Maya as well, for example hun “one” can be realized as [hum] (Bolles and Bolles 2001). Final -m appears to be a relatively recent innovation, not appearing in early studies of the dialect (such as Barrera Vásquez 1937, Nykl 1938, Mediz Bolio 1951, but see Ramos i Duarte’s (1895: 386) brief comment on pam), and is increasing in use among younger speakers of Yucatan Spanish (Yager 1989, Michnowicz 2007, 2008). Final - m is also more frequent among women than men (Yager 1989, Pfeiler 1992, Michnowicz 2007, 2008), and appears to be a linguistic marker of Yucatan identity. While most studies have attributed -m to Maya contact (e.g. Alvar 1969, Yager 1982, Lope Blanch 1987), the presence of -m in other dialects (i.e. in particular parts of Colombia; Lipski 2004; Canfield 1981) leads Cassano (1977) to outright reject any influence from Maya. More recent quantitative studies disagree with respect to direct Maya influence. Yager (1989) found no significant difference between language groups, while Michnowicz (2007, 2008) showed that Maya speakers produced significantly more -m than did Spanish monolinguals (32% -m vs. 19%, respectively; Michnowicz 2008: 292). The most likely scenario is probably that outlined by Lope Blanch (1987: 62-63), who argued that -m in Yucatan is the result of an internal process in Spanish that has been favored by Maya contact, arising first among bilingual speakers and then passing to monolinguals through contact. This feature has also been attested in varieties in contact with Guaraní in Paraguay, although it is a declining form in that country (Granda 1982; see Klee and Lynch 2009 for a summary). 3.1.2 /bdg/ Another feature of Yucatan Spanish frequently attributed to Maya contact is the occlusive realization of /bdg/ in positions that would normally favor approximants, including intervocalically, for example todo [ˈ] or yo vivo [ˈʝo.ˈ]


Jim Michnowicz

(Michnowicz 2009, 2011, 2012; Alvar 1969; Cassano 1977; Yager 1982; García Fajardo 1984; Lope Blanch 1987, 1990). Previous reports on /bdg/ are inconsistent, ranging from Alvar (1969: 165) who noted almost categorical occlusives, to later studies such as Yager (1982: 58) who reported high rates of stops, but also a slight preference for approximants. Complicating the issue, some researchers also report occasional relaxed articulations (Suárez 1979, García Fajardo 1984). Most recently, Michnowicz (2011: 198) found 42% [b], 32% [d] and 28% [g] in spontaneous speech. Regarding Maya influence, most studies propose a combination of factors, including Spanish-internal processes and possible indirect influence from Maya (Cassano 1977, Yager 1982). Michnowicz (2009) argued that stop [bdg], common in L2 varieties of Spanish around the world, is a case of a fossilized L2 feature that, like -m, has spread from bilinguals to the monolingual populace, through processes of large-scale shift (Thomason and Kaufman 1988). Michnowicz (2011: 201-202) found that fluent Maya-speakers produced significantly more stops than Spanishmonolinguals, by an average of 15%. Likewise, speakers over age 30, exposed to more Maya and Maya-influenced Spanish than younger speakers, also produced significantly more stops than younger speakers (Michnowicz 2011: 202). 3.1.3 /ptk/ In both bilingual and monolingual Yucatan Spanish, the voiceless stops /ptk/ are often aspirated to levels not ordinarily found in Spanish, for example pan [ˈphan] or tomas [ˈtho.ˈmas] (Michnowicz 2012; Michnowicz and Carpenter 2013; Nykl 1938; Suárez 1979; Alvar 1969; Yager 1982; García Fajardo 1984; Lope Blanch 1987, 1990). Many studies, noting the existence of both aspirated and ejective stops in Maya, argue for direct Maya influence (Nykl 1938, Yager 1982), while others suggest that, like /bdg/, aspirated /ptk/ are likely due to indirect influence (Coupal and Plante 1977; Lope Blanch 1987). Studies also report differences in the degree of aspiration in Yucatan Spanish. Suárez (1945) stated that aspiration is one of the characteristic features of the dialect, while Alvar (1969) reported that while aspiration is greater than that typical of Spanish, it does not reach levels of aspiration in English2. Coupal and Plante (1977: 150), one of very few instrumental studies of Yucatan /ptk/, noted that aspiration (VOT) is not as long as some would suggest, but that it is greater than that found in the Caribbean or Spain. More recently, Michnowicz and Carpenter (2013) found mean VOTs of 17ms, 22ms and 34ms for /ptk/ respectively, somewhat longer than many other varieties of Spanish, but not as aspirated as some early studies would suggest. In agreement with Lope Blanch (1987), who reports a wide range of individual difference


Lisker and Abramson (1964: 394) report mean values of 58ms for /p/, 70ms for /t/ and /80/ ms for /k/ in English.

Maya-Spanish Contact in Yucatan, Mexico


for aspiration, Michnowicz and Carpenter (2013) found VOT ranges much larger than other dialects; between 75ms and 122ms for Yucatan, compared with ranges between 21ms and 43ms for Castilian Spanish (Rosner et al. 2000). Thus while the overall mean VOT value is only slightly greater than in other dialects, the wide range of values indicates that at times /ptk/ is aspirated to a great degree, but not consistently across tokens (see also Yager 1982). Younger speakers generally have shorter (more like standard Spanish) VOTs than older speakers (Michnowicz and Carpenter 2013). Regarding direct Maya influence, Michnowicz and Carpenter (2013) found no consistent pattern for VOT measurements by language, and language background by itself was not a significant predictor. Language did interact significantly with age, however. Maya-speakers showed no significant differences across age group, whereas younger Spanish-speakers produced significantly shorter VOTs than older speakers. Based on this and other evidence, the authors do not discard the possibility of Maya influence, and note that such influence may have occurred in the past. If Maya has played a role in aspirated /ptk/, then the process of transfer to monolingual speakers may be further along than for /bdg/, thereby obscuring the effect for bilingualism that may have existed at an earlier period. 3.1.4 Glottalizations and /ʔ/ insertion The literature on Yucatan Spanish distinguishes between glottalized (or ejective) consonants, and the insertion of a glottal stop /ʔ/ in a variety of contexts, but most frequently between two vowels of different quality, for example me iba [me.ˈʔi. ba] or doce años [ˈˈʔa.ɲos] (Lope Blanch 1987: 115). Stress also plays a role, with stressed vowels favoring /ʔ/ insertion (Yager 1982: 88). Truly ejective consonants, common in Maya, rarely surface in Yucatan Spanish (Lope Blanch 1987: 103), with the exception of names of flora and fauna of Maya origin, and even then only among bilingual speakers. The insertion of /ʔ/, however, is more common, and is the focus of the little research done to date. Most researchers have concluded that his phenomenon is one of the strongest candidates for direct Maya influence on Yucatan Spanish (Barrera Vásquez 1937; Nykl 1938; Suárez 1979; Yager 1982; García Fajardo 1984; Lope Blanch 1987, 1990), based on the fact that a) /ʔ/ insertion is a process lacking in most varieties of Spanish; b) Maya does not permit vowel initial words, and therefore c) vowel-initial borrowings from Spanish are transferred to Maya with an epenthetic /ʔ/;for example amigo is borrowed as [ʔáamiɡóoh] (Frasier 2009:23). Although detailed sociolinguistic analyses of /ʔ/ do not yet exist, some preliminary details are available. Lope Blanch (1987: 106) found /ʔ/ insertion “with notable regularity in the Spanish of a good number of Yucatecans” (my translation), and García Fajardo (1984: 86-86) found that /ʔ/ occurs most frequently among lower class speakers. Finally, Michnowicz (2012), in case studies examining two families, found /ʔ/ in vowel initial words an average


Jim Michnowicz

of 19% for Maya speakers, compared with 7% for Spanish speakers, a significant difference. Likewise, older speakers produced significantly more /ʔ/ than younger speakers, with the youngest speakers in both the Maya speaking and the Spanish monolingual families producing almost no cases of /ʔ/. Michnowicz (2012) suggests that this variant may disappear as Maya speakers continue to shift to Spanish in ever greater numbers (see section 4 for further discussion). A similar pattern of /ʔ/ insertion has been reported for Guaraní-Spanish bilinguals in Paraguay and parts of Argentina, a feature also attributed to phonological patterns in the indigenous language (Granda 1982). 3.1.5 /ʝ/ Yucatan Spanish, as is common across Spanish varieties, permits a wide variety of realizations of /ʝ/. Using Solomon’s (1999) terms, these realizations range from “weak” articulations, such as deletion or approximant [j], to “strong” allophones, such as [ʒ] or [ʤ]. One of the most frequently commented forms is the elision of /ʝ/ intervocalically, most frequently in contact with /i/, as in anillo [a.ˈni.o] or gallina [ga.ˈ] (see Alvar 1969, García Fajardo 1984). Suárez (1979) and García Fajardo (1984) both indicated that deletion of /ʝ/ is most common among older, less educated speakers, and Maya bilinguals. Solomon (1999) compared weak forms (including deletion) to strong forms (primarily [ʤ], among others) for speakers in Valladolid, Yucatan. Her findings, consistent with earlier reports, showed a majority of weak variants (96%), although most of these represent a weak approximate [j] rather than true deletion (1999: 157). Weak variants were significantly favored by male gender, lower social class, and older age (1999: 171). Regarding possible Maya influence, Solomon (1999: 192) notes that in her recordings of Maya, speakers consistently employed a weak /j/ in that language, suggesting that “fluent Maya speakers might therefore be more likely to use [j] in their Spanish”. Likewise, many of her informants attributed this pronunciation in Yucatan Spanish to Maya influence (1999: 192). The author points out, however, that language background and other social factors, such as age and social class, strongly overlap, and thus it is difficult to disentangle the possible effects of Maya influence from that of other factors (1999: 193). Solomon (1999: 194) closed by stating that, given that weakened /ʝ/ is common in other areas of Mexico and Central America, Maya contact may have supported the use of weak variants, but “...there is no convincing evidence that Maya [/ʝ/] has had a central influence” on /ʝ/ in Yucatan Spanish. 3.1.6 Intonation Intonation is one of the most frequently commented, if least studied, possible contact features in Yucatan Spanish. Speakers of the dialect are conscious of differing intonational patterns that separate Yucatan Spanish from other varieties, and use

Maya-Spanish Contact in Yucatan, Mexico


terms such as pujado “pushed” to describe their accent. Yucatan Spanish intonation is described in the literature as “peculiar” (Barrera Vásquez 1945/1977: 341) and “slow and halting” (Suárez 1979: 77; my translations). Early researchers were also quick to attribute intonational differences to Maya contact (Nykl 1938; Barrera Vásquez 1937, 1945/1977; Mediz Bolio 1951; Suárez 1979). Mediz Bolio (1951: 19) observed that Yucatan Spanish intonation “is nothing else but a consequence of the original Maya accent” (my translation), also noting that this Maya-influenced pattern has permeated the monolingual Spanish-speaking population. Suárez (1979: 77) called Yucatan Spanish intonation “reflections of Maya phonetics” (my translation). Others have been more reserved, such as Lope Blanch (1987: 39), who argued that not enough is known about either Maya or Yucatan Spanish intonation to establish such a connection. Unfortunately, knowledge about Maya phrasal intonation is still limited, as most research has focused on the patterning of lexical tones (see Gussenhoven and Teeuw 2008, among others). An initial attempt at exploring Yucatan Spanish intonation in spontaneous speech is found in Michnowicz and Barnes (2013). They found that spontaneous Yucatan Spanish does display a higher rate of early F0 peaks (that is, where the intonational peak is aligned with the stressed, rather than the post-tonic, syllable) than many other varieties of Spanish, with 64% of peaks aligning with the stressed syllable. As a point of comparison, Face (2003) found 25% early peaks in spontaneous Castilian Spanish. This peak alignment pattern, common in many contact varieties (Cuzco Spanish - O’Rourke 2004; Buenos Aires Spanish - Colantoni and Gurlekian 2004), is employed in other dialects to mark focus or emphasis, and can give a staccato feel to an utterance (Face 2003). This result, then, in part accounts for the emphatic, pausing nature of Yucatan Spanish intonation reported in earlier studies and noted by visitors and native speakers alike. Regarding direct Maya influence, a preliminary statistical analysis did not find a significant difference regarding peak alignment between Maya bilinguals and Spanish monolinguals in Yucatan (Michnowicz and Barnes 2013), a finding that supports Barrera Vásquez’s (1937) observation that these two groups do not differ in their intonation. At the same time, the authors outlined a possible scenario in which Maya could have indirectly influenced peak alignment, noting that Maya, unlike Spanish, does not express focus or contrast through changes in prosody (see Frazier 2009, among others). This, combined with the observation that Maya high tones associate with phrase internal stressed syllables (see Gussenhoven and Teeuw 2008), leads to the possibility that shifting Maya speakers, when faced with prosodic input that did not match their native language, could have adopted early peaks as a general strategy of simplification (Michnowicz and Barnes 2013; see also Colantoni 2011). This pattern, like /ptk/ or /bdg/ outlined above, could then have spread to the monolingual population through large scale shift. More research is needed to confirm or refute this possibility.


Jim Michnowicz

3.2 Morpho-syntax Compared with phonetic/phonological features, there has been relatively little research done on morphosyntactic features in Yucatan Spanish. Many of the features that have been mentioned in the literature occur only among the bilingual population, and most likely are simple cases of L2 forms employed by second language speakers of Spanish around the world (see Michnowicz 2012). These include variation in prepositions, such as voy con el doctor for voy al doctor (Barrera Vásquez 1937); lack of gender agreement, as in nos gusta ver nuestra ciudad limpio (Michnowicz 2012: 106); non-standard use of auxiliary verbs, such as tengo ido a Cozumel (Michnowicz 2012: 106); variation in the placement of clitics, for example la paciente tuvo que se internar a la urgencia (Michnowicz 2012: 107); and pleonastic lo, as in ¿no te lo da vergüenza? (Suárez 1979: 180). None of these features were found among monolingual Spanish speakers, and likely represent general interlanguage forms found in many L2 contact situations around the world, rather than features unique to Yucatan Spanish or direct transfer from Maya (Michnowicz 2012; compare similar forms in Andean Spanish (Escobar 2011), Guarani-Spanish (Gynan 2011), and L2 and Heritage Spanish in the United States (Montrul, Foote and Perpiñán 2008)). Other features are more common, appearing also in the speech of some monolinguals, such as the use of pleonastic possessive pronouns, as in te cortaste tu dedo and su casa de Juan (examples from Suárez 1979: 179). Suárez (1979: 179) attributes this usage to transfer of the Maya genitive construction, which requires the possesive U before the possessed noun; i.e. U NOK’ IN SUKUUM ~ su ropa de mi hermano, although similar constructions also appear elsewhere in Spanish dialects. This feature, like the others outlined above, has never been systematically studied, and the discussion below will address three morphosyntactic features that have been addressed in more detail in the literature. 3.2.1 Subject pronoun expression Yucatan Spanish, along with other varieties of Spanish, is a pro-drop language, in which a subject may or may not be expressed according to a variety of contextual, syntactic and other factors (see Otheguy and Zentella 2012 for a detailed description of pro-drop in Spanish); for example, both hablo and yo hablo are acceptable structures in Spanish. Various researchers have proposed that contact with a language that requires a subject pronoun, such as English, can lead to higher pronoun rates (Otheguy and Zentella 2012 provide an overview). Additionally, other studies have suggested that contact with any language, even another pro-drop language, can also lead to higher pronoun rates, as bilingual speakers employ more subject pronouns as a simplification strategy (e.g. Sorace 2004). Other studies, however, have not found such an effect (de Prada 2009). As Maya, like Spanish, is a pro-drop language with a very low rate of overt subject expression (1% in a very small pilot study in Solomon

Maya-Spanish Contact in Yucatan, Mexico


1996, summarized in Solomon 1999: 225), data from Yucatan Spanish can contribute to the debate in the literature. Solomon (1999) demonstrates that the overall rate of expressed pronouns in her data from Valladolid, 19%, does not differ greatly from the rate found for other mainland varieties (cf. 24% expressed pronouns for recently arrived mainlanders in New York City - Otheguy and Zentella 2012: 108). Michnowicz (in press), in a study carried out with speakers from in and around Merida, also found a similar rate of 20% overt pronouns. Additionally, many of the same linguistic factors found to favor expressed pronouns in other varieties, such as switch reference, were also significant predictors in both Valladolid and Merida (Solomon 1999: 250; Michnowicz in press). Regarding direct Maya influence, studies differ with regards to differences between bilinguals and monolingual Spanish-speakers. While Solomon (1996) found no significant differences between bilinguals and monolingual Spanish-speakers with regard to pronoun expression, Michnowicz (in press) did find a significant effect for language, with Maya bilinguals producing significantly more overt pronouns than Spanish monolinguals (24% vs. 16% overt respectively). This difference is due almost entirely to cases of coreference with a preceding object, which disfavors overt pro for Spanish-speakers, but favors overt pro for bilinguals. Following Sorace (2004), Michnowicz (in press) attributes this difference in pronoun rate to a possible simplification strategy on the part of Maya speakers. Specifically, while bilingual speakers follow monolingual norms regarding the major distinction between CompleteSwitch and NoSwitch, they fail to apply the same constraints on a coreferential object. 3.2.2 Impersonal verb constructions with agentive por Another structure addressed briefly in previous literature is the existence of the plural impersonal verb with agentive por-phrase construction. Se lo llevaron por el viento for se lo llevó el viento Lo castigaron por su papá for lo castigó su papá (examples from Barrera Vásquez 1937: 9)

Barrera Vásquez (1937: 9) attributes this construction to a “literal translation of Maya syntax” (my translation), viewing the Yucatan Spanish se lo dijeron por su papá as a direct gloss of the Maya ‘álab ti’ tumen u tata’ (Barrera Vásquez 1937: 94). Lema (1991: 1283-1285) provides the following analysis of this phrase in Maya: al







‘to say’

reduplicative vowel

causal morpheme

indefinite pronoun [-1st per; - 2nd per]

causal preposition



Table 1. Analysis of phrase “álab ti’ tumen u tata” (Lema 1991).


Jim Michnowicz

Lema (1991) argues that the influence from Maya in the previous example is not syntactic, but rather semantic, in nature. For Lema (1991: 1284), this construction is related to the notion of causality in Maya, where a phrase such as “they are killing me” (“me matan” in Spanish) is expressed in Maya as “I am being affected by someone who is causing me to die” (my translation, following Tozzer 1977). The causal morpheme ‘b’ and preposition ‘tumen’ in Maya are identified with the existing (nonstandard) Spanish construction with por, as in se firmó la paz por los embajadores (citing Gili Gaya 1976), which allows for a prepositional phrase headed by por to act as the subject of an impersonal phrase with se. Also, the verb in the Yucatan Spanish construction appears in the plural, reflecting “the indefiniteness of the referent of the syntactic subject, without a morpheme in Maya, which indicates [-1st] and [-2nd] person. This indeterminacy in Spanish is marked with the plural ending on dijeron” (Lema 1991: 1285, my translation). It is noted, therefore, that “causality between Maya and Spanish is reinforced, the indefiniteness of the subject is transmitted, and the importance of the indirect is underlined” (Lema 1991: 1285). For this construction, it is argued that contact with Maya led to a reinforcement of an already (marginally) existing pattern in Spanish. Additionally, while Suárez (1979: 181) observes that this construction is only found among lower class rural speakers, Lema (1991) argues that it occurs frequently across all social and demographic groups, therefore suggesting that this is not unique to bilingual populations. 3.2.3 The pluralization of impersonal haber In normative Spanish, existential or impersonal haber prescriptively does not agree with its accompanying nominal phrase, which syntactically is analyzed as a direct object rather than a subject. In spoken Spanish, however, this construction is often reanalyzed by speakers, with the verb agreeing in number with what speakers perceive to be the subject. Compare standard había muchas personas with colloquial habían muchas personas. This feature is not unique to Yucatan, and occurs throughout the Spanish-speaking world (see Bentivoglio and Sedano 2011). Castillo-Trelles (2007: 80) presents a quantitative analysis of this phenomenon in Yucatan Spanish, where she finds an overall rate of 53% pluralization in conversational data, similar to rates found in other varieties (see Bentivoglio and Sedano 2011: 173 for an overview). Although language background was not a significant factor in the oral data, Maya bilinguals did significantly favor pluralization over Spanish monolinguals in a questionnaire task, in which the participants had to indicate which form (singular or plural) they would use in a series of phrases (Maya bilinguals = 65% pluralization; Spanish monolinguals = 42%) (Castillo-Trelles 2007: 81). Although CastilloTrelles does not hypothesize about the possible causes of the observed differences, it is likely that language background overlapped significantly with education level, which was not selected as significant in the statistical analysis. As the pluralization

Maya-Spanish Contact in Yucatan, Mexico


of haber is a universal phenomenon in Spanish, it is unlikely that the observed pattern can be attributed to either direct or indirect Maya influence. These findings do illustrate, however, the ways in which language background and social class/education level frequently interact in modern day Yucatan. 3.3 Lexical borrowing Lexical borrowing is especially prevalent in situations of language contact, even when that contact is not particularly intense, given that speakers do not even need to be functionally competent in the lending language in order to borrow words (Winford 2003: 30-31). This is particularly true for items or concepts that are foreign to the receiving language, such as terms for local flora and fauna. Both of the contact languages in Yucatan have undergone heavy cross-linguistic influence in their lexicons, although Maya has borrowed significantly more from Spanish than vice-versa, including even basic terms, such as numbers above four (Bolles and Bolles 2001). The Spanish of Yucatan, in contrast, has not adopted Maya terms for any of its core vocabulary - or at the very most when a Maya equivalent has been borrowed (such as Chichi for grandmother), the native Spanish form (abuela) continues to exist, often as a formal variant alongside the indigenous borrowing (see also Kolmer 2006). Suárez (1979) provides a detailed summary of borrowings into Yucatan Spanish, a few of which are presented in Table 2. Food Cooked underground; ex. pollo pibil





A regional stew

Papasul < Papak’sul

A regional dish with tortillas, eggs and sauce made from squash

Flora and Fauna Ch’om



A type of songbird




A type of large moth

Local customs/folklore Hanalpixán

Yucatan version of Mexican Day of the Dead

Mukbipollo < Maya Mukbil “cooked underground”; Spanish “pollo”

A type of tamal, eaten especially during Hanalpixán


Legendary female spirit in Yucatan; similar to La Llorona in Mexico


Small, legendary spirits or elves that inhabit Maya ruins.

Table 2. Sample lexical borrowings from Maya in Yucatan Spanish


Jim Michnowicz

In addition to the expected cultural borrowings above, common in most, if not all, cases of language contact, Maya terms have also entered colloquial Yucatan Spanish, and are used by all levels of society, regardless of language background. Kolmer (2006) observes that these Maya loans are becoming increasingly popular among speakers of Yucatan Spanish, perhaps as a way of expressing a sense of local pride and identity (a trend also seen with some other features of Yucatan Spanish; see for example the discussion on final -m above; Michnowicz 2008). Examples of these common terms include (all examples from Kolmer 2006: 185-186). Nouns such as tuch “bellybutton”, as in hoy Juan se rasca el tuch (lit. “today Juan is scratching his bellybutton”; i.e. Juan is not doing anything today). Adverbs, such as chan “a little” - ella lo chan quiere (she loves him a little); or hach “very much” - me hach gusta ese traje (I like that suit very much). Interestingly, these adverbs directly precede the verbs they modify, copying their preferred position in Maya (Kolmer 2006: 185), even intervening between the verb and the clitic pronoun, a position not normally licensed in Spanish3. Adjectives, such as kuch “tasteless; tacky” - está muy kuch tu sombrero (your hat is very tacky) Interjections, such as ¡huy! to express fear or anxiety; or chuch, used to reinforce an expression, as in ¡Chuch, qué bueno! Another interesting category of lexical borrowing is the use of hybrid MayaSpanish verb phrases constructed with the Spanish verb hacer (Suárez 1977: 130). Common expressions in this category include: hacer loch “to hug”, as seen on a popular t-shirt ¿Me haces loch?; hacer puts’ “to flee from an obligation”, as in hoy hago puts’ trabajo (today I’m skipping work; example from Kolmer 2006: 187)4. My own fieldwork suggests that Maya phrases such as these are often employed by younger speakers of the middle and upper social classes (see also Kolmer 2006). One informant, a 25 year old college graduate, notes that among his group of friends, a stereotypical Yucatan accent and Maya phrases are often used for comic effect: Eh, no sé un acento muy chistoso, muy alegre, muy risueño. Y al mezclarlo también con palabras mayas y todo, pues da risa, y es muy agradable. “Um, I



An anonymous reviewer points out the possibility that these adverbial expressions are borrowed as chunks or as fixed parts of an expression, rather than as a direct transfer of a Maya adverb to Spanish. More research is needed. Note that this construction is not unlike the hacer + English stem form reported for Spanish in the United States, e.g. “…no quería hacer open…” “…she did not want to open [the gate]…” (Toribio 2011: 538-539).

Maya-Spanish Contact in Yucatan, Mexico


don’t know…it’s a very funny accent, very happy, very cheerful. And when it is mixed with Maya words and such, well it makes us laugh, and it is very nice”. At the same time, this speaker notes that the use of Maya terms can also be a way to signal in-group identity, in the face of increased contact with those from outside of Yucatan (see Michnowicz 2011, 2012). ...y este igual nos da mucha gracia, que gente del exterior no las entiende [palabras mayas] y nosotros nos reímos. “…and we also think it is funny, that people from elsewhere do not understand Maya words, and we laugh”. 4. The future of Maya-Spanish contact in Yucatan When considering the future of Maya-Spanish contact in Yucatan, it is important to keep in mind that there are at least two factors involved. First, the demographics of the Maya-speaking population, as obviously sustained language contact is dependent on the continued presence of both languages in Yucatan. This of course includes not only Maya monolinguals, but also Maya-Spanish bilinguals that can serve as a locus for cross-linguistic influence; Second is the extent to which Mayainfluenced forms have directly or indirectly impacted the Spanish spoken by the monolingual majority, and therefore may continue on as vestiges of contact independent of the demographics. As noted in section 2, the number of Maya speakers in the State of Yucatan has steadily declined over the past half century (INEGI), although differing methodologies and questionnaires in different years of the Mexican census make it difficult to know exactly how many Maya speakers remain. The latest census (2010) reports that approximately 660,000 people in the State of Yucatan speak an indigenous language (overwhelmingly Maya). This figure represents 34% of the total population of Yucatan, up slightly from the 2005 census numbers, but in-line with those reported for 2000 (see Figure 2). Still, the overall trend is one of decline, as more and more speakers become bilingual, followed by a shift to Spanish by the children of these bilingual speakers (see Pfeiler and Zámišová 2006). Cifuentes and Moctezuma (2006: 191) use a series of four indicators of language vitality to examine questions of indigenous language maintenance and shift in Mexico: 1) the residence of indigenous language speakers in their ancestral homelands; 2) the rate of growth seen in the indigenous speaking community; 3) the levels of Spanish-indigenous language bilingualism; 4) the use of the indigenous language within the home. According to Cifuentes and Moctezuma (2006), Maya does fulfill the first characteristic, with a vast majority of Maya speakers residing in their native homelands on the Yucatan peninsula (2006: 220), thereby lending support to the maintenance of Maya. The other three indicators, however, point


Jim Michnowicz

toward language shift to Spanish, and they classify Maya as one of a group of indigenous languages in which “[w]e notice the last stages of bilingualism and the initial stages of shift towards Spanish monolingualism” (2006: 240). One of the factors that lead Cifuentes and Moctezuma (2006) to this conclusion is the high rate of bilingualism present in this group of shifting languages, which is as high as 91% for Maya (2006: 240; see also INEGI). While it is still possible to find Maya monolinguals (mostly older adults) in rural zones of the peninsula, the situation in the cities is quite different. In cities such as Merida, for example, many Maya-speakers use that language only rarely, as Maya use is restricted to familial and other semi-private contexts (Güémez Pineda 1994). At the same time, other researchers have noted the vibrancy of Maya compared to other indigenous languages in Latin America (see Lope Blanch 1987). As noted above, the fact that most Maya speakers live in their ancestral homeland on the Yucatan peninsula rather than being scattered throughout the country due to internal migrations can reinforce both the use of Maya (Cifuentes and Moctezuma 2006), as well as the level of prestige attributed to the indigenous language (Lope Blanch 1987). In fact, the urban pattern of Maya use mostly in the home domain, noted by Güémez Pineda (1994) as a potential weakness, is seen by others as a sign of possible vitality in the future, or at least as a constraint on the ultimate shift to Spanish (Cifuentes and Moctezuma 2006). In light of the slow but sure decline of Maya in Yucatan, some attempts have been made to revitalize the Maya language through various educational and cultural programs. Beginning in the 1940s, the government, along with civil associations, has undertaken various projects to promote the use of Maya in Yucatan, with mixed results (Pfeiler and Zámišová 2006). For example, the Academia de la Lengua Maya de Yucatán, founded in 1937 by Alfredo Barrera Vásquez, lists as one of its main goals “the promotion of oral and written expression in Maya5” (my translation), and its efforts include the recent publication of the Diccionario Maya Popular. In 2003, the Mexican government promulgated the General Law on Linguistic Rights of Indigenous Peoples (Ley General de Derechos Lingüísticos de los Pueblos Indígenas), which guarantees the right of speakers to employ indigenous languages “for any matter or transaction of a public nature” (see Althoff 2006; Pellicer, Cifuentes and Herrera 2006 for discussion). At the same time, in spite of these efforts and legal protections, the use of Maya, along with other indigenous languages in Mexico, continues to decline. Bilingual education, as implemented in Yucatan, has been shown to lead to both maintenance of Maya and increased shift to Spanish, depending on the way the


Maya-Spanish Contact in Yucatan, Mexico


programs are designed. Pfeiler and Zámišová (2006) compared two different approaches to bilingual education in Yucatan. The first of these, established by the Dirección General de Educación Indígena (DGEI), is the larger of the two programs examined, with almost 14,000 children enrolled in 2003 (Pfeiler and Zámišová 2006: 291). Pfeiler and Zámišová (2006) are critical of this program, however, noting that its primary goal is one of Hispanization of the indigenous population; in other words, the goal of the program is not to promote Maya language use per se, but rather to provide a bridge to allow Maya speaking students to master Spanish, thereby leading to increased shift (2006: 294). The authors contrasted this program with another implemented by the Consejo Nacional de Fomento Educativo (CONAFE), which first instructs indigenous students in their native language, before exposing them to the L2 (Spanish). The goal of the CONAFE program is “conscious bilingualism”, in which “the student values both his/ her mother tongue and its surroundings: [and] the Spanish language within its own context” (Pfeiler and Zámišová 2006: 293). The authors noted that programs of this type lead to increased maintenance of Maya, but unfortunately do not receive the same support as the DGEI program, only enrolling 749 children during 2003 (2006: 291). Bilingual education, then, in and of itself is necessary but not sufficient for Maya to be maintained in the long term. Thus, while the sheer number of speakers ensures that Maya is unlikely to disappear in the near future, the high levels of Maya-Spanish bilingualism among the indigenous population suggests that the shift to Spanish has in large part already taken place (see Cifuentes and Moctezuma 2006). In the short term, increased bilingualism also intensifies, in a technical sense, the level of language contact, since it is ultimately in the minds and speech of bilinguals that systematic interference can arise (as opposed to lexical borrowing, for example, which does not require bilingualism; see Winford 2003: 23-24 for an overview of different outcomes of language contact). At the same time, however, educational opportunities - in Spanish - are reaching the indigenous population at levels unthinkable several decades ago. Whereas previously bilingual speakers in small towns or rural areas would have had few opportunities to interact with or acquire monolingual varieties of Spanish, today young people of Mayan background are exposed to urban, standard Spanish through media, education, and face to face interactions with monolingual Spanish-speakers. One result of this exposure, as demonstrated in Michnowicz (2012), is that young educated speakers from Maya-speaking families have largely abandoned the L2 interlanguage forms employed by their parents’ and grandparents’ generations. The second aspect of Maya-Spanish contact is the use of possibly Maya influenced features by middle or upper class, monolingual Spanish-speakers in cities like


Jim Michnowicz

Merida. Not surprisingly, given the patterns of standardization for Maya youth outlined above, younger speakers of Yucatan Spanish are also quickly abandoning many of the regional variants that define traditional Yucatan Spanish as described in the literature on the dialect. This has been shown to be the case for aspirated / ptk/ (Michnowicz and Carpenter 2013; Michnowicz 2012); occlusive /bdg/ (Michnowicz 2012, 2011, 2009); and /ʔ/ insertion (Michnowicz 2012). Surprisingly, it has been shown that young speakers, whether from Maya or Spanish backgrounds, produce very similar, low rates for these features, showing that standardization appears to be happening in a similar fashion (and likely for similar reasons) across Yucatecan society (Michnowicz 2012). On the other hand, while some contact forms are quickly disappearing, research suggests that other variants possibly attributable to Maya contact are holding steady (such as peak alignment in intonation6) or increasing among some segments of the population, either as an indicator of identity or as a playful stereotype (such as final -m; see Michnowicz 2006, 2007, 2008; Yager 1989). In conclusion, Maya-Spanish contact in Yucatan has produced a unique variety of Spanish, spoken by both bilinguals and monolinguals alike. Many traditional features of Yucatan Spanish, frequently attributed to Maya contact, are also standardizing at a rapid rate. This standardization, brought about by a confluence of social, political, and demographic factors, is reducing or eliminating many of the contact-induced forms from monolingual speech (see Michnowicz 2012, 2011; Klee 2009). At the same time, an increased shift of Maya speakers to Spanish is putting the contact situation that has developed over the last few hundred years at risk of extinction. More studies are needed that examine how these processes of standardization and language shift are playing out across Yucatan. Analyses of this situation, and others like it throughout Latin America, provide a window through which we can observe how speakers of indigenous languages shift to Spanish and the linguistic consequences thereof, a process that must have occurred throughout most of Latin America in the past, but is now only observable on a large scale in a few remaining enclaves of indigenous language use. At the same time, further attention from scholars can help shed light on how to best help Maya-speaking communities maintain their language for the foreseeable future.


See Michnowicz and Barnes (2013), and compare intonation in Buenos Aires Spanish for insight on how contact-induced intonation can persist long after the contact situation has ceased (Colantoni 2011, among many others).

Maya-Spanish Contact in Yucatan, Mexico


REFERENCES Althoff, F. D. (2006). “Centralization vs. Local Initiatives. Mexican and U.S. Legislation of Amerindian Languages”. In M. Hidalgo (ed.), Mexican Indigenous Languages at the Dawn of the Twenty-First Century. Berlin: Mouton de Gruyter, pp. 167-190. Alvar, M. (1969). “Nuevas Notas Sobre el Español de Yucatán”. In Iberoromania I, pp. 159-189. Barrera Vásquez, A. (1937). “Mayismos y Voces Mayas en el Español de Yucatán”. Investigaciones Lingüísticas 4, pp. 9-35. Barrera Vásquez, A. (1945/1977). “La Lengua Maya de Yucatán”. In F. Luna Kan, L. Hoyos Villanueva, and C. Echánove Trujillo (eds.), Enciclopedia Yucatanense. Volume II. “Yucatán actual”. 2nd. ed. Mérida, Yucatán: Fondo Editorial de Yucatán. Bentivoglio, P./Sedano, M. (2011). “Morphosyntactic Variation in Spanish-Speaking Latin America”. In M. Díaz-Campos (ed.), The Handbook of Hispanic Sociolinguistics. Oxford: Wiley-Blackwell, pp. 168-186. Bolles, D./Bolles, A. (2001). A Grammar of the Yucatecan Mayan Language. Lancaster, CA: Labryrinthos. Canfield, D. L. (1981). Spanish Pronunciation in the Americas. Chicago: The University of Chicago Press. Cassano, P. (1977). “La Influencia del Maya en la Fonología del Español de Yucatán”. In Anuario de Letras 15, pp. 95-113. Castillo-Trelles, C. (2007): “La Pluralización del Verbo Haber Impersonal en el Español Yucateco”. In J. Holmquist, A. Lorenzino, and L. Sayahi (eds.), Selected Proceedings of the Third Workshop on Spanish Sociolinguistics. Somerville, MA: Cascadilla, pp. 74-84. Cifuentes, B./Moctezuma, J. L. (2006). “The Mexican Indigenous Languages and the National Censuses: 1970-2000”. In M. Hidalgo (ed.), Mexican Indigenous Languages at the Dawn of the Twenty-First Century. Berlin: Mouton de Gruyter, pp. 191-245. Colantoni, L. (2011). “Broad-Focus Declaratives in Argentine Spanish Contact and NonContact Varieties”. In C. Gabriel, and C. Lleó (eds.), Intonational Phrasing at the Interfaces: Cross-linguistic and Bilingual Studies in Romance and Germanic. Amsterdam: John Benjamins, pp. 183-212. Colantoni, L./Gurlekian, J. (2004). “Convergence and Intonation: Historical Evidence from Buenos Aires Spanish”. In Bilingualism: Language and Cognition 7, 2, pp. 107-119. Coupal, L./Plante, C. (1977). “Las Oclusivas Sordas Yucatecas: /ptk/: ¿Fuertes, Aspiradas, Glotalizadas?”. In Langues et Linguistique 3, pp. 129-176. De Prada, A. (2009). Subject Expression in Minorcan Spanish: Consequences of Contact with Catalan. Unpublished Doctoral Dissertation: Penn State University, State College, PA. Escobar, A. M. (2011). “Spanish in Contact with Quechua”. In M. Díaz-Campos (ed.), The Handbook of Hispanic Sociolinguistics. Oxford: Blackwell, pp. 323-352 Face, T. L. (2003). “Intonation in Spanish Declaratives: Differences between Lab Speech and Spontaneous Speech”. In Catalan Journal of Linguistics 2, pp. 115-131.


Jim Michnowicz

Frazier, M. (2009). The Production and Perception of Pitch and Glottalization in Yucatec Maya. Unpublished Doctoral Dissertation: The University of North Carolina - Chapel Hill. García Fajardo, J. (1984). Fonética del Español de Valladolid, Yucatán. México, DF: Universidad Nacional Autónoma de México. Gili Gaya, S. (1976). Curso Superior de Sintaxis Española. Barcelona: Vox. Granda, G. de (1982). “Observaciones Sobre la Fonética del Español en el Paraguay”. In Anuario de Letras XX, pp. 145-194. Güémez Pineda, M. A. (1994). “La Lengua Maya en Yucatán: Una Perspectiva Sociodemográfica”. In I’inaj, Semilla de Maíz, Revista de Divulgación del Patrimonio Cultural de Yucatán, August, 1994. Online version: miguel.html Gussenhoven, C./Teeuw, R. (2008). “A Moraic and a Syllabic H-Tone in Yucatec Maya”. In Z. Herrera, and P. M. Butragueño (eds.), Fonología Instrumental: Patrones Fónicos y Variación. México, DF: El Colegio de México, pp. 49-71. Gynan, S. N. (2011). “Spanish in Contact with Guaraní”. In M. Díaz-Campos (ed.), The Handbook of Hispanic Sociolinguistics. Oxford: Blackwell, pp. 353-373. Instituto Nacional de Estadística y Geografía (n.d.): Censos y Conteos de Población y Vivienda.

Klee, C. A. (2009). “Migration and Globalization: Their Effects on Contact Varieties of Latin American Spanish”, In M. Lacorte, and J. Leeman (eds.), Español en Estados Unidos y otros contextos de contacto. Sociolingüística, ideología y pedagogía. Madrid/Frankfurt: Iberoamericana/Vervuert, pp. 39-66. Klee, C. A./Lynch, A. (2009). El Español en Contacto con Otras Lenguas. Washington, D.C.: Georgetown University Press. Kolmer, K. (2006). “¡Chuch, Qué Bueno! Vom Wiederaufblühen der Maya-Kultur und ihrer Präsenz im Spanischen von Mérida (Yucatán, Mexiko)”. In Romanistik in Geschichte und Gegenwart 12, 2, pp. 179-192. Lema, R. (1991). “La Estructura ‘Se lo Dijeron por su Papá’ del Español Yucateco. ¿Simplemente Transportación de la Sintaxis Maya al Español?”. In C. Hernández, G. de Granda, C. Hoyos, V. Fernández, D. Dietrick, and Y. Carballera (eds), El español de América, Actas del III Congreso Internacional de El Español de América. Salamanca: Junta de Castilla y León, pp. 1279-1285. Lewis, M. P. (ed.) (2009). Ethnologue: Languages of the World, (16th ed.). Dallas, Texas: SIL International. Online version: Lipski, J. (2004). El español de América (3rd ed.). Madrid: Cátedra. Lisker, L./Abramson, A. S. (1964). “A Cross-Language Study of Voicing in Initial Stops: Acoustical Measurements”. In Word, 20, 3, pp. 384-422. Lope Blanch, J. M. (1987). Estudios Sobre el Español de Yucatán. México, DF: Universidad Nacional Autónoma de México. Lope Blanch, J. M. (1990). Atlas Lingüístico de México (dir). México, DF: El Colegio de México – Universidad Nacional Autónoma de México. Mediz Bolio, A. (1951). Interinfluencia del Maya con el Español de Yucatán. Mérida: Zamná.

Maya-Spanish Contact in Yucatan, Mexico


Michnowicz, J. (2006). “Final -m in Yucatan Spanish: A Rapid and Anonymous Survey”. In J. P. Montreuil (ed.), New Perspectives on Romance Linguistics. Vol. 2: Phonetics, Phonology, and Dialectology: Selected Papers from the 35th Linguistic Symposium on Romance Languages (LSRL). Amsterdam: John Benjamins, pp. 155-166. Michnowicz, J. (2007). “El Habla de Yucatám: Final -m in a Dialect in Contact”. In J. Holmquist (ed.), Selected proceedings of The Third Workshop on Spanish Sociolinguistics (WSS3). Somerville, MA: Cascadilla, pp. 38-43. Michnowicz, J. (2008). “Final Nasal Variation in Merida, Yucatan”. In Spanish in Context 5, pp. 278-303. Michnowicz, J. (2009). “Intervocalic Voiced Stops in Yucatan Spanish: A Case of Contact Induced Language Change?”. In M. Lacorte, and J. Leeman (eds.), Español en Estados Unidos y otros contextos de contacto. Sociolingüística, ideología y pedagogía. Madrid/Frankfurt: Iberoamericana/Vervuert, pp. 67-84. Michnowicz, J. (2011). “Dialect Standardization in Merida, Yucatan: The Case of (b d g)”. In Revista Internacional de Lingüística Iberoamericana, 18, pp. 191-212. Michnowicz, J. (2012). “The Standardization of Yucatan Spanish: Family Case Studies in Izamal and Mérida”. In K. Geeslin, and M. Díaz-Campos (eds.), Selected Proceedings of the 14th Hispanic Linguistics Symposium. Somerville, MA: Cascadilla, pp. 102-115. Michnowicz, J. (2015). “Subject Pronoun Expression in Yucatan Spanish”. In A.M. Carvalho, R. Orozco, and N. L. Shin (eds.), Subject Pronoun Expression in Spanish: A Cross-Dialectal Perspective. Georgetown: Georgetown University Press. Michnowicz, J./Barnes, H. (2013): “A Sociolinguistic Analysis of Pre-Nuclear Peak Alignment in Yucatan Spanish”. In C. Howe, S. E. Blackwell, and M. Lubbers Quesada (eds.), Selected Proceedings of the 15th Hispanic Linguistics Symposium. Somerville, MA: Cascadilla, pp. 221-235. Michnowicz, J./Carpenter, L. (2013). “Voiceless Stop Aspiration in Yucatan Spanish”. In Spanish in Context 10, 3, pp. 410-437. Montrul, S./Foote, R./Perpiñán, S. (2008). “Gender Agreement in Adult Second Language Learners and Spanish Heritage Speakers: The Effects of Age and Context of Acquisition”. In Language Learning 58, 3, pp. 503-553. Mosely, E. H. (1980): “From Conquest to Independence: Yucatan under Spanish Rule, 1521-1821,” In E. Mosely, and E. Terry (eds.), Yucatan: A World Apart. Alabama: University of Alabama Press, pp. 83-121. Norcliffe, E. (2009). “Revisiting Agent Focus in Yucatec”. In New Perspectives in Mayan Linguistics 59, pp. 135-156. Nykl, A.R. (1938). “Notas Sobre el Español de Yucatán, Veracruz y Tlaxcala”. In P. Henríquez Ureña (ed.), El Español en Méjico, Los Estados Unidos, y la América Central. Buenos Aires: La Universidad de Buenos Aires, pp. 207-225. O’Rourke, E. (2004). “Peak Placement in Peruvian Spanish”. In J. Auger, J. C. Clements, and B. Vance (eds.), Contemporary Approaches to Romance Linguistics: Selected Papers from the 33rd Linguistic Symposium on Romance Languages (LSRL), Bloomington, Indiana, April 2003. Amsterdam: John Benjamins, pp. 321-341. Otheguy, R./Zentella, A. C. (2012). Spanish in New York: Language Contact, Dialect Leveling and Structural Continuity. Oxford: Oxford University Press.


Jim Michnowicz

Pellicer, D./Cifuentes, B./Herrera, C. (2006). “Legislating Diversity in Twenty-First Century Mexico”. In M. Hidalgo (ed.), Mexican Indigenous Languages at the Dawn of the Twenty-First Century. Berlin: Mouton de Gruyter, pp. 128-166. Pfeiler, B. (1992). “Así som, los de Yucatán: El Proceso Fonológico Vn>m/___(#, C) en Dos Lenguas en Contacto”. In Memorias del Primer Congreso Internacional de Mayistas, vol. 1. México, DF: Universidad Nacional Autónoma de México, pp. 110-122. Pfeiler, B./Zámišová, L. (2006). “Bilingual Education: Strategy for Language Maintenance or Shift of Yucatec Maya?”. In M. Hidalgo (ed.), Mexican Indigenous Languages at the Dawn of the Twenty-First Century. Berlin: Mouton de Gruyter, pp. 282-300. Quezada, S. (2001). Breve Historia de Yucatán. México, DF: El Colegio de México. Ramos i Duarte, F. (1895). Diccionario de Mejicanismos: Colección de Locuciones i Frases Viciosas con sus Correspondientes Críticas i Correcciones Fundadas en Autoridades de la Lengua: Máximas, Refranes, Provincialismos i Remoques Populares de Todos los Estados de la República Mejicana. México, DF: Eduardo Dublán. Rosner, B. S./López-Bascuas, L. E./García-Albea, J. E./Fahey, R. P. (2000). “VoiceOnset Times for Castilian Spanish Initial Stops”. In Journal of Phonetics 28, pp. 217-224. Solomon, J. (1996). “Subject Expression in Yucatec Spanish: Accounting for similarities and differences in a contact variety”. Paper presented at NWAVE-XXV, Las Vegas. Solomon, J. (1999). Phonological and Syntactic Variation in the Spanish of Valladolid, Yucatán. Doctoral Dissertation. Palo Alto, CA: Stanford University. Sorace, A. (2004). “Native Language Attrition and Developmental Instability at the Syntax-Discourse Interface: Data, Interpretations and Methods”. In Bilingualism: Language and Cognition 7, pp. 143-145. Suárez, V.M. (1979). El Español que se Habla en Yucatán: Apuntamientos Filológicos (2nd ed.). Mérida: La Universidad de Yucatán. Thomason, S. G./Kaufman, T. (1988). Language Contact, Creolization and Genetic Linguistics. Berkeley: University of California Press. Toribio, A. J. (2011). “Code-Switching among US Latinos”. In M. Díaz-Campos (ed.), The Handbook of Hispanic Sociolinguistics. Oxford: Wiley-Blackwell, pp. 530-552. Tozzer, A. M. (1977). A Maya Grammar. New York: Dover. Weber, I. L. (1980). “Social Organization and Change in Modern Yucatan”. In E. Mosely, and E. Terry (eds.), Yucatan: A World Apart. University of Alabama: University of Alabama Press, pp. 172-201. Yager, K. (1982). Estudio del Cuadro Consonántico del Español de Mérida, Yucatán con Consideraciones de Posible influencia maya. University of California: Santa Barbara: Unpublished Master’s Thesis. Yager, K. (1989). “La -m Bilabial en Posición Final Absoluta en el Español Hablado en Mérida, Yucatán (México)”. In Nueva Revista de Filología Hispánica XXXVII, 1, pp. 83-94. Winford, D. (2003). An Introduction to Contact Linguistics. Oxford: Blackwell. 


1. Introduction The autochthonous population of Easter Island has seen the intergenerational transmission of their native language (Rapanui) succumb to the pressures of the official language (Spanish) since the annexation of the insular territory to the Republic of Chile in 1888. The current situation of Rapanui is akin to the one documented in Rindstedt and Aronsson (2002) for Quechua, in which even when children are addressed in the local minority language, they reply using the majority language (Spanish). Moreover, as described in Makihara (2007), adults tend to accommodate the children’s and young adults’ preference for Spanish by often switching from Rapanui to Spanish. In truth, Makihara (2005a) notes that informal Chilean Spanish is the preferred language among Rapanui teenagers and young adults. Rapanui has been marginalized (Makihara 2001, 2005a, 2005b, 2007; Makihara and Schieffelin 2007; Weber and Weber 1990) and, as a result, it is currently classified as “severely endangered” (Moseley 2010). In fact, in spite of some recent efforts to revitalize and promote the local Polynesian language, there is strong evidence that a language shift has been taking place (Faundes Peñafiel 2004; Makihara 2007; Wurm 2000). For instance, because the language is not being transmitted to the younger generations (for the most part), only older speakers are fluent and dominant in Rapanui. As pointed out in Gómez Macker (1982), various degrees of Spanish-Rapanui bilingualism are present in individuals over the age of 60. There are no monolingual speakers of Rapanui, as reported in Makihara (2005a), while the majority of Rapanui children, teenagers and even young adults cannot speak the local language fluently (Du Feu 1996; Makihara 2001, 2005a, 2005b, 2007; Thiesen de Weber and Weber 1998) – however, Fisher (2008) found indices of revitalization. 1

I am indebted to my informants for their time and patience. I am also very grateful for the help, advice, and hospitality that Nancy and Roberto Weber offered me while I was visiting the Island. Finally, I need to thank Jim Michnowicz and three anonymous reviewers for their helpful comments and suggestions. The usual disclaimers apply.


Verónica González López

Given this situation, it is unsurprising that the vast majority of research focuses on the unidirectional influence that the official language exerts on the local minority language and on the various linguistic outcomes of such prolonged contact (see, for instance, Makihara 2001). With the exception of Makihara (2005a, 2005b, 2007) and Weber (1988), however, there is a considerable lack of studies that examine the possibility of bidirectional influence. As mentioned, Spanish and Rapanui have been in contact for an extended period of time; nevertheless, in spite of the asymmetrical relationship between the two languages in terms of prestige and representation in the public sphere, it is quite likely that the Spanish spoken on Easter Island displays features of Rapanui. The present study examines the Rapanui influence on the Spanish spoken on Easter Island and offers a systematic account of certain morphosyntactic characteristics of this variety. For the current study, several naturalistic (i.e. non-formal, untutored) interviews were conducted with 21 Rapanui-Spanish childhood bilinguals (10 men, 11 women) between the ages of 30 and 74. The results from the interviews suggest a shift toward Spanish. While older individuals present a variety of Spanish that consistently shows Rapanui influence, younger speakers display a monolinguallike command of Chilean Spanish. The data obtained in this study are consistent with a number of linguistic phenomena that have been amply documented in similar cases of creolization, language shift and language death (Clements 2009a,b; Dorian 1981; Escobar 2000; Martínez 2010; Lipski 2007; Otheguy and Stern 2010; Schmidt 1985). Therefore, the data presented here further support the existence of a series of cross-linguistic mechanisms that are at work in all cases of language contact. Section 2 provides a succinct description of the bilingual continuum currently found on Easter Island and the role played by the educational system in the formation and development of such a linguistic scenario. Section 3 offers a description of the informants and of the methodology and techniques employed to elicit the grammatical structures under examination. Section 4 includes a definition of gender assignment as a morphological manifestation of syntactic agreement, and a description of gender agreement mechanisms in Spanish. This section also provides a succinct description of the Rapanui NP and the various syntactic functions and semantic notions encoded by the Rapanui nominal morphology. Lastly, the gender agreement system in Easter Island Spanish (EIS) is examined. Section 5 focuses on the verbal morphology of Spanish and Rapanui and shows that the lack of person and number morphemes in Rapanui favors the appearance of the default verbal morphology in EIS. Section 6 presents clitic reduplication in Spanish and EIS, and claims that even though this pattern is not exclusive to EIS, the reduplication mechanisms found in Rapanui may be facilitating and/or solidifying clitic

Rapanui Features in the Morphosyntactic System of Easter Island Spanish


reduplication phenomena in EIS. Finally, Section 7 offers a brief summary and concludes with some general observations. 2. Bilingualism on Easter Island Rapanui and Spanish have been in contact well before Easter Island’s annexation to Chile in 1888. Given the unequal status of Spanish and Rapanui with respect to prestige, usage, and representation, it is expected that some of the natural linguistic processes that arise in similar situations of extended contact, such as mixing or code-switching, may be perceived as undesirable and therefore suppressed or stigmatized. Nevertheless, the locals’ attitudes towards language mixing vary. Stanton (2001), for instance, provides examples in which native speakers of Rapanui (particularly older speakers) consider code-switching and mixing unacceptable and disgraceful, and therefore reject it. On the other hand, some of the younger speakers interviewed for the current study (young adults in their 30s in particular) report switching between Spanish and Rapanui regularly and accept it as a form of speech that characterizes the way in which the members of their generation communicate (see also Makihara 2001, 2005a for a similar account and examples). The majority of the Rapanui adult population can be classified as bilingual; however, there are varying degrees of individual bilingualism that fall along a continuum (Makihara 2001, 2005a). For instance, older speakers are reported to be more dominant in Rapanui and some even have serious communicative deficiencies in Spanish (cf. Gómez Macker 1982: 97). In fact, older speakers’ knowledge of Spanish barely satisfies daily communicative needs. On the contrary, young Rapanui adults, teenagers, and children tend to be more dominant or even monolingual in Spanish (cf. Gómez Macker 1982: 97). The age gap mentioned in Gómez Macker (1982) and Makihara (2005a), and observed in the present study as well, can be easily explained by examining the past educational conditions on the Island. The majority of the informants acquired Rapanui from birth and learned Spanish later in school. Spanish was the only language used for formal instruction until 1976 (see Gómez Macker 1982; Makihara 2005a, 2005b, 2007; Weber and Weber 1990). In addition, the school in Hangaroa (the only settlement on Easter Island) offered only grades K-6 until 1953, and grades K-8 until 1976, the year in which grades 9 and 10 were instituted. Both public and recently opened private schools have offered levels K-12 since 1989 (Makihara 2005a). Thus, the informants who were around age 60 by the time the interviews were conducted were born in the 1950s and instructed solely in Spanish during the 1950s and 1960s (some during the early 1970s). Most importantly, up until the early 1970s, the majority of teachers were continental Chileans whose na-


Verónica González López

tive and only language was Spanish. Therefore, classes were taught, planned and designed in Spanish specifically for native, monolingual speakers of Spanish. As pointed out in Gómez Macker (1982: 98), the acquisition of the Spanish language occurred in an unsystematic and inadequate way, leading to deficient (or non-target-like) results, linguistically speaking. In fact, students used to leave school with limited dominance of the informal, oral register of Spanish and almost no knowledge of the formal, written registers of the language (cf. Gómez Macker 1982: 98). Not surprisingly, according to a questionnaire administered in 1973, 85% of those interviewed reported being L1 Rapanui speakers, and 90% reported using Rapanui exclusively to speak with their parents and siblings (cf. Gómez Macker 1982: 95). To summarize, two major issues have affected the linguistic outcome of the modern EIS variety. On the one hand, the lengthy co-existence of Rapanui and Spanish has resulted in a variety that exhibits features usually found in other language contact settings. On the other hand, with the additional factor of the inadequate educational system in which Rapanui children receive instruction in monolingual Spanish in a system designed for monolingual Spanish children, the resulting variety shows characteristics akin to other cases of L2 acquisition. It is difficult, then, to determine whether EIS is a contact variety or an L2 variety of Spanish with various fossilized features in its morphosyntactic system. Therefore, explanations from a variety of contact and L2 acquisition studies are used to account for the phenomena under examination. 3. Methodology Sociolinguistic interviews were conducted with 21 Rapanui-Spanish bilinguals. The gender of the informants was distributed as follows: 10 male and 11 female. Their ages ranged between 30 and 74 years (mean age for all informants = 46.23; mean females= 51; mean males= 41). Based on self-reported estimates of language use, the informants had varied degrees of fluency and dominance in Rapanui and Spanish. The 10 male informants reported Rapanui as the language they learned from birth, and Spanish as the language they learned later on in their childhood, specifically at the onset of formal schooling. The 11 female informants, on the other hand, were a more heterogeneous group: 8 of them reported Rapanui as the language they learned from birth, followed by Spanish when they started formal schooling; the remaining 3 female speakers reported growing up in a bilingual household in which both Rapanui and Spanish were spoken, therefore acquiring both languages since birth. The Spanish to which these 3 female informants were exposed is considerably different from the contact variety of Spanish acquired by the rest of the participants. In particular, these 3 informants grew up with one monolingual Spanish-speaking parent and one bilingual Rapanui-Spanish speak-

Rapanui Features in the Morphosyntactic System of Easter Island Spanish


ing parent. In all other cases, the informants grew up in monolingual Rapanui households and were exposed to both EIS and Chilean Spanish outside of home. Data show that the amount of exposure to and production of both Spanish and Rapanui vary depending on the age of the informants. In line with Makihara (2005a, 2005b) and in accordance with the description provided earlier, only adults over the age of 60, who are fluent and/or dominant in Rapanui, exhibit morphosyntactic features that systematically diverge from standard Spanish. Younger adults, who may be fluent in Rapanui, display a monolingual-like command of Chilean Spanish with monolingual-like morphosyntactic features. The current study will focus exclusively on the EIS used by the older group. The linguistic features examined in this study were elicited and extracted from one-on-one interviews in a naturalistic setting. The interviews were conducted by allowing the informants to talk about a variety of topics such as personal anecdotes about living and growing up on the island, daily and favorite activities, and local customs and folklore, among others. With the purpose of reducing the Observer’s Paradox (Labov 1972) as much as possible, follow-up questions were formulated during each interview. This interviewing technique follows the Principle of Tangential Shift (Labov 1984: 37), which indicates that the range of topics covered in an interview does not need to follow a previously arranged script or order. Nevertheless, the Principle of Tangential Shift advises that each interview should begin with general questions and proceed toward more personal topics. Crucially, the changing of topics should progress as smoothly as possible, be initiated by the speaker, and guided by follow-up questions based on the information provided by the informant. The interviews lasted between 15 and 117 minutes, with an average of 52 minutes each. There are not many studies that focus on Rapanui Spanish. The existing few examine grammatical characteristics and linguistic processes that can also be observed in other contact varieties in similar contexts regarding unequal social prestige and use (see, for instance, Otheguy and Stern 2010 for a brief description of the current situation of popular US Spanish, or Zimmermann 1986 for Otomí Spanish in Mexico). The most outstanding and comprehensive study of the Rapanui-Chilean Spanish continuum from an anthropological perspective can be found in Makihara (2005a), who points out that EIS is characterized by “linguistic simplification and ‘interference’ of Rapanui features and other contact phenomena.” (Makihara 2005a: 729). The subsequent sections examine those areas of the EIS morphosyntactic system that display non-target-like characteristics. In particular, I will focus on gender agreement processes in the DP, subject-verb agreement, and clitic placement. The distribution and configuration of these constructions and features in monolingual Spanish, monolingual Rapanui, and EIS are detailed in the following sections.


Verónica González López

4. Gender assignment and gender agreement Gender is the morphological manifestation of a syntactic agreement mechanism (Fodor 1959). As indicated in Dahl (2000) and Neumann-Holzschuh (2006), gender is primarily a morphological property of a noun that extends to the whole DP. In minimalist terms (Chomsky 2001), gender is an uninterpretable feature that triggers agreement during the course of the derivation; Picallo 2008, however, argues that gender is the grammatical manifestation of an interpretable feature. Gender agreement, then, obtains from a feature valuation process that occurs in certain specific syntactic configurations (Pesetsky and Torrego 2007; Picallo 2008). Gender agreement is defined as the sharing of nominal features by the lexical (i.e. adjectives) and/or functional items in the DP (e.g., determiners). Even though there are many languages whose grammatical systems manifest gender morphologically, gender is not a universal linguistic category (cf. Ibrahim 1973). On the contrary, there are many languages that lack gender assignment and agreement. Notably, isolating languages like Rapanui, in which each separate word tends to represent a single morpheme, do not classify their nouns according to gender; hence they lack gender agreement mechanisms. Languages whose grammatical systems instantiate gender, on the other hand, may do so by means of two different operations (cf. Corbett 1991): some languages display a semantic system in which both gender assignment and gender agreement depend on the meaning of the noun; some other languages are endowed with a formal system that relies on the morphological form of the noun regardless of its meaning. 4.1. Gender assignment and agreement in Spanish Spanish follows primarily a binary formal gender assignment system (Ambadiang 1999; Bergen 1978; Gómez Torrego 1998; Harris 1991, among others) in which all nouns are obligatorily assigned either masculine or feminine gender morphology. Gender assignment is somewhat arbitrary since “there is no correlation with either meaning […] or phonological shape of the stem” (Harris 1991: 36). Some cases of semantic gender assignment can be observed in Spanish, in particular when grammatical gender matches the biological sex of animate referents (cf. Harris 1991), such as the following examples (see Ambadiang 1999 for a detailed account and examples of Spanish noun classification according to gender): (1) El hombre; la mujer. ‘the man; the woman.’ (2) El caballo; la yegua. ‘the horse; the mare.’

Rapanui Features in the Morphosyntactic System of Easter Island Spanish


(3) El toro; la vaca. ‘the bull; the cow.’

In general, morphological masculine gender manifests as the ending /-o/ in both gender assignment and agreement, as in (4), while morphological feminine gender surfaces as /-a/, as in (5) (see Ambadiang 1999 for a more exhaustive description of Spanish gender morphology). There are, however, exceptions to this generalization, such as la mano (the hand), which is a feminine noun that ends in the prototypically masculine marker /-o/, or el problema (the problem), a masculine noun that ends in /-a/.2 Nevertheless, agreement is still –o for masculine and –a for feminine, as in (6-7) respectively. (4) El martillo es negro (*negra). ‘The hammer is black.’ (5) La mesa es negra (*negro). ‘The table is black.’ (6) La mano es pequeña (*pequeeño). ‘The hand is small.’ (7) El problema es pequeño (*pequeña). ‘The problem is small.’

4.2. Gender assignment and agreement in Rapanui The Rapanui noun phrase (NP) has four positions that instantiate a variety of semantic concepts and grammatical functions. As illustrated in Du Feu (1996), there are two syntactic positions preposed to the noun: “position two” (located right before the noun), and “position one” (located to the left of the marker in position two). The syntactic markers in position one represent semantic concepts such as agent/instrument, possession, location, or syntactic functions such as subject/object, vocative, etc.; the discourse markers that occupy position two (right before the noun) represent semantic concepts like specificity, distance or visibility, or grammatical functions such as plurality or possession.3 For instance, in example (8) the vocative particle e…..e occupies position one in the NP. The NP i te moai



Exceptions like the ones mentioned above and similar cases are one of the reasons why, for some, the morphological markers of grammatical gender in Spanish are the determiners “el/ los” (masculine singular/plural) and “la/las” (feminine singular/plural), and not the endings that attach to the noun stem (Alarcos Llorach 1994; Alcina and Blecua 1975). See Du Feu (1996) for a full list of the particles that can appear in the Rapanui NP and the semantic concepts and grammatical functions that they instantiate.


Verónica González López

in example (9) shows how the two positions to the left of the noun may be filled. Thus, the syntactic marker i in position one illustrates the semantic notion of “relation”, while the discourse marker te in position two represents the semantic notion of “specificity.” The possessive ta’aku is exemplified in the NP ta’aku puka in (10), (128) in the original (Du Feu 1996: 33). (8) E Vero e!4 VOC5 Vero ‘Vero!’ (9) E tarai ro ‘a raua i te moai.6 STA carve +REA RES 3p RLT +SPE statue ‘They are carving statues.’ (10) Ko rehu ‘a i a au ta’aku puka.7 PFT forget RES RLT PRS 1s POS1sa book ‘I have forgotten my book.’

The third position in the NP is always occupied by the noun, while markers in the fourth (and last) position, always postposed to the noun, instantiate semantic concepts such as distance or specificity, as in (11): (11) Ko ŋaro’a ‘a e au tu poŋeha era.8 PFT hear RES AG 1s +SPE noise PPD ‘I’ve been hearing that ding going on.’

In (11), the NP tu poŋeha era contains the postpositive determinant era that signals that the lexical item is not nearby (i.e. distance) and that it is known (see Du Feu 1996: 12). As can be surmised from these examples, Rapanui NPs express a variety of syntactic functions and semantic notions that are encoded in strictly ordered particles to the left and right of the nominal lexical item. Crucially, unlike Spanish

4 5

8 6 7

Example (128) in Du Feu (1996: 33). The following abbreviations are used in all the examples (from Du Feu 1996: 6-8): 1s = 1st person singular; 3p = 3rd person plural; AG = agentive; BEN = benefactive; DUP = reduplication; MOM = momentary; PFT = perfect tense; POS1sa = possessive 1st person singular alienable; PPD = postpositive determinant; PRS = person singular; +REA = realized action; RES = resultative; RLT = relational particle; +SPE = specific; STA = state (verbal); TOW = towards subject; VOC = vocative. Example (540) in Du Feu (1996: 113). Example (544) in Du Feu (1996: 114). Example (543) in Du Feu (1996: 113).

Rapanui Features in the Morphosyntactic System of Easter Island Spanish


nouns, Rapanui nouns are not inflected for grammatical gender,9 as illustrated in Du Feu (1996). Consequently, the Rapanui grammatical system does not show gender agreement between nouns and other elements like articles or adjectives. In fact, since nouns are not classified in terms of gender,10 there is no noun-adjective agreement in Rapanui, as indicated in Du Feu (1996). 4.3. Gender agreement in EIS As mentioned earlier, Rapanui is an isolating language whose grammatical system lacks gender assignment, hence the lack of gender agreement mechanisms. Considering that all of the older speakers interviewed acquired Spanish after their Rapanui system had already been acquired, and, in some cases, as a second language through formal schooling, it is hypothesized that the lack of gender assignment in Rapanui may be causing the agreement mismatches observed between nouns and adjectives (12) and determiners and nouns (13) in EIS. Moreover, taking into account that gender agreement mismatches are present in some varieties of monolingual Southern Cone Spanish, which includes Chilean Spanish (see examples in Martínez 2010), it is possible that the extended contact between Spanish and a language that does not encode gender morphologically is only accelerating or increasing the emergence of an already existing tendency in Spanish. (12) a. Mi abuela materno era francés. (M62)11 my grandmother-fem maternal-masc was French-masc ‘My maternal grandmother was French.’ b. Hay una piscina cerrao. (F57) there-is a swimming-pool-fem closed-masc ‘There is a closed swimming pool.’ (13) a. Una camino ahí. (M67) a-fem road-masc over-there ‘A road over there.’ huevo de manutara. (M65) b. La the-fem egg-masc of manutara ‘The Manutara egg.’

Fuentes (1960) claims that Rapanui nouns are inflected for gender and number, just like Spanish nouns. However, the examples provided here indicate the opposite and further corroborate the information presented in Du Feu (1996). 10 Du Feu (1996) states that the only classification that Rapanui nouns undergo is in terms of alienable vs. inalienable. This, however, does not affect either article-noun agreement or noun-adjective agreement (or lack thereof). 11 The letter-number coding combination included after each example indicates the gender of the informant (M= male; F= female) followed by his/her age. 9


Verónica González López

In (12a), the noun abuela (grandmother) is feminine, and thus requires feminine agreement with other elements both inside and outside the DP. In this case, however, both the adjective in the DP materno (maternal) and the predicate adjective francés (French) show the corresponding masculine forms in Spanish. Likewise, the feminine noun piscina (swimming pool) in (12b) would trigger feminine agreement with the relevant lexical and functional items. In example (12b), while the determiner una (a) is feminine and hence matches the gender of the noun, the adjective cerrao (closed) surfaces in the masculine form, as evidenced by the ending -o. The examples in (13), on the other hand, show gender agreement mismatch between the indefinite determiner una (one, a) and the masculine noun camino (road) (13a), and the definite determiner la (the) and the masculine noun huevo (egg) (13b). Unfortunately, there are not enough examples of lack of gender agreement between definite determiners and nouns in the dataset to be able to determine if semantic features like definiteness may influence the gender agreement system of EIS. In addition, the only examples of gender mismatch observed between noun and adjectives are of the type illustrated in (12), in which a feminine noun is followed by masculine adjectives; no examples of masculine nouns followed by feminine adjectives were found. As will be seen shortly, this may be an example of the emergence of the masculine gender as the default or underspecified morphological marking. There is a substantial amount of research and language documentation from a variety of contact situations and second language acquisition (SLA) studies that can help explain the gender mismatch phenomena observed in this and other contact varieties of Spanish such as Otomí Spanish (Zimmermann 1986), Afroyungueño (Lipski 2006, 2009; Sessarego and Gutiérrez-Rexach 2011), Mayan Spanish (Michnowicz 2012), illustrated in (14-15) respectively, Spanish in contact with Quechua (Escobar 2000; Martínez 2010), and with Aymara (see Martínez 2010 and references therein for examples in written and spoken Spanish). The majority of SLA studies focus on gender agreement within the DP in a variety of languages (some of the most recent studies include Bruhn de Garavito and White 2002; Franceschina 2001, 2005; Montrul, Foote and Perpiñán 2008, among others), but there are some that examine the acquisition of long-distance gender agreement (Alarcón 2006). In general, these studies conclude that gender agreement appears to be a vulnerable area of the grammar that is acquired late by native speakers, and sometimes not even fully acquired, as might be the case for bilinguals (both early and late) and non-native speakers. Therefore, gender agreement mismatches are expected to surface in cases of naturalistic, unmonitored L2 acquisition.

Rapanui Features in the Morphosyntactic System of Easter Island Spanish


(14) a. La pulmón.12 the-fem lung-masc ‘The lung.’ b. La patio. the-fem patio-masc ‘The patio’ (15) a. Nos gusta ver nuestra ciudad limpio.13 we like to-see our-fem city-fem clean-masc ‘We like to see our city clean.’ b. Los fiestas. the-masc parties ‘The parties.’

As indicated in Lipski (2006, 2009), traditional Afroyungueño shows lack of gender agreement between nouns and adjectives (see also Sessarego and GutiérrezRexach 2011 for examples and a detailed analysis of gender agreement across generations in this variety of Spanish), and also between determiners and nouns, as illustrated in (14). Likewise, the grammatical system of Mayan Spanish also displays a lack of gender agreement between nouns and adjectives (15a) and determiners and nouns (15b). Notably, the examples in Mayan Spanish parallel those of EIS (12b) since feminine agreement is observed between the possessive nuestra (our) and the noun ciudad (city) in (15a), but not between the feminine noun ciudad and the masculine adjective limpio (clean). Unfortunately, the dataset examined in this study does not contain enough examples to extrapolate that this is a consistent pattern of gender agreement in EIS that may lead to a generalization.14 The examples of gender agreement in EIS, (12) above, show the emergence of the masculine morphology in adjectives that modify feminine nouns. It is commonly assumed that masculine is the default or underspecified gender (see Corbett and Fraser 2002 for a cross-linguistic sample and discussion; for Spanish, see, among many others, Alarcón 2006; Ambadiang 1999; Franceschina 2001, 2005; or Harris 1991). In fact, there exists a vast amount of evidence from a variety of studies confirming this, ranging from language acquisition in children and adults or L1 attrition, to research on creolization and pidginization processes. For example, stud Examples quoted from Lipski (2009) and adapted by Sessarego and Gutiérrez-Rexach (2011: 478). 13 Both examples from Michnowicz (2012). 14 In spite of the amount of data collected for the present study, the examples that illustrate non monolingual-like Spanish nominal, verbal, and pronominal constructions are scarce. Nevertheless, it should be noted that the examples included in this study are not idiosyncratic to a few speakers. On the contrary, these examples illustrate the spontaneous speech of a specific cross-section of the Easter Island autochthonous population described earlier. 12


Verónica González López

ies on the L2 acquisition of Spanish gender morphology in children show a strong preference for the masculine in both functional and lexical items (Plann 1979). Other studies, however, find no preference for the masculine gender in adjectives (Andersen 1984; Boyd 1975; Gathercole 1989). Nevertheless, the overgeneralization of the masculine gender morphology is observed across the board in spite of a few controversial findings (see Antón-Méndez 1999; Antón-Méndez, Nicol and Garrett 2002 and references therein). In the case of adult language acquisition, bilingualism, and attrition, the majority of studies corroborate the hypothesis that the masculine gender operates as the default morphological marking that is overgeneralized to the functional and lexical items in the DP, as well as to other elements outside of the DP in cases of long-distance agreement – as in predicative adjectives – (Alarcón 2006; Bruhn de Garavito and White 2002; Cain, Weber-Olsen and Smith 1987; Fernández-García 1999; Finnemann 1992; Liceras et al. 2008; Montrul, Foote and Perpiñán 2008; Schlig 2003; White et al. 2003; White et al. 2004). In addition, as tends to be the case in child production, adult production of masculine gender morphology is generally more accurate than that of the feminine gender morphology. Finally, Alarcón (2006) notices that participants’ level of proficiency seems to play a role in the adoption of the masculine morphology as the underspecified gender. While beginners may adopt either the masculine or the feminine gender markings as the default, more proficient learners generally default to the masculine morphological markings. The evidence present in studies that examine morphological gender in creole languages and language death and shift further solidifies the hypothesis that semispeakers tend to adopt the masculine gender as the default or underspecified morphological marking (Andersen 1982; Campbell and Muntzel 1989; Dorian 1973, 1981; Janse 2003; Lipski 1993; Romero 2011; Schmidt 1985). Even though creoles tend to lack gender morphology and gender agreement mechanisms, the functional modifiers that accompany nouns (determiners, quantifiers, etc.) are usually derived from the masculine forms of the lexifier language (see examples in Baptista 2002; Baxter 2010; Baxter, Lucchesi and Guimarães 1997; Caid-Capron 1996; Clements 2009a; Clements et al. 2006; Friedemann and Rosselli 1983; Holm 2000; Klinger and Dajko 2006; Lumsden 1999; Neumann-Holzschuh 2006). In general, the examples from the data show the emergence of masculine features as the default gender marking. Nevertheless, it is still unclear why feminine determiners sometimes precede masculine nouns, as in (13) above. Makihara (2005a) provides parallel examples in which the feminine third person strong personal pronoun is used to make reference to third person masculine extra-linguistic referents. She argues that “the use of the feminine third-person pronoun eia (ella “she”) instead of él “he” probably arises from the combination of rule-generalization

Rapanui Features in the Morphosyntactic System of Easter Island Spanish


and the indirect interference of the Rapa Nui15 pronoun system, in which gender distinction is not made.” (Makihara 2005a: 734). This explanation, however, is not without problems. For instance, the nature and configuration of the “rule-generalizations” mentioned in Makihara (2005a) are vague, as she fails to provide further explanation or clarification on this matter. Overall, there are not enough examples in the data to formulate a generalization that accounts for cases of gender agreement mismatch with masculine nouns. As mentioned earlier, gender agreement seems to be a vulnerable area of the grammar prone to optionality, fossilization, and incomplete acquisition (see Montrul 2008 and references therein). In fact, agreement in general seems to be permeable. It will be seen next that EIS is also characterized by verbal agreement mismatches that default to the third person singular morphology. 5. Verbal morphology This section focuses on how different features of Spanish and Rapanui verbal morphology combine to produce the non-monolingual-like characteristics observed in the EIS verbal system. The section offers an examination of the EIS morphosyntactic features elicited from the dataset that differ from monolingual Spanish varieties. A combination of factors account for the phenomena observed, among which Rapanui influence is scheduled. 5.1. Spanish verbal morphology The Spanish verbal system includes a variety of morphemes that encode several morphosyntactic and semantic concepts such as tense, mood, aspect, person, and number, as illustrated in the following example: (16) Cantábamos. sang-TMA-1pl ‘We used to sing.’

The verbal form in (16) contains the following morphemes: i) the stem cant-, which encodes the lexical meaning of the verb; ii) the thematic vowel -a-, which indicates the class to which a particular verb belongs;16 iii) the tense, mood, and Both spellings “Rapanui” and “Rapa Nui” are found in the literature to refer to the language as well as the ethnicity. Some authors, however, reserve “Rapanui” for the language, and “Rapa Nui” for the ethnicity. I do not make that distinction here. 16 There are three thematic vowels in Spanish: -a, -e, -i, each of which indicates the class that a verb belongs to and surfaces in the infinitive form of the verb, as in “hablar” (to speak), “comer” (to eat), and “escribir” (to write). 15


Verónica González López

aspect morpheme -ba-, which in this case encodes past tense, indicative mood, and progressive aspect; and iv) the person and number morpheme –mos, which in this particular form is 1st person and plural. It should be pointed out that not all verbal morphemes are always present in all verbal forms, and that irregular forms may display some variation. 5.2. Rapanui verbal morphology Rapanui verbal morphology includes prepositive particles that encode grammatical tense, aspect, mood, and negation (Du Feu 1996: 111). Unlike Spanish, Rapanui verbs are not inflected for person or number, as illustrated in the following contrast (see also Fuentes 1960): (17) a. E tunu au i te kai mo ta’aku poki.17 STA cook 1s RLT +SPE food BEN POS1sa child ‘I must cook a meal for my children.’ b. E tunu ‘a Nua i te kai ‘iroto i te hare. STA cook RES Nua RLT +SPE food within RLT +SPE house ‘Nua is cooking the meal inside.’

As can be observed in (17a) and (17b), the verbal form tunu appears with both the 1st person subject pronoun (au) and the 3rd person noun Nua. Most importantly, there are no preposed or postposed particles that signal any kind of number or person agreement with any other elements in the sentence (i.e. subjects, objects, etc.). Nevertheless, plurality may be morphologically encoded in the verb, especially when there is contextual ambiguity due to the lack of number inflection in nouns as well. In the case of intransitive verbs, plural subjects may be indicated by reduplication of the initial syllable of two-syllable verbs, as in (18a), or by reduplication of the final syllable(s) of three (or more)-syllable verbs. The same mechanism is observed to indicate the plurality of the object in the case of transitive verbs, as in (18b). (18) a. Ko tetere ‘a te hoi.18 PFT DUPrun PRES +SPE horse ‘The horses ran off.’ b. Ka momore mai te maika! MOM DUPcut TOW +SPE banana ‘Go and pick the bananas.’

17 18

Examples (702) and (704) respectively in Du Feu (1996: 154-155). Examples (769) and (771) respectively in Du Feu (1996: 167).

Rapanui Features in the Morphosyntactic System of Easter Island Spanish


5.3. EIS verbal morphology It has been documented that EIS displays verbal agreement mismatches (Makihara 2005b) in terms of person and number. This morphosyntactic behavior was present in the informants’ spontaneous speech samples obtained for this study and is illustrated in the following examples: (19) a. Yo sabe mucho.19 (F68) I know-3sg a-lot ‘I know a lot.’ b. Yo e(s) rapanui.20 (M70) I is Rapa Nui ‘I am Rapanui.’ (20) Dimes tú. (F68)21 tell-me you ‘Tell me.’

The examples in (19) show verbal agreement mismatch between the 1st person pronominal subject yo (I) (19a) and the 3rd person singular verbal morphology in the present tense (19b). 22 The example in (20), however, is not as straightforward as (19). In (20), the imperative form dime “tell me” contains an epenthetic final [s] sound that is not present in standard Spanish. Similar processes in other monolingual and contact varieties of Spanish offer a plausible explanation. On the one hand, examples like the verbal form observed in (20) above illustrate paradigm leveling (cf. Otheguy and Stern 2010) or erosion (cf. Bullock and Toribio 2006). Thus, by analogy with regular 2nd person singular forms, 2nd person imperatives show the same ending. A similar phenomenon has been documented in the Mexican Spanish of migrant workers in the US, among other varieties. As discussed in Bullock and Toribio (2006), the 2nd person singular form of the preterite in this variety of Spanish sometimes has a final [s]. The authors argue that contact with English (in this case) facilitates the grammatical erosion of low-frequency and/or irregular features, which results in paradigm leveling. Taking this into account, it could be argued that this case exemplifies an extension of an agreement pattern rather than verbal feature mismatch. EIS seems to favor 3rd person singular verbal forms “regardless of contextual person and number” (Makihara 2005a: 734). Taking into account evidence from a va-

21 22 19 20

Monolingual Spanish: ‘Yo sé mucho.’ Monolingual Spanish: ‘Yo soy rapa nui.’ Monolingual Spanish: ‘Dime (tú).’ There were no instances of subject-verb agreement mismatches in other grammatical tenses or verbal forms besides the present and the 2nd person singular imperative.


Verónica González López

riety of studies on SLA, bilingualism, and attrition, this is somewhat expected, as 3rd person singular verbal morphology has been argued to constitute the unmarked form (cf. Chomsky and Lasnik 1977), hence the default or underspecified form (examining results from various theoretical approaches, underspecified forms may be equated to unmarked forms, see Noyer 1992; Harley and Ritter 2002; CarstairsMcCarthy 1998, among many others). Even though there are some discrepancies and conflicting results in the relevant fields, 3rd person singular has generally been identified as the unmarked or underspecified verbal form cross-linguistically, as suggested already in Greenberg (1966). In the case of Spanish and closely related languages, the same generalization applies. Prévost and White (2000), for example, mention the emergence of 3rd person singular morphology in the acquisition of L1 French (see Ferdinand 1996). Meisel (1994), as discussed in Liceras, Valenzuela and Díaz (1999), notes that Spanish 3rd person singular indicative may be taken as the unmarked or “root” form due to its lack of person and number morphemes. In line with previous findings, the fact that the Rapanui verbal system does not encode grammatical person and number may facilitate the emergence of the unmarked or default verbal morphology in EIS. 6. Clitic reduplication The last morphosyntactic characteristic of EIS to be discussed here, clitic reduplication, is also present in monolingual varieties of Spanish, like Chilean Spanish, and other closely related languages (see González López 2008 for examples). Even though clitic reduplication in EIS cannot be accounted for solely by contact with Rapanui, there are morphological mechanisms in Rapanui that may be facilitating the appearance of this phenomenon. 6.1. Spanish pronominal clitics There are a multitude of analyses and descriptions of Spanish pronominal clitics (see González López 2008, 2012 for a review of the literature and examples). Cross-linguistically, pronominal clitics are characterized by a specific set of phonological, morphosyntactic and semantic features that differentiate them from other types of pronouns (see Cardinaletti and Starke 1999; Kayne 1975; Zwicky 1977; Zwicky and Pullum 1983). For instance, phonologically and syntactically, Spanish clitics are unstressed elements that do not appear in isolation without a host to cliticize to (Klavans 1982, 1995), as in (21) below: (21) ¿Viste el perro? Sí, *lo/ Sí, lo vi. saw-you the dog? Yes, it/ yes, it saw-I ‘Did you see the dog? Yes, I saw it.’

Rapanui Features in the Morphosyntactic System of Easter Island Spanish


As can be observed in (21), in the answer to the question ¿Viste el perro? (did you see the dog?) the pronominal clitic lo (it) needs to appear in a sentence that contains a verb, to which the clitic is claimed to attach, for the construction to be grammatical. Therefore, Spanish clitics are claimed to be functional markers that always surface attached to a verbal form (see González López 2008, 2012 and references therein). 6.2. Rapanui clitics If the phonological criteria for clitichood (which establishes, among other things, that clitic pronouns are phonologically unstressed elements that need to be attached to a stressed lexical item) are followed, then the unstressed nominal and verbal particles that appear in Rapanui NPs and VPs may qualify as clitics (Du Feu 1996: 112, 176). Other than these particles, there are no other elements in Rapanui that resemble the morphosyntactic and semantic behavior characteristic of pronominal clitics cross-linguistically in terms of position in the structure, cliticization processes, or types of semantic features instantiated by clitics (such as specificity or agreement). Thus, it can be assumed that Rapanui does not have clitic pronouns, as also claimed in Fuentes (1960). 6.3 Clitic reduplication in EIS The last feature of EIS that helps characterize this variety as distinct from others is clitic reduplication, illustrated in the following example: (22) Yo te voy a mostrarte. (F67) I cl go to show-cl ‘I am going to show you.’

As in example (22) above, the 2nd person singular pronominal clitic te appears twice in the structure, before the conjugated verb voy (I go), as is canonical with finite verbs, and attached to the right of the infinitive mostrar (to show), as is common with non-finite verbs. Since this phenomenon of clitic reduplication has been documented in Chilean Spanish (notably Kany 1945; Lipski 1990; Luján 1987), it is not entirely clear whether clitic reduplication in EIS may be a product of contact between Spanish and Rapanui whose grammatical manifestation parallels other contact and monolingual varieties, or simply a dialectal feature from the monolingual variety that Rapanui is in contact with. Nevertheless, as illustrated in example (18) earlier (see also Weber, N. 2003; Weber, R. 2003), reduplication is an available and pervasive mechanism in Rapanui that may be facilitating and even triggering the appearance of more than one co-referential clitic pronoun in the structure.


Verónica González López

Certain semantic and morphosyntactic concepts may be instantiated via reduplication in Rapanui (see Du Feu 1996; Fuentes 1960 for a variety of examples). Assuming that Spanish pronominal clitics are functional markers that also encode specific morphosyntactic and semantic information, it is quite likely that the grammatical system of EIS reanalyzes pronominal clitics and the information that they encode in a manner that parallels mechanisms of reduplication. Since the Rapanui verbal frame contains a great number of particles whose functions may parallel those of pronominal clitics, it is likely that the clitic is being reanalyzed as part of the verbal form. With the exception of recently introduced loanwords, Rapanui syllables are formed exclusively by a consonant followed by a vowel (Du Feu 1996; Fuentes 1960). Therefore, the similarities that exist between the Spanish and Rapanui syllable structures aid in this possible reanalysis. Coincidentally, reduplication in Rapanui is syllable-based, as explained in Du Feu (1996) and Fuentes (1960). Crucially, the majority of syllables in Spanish follow the CV distribution, which is also the pattern observed in clitic pronouns.23 Thus, the combination of semantic, morphosyntactic, and phonological mechanisms may be contributing to the manifestation of clitic reduplication in this variety of Spanish. 7. Summary and conclusions The examination of the various morphosyntactic features and phenomena that characterize EIS as a contact variety of its own reveals a tendency towards morphological simplification (as in Otheguy and Stern 2010). In general, the present study shows a reduction of Spanish grammatical paradigms (as compared to monolingual varieties), akin to other contact situations (Bullock and Toribio 2006; Otheguy and Stern 2010; Zimmermann 1986). Taking into account studies that focus on the acquisition of morphology (both in children and adults), this is unsurprising. Prévost and White (2000), for instance, observe that “L2 learners have difficulty with the overt realization of morphology” (2000: 104, 128). The same generalization appears to hold true for naturalistic language acquisition, since similar outcomes to the ones analyzed here have also been attested and documented in other instances of untutored Spanish acquisition (see Escobar 2000 and Martínez 2010 for Spanish in contact with Quechua; see Lastra 1990 for Spanish in contact with Otomí; see Lipski 2007 and Michnowicz 2012 for examples of Spanish in contact with Mayan; see Lipski 1994 and Ortiz López 1998 for Bozal Spanish; see Martínez 2010 and Mendoza 1991a,b for Spanish in contact 23

An anonymous reviewer points out that the syllabic structure of the plural forms of clitics in Spanish is CVC. Even though all the examples of clitic reduplication in the dataset involve the singular form, both Chilean Spanish and EIS show coda /s/-aspiration/deletion, which could further solidify the influence of the CV pattern mentioned.

Rapanui Features in the Morphosyntactic System of Easter Island Spanish


with Aymara). This further supports the existence of universal mechanisms that emerge in cases of unmonitored language acquisition. In fact, DeGraff (1999) states that “deviant” (i.e. non monolingual-like) linguistic systems may inform different stages of language change and unveil mechanism of language acquisition. Finally, the data shows that only speakers aged 50 and older produce the non-monolingual morphosyntactic features examined here, which further confirms a language shift in the direction of Spanish. REFERENCES Alarcón, I. V. (2006). The Second Language Acquisition of Spanish Gender Agreement: The Effects of Linguistic Variables. München: Lincom. Alarcos Lorach, E. (1994). Gramática de la Lengua Española. Madrid: Espasa Calpe. Alcina Franch, J./Blecua. J. M. (1975). Gramática Española. Barcelona: Ariel. Ambadiang, T. (1999). “La Flexión Nominal. Género y Número”. In I. Bosque, and V. Demonte (eds.), Gramática Descriptiva de la Lengua Española. Madrid: Espasa Calpe, pp. 4843-4913. Andersen, R. (1982). “Determining the Linguistic Attributes of Language Attrition”. In B. F. Freed, and R. Lambert (eds.), The Loss of Language Skills. Rowley: Newbury House, pp. 83-118. Andersen, R. (1984). “What’s Gender Good for, Anyway?”. In R. Andersen (ed.), Second Languages: A Cross-Linguistic Perspective. Rowley, MA: Newbury House, pp. 77-99. Antón-Méndez, I. (1999). “Gender and Number Agreement Processing in Spanish”. Doctoral Dissertation. Tucson: University of Arizona. Antón-Méndez, I./Nicol, J. L./Garrett, M.F. (2002). “The Relation Between Gender and Number Agreement Processing”. In Syntax 5(1), pp. 1-25. Baptista, M. (2002). The Syntax of Cape Verdean Creole. Amsterdam/Philadelphia: John Benjamins. Baxter, A. (2010). “Vestiges of Etymological Gender in Malacca Creole Portuguese”. In Journal of Pidgin and Creole Languages 25(1), pp. 120-154. Baxter, A. N./Lucchesi, D./Guimarães, M. (1997). “Gender Agreement as a “Decreolization” Feature of an Afro-Brazilian Dialect”. In Journal of Pidgin and Creole Languages 12(1), pp. 1-57. Bergen, J. (1978). “A Simplified Approach for Teaching the Gender of Spanish Nouns”. In Hispania 61, pp. 865-876. Boyd, P. A. (1975). “The Development of Grammar Categories in Spanish by Anglo Children Learning a Second Language”. In TESOL Quarterly 9(2), pp. 125-135. Bruhn de Garavito, J./White, L. (2002). “The Second Language Acquisition of Spanish DPs: The Status of Grammatical Features”. In A. T. Pérez-Leroux, and J. M. Liceras (eds.), The Acquisition of Spanish Morphosyntax: The L1/L2 Connection. Dordrecht: Kluwer Academic Publishers, pp. 153-178.


Verónica González López

Bullock, B./Toribio, A. J. (2006). “Intra-system Variability and Change in Nominal and Verbal Morphology”. In R. Gess, and D. Arteaga (eds.), Historical Romance Linguistics: Retrospective and Perspectives. Amsterdam: John Benjamins, pp. 305-325. Caid-Capron, L. (1996). “La Classe Adjectivale en Créole Mauricien et Réunionnais”. In D. Véronique (ed.), Matériaux pour l’Etude des Classes Grammaticales dans les Langues Créoles. Aix-en-Provence: Université de Provence, pp. 163-192. Caid, J./Weber-Olsen, M./Smith, R. (1987). “Acquisition Strategies in a First and Second Language: Are They the Same?”. In Journal of Child Language 14(2), pp. 333-352. Campbell, L./Muntzel, M. C. (1989). “The Structural Consequences of Language Death”. In N. C. Dorian (ed.), Investigating Obsolescence: Studies in Language Contraction and Death. Cambridge: Cambridge University Press, pp. 181-196. Cardinaletti, A./Starke, M. (1999). “The Typology of Structural Deficiency: A Case Study of the Three Classes of Pronouns”. In H. van Riemsdijk (ed.). Clitics in the Languages of Europe. Berlin: Mouton de Gruyter, pp. 146-233. Carstairs-McCarthy, A. (1998). “How Lexical Semantics Constrains Inflectional Allomorphy”. In G. Booij, and J. van Marle (eds.), Yearbook of Morphology 1997. Dordrecht: Kluwer, pp. 1-24. Chomsky, N. (2001). “Derivation by Phase”. In M. Kenstowicz (ed.), Ken Hale, a Life in Language. Cambridge, MA: The MIT Press, pp. 1-52. Chomsky, N./Lasnik, H. (1977). “Filters and Control”. In Linguistic Inquiry 8(3), pp. 425-504. Clements, J. C. (2009a). “Accounting for Some Similarities and Differences among the Indo-Portuguese Creoles”. In Journal of Portuguese Linguistics 8(2), pp. 23-47. Clements, J. C. (2009b). The Linguistic Legacy of Spanish and Portuguese. Cambridge: Cambridge University Press. Clements, J. C./Klinger, T. A./Piston-Hatlen, D./Rottet, K. J. (2006). History, Society and Variation. In Honor of Albert Valdman. Amsterdam/Philadelphia: John Benjamins. Corbett, G. (1991). Gender. Cambridge: Cambridge University Press. Corbett, G./Fraser, N. (2002). “Default Genders”. In B. Unterbeck, and M. Rissanen (eds.), From Syntax to Cognition; from Phonology to Text. Berlin: Mouton de Gruyter, pp. 297-339. Dahl, Ö. (2000). “Animacy and the Notion of Semantic Gender”. In B. Unterbeck (ed.), Gender in Grammar and Cognition: Part I. Approaches to Gender. Berlin: Mouton de Gruyter, pp. 99-116. DeGraff, M. (1999). “Creolization, Language Change, and Language Acquisition: A Prolegomenon”. In M. DeGraff (ed.), Language Creation and Language Change. Cambridge, MA: The MIT Press, pp. 1-46. Dorian, N. C. (1973). “Grammatical Change is a Dying Dialect”. In Language 49, pp. 413-438. Dorian, N. C. (1981). Language Death: The Life Cycle of a Scottish Gaelic Dialect. Philadelphia: University of Pennsylvania Press. Du Feu, V. (1996). Rapanui. London/New York: Routledge.

Rapanui Features in the Morphosyntactic System of Easter Island Spanish


Escobar, A. M. (2000). Contacto Social y Lingüístico: El Español en Contacto con el Quechua en el Perú. Lima: Pontificia Universidad Católica del Perú. Faundes Penafiel, J. J. (2004). “El Reconocimiento de los Pueblos Indígenas en Chile: Una Propuesta de Reforma al Estado”. In O. J. Aylwin (ed.), Derechos Humanos y Pueblos Indígenas: Tendencias Internacionales y Contexto Chileno. Temuco, Chile: Instituto de Estudios Indígenas/Universidad de la Frontera, pp. 299-315. Ferdinand, A. (1996). The Development of Functional Categories: The Acquisition of the Subject in French. The Hague: Holland Academic Press. Fernández-García, M. (1999). “Patterns of Gender Agreement in the Speech of Second Language Learners”. In J. Gutiérrez-Rexach, and F. Martínez-Gil (eds.), Advances in Hispanic linguistics: Papers from the 2nd Hispanic Linguistics Symposium. Somerville, MA: Cascadilla, pp. 3-15. Finneman, M. D. (1992). “Learning Agreement in the Noun Phrase: The Strategies of Three First-Year Spanish Students”. In IRAL 30, 2, pp. 121-136. Fisher, S. R. (2008). “Reversing Hispanization on Rapa Nui (Easter Island).” In T. Stolz, D. Bakker, and R. Salas-Palomo (eds.), Hispanization. The Impact of Spanish on the Lexicon and Grammar of the Indigenous Languages of Austronesia and the Americas. Berlin/New York: Mouton de Gruyter, pp. 149-165. Fodor, I. (1959). “The Origin of Grammatical Gender”. In Lingua 8, 1-41, pp. 186-214. Franceschina, F. (2001). “Morphological or Syntactic Deficits in Near Native Speakers? An Assessment of Some Current Proposals”. In Second Language Research 17, 3, pp. 213-247. Franceschina, F. (2005). Fossilized Second Language Grammars. Amsterdam: John Benjamins. Friedemann, N. S. De/Patino Roselli, C. (1983). Lengua y Sociedad en el Palenque de San Basilio. Bogotá: Caro y Cuervo. Fuentes, J. (1960). Diccionario y Gramática de la Lengua de la Isla de Pascua. Santiago de Chile: Editorial Andrés Bello. Gathercole, V. (1989). “The Acquisition of Sex-Neutral Uses of Masculine Forms in English and Spanish”. In Applied Linguistics 10, 4, pp. 401-427. Greenberg, J. (1966). Language Universals with Special Reference to Feature Hierarchies. The Hague: Mouton. Gómez Macker, L. (1982). “El Bilingüismo en Isla de Pascua”. In Signos XIV, 19, 91-99. Gómez Torrego, L. (1998). Gramática Didáctica del Español. Madrid: Ediciones SM. González López, V. (2008). “Spanish Clitic Climbing”. Doctoral Dissertation. State College, PA: The Pennsylvania State University. González López, V. (2012). “Spanish Clitic Placement: Direct Object Clitics are Markers of Epistemic Specificity”. In M. González-Rivera and S. Sessarego (eds.), Current Formal Aspects of Spanish Syntax and Semantics. Cambridge: Cambridge Scholars Publishing, pp. 42-77. Harley, H./Ritter, E. (2002). “Person and Number in Pronouns: A Feature-Geometric Analysis”. In Language 78, pp. 482-526. Harris, J. (1991). “The Exponence of Gender in Spanish”. In Linguistic Inquiry 22, 1, pp. 27-62.


Verónica González López

Holm, J. (2000). An Introduction to Pidgins and Creoles. Cambridge: Cambridge University Press. Ibrahim, M. H. (1973). Grammatical Gender: Its Origin and Development. The Hague: Mouton. Janse, M. (2003). “Language Death and Language Maintenance”. In M. Janse, and S. Tol (eds.), Language Death and Language Maintenance: Theoretical, Practical and Descriptive Approaches. Amsterdam: Benjamins, pp. ix-xvii. Kany, C. (1945). American-Spanish Syntax. Chicago: University of Chicago Press. Kayne, R. (1975). French Syntax: The Transformational Cycle. Cambridge, MA: The MIT Press. Klavans, J. (1982). Some Problems in a Theory of Clitics. Bloomington: Indiana University Linguistics Club. Klavans, J. (1995). On Clitics and Cliticization. New York: Garland. Klinger, T. A./Dajko, N. (2006). “Louisiana Creole at the Periphery”. In J. C. Clements, T. A. Klinger, D. Piston-Hatlen, and K. J. Rottet (eds.), History, Society and Variation. Amsterdam/ Philadelphia: John Benjamins, pp. 11-28. Labov, W. (1972). Language in the Inner City. Philadelphia: University of Pennsylvania Press. Labov, W. (1984). “Field Methods of the Project on Linguistic Change and Variation”. In J. Baugh, and J. Sherzer (eds.), Language in Use. Englewood Cliff: Prentice Hall, pp. 84-112. Lastra, Y. (1990). “Acerca del Español de los Ptomíes de Toluca”. In V. Demonte, and B. Garza Cuarón (eds.), Estudios de Lingüística de España y México. México. DF: Universidad Nacional Autónoma de México/El Colegio de México, pp. 561-570. Liceras, J. M./Valenzuela, E./Díaz, L. (1999). “L1/L2 Spanish Grammars and the Pragmatic Deficit Hypothesis”. In Second Language Research 15, 2, pp. 161-190. Liceras, J. M./Fernández Fuertes, R./Perales, S./Pérez-Tattam, R./Spradling, K. T. (2008). “Gender and Gender Agreement in Bilingual Native and Non-native Grammars: A View from Child and Adult Functional-Lexical Mixings”. In Lingua 118, pp. 827-851. Lipski, J. (1990). “Clitic Doubling and Direct Object Dropping in Andean Spanish”. Ms. University of Florida. Lipski, J. (1993). “Creoloid Phenomena in the Spanish of Transitional Bilinguals”. In J. Lipski, and A. Roca (eds.), Spanish in the United States: Linguistic Contact and Diversity. Berlin: Mouton de Gruyter, pp. 155-182. Lipski, J. (1994). Latin American Spanish. London: Longman. Lipski, J. (2006). “El Dialecto Afroyunqueño de Bolivia: En Busca de las Raíces del Habla Afrohispánica”. In Revista Internacional de Lingüística Iberoamericana 3, pp. 137-166. Lipski, J. (2007). “El Español de América en Contacto con Otras Lenguas”. In M. Lacorte (ed.), Lingüística Aplicada del Español. Madrid: Arco Libros, pp. 309-345. Lipski, J. (2009). Afro-Bolivian Spanish. Madrid/Frankfurt: Iberoamericana/Vervuert. Luján, M. (1987). “Clitic Doubling in Andean Spanish and the Theory of Case Absorption”. In T. Morgan, J. Lee and B. VanPatten (eds.), Language and Language Use. Studies in Spanish. New York: University Press of America, pp. 109-121.

Rapanui Features in the Morphosyntactic System of Easter Island Spanish


Lumsden, J. S. (1999). “Language Acquisition and Creolization”. In M. DeGraff (ed.), Language Creation and Language Change. Cambridge, MA: The MIT Press, pp. 129-157. Makihara, M. (2001). “Modern Rapanui Adaptation of Spanish Elements”. In Oceanic Linguistics 40, 2, pp. 191-223. Makihara, M. (2005a). “Rapa Nui Ways of Speaking Spanish: Language Shift and Socialization on Easter Island”. In Language in Society 34, pp. 727-762. Makihara, M. (2005b). “Being Rapa Nui, Speaking Spanish. Children’s Voices on Easter Island”. In Anthropological Theory 5, 2, pp. 117-134. Makihara, M. (2007). “Linguistic Purism in Rapa Nui Political Discourse”. In M. Makihara, and B. Schieffelin (eds.), Consequences of Contact. Oxford: Oxford University Press, pp. 49-69. Makihara, M./Schieffelin, B. (2007). “Cultural Processes and Linguistic Mediations”. In M. Makihara, and B. Schieffelin (eds.), Consequences of Contact. Oxford: Oxford University Press, pp. 3-29. Martínez, A. (2010). “Lenguas y Variedades en Contacto. Problemas Teóricos y Metodológicos”. In Revista Internacional de Lingüística Iberoamericana 8, 1, pp. 9-31. Meisel, J. (1994). “La Adquisición de la Negación en Euskera y en Castellano: Sobre la Separación Temprana de Sistemas Gramaticales por Niños Bilingües”. In J. Meisel (ed.), La Adquisición del Vasco y del Español en Niños Bilingües. Madrid/Frankfurt: Vervuert, pp. 151-180. Mendoza, J. (1991a). El Castellano Hablado en La Paz. Sintaxis Divergente. La Paz: Universidad Mayor de San Andrés. Mendoza, J. (1991b). “Aproximación Morfosintáctica al Castellano Paceño”. In C. A. Klee, and L. A. Ramos-García (eds.), Sociolinguistics of the Spanish-speaking World: Spain, Latin America and the United States. Tucson, AZ: The Bilingual Review Press, pp. 207-229. Michnowicz, J. (2012). “The Standardization of Yucatán Spanish: Family Case Studies in Izamal and Mérida”. In K. Geeslin, and M. Díaz-Campos (eds.), Selected Proceedings of the Hispanic Linguistics Symposium. Somerville, MA: Cascadilla, pp. 102-115. Montrul, S. (2008). Incomplete Acquisition in Bilingualism: Re-examining the Age Factor. Amsterdam/Philadelphia: John Benjamins. Montrul, S./Foot, R./Perpinán, S. (2008). “Gender Agreement in Adult Second Language Learners and Spanish Heritage Speakers: The Effects of Age and Context of Acquisition”. In Language Learning 58, 3, pp. 503-553. Moseley, C. (2010). Atlas of the World’s Languages in Danger. Paris: UNESCO Publishing. Neumann-Holzschuh, I. (2006). “Gender in French Creoles”. In J. C. Clements, T. A. Klinger, D. Piston-Hatlen, and K. J. Rottet (eds.), History, Society and Variation. Amsterdam/Philadelphia: John Benjamins, pp. 251-272. Noyer, R. (1992). “Features, Projections and Affixes in Autonomous Morphological Structure”. Doctoral Dissertation, MIT. Ortiz López, L. A. (1998). Huellas Etno-sociolingüísticas Bozales y Afrocubanas. Madrid/Frankfurt: Iberoamericana Vervuert.


Verónica González López

Otheguy, R./Stern, N. (2010). “On So-called Spanglish”. In International Journal of Bilingualism 15, 1, pp. 85-100. Pesestky, D./Torrego, E. (2007). “The Syntax of Valuation and the Interpretability of Features”. In S. Karimi, V. Samiian, and W. K. Wilkins (eds.), Phrasal and Clausal Architecture: Syntactic Derivation and Interpretation. Amsterdam: John Benjamins, pp. 262-294. Picallo, C. (2008). “Gender and Number in Romance”. In Lingue e Linguaggio 1, pp. 47-66. Plann, S. (1979). “Morphological Problems in the Acquisition of Spanish in an Immersion Classroom”. In Roger Andersen (ed.), The Acquisition and Use of Spanish and English as First and Second Languages. Washington, DC: TESOL, pp. 120-132. Prévost, P./White, L. (2000). “Missing Surface Inflection or Impairment in Second Language Acquisition?”. Evidence from Tense and Agreement. In Second Language Research 16, 2, pp. 103-133. Rindstedt, C./Aronsson, K. (2002). “Growing up Monolingual in a Bilingual Community: The Quichua Revitalization Paradox”. In Language in Society 31, pp. 721-742. Romero, R. (2011). “Mainland vs. Island: A Comparative Morphological Study on SpanishTurkish Contact”. In J. Michnowicz, and R. Dodsworth (eds.), Selected Proceedings of the 5th Workshop on Spanish Sociolinguistics. Somerville, MA: Cascadilla, pp. 50-56. Schlig, C. (2003). “Analysis of Agreement Errors Made by Third-year Students”. In Hispania 86, 2, pp. 312-319. Schmidt, A. (1985). Young People’s Dyirbal: An Example of Language Death from Australia. Cambridge: Cambridge University Press. Sessarego, S./Gutiérrez-Rexach, J. (2011). “A Minimalist Approach to Gender Agreement in the Afro-Bolivian DP: Variation and the Specification of Uninterpretable Features”. In Folia Linguistica 45, 2, pp. 465-488. Stanton, M. (2001). “I’m not a Chileno! Rapa Nui Identity”. In P. R. Spickard, and W. J. Burroughs (eds.), We Are a People. Philadelphia: Temple University Press, pp. 142-150. Thiesen de Weber, N./Weber, R. (1998). Diagnóstico Lingüístico de los Educandos de Kinder a Séptimo Año Básico de la Escuela “Lorenzo Baeza Vega” de Isla de Pascua, al Final del Año Escolar 1997: Informe y Recomendaciones. Programa Lengua Rapa Nui. Weber, N. (2003). Marca de Caso en Rapa Nui, Lengua Polinésica de Isla de Pascua. Valparaíso: Universidad Católica de Valparaíso. Weber, R. (1988). “Procesos Fonológicos en la Adquisición Naturalista del Español por los Rapa Nui de la Isla de Pascua”. In Revista Latinoamericana de Estudios Etnolingüísticos 5, pp. 21-31. Weber, R. (2003). Morfología Verbal del Rapa Nui, Lengua Polinésica de Isla de Pascua, y su Función en la Narración. Valparaíso: Universidad Católica de Chile. Weber, R./Weber, N. (1990). “¿Podrá Sobrevivir el Idioma Rapa Nui?”. In Revista Signos 23, 28, pp. 119-36. White, L./Valenzuela, E./Kozlowska-Macgregor, M./Leung, I. (2003). “Gender Agreement in L2 Spanish: Evidence Against Failed Features”. In McGill Working Papers in Linguistics 17, 2, pp. 153-182.

Rapanui Features in the Morphosyntactic System of Easter Island Spanish


White, L./Valenzuela, E./Kozlowska-Macgregor, M./Leung, I. (2004). “Gender and Number Agreement in Nonnative Spanish”. In Applied Psycholinguistics 25, pp. 105-133. Wurm, S. (2000). “The Language Situation and Language Endangerment in the Greater Pacific Area”. In M. Janse (ed.), Language Death and Language Maintenance. Amsterdam/Philadelphia: John Benjamins, pp. 15-47. Zimmermann, K. (1986). “El Español de los Otomíes del Valle del Mezquital (México), un Dialecto Étnico.” In J. Moreno de Alba (ed.), Actas del II Congreso Internacional Sobre el Español de América. México: Universidad Nacional Autónoma de México, pp. 234-240. Zwicky, A. (1977). On Clitics. Bloomington: Indiana University Linguistics Club. Zwicky, A./Pullum, G. (1983). “Cliticization vs. Inflection: English N’T”. In Language 59, 3, pp. 502-513.

THE FORMAL GUARANÍ AND SPANISH OF PARAGUAYAN BILINGUALS Shaw Nicholas Gynan, Ernesto Luís López Almada, Carlos Marino Lugo Bracho and María Eva Mansfeld de Agüero Western Washington University, Universidad Nacional de Itapúa and Universidad Nacional de Asunción

1. Introduction The following study reports on the mutual influence of Spanish and Guaraní, but with an emphasis on the intrusion of Spanish on Guaraní in a formal register of speech. We review the demography and history of Paraguayan Guaraní-Spanish bilingualism and de Granda’s theory of Guaraní-Spanish convergence (1988, 1990, 1996). The methodology used to elicit the language data analyzed here is described and then following traditional levels of linguistic analysis (phonological, morphological, syntactic, and lexical) we enumerate contact phenomena in Guaraní and Spanish. We conclude with a brief discussion of attitudinal dimensions of language contact in Paraguay, implications for language policy and recommendations for further research. 2. Paraguay’s Linguistic Demography Guaraní and Spanish have been in contact for almost five centuries. Paraguayan Guaraní-Spanish bilingualism has been documented since 1950 by the Paraguayan census as characterizing the majority of the country’s population. In 2002, the year of the last published census data, 98.8% of Paraguay’s population self-identified as non-indigenous, but 81.4% of Paraguay’s population reported speaking Guaraní (Gynan 2007). Paraguay is the only country in the New World where any significant proportion of the non-indigenous population speaks an indigenous language. Guaraní-Spanish bilinguals comprise 52.6% of the population, meaning that Paraguay has by far the highest rate of bilingualism of any country in the New World. Consequently, contact between Guaraní and Spanish is widespread. The long history of contact and the prevalence of individual bilingualism lead one to hypothesize that there are significant language contact phenomena.


Gynan, López Almada, Lugo Bracho and Mansfeld de Agüero

The volume in which this chapter is included has as its focus new perspectives on contact linguistics in the Americas. Here, we respond to an analysis of GuaraníSpanish contact that was developed by Germán de Granda (1990, 1996), who published extensively on Paraguayan bilingualism and proposed that there has been structural convergence between the two languages. Guaraní-Spanish language contact phenomena have been analyzed by de Granda and others, including, notably Thun (2002) and Zajíková (2009). By examining a corpus of systematically collected language data that is larger than what de Granda had at his disposal, this analysis affirms some of de Granda’s assertions and calls for revision of others. Clarification is made of the distinction between contact phenomena that are due to interference (the synchronic influence of one of a bilingual’s languages on the other, ascribed to imperfect learning by Bakker, et al. 2008) and those that are a product of convergence (a diachronic merging of grammatical aspects of two languages). This analysis also presents a preliminary analysis of de Granda’s claim of bidirectional influence of Guaraní and Spanish on one another. While we agree that the influence is mutual, our data indicate clearly that whereas the most important influence of Guaraní on Spanish is phonological, the greatest impact of Spanish on Guaraní is lexical. 3. Sample Characteristics and Study Methodology The data for the studies reported here were gathered by the authors from 2007 to 2012. The total number of individuals interviewed during that period is roughly 100. For the quantitative analysis of the influence of Spanish on Guaraní we refer mainly to a corpus developed by Lugo Bracho (2008). Of the 72 informants in Lugo Bracho’s study, 55 (74.4%) were teachers or otherwise formally educated at the university level and 17 (23.6%) had no university training. The majority of the informants, 62.5%, was female and 37.5% was male. The average age of the informants was 35 years, the youngest being 10 years old and the oldest 60 years old. The informants were interviewed in 15 cities and towns throughout the country. For the section on contact phenomena in Spanish, we refer to the data in López Almada (2011), since that has a more detailed analysis of the substrate influence of Guaraní on Spanish. Of the 21 informants in López Almada’s study, 15 (71.4%) were teachers or otherwise formally educated at the university level and 6 (28.6%) had no university training. The majority of the informants, 71.4%, was female and 28.6% was male. The average age of the informants was 35.8 years, the youngest being 16 years old and the oldest 71 years old. The informants were interviewed in rural and urban Itapúa and in Greater Asunción. This is a diversified convenience sample that includes individuals who are rural speakers of Guaraní for whom

The Formal Guaraní and Spanish of Paraguayan Bilinguals


Spanish is clearly a second language. Such people are critically important for documenting the status of the “purest” Guaraní, relatively uninfluenced by the standardization of formal education in Guaraní. Greater representation of Paraguayans untutored in Guaraní will be needed in future studies, as well as better representation of males. Since the focus of this research was primarily on Guaraní, less effort was made to interview urban youth whose control of Guaraní is weak or non-existent. We are able to point out some features of the Paraguayan Spanish of Spanish-dominant individuals, but future samples will need to include representatives from that group. For Lugo Bracho’s study, an instrument was devised to elicit an oral composition guided by illustrations that depict the development of agriculture in Paraguay from pre-colonial times to the present. That same instrument was used in Mansfeld de Agüero’s (2009) and López Almada’s (2011) studies. The instrument consisted of three pastel drawings. The first depicts indigenous men and women with their children hunting and gathering. The second shows a typical rural peasant house in front of which a woman prepares manioc and corn with her children and behind which a man is tilling a field with an ox-drawn plough. Cows, pigs and chickens complete the scene. The third shows a line of peasants with luggage in hand leaving for a city in the distance, crossing large fields which are being worked with large-scale machinery. The illustrations were assembled into two presentations, one in Guaraní and another in Spanish. The subject chooses the language in which he or she wishes to start, and then views the first picture, as voice explains the procedure in Guaraní or Spanish, saying that the subject will look at the illustrations and listen to the story, after which the illustrations will be displayed again so that the informant may develop his or her own version. After completing the narration, the informant heard instructions in the other language and then repeated the procedure in the second language. The narrations were recorded using the same computer that displayed the pictures. Narrations were subsequently transcribed following Lehmann’s conventions for interlinear morphemic glossing (2004). For all the recordings, canonical trilinear glosses have been developed, and for some, hesitation phenomena, phonological and phonetic levels of representation are included. Similar use of oral composition was pioneered by Wallace Chafe (1980), who used “The Pear Stories” to elicit comparable narratives in multiple languages. Here the technique of oral composition is used in an experimental way to control loosely for content and extent, excluding the variable of the interviewer. The result is typically a pair of two-minute narratives in the two languages. These kinds of data are relatively easily transcribed and quantified. There are drawbacks to a technique such as oral composition. Unlike a sociolinguistic interview, there is only one style of speech elicited and that is formal.


Gynan, López Almada, Lugo Bracho and Mansfeld de Agüero

Since it is a narration, there is a limited range of language structures produced. For instance, examples of person-number marking were largely limited to the third person and not one single instance of imperative or optative forms of the verb was documented. López Almada’s study of the speech of 21 Paraguayans does include a longer, structured interview in Guaraní. Those data are, as would be expected, more extensive and reveal more code-switching. The focus of Lugo Bracho’s study was the lexical influence of Spanish on Guaraní. Once the data were transcribed, we measured duration of the narrations, word production (total number of words as defined by the rules of orthography of the language), morphemic production (total number of lexical and grammatical morphemes), fluency (measured in words per minute), lexical density (the total of nouns, verbs, adjectives and adverbs divided by the total of words, as defined by Johansson (2008)), mean length of utterance (MLU) and Hispanisms in Guaraní. The focus of López Almada’s study was the Guaraní proficiency of the participants. The bilingual oral compositions produced by the 21 informants were transcribed with orthographic and phonological levels of representation and interlinear morphemic glosses (Lehmann 2004). Bilingual profiles were developed for each of the 21 informants using the same measures as in Lugo Bracho, with the addition of a measure of fluency in syllables per minute. Oral proficiency interviews in Guaraní only were administered to 17 of the 21 participants. These were transcribed orthographically. Preliminary quantitative analysis was performed, identifying duration of the interview, number of words and sentences produced and the mean length of the utterances. The full corpus of 100 oral compositions is still being transcribed at the phonetic level of representation. Once the phonetic representation is complete, we plan to apply the methodology detailed in Deuchar, Muysken and Wang (2007) to identify and categorize specific features of the language contact phenomena of the entire corpus. For the current study, however, our analysis of specific language contact phenomena refers to selected illustrative examples from the corpus. 4. De Granda’s Views on Guaraní-Spanish Convergence De Granda, based on Gumperz and Wilson (1971), defines convergence as “various convergent processes, which lead to the development of a common grammatical structure in communicative codes that live in a situation of intense and prolonged language contact” (de Granda 1996: 64). De Granda unifies two processes, transfer and convergence, as isogrammatism (Golab 1959, 1966). De Granda hypothesizes that Spanish-Guaraní isogrammatism is due not to “the chronologi-

The Formal Guaraní and Spanish of Paraguayan Bilinguals


cal duration of the process of contact between the two languages” but rather to “sociohistorical contingencies which determine, jointly, the lack of social distance among the speakers of the two communicative codes” (1996: 66). The thesis is challenging, because we believe that differential language contact phenomena in Guaraní and Spanish are reflective of a significant social distance between a historically, functionally monolingual, rural Guaraní-speaking population and a mostly bilingual, urban Spanish-speaking population. This linguistic division has characterized Paraguayan bilingualism since first contact (Solé 1991). (At the very time of the writing of this article, the overall level of Spanish monolingualism in Paraguay is exceedingly low, even in urban areas.) Despite the official status of Guaraní since 1992, the language has less prestige than Spanish. (By prestige we refer to the concept of standardization as developed by Ferguson (1966). Although both languages have standardized varieties, the Spanish standard is perceived to be closer to the spoken variety and the Guaraní standard is perceived to be further from the spoken variety (Gynan 2005). There is tremendous loyalty to and affection toward Guaraní, but Spanish is still heavily predominant in government, education and the mass media. The data reported on here indicate that there are important typological shifts underway in the Guaraní of some informants, which are due to long-term contact with Spanish. The typological shift referred to is in the Guaraní of rural people who are functionally monolingual. Since they are so clearly Guaraní-dominant, which we were able to confirm by analyzing the Spanish they produced during the bilingual oral composition task, we believe that we can eliminate the possibility of synchronic influence of their own Spanish on their Guaraní. 5. Phonological Contact Phenomena in Paraguayan Guaraní De Granda did not address in detail the issue of phonological convergence. Cassano (1973) concluded that Guaraní phonology has had relatively little influence on the segmental phonology of Paraguayan Spanish. Paraguayan Guaraní comprises 12 vowel phonemes and 21 consonantal phonemes. Our contention is that the phonological system of Guaraní has remained largely intact; however, the superstrate language has clearly made inroads in Guaraní. Certain infrequent sounds in Guaraní are found in Spanish loanwords, including the alveolar lateral /l/, and the alveolar trill /r/, which is actually pronounced as [ř], the voiced alveolar slit fricative. The voiced labiodental approximant /ʋ/ may be occasionally pronounced as a bilabial, which would be attributable to influence from Spanish, but preliminary analysis has found no convincing evidence for this (Gynan 2011). These infrequent sounds are found exclusively in loanwords and have not altered the phonology of the language.


Gynan, López Almada, Lugo Bracho and Mansfeld de Agüero

(1)a. Ha umi kuñakuéra katu (este) {mba’eicha piko ja’e} /ha.u.ˈmi.ku.ɲa.ˈkwe.ɾa.ka.tu./ [ha.u.ˈmi.kũ.ɲa.ˈkwe.ɾa.ka.tu.] ha umi kuña-kuéra katu conj woman-pl


(1)b. (este) ombyaty yva okaru haɡ̃ua hikuái /o.mɨ.a.ˈtɨ.ɨ.ˈʋa.o.ka.ˈɾu.ha.ˈw̃a.hi.ˈkwa.i./ [o.mbɨ.a.ˈtɨ.ɨ.ˈʋa.o.ka.ˈɾu.ha.ˈw̃a.hi.ˈkwa.i.] yva o-karu haɡ̃ua hikuái o-mbyaty 3pl 3.act-gather fruit 3.act-eat consec ‘And on the other hand the women gathered fruit in order to eat.’ (ER, Female, 45 years old, Encarnación, López Almada 2011).

Intonation and nasal harmony are two areas in which a reasonable case for convergence may be made. Fluent speakers of Guaraní have been recorded using questions with a rising final intonational contour typical of Spanish, but not of Guaraní. Evidence regarding nasal harmony is provided in (1) and (2). Both speakers from the interior are bilingual. The first speaker pronounces kuñanguéra ‘woman’ in (1) as [kũ.ɲa.ˈkwe.ɾa.]. The second speaker pronounces the same word in (2) as [kũ. ɲã.ˈw̃e.ɾa.]. Guaraní nasal spreading is missing in (1) and present in (2). The bilingual data from Lugo Bracho, Mansfeld de Agüero and López Almada include several examples of the lack of nasal spreading in the speech of individuals whose control of Spanish is very poor, constituting evidence that nasalization in Guaraní is converging with that of Spanish; however, at this stage, the presence of both forms in the speech of native speakers of Guaraní indicates that nasal harmony still exists, even in the partially converged Guaraní of untutored rural speakers. Among those informants who have received formal training, we can report no instances of complete lack of operation of the rule. Aside from these considerations, the phonological system of Paraguayan Guaraní appears to remain largely intact. (2)a. Ha kuñanguéra ha imembykuéra ndive /ha.ku.ɲa.ˈkwe.ɾɨ.ˈkwe.ɾʋe./ m [ha.kũ.ɲãˈwe.ɾa.ha.i.mẽ. bɨ.ˈkwe.ɾã.ndi.ʋe.] ˜ ha kuña-nguéra ha i-memby-kuéra conj woman-pl conj 3.inact-son_of_woman-pl

ndive comit

(2)b. ono’õ yva okaru haɡ̃ua avei /ˈʔõ.ɨ.ˈʋa.o.ka.ˈɾu.ha.ˈw̃ a.a.ʋe.ˈi./ [ˈʔõ.ɨ.ˈʋa.o.ka.ˈɾu.hã.ˈw̃ ã.a.ʋe.ˈi.] o-no’õ yva o-karu haɡũ a avei 3.act-gather fruit 3.act-eat consec also ‘And the women and with their children gathered fruit in order to eat also.’ (R, Female, 25 years old, Encarnación, López Almada 2011).

The Formal Guaraní and Spanish of Paraguayan Bilinguals


6. Morphosyntactic Contact Phenomena in Paraguayan Guaraní Guaraní is described by Tonhauser and Colijn (2010) as “mildly polysynthetic”. In (3), a typical example of Guaraní polysynthesis shows the noun pira ‘fish’ incorporated to the verb oikutu ‘they pierce’. (Guaraní grammarians will take note that the prefix remains aireal, despite the rule that changes personal prefixes from aireal to areal before incorporated nouns and reflexive prefixes.) These kinds of compound constructions are generally listed as entries in Guaraní dictionaries. (3) Oipirakutu /o.i.pi.ɾa.ku.ˈtu./ oi-pira+kutu 3.act-fish+pierce ‘They fished.’ (S, Female, 35 years old, Encarnación, Lugo Bracho 2008)

A similar example is provided in (4), where mymba ‘animal’ is incorporated to the verb ojuka ‘they kill’. (4) also provides an example of the agglutinative nature of Guaraní, where the verb ojuka ‘they kill’ has four concatenated grammatical suffixes. Krivoshein de Canese (1994) estimates that there are 70 grammatical morphemes in Guaraní. In the morphological analysis of the combined data from the three studies reported on here, roughly 75 grammatical morphemes have been identified. The most complex samples of Guaraní that we have elicited display rich grammatical morphology and polysynthesis, as in (4). (4)a. Ñepyrũrã ahecha mbohapy tekove ñande ypykuéra /ɲe.pɨ.ɾũ.ˈɾã.a.he.ˈʃˈpɨ.te.ko.ˈʋe.ɲɨ.pɨ.ˈkwe.ɾa./ ñe+pyrũ-rã a-hecha mbohapy t-ekove ñande ypy-kuéra refl+begin-dest 1sg.act-see three impr-person 1.inact ancestor-pl (4)b. omymbajukatavahína /o.mɨ.ma.dʝu.ka.ta.ʋa.ˈ o-mymba+juka-ta-va-h-ína 3.act-animal+kill-prosp-rel-3-prog (4)c. oikove haɡ̃ua hikuái ou mboyve umi españagua /o.i.ko.ˈʋe.ha.ˈw̃a.hi.ˈɨ.ˈʋe.u.ˈɲa.ˈwa./ oi-kove haɡ̃ua hikuái o-u mboyve umi españa+gua 3.act-live consec 3pl 3.act-come before Spain+adjr ‘At the beginning, I see three people, our ancestors who killed animals in order to live before the Spaniards came.’ (DO, male, 20 years old, Encarnación, López Almada 2011)

The data also include examples of transitive constructions which are more similar to Spanish syntax, as in (5), where the noun is not incorporated to the verb but


Gynan, López Almada, Lugo Bracho and Mansfeld de Agüero

instead follows it. This may be stylistic variation indicative of incipient conversion of Guaraní and Spanish that will eventually result in the loss of polysynthesis as a feature of Guaraní morphology. On the basis of the evidence in the corpus collected so far, we believe that Guaraní grammatical morphology has remained largely intact, and that formal education in the language reinforces its polysynthetic and agglutinative features. (5)a. Ñepyrũrãitépeakue ñande ypykuéra oikoakue /ɲe.pɨ.ɾũ.ɾãˈkwe.ɲɨ.pɨ.ˈkwe.ɾa.o.i.ko.a.ˈkwe./ ñe+pyrũ-rã-ité-pe-akue ñande ypy-kuéra refl+begin-dest-sup-loc-impv 1pi.inact ancestor-pl

oi-ko-a+kue 3.act-live-rel+post

(5)b. okaru haɡ̃ua ooo... ojuka mymba ka’aguy ha avei oikutu pira. /o.ka.ˈɾu.ha.ˈw̃a.o.dʝu.ˈka.mɨ.ˈma.kaʔ.a.ˈwɨ.ha.a.ʋe.ˈi.o.i.ku.tu.pi.ˈɾa./ o-karu haɡ̃ua o-juka mymba ka’a+guy Ha avei oi-kutu pira 3.act-eat consec 3.act-kill animal plant-subess conj also 3.act-cut fish ‘Our ancesters who lived here in the very beginning killed animals in the forests and also caught fish in order to eat.’ (RG, male, 45 years old, Concepción, Mansfeld de Agüero 2008)

7. Superstrate Influence on the Guaraní Determiner Phrase (DP) The Spanish feminine singular definite article la ‘the’ is found in Paraguayan Guaraní. De Granda and others identify a plural lo (based apparently on the masculine plural definite article los, from which plural -s has been eliminated). The indefinite article in Guaraní is peteĩ. De Granda concludes that the article system of Guaraní is now “parallel to that of Spanish” (1996: 66); however, according to Krivoshein de Canese “The articles lo and la in Guaraní have a different meaning than in Spanish. . . . There are many speakers who do not use these articles and they can be substituted by possessives or demonstratives” (1983: 17, our translation). Ortiz writes that the use in Guaraní of the definite article borrowed from Spanish: “has nothing to do with gender or any other feature of words, but is rather a type of crutch” (1990: 37, our translation). The Guaraní narrations produced by Paraguayans interviewed include many instances of la. Whether this use can be reversed is an important topic for future research, but those who have been educated in guaraniete ‘Academic Guaraní’ produce formal narrations that are completely devoid of Spanish determiners. In Lugo Bracho’s study, a preliminary analysis of the 72 bilingual oral compositions revealed that out of 7,729 words produced in the Guaraní narrations, la occurred 169 times (Table 1), an average of 2.5 times per narration; however, the standard deviation of 4.38 indicates considerable variation (Table 2). Some speakers

The Formal Guaraní and Spanish of Paraguayan Bilinguals


completely avoided inserting la in their narrations and others used it frequently, to a maximum of 24 times in the most extreme case. The presence of Hispanisms generally and the specific use of la is far lower in the speech of those who have received formal education in Guaraní (Table 3). Our sample of Paraguayan bilinguals includes several individuals from rural areas whose Spanish is rudimentary, and every one of them used la with greater frequency. This constitutes evidence that la in the Guaraní of functionally monolingual rural Paraguayans is incorporated to the DP of Guaraní, and not the result of interference from Spanish. Sentence (6) provides an example of the Spanish definite article in Guaraní, preceding soha ‘soy’. This example raises the question of whether the phrase constitutes a code-switch (referred to by some language practitioners in Paraguay as an instance of jehe’a ‘union’) or a lexical insertion (referred to as an instance of jopara ‘mixture’) (Paraguay 2001). Deuchar, Muysken and Wang (2007) provide a useful discussion of this issue and propose eliminating single-word switches in order to avoid mistakenly inflating the score for lexical insertion. Since the purpose of the studies summarized here was to document the influence of Spanish on Guaraní, we have counted all instances of Spanish words as Hispanisms, without distinguishing between loans and insertion. This would inflate the count of switches from Guaraní to Spanish, and in a later phase of the analysis we will seek to eliminate the most obvious loanwords using the dictionary, expert and nonexpert judgments, phonological and morphological incorporation, and hesitation phenomena. The answer to the question regarding the status of soha ‘soy’ is straightforward. Soy was unknown to pre-contact Guaraní. One may consult a dictionary to see if the word is listed, but Paraguayan lexicographers tend to be conservative. For instance, the acceptability of a loanword such as demokrásia ‘democracy’ instead of the Guaraní term tekojoja ‘democracy’ is indignantly rejected by defenders of guaraniete ‘Academic Guaraní’. Indeed, soha is absent from Trinidad Sanabria (2002), perhaps the most complete bilingual dictionary of Guaraní and Spanish to date. Our orthographic representation of the soha with ‘h’ instead of the ‘j’ used in Spanish spelling is consistent with our proposal that soha ‘soy’ is a loanword. (6) Ha upéi katu *oúma la. . .* oñotỹ la sohakuéra hikuái, /ha.u.ˈpe.i.ka.tu.o.ño.ˈtɨ.̃ˈkwe.ɾa.hi.ˈkwa.i./ ha upéi katu o-ñotỹ la soha-kuéra hikuái conj after advrs 3.act-cultivate def soy-pl 3pl ‘So then they cultivated soy.’ (A, male, 58 years old, Nueva Colombia, Lugo Bracho 2008)

If soha were a lexical insertion, the interlinear morphemic gloss (IMG) could be modified to indicate that la is a feminine definite article; however, the marking


Gynan, López Almada, Lugo Bracho and Mansfeld de Agüero

of soha with a plural suffix is evidence that the word has been morphologically incorporated. If that is the case, since Guaraní does not mark grammatical gender on inanimate nouns, the definite article is actually not inflected for gender. We also believe that la is not inflected for number, although again there is the possibility of an underlying /-s/ that has subsequently elided. We postulate that kamiõ ‘truck’ is a loanword. It is found in the dictionary spelled as presented here, indicating that the word no longer ends in a consonant, but rather with a nasal vowel. In Spanish, camión ‘truck’ is of masculine gender. The use of the feminine form of the definite article la ‘the’ in Guaraní could conceivably be in agreement with the feminine gender of kamiõ, but that is unlikely, since as mentioned above inanimate nouns in Guaraní have no gender and the original loanword is masculine. (7) Ojogua la kamiõ guasu ogueroja haɡ̃ ua pýpe. /o.dʝo.ˈˈmi.õ.wa.ˈsu.o.we.ɾo.dʝa.ˈha.ˈw̃ a.ˈpɨ.pe./ o-jogua la kamiõ guasu o-gueroja haɡ̃ua py+pe 3.act-buy def truck large 3.act-transport consec base+loc ‘They bought large vehicles in order to transport [it] there.’ (A, male, 58 years old, Nueva Colombia, Lugo Bracho 2008)

In (8c), la occurs twice (examples bolded for reference). The first instance of la in (8c) is glossed as a feminine, definite article, but it functions as a relative pronoun. It precedes not a noun, but a verb, and is derived from the Spanish nominalizing conjunction “la que”. We will return to the function of la as a relative pronoun below. (8)a. Ojehopa upévagui ambue tetãrekuéra /o.dʝe.ho.ˈpa.u.ˈpe.ʋˈe.te.tã.ɾe.ˈkwe.ɾa./ o-je-ho-pa upé+va-gui ambue tetã-re-kuéra 3.act-refl-go-cmpl dist+rel-abl other country-obl-pl (8)b. ikatu haɡ̃uáicha ...... ovivi porãmive rekávo /i.ka.ˈtu.ha.ˈw̃a.i.ʃa.o.ʋi.ˈʋi.po.ɾã.mi.ˈʋe.ɾe.ˈka.ʋo./ i-katu haɡũ á+icha o-vivi porã-mi-ve reká-vo 3.inact-poder consec+eqt 3.act-live good-dim-rel seek-sim (8)c. la oheka hikuái la irrecursokuéra /la.o.he.ˈka.hi.ˈře.kuɾ.so.ˈkwe.ɾa./ l-a o-heka hikuái l-a i-rrecurso-kuéra def-f 3.act-search 3pl def-f 3.inact-resource-pl ‘For that reason they all went to other countries in order to be able to live a little better, those who were seeking sustenance.’ (A, male, 58 years old, Nueva Colombia, Mansfeld de Agüero 2008)

The Formal Guaraní and Spanish of Paraguayan Bilinguals


The second instance of la in (8c) shows that it may precede plural nouns as well, thus la is not specifically singular. The second use of la is also unlike Spanish in that it precedes a noun that carries a third-person possessive prefix. In modern Spanish the definite article cannot precede a possessive pronoun. We have not found this kind of construction in our 100 samples of Paraguayan Spanish, but frequently in Guaraní, with la and other Guaraní determiners. In (9a) lo ‘’ is documented. We believe that the meaning of lo is plural because the reference in the original narration is to the arrival of men from Spain. In (9a) there is not a lot of context to justify the contention that lo does not mark masculine gender; however, in (10), from the speaker’s same narration, lo appears before a feminine noun marked for plural, confirming its plural meaning and lack of marking for gender. Lo is far less frequent than la in this corpus. Lo is never found with singular nouns, and may be substituted by la before plural nouns. In (8c) and (11), la precedes nouns marked with a possessive prefix. Although Modern Spanish bars the definite article from DPs containing a possessive adjective, at the time of first contact with Guaraní in the 16th century, DP allowed both (Batllori Dillet/Roca Urgell 1998), as in Modern Portuguese and Catalan. Furthermore, the Guaraní DP allows a variety of Guaraní specifiers in DP. There is therefore a landing site for insertion of the definite article, a transfer to somewhere (Andersen 1983). (9)a. Upéi katu oúma lo español ohecha /u.ˈpe.i.ka.tu.o.ˈˈɲol.o.he.ˈʃa./ upéi katu o-ú-ma lo español o-hecha then advrs 3.act-come-prf Spaniard 3act-see (9)b. ore yvy mba’éichapa romba’apo /o.ɾe.ɨ.ˈʋɨ.ma.ˈʔe.i.ʃɾʔa.ˈpo./ ore yvy mba’é+icha+pa ro-mba’apo 1pe.inact land thing+eqt+intr 1pe.act-work ‘But then the Spaniards came to see how we worked our land.’ (N, Male, 45 years old, Eusebio Ayala, Lugo Bracho 2008) (10)a. Upéi lo kuñakaraikuéra /u.ˈpe.i.lo.ku.ɲa.ka.ɾa.i.ˈkwe.ɾa./ upéi lo kuña+karai-kuéra then woman+lord-pl (10)b. ombyaty yva mimi imembykuérape. /o.mɨ.a.ˈtɨ.ɨ.ˈʋa.mi.ˈɨ.ˈkwe.ɾ o-mby+aty yva mi+mi i-memby-kuéra-pe 3.act-caus-collect fruit dim+dim 3.inact-offspring_of_woman-pl-loc ‘Then the women gather some fruit for their children.’ (N, Male, 45 years old, Eusebio Ayala, Lugo Bracho 2008)


Gynan, López Almada, Lugo Bracho and Mansfeld de Agüero

(11) Ha ombotuichave la ikokue /ʃa.ˈʋˈkwe./ ha o-mbo+tuicha+ve l-a i-kokue conj 3.act-caus+enormous+cmpr def-f 3.inact-farm ‘They made the farms bigger.’ (LA, Male, 55 years old, Edelira, López Almada 2011) (12)a. Ojama voi la ore rupa cósta-pe o-ja-ma voi l-a 3.act-approach-prf imm def-f

ore r-upa cósta-pe 1pe.inact rell-bed side-loc

(12)b. la oúva l-a o-ú-va def-f 3.act-come-rel ‘Those who had arrived approached the side of our bed.’ (Comisión Nacional de Rescate 1991: 12) (13) Ndaikuaái che mba’éicha-pa la añepyrũta nd>ai-kuaá proper nouns > [+human] > animates > inanimates), to the point of becoming an object marker (Company 2001: 149; see also Laca 2006). The variable argument structure of certain verbs, shifting from direct objects to indirect objects and vice versa, also supported the extension of DOM (Laca 2006: 427-428, 470). This is part of the widespread assumption that the dative ‘is invading’ the space of the accusative (Company 2001; FernándezOrdóñez 1993, 1999). A traditionally poorly understood aspect of Spanish syntax is the diachronically attested inherent variation of the 3rd person clitic paradigm as a referential system, termed leísmo, loísmo and laísmo8, as mentioned earlier. In traditional grammar the variation is based on a twofold distinction, either animate (personal) vs. inanimate (things), or on eliminating gender in favor of case distinctions. Loísmo is the use of lo for le, and represents an extension of the accusative into the dative. Loísta dialects refer to the male human direct object in (7) with the gender specific accusative clitic: (7) Ayer vi a Pepe y lo di el libro Yesterday see.1sg. past dom Pepe and docl.msg give.1sg. past det.msg book.msg ‘I saw Pepe yesterday, and I gave him the book’

Leísmo is a highly complex multisystem, showing distinctions based on geographical variation, contact with non-gender marking languages, different usage in written and oral language and finally the actual use compared to the educated “standard” use. Leísmo personal (personal leísmo) in (8) is the use of the dative le for the accusative lo referring to mainly singular male humans in Peninsular Spanish since the 16th century: (8) Ayer lei vi a Pabloi yesterday saw.1sg iom/dom Pablo ‘Yesterday I saw Pablo’

In Latin American Spanish, on the other hand, leísmo is case-based (Palacios 2005; Mayer 2010) and extends to female humans in (9):


See Fernández-Ordóñez (1999) for an extensive discussion and overview.

Continuity and Innovation in Peruvian Spanish


vi a Anai (9) Ayer lei yesterday saw.1sg iom/dom Ana ‘Yesterday I saw Ana’

Note the ambiguous role of a, which in both (8) and (9) can be interpreted as a (reinterpreted) indirect object case marker (IOM) or as a case of DOM with leísmo. Personal a in standard Spanish is already ambiguous with regard to objective case. In the contact variety discussed here, synchronic variation of canonical and noncanonical object marking, morphological marking of both objects with a case syncretic marker, and covariation of loísmo and leísmo produce conflictive analyses with regard to the syntactic distinction of objects (see Bossong 1991 for the creation of a new accusative case marker; Alsina 1996 for a binary distinction into ±dative case). Finally, under extensive leísmo, DOM and leísmo doubling extends to inanimate objects (10) in Ecuadorian Spanish (Suñer 1988: 512): (10) Ya lei veo a la camionetai already IOCL.SG see-1SG OM DET.FSG minibus.fsg ‘I can already see the minibus’

Extensive leísmo is a significant feature of Ecuadorian Andean and Paraguayan Spanish and a strong sign of the struggle of both internal objects for primary object status as well as a significant indicator of a language change in progress. Based on extensive documentation of the reanalysis of the locative preposition a in Mexican Spanish, Company (2003) links the struggle between dative and accusative for primary object status to the reanalysis of the preposition from a case marker of “cannibalistic” datives to the spread of DOM to inanimates. In her analysis, DOM marking for [+human] and [+anim] objects is only a transitory stage leading to extensive [Dat+] marking through extending DOM to [-anim] objects. This analysis can be linked to the [Dat±] hypothesis by Alsina (1996). c. Topic marking. In addition to semantics, pragmatics and information structure have also influenced the formation of DOM. Pensado (1995) remarks that the Romance origin of DOM marking originates in the use of the preposition ad with tonic personal pronouns, as in ad mihi (Pensado 1995: 191), which were highly likely to be used as topics (dislocation for topicalization). The loss of dative case morphology in the pronouns and its substitution by the prepositional phrase with ad, with the help of analogical generalizations, help DOM extend to non-pronominal uses as well. This idea of topic is crucial in Givón (1976) and, more recently, in Darlymple and Nikolaeva (2007), who locate the rationale for DOM in secondary topic marking.


Elisabeth Mayer and Manuel Delicado Cantero

The interaction between DOM and CLD is attested as early as in El Cid (Melis 1995; Laca 2006: 428). From a syntactic/morphological point of view, the question of the categorical nature of these duplicating clitics has been argued extensively in the literature. Rivero (1986) argues for a full NP (DP, in current terms) nature of those clitics in the old language, as opposed to their later re-categorization as agreement markers (objective morphology). Such syntactic representation is captured in Suñer (1988) and Franco (2000), among others. The typical examples of IO and DO reduplication in Spanish are semantically constrained by animacy, specificity and/or definiteness. Franco (2000: 167) notes that “feature erosion in pronominal affixes is a characteristic typical of agreement systems”. Interestingly, feature loss is the final necessary step for CLD (agreement markers) to become proper object markers (OM). This is what we find in Ecuadorian Andean Spanish and Peruvian Spanish contact varieties (Bossong 2003: 38; Palacios 1998, 2006, among others). At this point, the “clitic” suffices as object marker, which forces the redundancy of DOM a. Previously applicable semantic constraints, such as animacy/referentiality, are erased. We obtain the following dialectal divergence:

1. DOM due to semantic restrictions [+def], [+spec] (Suñer 1988; Torrego 1999; von Heusinger and Kaiser 2005; among many others). This still applies to standard Spanish and River Plate CLD + DOM. 2. DOM > OM; no semantic restrictions on the object (no animacy or definiteness/ specificity constraints).

3. DOCLD with non-agreeing clitics in contact varieties or non-standardized varieties Previous accounts reflect the interactions of contact and ongoing grammaticalization processes. Apart from Mayer (2003, 2006), Sánchez (2005) is the only study on Standard Limeño Spanish (StLS) optional clitic doubling with fully referential agreeing clitics. Work on optional direct object clitic doubling with agreeing clitics in liberal CLD varieties focuses on semantic and pragmatic strategies where DOM and CLD are licensed by the interaction of the referential categories animacy, definiteness and specificity. Recall that the norm for standard Spanish does not include CLD and DOM of inanimate and non-specific lexical direct objects (see examples in (3) above). Liberal CLD varieties such as River Plate (RP) and Limeño Spanish (LS) extend CLD and DOM to inanimates with agreeing (coreferential) clitics (11). This constitutes the first Latin American (LA) Spanish innovation regarding CLD and can be linked to marking the topical or salient object (Silva-Corvalán 1995; Estigarribia 2005).

Continuity and Innovation in Peruvian Spanish


a la cebollai? [-anim][+spec][+agr][+ref] (11) ¿Lai frío top2 fry.1sg dom det.fsg onion.fgs ‘Shall I fry the onion?’ (RP, StLS)

On the other hand, non-standardized varieties such as LSCV show two additional co-occurring innovations. The first innovation in (12) exhibits the same extension to inanimates as in (11), but with the non-agreeing (nonreferential) clitic lo:9 (12) Loi frío a la cebollai top2 fry.1sg dom det.fsg onion.fsg ‘I fry the onion’ (LSCV, Mayer 2010)


The second innovation is seen in (13), which shows that in cases of doubling DOM a has become redundant. Both innovations (12) and (13) occur in contact varieties and covary with the contact specific variants of loísmo and leísmo, where strange lo as an instance of lo-doubling in (13a) covariates with a specific variant of leísmo doubling (13b), both violating Kayne’s Generalization: (13) a. No, y no lo pronunciaba bien la erre no and no top2 pronounce.3sg.past well det.fsg r.fsg ‘No, and (s)he did not pronounce well the letter ’ b. No le saben pronunciar bien las erres not IOCL.SG know.3PL pronounce.INF well DET.FPL r.fpl ‘They don’t know how to pronounce the letter ’ (LSCV, Mayer 2010)

In non-standardized contact varieties we expect to find unstable and transitory stages where several strategies (old and new) coexist. In both cases the inanimate object is quite specific, if there were a pause after the bien, we could think of right dislocation, but that is not the case in the data in question. The preverbal, non-agreeing and non-referential clitic lo in (12) is known in the literature as strange lo, as mentioned above (Cerrón-Palomino 2003, among others). Strange lo is a featureless and invariant form “that can act as a surrogate for the entire category” (Greenberg, 1966; Bresnan 2001: 61) producing interesting morphosyntactic variation which is not restricted to a particular geographic region or dependent on extra-linguistic factors alone.


This combination is very unusual and highly stigmatized. However, according to personal communication from native speakers of River Plate in the greater Buenos Aires region, the same variation can be found in nonstandard River Plate Spanish.


Elisabeth Mayer and Manuel Delicado Cantero

Additionally, contact clitics exhibit morphological simplification in (14) as a sign of an ongoing reanalysis process: (14) Lei entregamos el regalo *(a) los niñosi give.1pl det.msg gift iom det.msg children.msg ‘We give the gift to the children’

An originally Andean Spanish phenomenon also found in LSCV nowadays is locative doubling, as illustrated in (15). It can be analyzed either as an intransitive goal that has been reached (Ritter and Rosen 2001), or as a boundedness delimitation of an event, secondary agreement of an oblique10 (Kiparsky 1998: 266): (15) Lo llegaron a este pueblo docl.msg arrive.3pl loc dem.msg village.msg ‘They arrived at this village’ (Andean Spanish, Cerrón-Palomino 2003: 168-170)

All innovations exposed for LSCV exist in an Andean continuum, which is synonymous with rural conditions, and much more precisely expressed by Cerrón-Palomino (2003: 190) as español rural, which has been serving as the only language of communication in those areas for almost 500 years. Here we find the same inter- and intra-speaker variation situated at different stages of evolution based on extra-linguistic factors favoring and accelerating an ongoing reanalysis process in the pronominal paradigm for 3rd person. 4. A pragmatic analysis Our data from Peruvian Spanish contact varieties corroborate the commonly accepted idea that language change proceeds gradually, since not all linguistic features shift at the same time and sometimes different variants coexist even in the same text (Kroch 1994; Croft 2000; Joseph and Janda 2003, among many others). This is evident in LSCV, where coexistence of variant forms is present in interspeakers and intra-speakers.11 4.1. Object marking systems The variation over time and space is summarized in Table 1:

The distinction between oblique and indirect object is debatable. For more evidence for that see Allen (1995) on the gradual loss of dative experiencer subjects from Old to Early Modern English and Allen (2000) for a gradual shift from OV to VO in Early Middle English.

10 11

Continuity and Innovation in Peruvian Spanish




standard clitic

contact clitic (Peru)

Ecuador/ Paraguay


dative a





locative a





M: lo(s), F: la(s)



Table 1. Object marking systems

The striking case syncretism in column 1 finds its counterpart in the Quechua case marker -ta. Feature-specifying standard clitics follow a semantic strategy distinguishing unambiguously between indirect and direct object in anaphoric agreement and subject to Kayne’s Generalization, in grammatical agreement. As shown above, River Plate (Silva-Corvalán 1995; Estigarribia 2005) and Limeño Spanish (Mayer 2003, 2006; Sánchez 2005) varieties show the first reanalysis processes in terms of liberal direct object CLD marking highly topical direct objects. This constitutes a first move from semantic object marking strategies to pragmatically motivated object marking in contact clitics. As previously mentioned, contact fast-tracks inherent changes in clitics through a reanalysis process from feature-specifying agreement markers to featureless object markers. In that respect, the Quechua topic marker -qa plays an important role in the Peruvian Andes. Finally, a glimpse of an endpoint is the extensive leísmo in Ecuador –also in Paraguay, in contact with Guarani– where lo survives as a propositional anaphor in certain regions, e.g. in Ecuadorian Highland Spanish (Suñer and Yépez 1988; Haboud 1998; Palacios 2002). 4.2. The role of secondary topic and its relation to primary object We link strange lo to the pragmatic notion of secondary topic marking (TOP2) following a new DOM theory by Dalrymple and Nikolaeva (2007). As the exponent of TOP2, this featureless and invariant form cross-references animate and inanimate objects in LSCV that are topic-worthy (Dalrymple and Nikolaeva, 2007). The notion of topic here is restricted to sentence or clause topics, including topical participants; it does not extend to discourse topics (Lambrecht 1994). In defining sentence topic as “the thing which the proposition expressed by the sentence is about” (Lambrecht 1994: 118), and in taking the communicative context into account, the ‘aboutness’ relation combines the topic referent and the object referent and the resulting propositions in a given discourse. The pragmatic notion of secondary topic is based on an opposition of direct and indirect argument functions active in many languages. Based on the assumption that the subject is the primary topic, Givón (1983) proposed the term secondary


Elisabeth Mayer and Manuel Delicado Cantero

topic for the grammatically marked direct object. Based on work on Ostyak, Nikolaeva (2001: 26) defines the secondary topic as “an entity such that the utterance is construed to be ABOUT the relationship between it and the primary topic”. In a new DOM theory, Dalrymple and Nikolaeva (2007) analyze the differences between marked and unmarked objects as different information structures expressed in syntactic terms representing two different grammatical functions. According to their theory, marking is preferred “in contexts where the object is salient and the utterance updates the addressee’s knowledge about the relationship that holds between the subject and the object referents” (Dalrymple and Nikolaeva 2007: 37). In these terms, marked direct objects in some languages can be analyzed as the primary object and the secondary topic. Crucial for the distinction between TOP and TOP2 in LSCV is the resulting implication that TOP can be either the subject or another grammatical relation in the proposition. Another important distinction is the difference between natural topicality (Croft 1991) and topic-worthiness (Comrie 2003). The former reflects the pragmatic relation between the referent and the proposition, and the latter represents the potential to become a topic by the degree of pragmatic saliency they show for the interlocutors. Specifically highly agentive subjects are in charge of monitoring the close relationship between primary and secondary topic (see Givón 1976; Darlymple and Nikolaeva 2007), in other words, topic-worthiness depends on the primary topic’s selection to specifically mark the strong relationship between itself and the object, a strategy also known as double actant marking (Bossong 2003). The distinction between TOP and TOP2 constructions is shown in examples (16) and (17). TOP or topical constructions in liberal CLD standard varieties such as River Plate in (16a) and Limeño standard Spanish in (16b) differ from TOP2 constructions in several ways. Liberal CLD extends to inanimates and quantifiers albeit with feature specifying clitics as agreement markers; these constructions are part of syntactic functions and have also made their way into written language: (16) a. La frío a la cebolla docl.fsg fry (d)om det.fsg onion.fsg ‘I fry the onion’ (River Plate Spanish, Rodríguez Louro pc) b. No lo vieron a nadie en esta playa not docl.msg saw om nobody prep demonstr.fsg beach ‘They didn’t see anyone on this beach’ (Limeño Standard Spanish, Mayer 2010: 96)12 12

See also Merma Molina (2007: 223).


Continuity and Innovation in Peruvian Spanish


Secondary topics in LSCV, as shown in (17a) and (17b), are referents that are more marked; they show [±DOM] and the featureless form lo, as part of object marking morphology: (17) a. Lo frío a la cebolla docl.msg fry (d)om det.fsg onion.fsg ‘I fry the onion’ (LSCV, Mayer 2010: 210) b. No, y no lo pronunciaba bien la erre no and no docl.msg pronounce well det.fsg r.fsg ‘No, and (s)he did not pronounce well the letter ’ (LSCV Mayer, fieldwork, 2006)

Secondary topics or TOP2 also show a very close relationship to the primary topic, and they are the central point of attention of the primary topic (see Bossong 2003 for his description of double actant marking). TOP2 are never new information (that is, they are not FOC), and they are the targets of several actions performed by the primary topic: they are activities, events or goals accompanied by body language (gestures for example); the attention shifts between primary and secondary topic with an intent to assure concentration on the object (the secondary topic) that is talked about, worked with, etc. The expression of secondary topics as illustrated above is attested in oral communication rather than in written language; it is very much part of oral language in contact. Independent justification for these pragmatic structures comes from extensive research in Peruvian Andean Spanish and Quechua (Weber 1989, 1993; CerrónPalomino 2003, among many others), where TOP2 marking can be linked to contact with Quechua. The next section provides some evidence of the role of contact in the general evolution of the pronominal system in this part of the Spanish-speaking world. 5. The role of contact The remaining question to answer is the role contact plays in relation to secondary topic marking in the Peruvian Andes and in extensive leísmo in Ecuadorian Andean Spanish. 5.1. Early attestations of contact effects on pronouns The literature on the contact effects on the pronominal system of Andean (Ecuadorian/Peruvian) Spanish is extensive, particularly on the simplification of 3rd person pronoun features (Palacios 1998, 2006; Cerrón Palomino 2003).


Elisabeth Mayer and Manuel Delicado Cantero

In the case of extensive use of leísmo with doubling, early evidence in texts produced by bilingual speakers is found in Guamán Poma’s Nueva corónica y buen gobierno (1615).13 Notice the use of le for direct object and with plural referents in (18a) and (18b); as well as the use of DOM with inanimate objects in (19c), where the direct object is a corpse: (18) a. y le mató a los susodichos dos ermanos Topa Amaro, Uari Tito Ynga and iocl.msg killed dom the said two brothers Topa Amaro, Uari Tito Ynka ‘And he killed the aforementioned two brothers Topa Amaro, Uari Tito Ynka’ b. Dizen que aquello le espantó al Ynga y a los Say that that iocl.msg frightened dom.the.msg Inka and dom det.mpl yndios questauan en los dichos baños de Caxamarca Indians that.were in the said baths of Cajamarca ‘They say that that frightened the Inka and the Indians that were in the aforementioned baths of Cajamarca’ c. Y ací le uenció y le prendió al cuerpo de Uascar Ynga and thus iocl.msg defeated and iocl.msg grabbed dom.det.msg body of Uascar Inka ‘And thus he defeated him and grabbed Uascar Inka’s body’

It is also possible to find early evidence of the special use of lo in texts written by bilingual speakers (Rivarola 2000): (19) a. an benido juntos, lo están en vna misma casa have come together docl.neut are in a same house ‘They have come together; they are in the same house’ (Memorial que presenta el fiscal Agustín Capcha, Nov 14th 1662, 17th c.) (Rivarola 2000: 77) b. y que a esto lo dixo su dicha mugger and that to this docl.neut said his said wife ‘And to this answered his wife’ (Memorial que presenta el fiscal Agustín Capcha, Nov 20th 1662, 17th c.) (Rivarola 2000: 85)

Equivalent examples of these special uses of clitics le(s) and lo(s) are attested in the chronicle written by the bilingual speaker Juan Santa Cruz Pachacuti (17th c.), particularly rich in contexts of doubling, which Palacios (1998: 137, 139) analyzes as evidence of the grammatical role of le/lo as agreement objective morphology. For Palacios the bilingualism of Santa Cruz Pachacuti (Quechua/Spanish) is an important cause behind the attested pronominal system, favored nevertheless by the pre-existing pronominal flexibility attested in old Spanish (Palacios 1998: 142). The transmission of Quechua patterns of object duplication morphology into Spanish by Quechua native speakers with Span The Guamán Poma Website: .


Continuity and Innovation in Peruvian Spanish


ish as L2 and the syntactic possibilities of the old language provide, according to Palacios (1998: 143), a basis for the initial steps in the simplification of the pronominal system of the Andes. In any case, contact was a necessary mechanism for this change. 5.2. Fossilization of lo in the Peruvian Andes The lack of 3rd person object clitics in Quechua is generally assumed to be the reason for the apparent free variation of 3rd person clitics in the Peruvian Andean region. Camacho and Sánchez (2002: 37) show in an OT framework that competing case in both grammars allows for permeability and transfer concluding that “dialects in contact with Quechua can simplify either in favor of the dative or the accusative”. On the other hand we have a fossilized clitic lo which we analyzed as secondary topic marker in the previous section. The comparison of the Quechua topic marker -qa and the TOP2 maker lo is given in Table 2: PROPERTIES



phonologically dependent specific locality conditions co-occurrence with case marker multiple occurrence grammatical agreement head-marking cataphoric reference no object drop

yes yes yes no yes yes yes yes

yes yes yes yes yes yes yes no

Table 2. Quechua and LSCV TOP makers

Adjacency conditions for both are specific and different as to the host category; however, both are restricted to main clause constituents only; -qa can attach to a number of syntactic categories which lo cannot use as a host; both can co-occur with a case marker. Both are part of grammatical agreement and, as cataphoric agreement, point towards the relevance of the subject and object and their relationship in the context. By that Weber (1989: 404) addresses “the set of propositions that the speaker assumes the hearer to know at the point at which he says the sentence.” Weber emphasizes the high frequency of use of -qa in folktales as opposed to a low one in personal narratives, ascribing this to the pointing effect when shifting speakers and events. This is the reason for the cataphoric reading in the chart.


Elisabeth Mayer and Manuel Delicado Cantero

The high degree of affinity between the Quechua topic marker -qa and the LSCV TOP2 marker lo as shown in Table 2 is illustrated in the following examples.14 Primary topics in (20) are morphologically marked and stay in situ. LSCV monotransitive clauses such as (17a) are very similar in extending DOM to topic inanimates, whereas standard Spanish restricts DOM to specific and animate objects except for topicalization purposes: (20) Wasi hunt’s-ta riku-ni house full.acc see.1sg ‘I see a full house’ (Lefebvre and Muysken 1988: 105)

Co-occurrence of the TOP marker -qa and the object maker -ta is restricted to main clause constituents in (21): (21) Hatun wasi-ta-qa muna-: big house-acc-top want.1sg ‘I want a big house’ (Weber 1989: 395)

Secondary topics as in (22) can appear either on the right or on the left periphery of the verb, and co-occurrence of a co-indexed lexical pronoun and the case marker -ta in the same clause is constrained to the pronoun in topic position. This is very similar to LSCV TOP2 marking and can be linked to double actant marking (Bossong 2003): (22) Hamu-q warma-(ta)-qa, Santiyagu riku-n come.a girl.(acc).top Santiago see.3 ‘Santiago sees the girl that is coming’ (Lefebvre and Muysken 1988: 138)

The typological differences, and even more so, the similarities of both languages are of particular importance with regard to argument marking and potential transfer via Andean Spanish to LSCV. Hence, we can assume that prolonged contact has facilitated innovative structures through lexical and semantic borrowings, yielding a specialized pragmatic strategy appropriate for communication in a multilingual environment.


For the compilation of the Quechua parameters we relied on Weber’s (1989) grammar of Huallaga (Huánuco) Quechua.

Continuity and Innovation in Peruvian Spanish


5.3. Extensive leísmo under contact The main extra-linguistic difference between Peru and Ecuador in relation to contact is the fact that Cuzco, home to administrative offices and as such to highly educated standard Spanish speakers, was exposed to European Spanish and Andean Spanish for centuries. This was not the case in Ecuador; however, in both countries contact with Spanish was mainly mediated by Andean Spanish.15 Under extensive leísmo in Ecuadorian Andean Spanish the theme object in (23a) receives primary and secondary agreement16 and fulfills all the requirements for primary objecthood including passivization in (23b): (23) a. Ya le veo a la camioneta already see om det.fsg minibus.fsg ‘I can already see the minibus’ (Ecuador, Suñer and Yépez 1988: 512) b. La camioneta es vista (por mí) det.fsg minibus.fsg is.3sg seen (prep me) ‘The minibus is seen by me’ (Haboud, pc)

These apparent ungrammatical phenomena are based on two concurrent and independent processes: the constant evolution of DOM and the reanalysis process of clitics, as visualized in table 3: STAGE












lo/le/ → le


Table 3. Three stages of DOM and CLD evolution

Standard Spanish is representative for stage I. Liberal clitic doubling varieties (non-contact varieties such as River Plate and Standard Limeño) for stage II. The final stage III shows the transient (and maybe fossilized) state of nonstandardized contact varieties such as LSCV and also non-standard River Plate. Stage III also shows the struggle for primary object status through TOP2 mark According to Muysken (1984), the most plausible explanation for the lack of fossilized lo in Ecuador is a crucial syntactic difference in subject-predicate agreement between Southern Quechua spoken in Peru and Ecuadorian Quichua. Whereas Southern Quechua shows agreement, Ecuadorian Quichua lacks it. 16 Primary agreement here refers to DOM and secondary agreement to CLD. 15


Elisabeth Mayer and Manuel Delicado Cantero

ing and the co-variation of leísmo and strange lo. The arrows point to varieties exhibiting extensive leísmo, such as Quiteño and Paraguayan Spanish, where the single case clitic paradigm signals completed change. 6. Conclusion The variation found in non-standardized varieties of Peruvian contact Spanish shows a combination of archaic features, continuing on from old Spanish, and “peculiar” innovations which can be related to extra-linguistic factors such as remoteness, high levels of illiteracy and under-education. The intra-linguistic factors involved include the well-known diachronic variation of the clitic paradigm (the struggle between DAT and ACC), and extended DOM regulated by pragmatic strategies taking precedence over semantic strategies (as opposed to River Plate Spanish and other such varieties, where semantic restrictions are very much in place). The effects of primary contact between Spanish and Amerindian languages gave rise to Andean Spanish. Today’s Peruvian Spanish contact varieties are the result of additional contact between Andean Spanish and Standard Peruvian Spanish. In terms of morphosyntax, clitics turned into morphological agreement markers (via feature loss) have evolved into topicality and transitivity markers (mere object markers, thus [-referential]) denoting secondary topics. Furthermore, CLD as a secondary topic/OM construction does not necessarily show a, which is optional once the ‘clitic’ becomes the OM and thus makes DOM a redundant. In this sense, we find that, contra Kayne (1975), there is no need for DOM a for doubling; in contrast, “regular” CLD falls under Kayne’s generalization. REFERENCES Allen, C. (1995). Case-Marking and Reanalysis: Grammatical Relations from Old to Early Modern English. Oxford: Oxford University Press. Allen, C. (2000). “On the development of a friend of mine”. In T. Fanego, Teresa, M. J. López-Couso, and J. Pérez Guerra (eds.), English Historical Syntax and Morphology. Selected papers from 11 ICEHL. Amsterdam and Philadelphia: John Benjamins, pp. 23-41. Alsina, A. (1996). The Role of Argument Structure in Grammar. Evidence from Romance. Stanford: CSLI. Bello, A. (1984). Gramática de la Lengua Castellana. Madrid: EDAF. Bresnan, J. (1998). “Morphology Competes with Syntax: Explaining Typological Variation in Weak Crossover Effects”. In P. Barbosa, D. Fox, P. Hagstrom, M. McGinnis,

Continuity and Innovation in Peruvian Spanish


and D. Pesetsky (eds.), Is the Best Good Enough? Optimality and Competition in Syntax. Cambridge, MA: MIT Press and MIT Working Papers in Linguistics, pp. 59-92. Bresnan, J. (2001). Lexical-Functional Syntax. Oxford: Blackwell Publishers. Bossong, G. (1991). “Differential Object Marking in Romance and Beyond”. In D. Wanner, and D. A. Kibbee (eds.), New Analyses in Romance Linguistics. Amsterdam/ Philadelphia: John Benjamins, pp. 143-185. Bossong, G. (2003). “Nominal and/or Verbal Marking of Central Actants”. In G. Fiorentino (ed.), Romance Objects. Transitivity in Romance Languages. Berlin: Mouton de Gruyter, pp. 17-49. Camacho, J./Sánchez, L. (2002). “Explaining Clitic Variation in Spanish”. In M. Amberber, and P. Collins (eds.), Language Universals and Variation. Westport, CT: Praeger Publishers, pp. 21-41. Cerrón-Palomino, R. (2003). Castellano Andino. Aspectos sociolingüísticos, pedagógicos y gramaticales. Lima: Pontificia Universidad Católica del Perú. Company, C. (2001). “Multiple Dative-Marking Grammaticalization. Spanish as a Special Kind of Primary Object Language”. In Studies in Language 25, 1, pp. 1-47. Company, C. (2003). “Transitivity and Grammaticalization of Object. The Struggle of Direct and Indirect Object in Spanish”. In G. Fiorentino (ed.), Romance Objects. Transitivity in Romance Languages. Berlin: Mouton de Gruyter, pp. 217-260. Comrie, B. (2003). “When agreement gets trigger-happy”. In Transactions of the Philological Society 101, 2, pp. 313-337. Croft, W. (1991). Syntactic Categories and Grammatical Relations. Chicago: University of Chicago Press. Croft, W. (2000). Explaining Language Change: An Evolutionary Approach. London: Longman. Dalrymple, M./Nikolaeva, I. (2007): Topicality and Nonsubject marking: Agreement, Casemarking and Grammatical Function. ms. Oxford University. Estigarribia, B. (2005). “Why Clitic Doubling? A Functional Analysis for Rioplatense Spanish”. In M. Butt, and T. Holloway King (eds.), Proceedings of the LFG05 Conference. Stanford, CA: CSLI Publications, pp. 116-135. Fernández-Ordóñez, I. (1993). “Leísmo, laísmo y loísmo: estado de la cuestión”. In O. Fernández Soriano (ed.): Los pronombres átonos. Madrid: Taurus, pp. 63-96. Fernández-Ordóñez, I. (1999). “Leísmo, laísmo y loísmo”. In: I. Bosque, and V. Demonte (eds.), Gramática descriptiva de la lengua española. Madrid: Espasa Calpe, pp. 1317-1397. Franco, J. A. (2000). “Agreement as a Continuum. The Case of Spanish Pronominal Clitics”. In F. Beukema, and M. den Dikken (eds.), Clitic Phenomena in European Languages. Amsterdam/Philadelphia: John Benjamins, pp. 147-190. Givón, T. (1976). “Topic, Pronoun, and Grammatical Agreement”. In C. Li (ed.), Subject and Topic. New York: Academic Press, pp. 149-188. Givón, T. (ed.) (1983). Topic Continuity in Discourse: A Quantitative Cross-Language Study. Amsterdam/Philadelphia: John Benjamins. Greenberg, J. H. (ed.) (1966). Universals of Language. 2nd edition. Cambridge, MA: MIT Press.


Elisabeth Mayer and Manuel Delicado Cantero

Guamán Poma, F. (1615). Nueva corónica y buen gobierno, available at: permalink/2006/poma/info/en/frontpage.htm Haboud, M. (1998). Quichua y castellano en los Andes ecuatorianos. Los efectos de un contacto prolongado. Quito: Abya-Yala. von Heusinger, K./Kaiser, G. (2005). “The evolution of differential object marking in Spanish”. In K. von Heusinger et al. (eds.), Proceedings of the Workshop “Specificity and the Evolution/Emergence of Nominal Determination Systems in Romance”, Konstanz: Universität Konstanz, pp. 33-70. Joseph, B./Janda, R. (eds.) (2003). The Handbook of Historical Linguistics. Malden, MA: Blackwell. Kayne, R. (1975). French Syntax. The Transformational Cycle. Cambridge, MA: The MIT Press. Kiparsky, P. (1998). “Partitive case and aspect”. In M. Butt, and W. Geuder (eds.), Projecting from the Lexicon. Stanford: CSLI, pp. 265-307. K roch , Anthony (1994). “Morphosyntactic Variation”. In K. Beals et al. (eds.), Papers from the 30th Regional meeting of the Chicago Linguistics Society: Parasession on Variation and Linguistic Theory. Available at: http://www.ling.upenn. edu/~kroch/ Laca, B. (2006). “El objeto directo. La marcación preposicional”. In C. Company Company (ed.), Sintaxis histórica de la lengua española. Vol. 1. México DF: Universidad Nacional Autónoma de México, pp. 421-475. Lambrecht, K. (1994). Information Structure and Sentence Form. Topic, Focus and the Mental Representations of Discourse Referents. Cambridge: Cambridge University Press. Lapesa, R. (2000). Estudios de morfosintaxis histórica del español. Madrid: Gredos. Lefebvre, C./Muysken, P. (1988). Mixed Categories. Nominalizations in Quechua. Studies in Natural Language and Linguistic Theory. Dordrecht: Kluwer Academic Publishers. Mayer, E. (2003). Clitic Doubling in Limeño. A Case Study in LFG. MA Ling thesis, Canberra, Australian National University. Mayer, E. (2006). “Optional Direct Object Clitic Doubling in Limeño Spanish”. In M. Butt, and T. Holloway King (eds.), Proceedings of the LFG06 Conference. Stanford: CSLI Publications, pp. 328-342. Mayer, E. (2008). “Clitics on the Move: from Dependent Marking to Split Marking”. In M. Butt, and T Holloway King (eds.), Proceedings of the LFG08 Conference. Stanford: CSLI Publications, pp. 352-372. Mayer, E. (2010). Syntactic variation of object arguments in Limeño Spanish contact varieties. PhD dissertation, Canberra, Australian National University. Melis, C. (1995). “El objeto directo personal en el Cantar de Mío Cid: estudio sintácticopragmático”. In: C. Pensado (ed.), El complemento directo preposicional. Madrid: Visor, pp. 133-163. Merma Molina, G. (2007). Contacto lingüístico entre el español y el Quechua: un enfoque cognitivo-pragmático de las transferencias morfosintácticas en el español andino peruano. PhD dissertation, Universidad de Alicante.

Continuity and Innovation in Peruvian Spanish


Muysken, P. (1984). “The Spanish that Quechua Speakers Learn: L2 Learning as Normgoverned Behaviour”. In R. W. Andersen (ed.), Second Languages: a Crosslinguistic Perspective. Rowley, MA: Newbury House, pp. 101-119. Nichols, J. (1986). “Head-Marking and Dependent-Marking Grammar”. In Language 62, 1, pp. 56-119. Nikolaeva, I. (2001). “Secondary Topic as a Relation in Information Structure”. In Linguistics 39, pp. 1-49. Palacios, A. (1998). “Santacruz Pachacuti y la falsa pronominalización del español andino”. In Lexis 22, 2, pp. 119-146. Palacios, A. (2002). “Leísmo y loísmo en el español ecuatoriano: el sistema pronominal del español andino”. In E. Hopkins Rodríguez (ed.), Homenaje al Dr. Luis Jaime Cisneros, vol. I. Lima: Pontificia Universidad Católica del Perú, pp. 389-408. Palacios, A. (2005). “Aspectos teóricos y metodológicos del contacto de lenguas: el sistema pronominal del español en áreas de contacto con lenguas amerindias”. In: V. Noll, K. Zimmermann, and I. Neumann-Holzschuh (eds.). El español en América: Aspectos teóricos, particularidades, contactos. Madrid/Frankfurt: Iberoamericana/ Vervuert, pp. 63-94. Palacios, A. (2006). “Cambios inducidos por contacto en el español de la sierra ecuatoriana: la simplificación de los sistemas pronominales (procesos de neutralización y elisión)”. In Tópicos del seminario 15, pp. 197-229. Pensado, C. (1995). “La creación del complemento directo preposicional y la flexión de los pronombres personales en las lenguas románicas”. In C. Pensado (ed.), El complemento directo preposicional. Madrid: Visor, pp. 179-233. Pensado, C. (ed.) (1995). El complemento directo preposicional. Madrid: Visor Real Academia Española (1985). Esbozo de una nueva gramática de la lengua española. Madrid: Espasa-Calpe. Ritter, E./Rosen, S.T. (2001). “The Interpretive Value of Object Agreement”. In Language Sciences 23, 425-451. Rivarola, J. L. (2000). Español andino. Textos bilingües de los siglos XVI y XVII. Madrid/Frankfurt: Iberoamericana/Vervuert. Rivero, M. L. (1986). “Parameters in the Typology of Clitics in Romance and Old Spanish”. In Language 62, pp. 774-807. Sánchez, L. (2005). “Clitic-Doubling and the Checking of Focus”. Rutgers University, unpublished manuscript. Available at: Silva-Corvalán, C. (ed.) (1995). Spanish in Four Continents: Studies in Language Contact and Bilingualism. Washington, D.C.: Georgetown University Press. Suñer, M. (1988). “The role of agreement in clitic-doubled constructions”. In Natural Language and Linguistics Theory 6, 3, pp. 391-434. Suñer, M./Yépez, M. (1988). “Null Definite Objects in Quiteño”. In Linguistic Inquiry 19, pp. 511-519. Torrego, E. (1999). “El complemento directo preposicional”. In I. Bosque, and V. Demonte (eds.), Gramática descriptiva de la lengua española. Madrid: Espasa Calpe, pp. 1779-1806.


Elisabeth Mayer and Manuel Delicado Cantero

Weber, D. J. (1989). A Grammar of Huallaga (Huánuco) Quechua. Berkeley: University of California Press. Weber, D. J. (1993). The Binding Properties of Quechua Suffixes. Work Papers of the Summer Institute of Linguistics vol. 37. Grand Forks, ND: University of North Dakota.


1. Introduction Baure and Paunaka are two Bolivian Arawakan languages, both seriously endangered and currently being documented by the authors. Both languages have undergone long-term Spanish contact since the colonization of Bolivia and for both linguistic communities Spanish is the dominant means of communication. Nevertheless, we find important differences between the two languages when it comes to strategies of clause combining. While in Baure, TMA affixes on the predicates provide clues on how the semantic connection between clauses is to be understood, Paunaka hardly makes use of this strategy. Nominalization plays a role to mark subordinate verbs in Baure, whereas in Paunaka there is a specialized verbal affix to indicate subordinate verbs. And last, but not least, Paunaka frequently employs connectives borrowed from Spanish, where Baure uses native particles and applies them on a Spanish syntactic model. In this paper, we compare the different strategies used in the two languages. We argue that the loss of TMA marking on the predicates, the frequent employment of connectives, and most obviously, the use of borrowed connectives in Paunaka is due to contact-induced change. We then try to give an explanation as to why two very closely related languages in a similar contact scenario re-structured their systems of clause combining strategies differently. This chapter is structured as follows: first, the sociolinguistic settings of Baure and Paunaka are described and the general terminology concerning clause combining is introduced (Section 2). In Section 3, the different clause combining patterns found in Baure are illustrated and compared, with respect to more native constructions (3.1) that are still in use today, and clause combining constructions that have been modelled on Spanish syntax (3.2). In 3.3, the role of Spanish contact on Baure is summarized. In Section 4, Paunaka clause combining constructions are illustrated, again first focusing on native patterns (4.1) and then moving to construction borrowed from Spanish (4.2). The process of continuing replacement of


Swintha Danielsen and Lena Terhart

native patterns in Paunaka by contact-induced patterns is explained in 4.3. Finally, section 5, summarizes the similarities found in Baure and Paunaka clause combining patterns (5.1) and tries to give some explanations for the diverging pictures we have of these two closely related languages (5.2). 2. Baure and Paunaka 2.1 The Baure language Baure is a Southern Arawakan language spoken in Bolivian Amazonia, in the town of Baures, department Beni, close to the border with Brazil. To be more precise, the name Baure serves as a cover term for the three varieties of what should more properly be called the Baure language group, which consists of the extinct Joaquiniano language, the almost extinct Carmelito dialect (with four speakers), and the Baure dialect (with 54 speakers and semi-speakers). In this paper, only the Baure dialect is considered. It is severely endangered, being spoken only by the grandparental generation (cf. Krauss 2007:6) and having no active transmission. However, in the past few years and with the assistance of the Baure Documentation Project, children have begun to learn Baure again, in school, in special language classes, or from their grandparents.1 Baure is most closely related to the other Bolivian Southern Arawakan languages of the Moxo group (Trinitario and Ignaciano) and Paunaka. Languages in the surroundings of Baures were Itonama (Isolate) and Siriono (TupiGuaraní), but contact with these groups ceased in the first half of the 20th century at its latest. Baures was founded by the Jesuits in 1702, and thus, the Baure language has been in contact with Spanish for more than three centuries. Many grammatical characteristics result from Spanish language contact on Baure, among others, preverbal particles (cf. Danielsen 2015), the loss of complex verb structure, and also the combination of clauses by means of conjunction-like particles. Typologically, Baure is a polysynthetic, agglutinating, mainly head-marking language with an extensive but only partly grammaticalized classifier system (cf. Terhart 2009). Baure is most complex in its verbal morphology, as example (1) shows: (1)



pimowanoeyoworon tech pitir koromok? pi=imo-wano-iy-wo=ro-no tech pitir koromok 2sg=put-dep-loc-cop=3sgm-nmlz dem.m 2sg.poss cloth ‘Where did you leave your cloth (on passing)?’ (DC-060322S) 2

The Baure Documentation Project was financed by DobeS (Documentation of endangered languages ‒ VW foundation) and was running from 2008 till 2013. For details on the teaching efforts and materials developed within the project, cf. Examples have a reference to the recording in the corpus. The abbreviations for the Baure data

Borrowed Clause Combining Patterns in Two Arawakan Languages


Verbal predicates show obligatory SA marking by a personal proclitic and pronominalizing O marking (of up to two objects) by enclitics. Subject marking may be dropped in relative clause constructions, which are marked by the nominalization of the verb and do not need any subject marking. Non-verbal predicates are stative and mark their SO by a personal enclitic, a phenomenon generally referred to as split-S marking.3 Ditransitive verbs are generally derived by some kind of applicative affix (general applicative, causative, or benefactive). Verbal morphology furthermore includes a variety of aspect and mood markers, whereas tense is unmarked. The basic word order is VSO with nominal subjects and SVO with pronominal subjects. Nouns are not as morphologically complex as verbs, but they are frequently used as predicates with verbal morphology, as in (2) with the possessed noun vimoestar ‘our teacher’ being used as a predicate (with SO marking). (2) tin eton nga vimoestarwori. tin eton nga vi=moestar-wo=ri dem.f woman neg 1pl=teacher-cop=3sgf ‘That woman isn’t our teacher.’ (JI-030822S)

Data for this chapter have been taken from Baure recordings collected in the field by Danielsen and her team since 2003. In addition, linguistic material from the 18th century is cited for comparison. This material is referred to as “historical Baure” (Magio [1749]; Asis Coparcari [1767], published in Adam and Leclerc 1880). 2.2 The Paunaka language Paunaka is a Southern Arawakan language spoken in the Chiquitanía in eastern Bolivia, south of humid Amazonia and north of the dry Gran Chaco (see Danielsen and Terhart 2014). There are only ten speakers and semi-speakers left, who live in and around Concepción de Chiquitos. The youngest speaker is around 50 years of age, and the language is no longer taught to the children; thus Paunaka can be classified as critically endangered (cf. Krauss 2007: 6). All speakers are trilingual with Paunaka, Spanish and Bésɨro (or Chiquitano)4. Chiquitano developed into a lengua general

3 4

contain the following information: two-letter code for the speaker(s) – genre abbreviation (no abbreviation = elicitation) – date (yymmdd) – interviewer(s). The references for Paunaka data are similar, only that the initial abbreviation is always a three-letter code, with a one-letter abbreviation for the speaker(s) and possibly x as a placeholder; there is always a genre abbreviation, also for elicitation. This has been discussed in detail in Danielsen and Granadillo 2008. Bésɨro is the name that some Chiquitano people use to refer to their language, whereas Chiquitano is the Spanish denomination. It is only used as a self-denomination regionally, (e.g. by people in and around Concepción, where the Paunaka data come from) whereas in other regions speakers prefer the Spanish name. The language probably belongs to the Marco-Jê linguistic family, according to Sans (2010: 3).


Swintha Danielsen and Lena Terhart

during the period of the Jesuit missions in the Chiquitanía (1672-1767), when Chiquitano speakers5 were in the majority in the multilingual mission towns (Tomichá 2002: 278, 291). Today, the Bésɨro variety is the dominant indigenous language of the region, although Spanish has certainly taken over the role of the most important language. The Paunaka language is currently under investigation within the Paunaka Documentation Project.6 Paunaka is genetically most closely related to the Moxo and Baure languages. It is a synthetic and agglutinating language, predominantly headmarking, but with some dependent-marking characteristics. Paunaka has a classifier system with a reduced productivity. It has a split S-marking system that differentiates between verbal and nominal predicates and is similar to the Baure one described in Section 1.1. The basic word order is SVO if both arguments are expressed by NPs; however, clauses with only one argument expressed by an NP are found more frequently, and those clauses are usually verb-initial. The main word classes are nouns and verbs. Verbs can be fairly complex, with affixes for aspect and mood, but most of the times, the verb structure is simple. The only categories being obligatorily marked on verbs are subject person and reality status (cf. Elliott 2000), i.e. realis vs. irrealis. The subject is marked by a prefix. One object may be marked by a suffix. To generalize, there are two 3rd person prefixes, chi- is used on transitive verbs when a 3rd person acts on a human or emphasized 3rd person, and ti- is used in most other cases. Generally, the verbal irrealis is marked by changing the last /u/ of the stem or of a following suffix to /a/, while the realis is marked by absence of /a/, see example (3). A few verbs show an irrealis prefix a-, instead. Irrealis can be marked on other word classes, too, with a suffix -ina (which is also lexicalized in the negative particle kuina, in (3b)). (3) a. niniku b. kuina ninika ni-niku kuina ni-nika 1sg-eat neg 1sg-eat.irr ‘I eat’ ‘I don’t eat’

One reason for a simpler verb structure, compared to other Southern Arawakan languages, is that many obliques are not marked by applicatives, but with adposition-like particles outside the verb, like the beneficiary of an action in example (4).



There were several varieties of Chiquitano by the time the missions were established (see list of ethnic groups and languages in Matienzo et al. 2011: 433-441) and in the different mission towns different dialects may have been spoken by that time. Nevertheless, the Jesuits’ language policy during that time was one of equalling dialectal differences (cf. Saito 2009). The differences that we find in spoken Chiquitano today may well be of a later development instead of reflecting the varieties that existed in the 17th century. The Paunaka Documentation Project (running from 2011 – 2013) was financed by the Endangered Languages Documentation Programme of the Hans Rausing Endangered Languages Project, funded by Arcadia. For details on the project, cf.

Borrowed Clause Combining Patterns in Two Arawakan Languages


(4) binejika eka ajumerku pitÿpi bi-nejika eka ajumerku pi-tÿpi 1pl-leave.irr dem paper 2sg-ben ‘We will leave you this page.’ (mux-c110810l)

2.3 Clause combining terminology The terminology on clause linking is sometimes a bit confusing, and we therefore want to make some general statements on the terminology we apply in this chapter. We use the term juxtaposition of elements or clauses, when there is no syntactic linker involved and both clauses can be regarded as complete and work on the same level. Thus, there is no syntactic dependency. Nonetheless, such clauses may have a semantic relation of subordination. We have tried to make these distinctions clear where relevant. Juxtaposition clauses could also be addressed as “coordination”, but we try to avoid this term where unnecessary. In general, (syntactic) subordination has to be marked by either a connective (conjunction) or some subordinating morphological affix. The distinction between juxtaposition and subordination is blurred where certain verbal suffixes may well indicate a semantic relation (of subordination) of two clauses, but the same affixes can also be applied without the goal of subordination. One example is the Baure irrealis suffix -sha, which can be used to mark conditional clauses, as C1 given in (5):7 C1 C2 (5) piki’inasha nipapi to peroserokoch. pi=ki’in-a-sha ni=pa=pi to pi=eroserokoch 2sg=want-lk-irr 1sg=give=2sg art 2sg=lasso ‘If you want, I give you your lasso(es).’ (RP-N030929S-2)

The verb piki’inasha can be translated as ‘if you want’, but also as ‘whenever you want’, ‘do you want?’, as in asking politely for a favor. The verb can be used as an independent clause, since it has person marking and is a fully inflected form, so that it is difficult to argue for syntactic dependency and subordination here. We would therefore conclude that these morphemes can occur in clause juxtaposition, but we may also address the clause complex as subordinating. Furthermore, subordination should be distinguished from embedding. Embedding means that a clause is turned into a constituent of the clause, be it a main argument or a modifier within an NP referring to an argument in the clause. The occurrence of marked predicates in what is translated as adverbial clauses may then also be regarded to be embedding as oblique arguments. Here we are again confronted with borderline cases, because thus marked clauses may also be used as independent clauses. We 7

The abbreviation “C” stands for clause here.


Swintha Danielsen and Lena Terhart

thus only speak of embedding in cases of morphosyntactic marking of originally main predicates as either S or O arguments − including complements − in the clause or part of the NPs referring to the core arguments.8 Adverbial clauses are consequently not embedded but subordinated clauses. The terms clause combining and clause linking are used interchangeably throughout the chapter and serve as a hypernym of all kinds of combinations of two clauses. 3. Clause combining in Baure Baure has a number of different constructions for the combining of two or more clauses. Since we also have historical Baure data from the 18th and 19th century (Magio 1880 [1749]; Asis Coparcari 1880 [1767]; Cardús 1886), we can identify constructions that may have spread because of the long-term contact with the Spanish language. Details on Baure clause combining are given in Danielsen (2011) and will be summarized and elaborated with a different focus in 3.1. In section 3.2, we highlight those constructions that may well have existed in Baure but which show the influence of language contact with Spanish. The process of the increasing spread of contact-induced constructions in Baure is summarized in 3.3. Baure is a very verby language, and many meanings that we are used to see being expressed by some kind of particle, like conjunctions or adpositions, are in fact verbal constructions in Baure. As an example we want to point out a construction that came out of use in current Baure in (6). Here we are dealing with clause juxtaposition, the first clause being ‘I killed the jaguar’, and the juxtaposed clause ‘I used an arrow’. The whole utterance could also be analyzed as an instance of semantic subordination, especially when considering that the predicate of the second clause may refer to the same event as the one in the first clause. The semantically subordinate but syntactically juxtaposed second clause is translatable as a prepositional phrase into English ‘with an arrow’9 or Spanish con una flecha. (6)

For more discussion on the specific details of clause combining, please cf. Danielsen (2007: 381ff.). 9 Alternatively, in English we could also have a gerund construction ‘using an arrow’. 10 The orthography of the original has been adapted here for the matter of comparability. 8

ndi’ niyoka te chuwuikana ninaw te kariraka.10 [historical Baure] ndi’ ni=yoka te chuwuikana ni=ina-wo te karinaka 1sg 1sg=pierce dem.m jaguar 1sg=use-cop dem.m arrow ‘I killed the jaguar with an arrow (lit. I killed the jaguar, I used an arrow).’ (Cardús 1886: 319)

Borrowed Clause Combining Patterns in Two Arawakan Languages


The verb -ina- ‘use’ seems to be cognate with an applicative suffix -ino/-ina of historical sources, where instrumental was one of its functions. Nowadays, Baure has a benefactive suffix in the strict sense -ino ‘ben’, and this coexists with the verb -ina- ‘use’, which is hardly used for clause linking like the one in (6). Likewise, comparative constructions are constructed by means of verbs11, cf. (7): (7) pinokowoni piti’ mejowokon. pi=inoko-wo=ni piti’ mejewokon 2sg=resemble-cop=1sg 2sg bad ‘You resemble me in being bad (lit. You resemble me, you are bad).’ (GP_LO-040801S)

Our focus on the verbiness of Baure shows that it can generally be presumed that most clause linking consisted of juxtaposition of predicate phrases. 3.1 Native clause combining constructions in Baure The typical clause combining constructions in Baure include embedding, subordination, and the juxtaposition of clauses. The lone marking of coordination or subordination by some conjunction-like particle (connective) is less typical and will be referred to in 3.2. Embedded clauses consist of three different derivation types (two types of relative clauses and one complement clause construction). One example of a relative clause and one of a complement clause are given in (8) and (9), respectively12. In example (8), the embedded clause nikon chindinev ‘who eats people’ is marked by the nominalized predicate. In addition, the predicate lacks subject marking, which should be ri= ‘3sgf’ in congruence to marip ‘witch’. Subject marking is obligatory on verbal predicates in main clauses, but it is dropped in modifying relative clauses in which the subject is identical with the head. In these relative clauses, however, the verbal predicate is nominalized, and therefore also has a different syntactic status than a bare verb. The predicate nikon ‘eater (of)’ has been embedded as a modifier that is part of the subject NP of the main clause, the witch. [ C1 [ C2 ] ] (8) kwe’ tin marip nikon chindinev. kwe’ tin marip niko-no chindi-nev exist dem.f witch eat-nmlz person-pl ‘There is that witch who eats people.’


In (9) the complement clause is marked by the participle constructions piwoiykoch shep ‘your making of manioc flour’. The complement clause always has subject There is one verb -ino(k)- ‘be like, resemble’ and another one -ero- ‘be more than, surpass’. The square brackets denote the limits of the clause.




Swintha Danielsen and Lena Terhart

marking, but as (9) illustrates, the subject may be a different one in the complement clause than in the matrix clause. (9)

[ C1 [ C2 ] ] noka nijinokowovi piwoiykoch shep. noka ni=jinoko-wo=pi pi=woyiko-cho shep neg 1sg=see-cop=2sg 2sg=make-ptcp manioc.flour ‘I haven’t seen you making manioc flour (lit. your making of manioc flour).’ (HC-060308S)

In the described constructions, no conjunctions or clause introducing particles are involved. The clause combinations, rather, have in common the fact that the clauses are embedded into the matrix clause by means of derivational morphology. Participles can also be used for various adverbial clauses, similar to the English gerund derived by -ing. If the participle is not a complement, it is part of a syntactically marked subordinate clause, as in (10): (10) jeni, tiow te vinanaw te vejmoekchow. jeni tiow te vi=ina~na-wo te vi=ejmoek-cho-wo yes cleft dem1m 1pl=use~ints-cop dem1m 1pl=wash.clothes-ptcp-cop ‘yes, this is what we use for washing clothes (the lard they make the soap of)’ (HC-D120509S-1.056)

Other adverbial clauses are marked by one of the specific suffixes for subordination in Baure: -pi ‘qnmlz’ for causal clauses (‘because of’), -iyo ‘loc’ for locative clauses (‘where’), and -ro ‘temp’ for temporal backgrounding clauses (‘when’ (in the past)). The locative suffix is not only productive for locative subordination, as in (11), but it is mostly used for interrogative clauses asking for the place of an event. Thus, the same marked predicate clause (C2 in (11)) can be an interrogative clause with the rising intonation. [ C1 [C2 ] ] (11) te noemonoporeiy noiy noemonoekiyow kanikon. te no=imono-poreiy noiy no=imonoek-iy-wo kanikon dem.m 3pl=buy-rep there 3pl=sell-loc-cop food ‘They will buy again there where they are selling food.’ (AD-090807S)

All of the described marked clauses, embedded and subordinate, can also be introduced by a demonstrative pronoun or the sequence to ka ‘that which’, both of which seem to function as a relative pronoun or complementizer, see example (12):13

The abbreviation “rel” stands for relativizer here.


Borrowed Clause Combining Patterns in Two Arawakan Languages rel [ C1 (12) moej nipeiy tekow to ka moej ni=pa=yi tekow to ka cert 1sg=give=2pl all art ind ‘I will certainly give you all that you need.’


[ C2 ] ] yikamoseronow. yi=kamosero-no-wo 2pl=need-nmlz-cop (SIL14-B1966)

As mentioned above, in Baure, clauses are frequently juxtaposed without any connecting element. Clause juxtaposition is predominantly used for coordination and refers by default to the chronological listing of events of the type veni vidi vici. One example of such an asyndetic coordination is given in (13): C1 (13) ver noti’ noshim noiy ver noti’ no=shim noiy prf 3pl 3pl=arrive there

C2 novejshachow, no=vejshacho-wo 3pl=change.clothes-cop

C3 nokachow nojaviak, no=kacho-wo no=javiak 3pl=go-cop 3pl=swim

C4 nojaviakow. no=javiako-wo 3pl=swim-cop ‘They arrived there and changed clothes, went to swim, and swam.’ (RP-N030917S)

The clauses in (13) are all independent, and the temporal order is derived from the chronological order of the expressed events. However, the juxtaposition of clauses may also have some other interpretations, depending on the morphology found on the predicates in the clauses. Examples of suffixes causing such effects are -pa ‘intl’, -sha ‘irr’, and -wana ‘dep’. Syntactically, these clauses are independent, only semantically there may be some kind of subordination. In clause juxtaposition, the second event can be interpreted as a purposive or consecutive clause if the predicate is marked by -pa ‘intl’ (14). That the marking by -pa ‘intl’ is not necessary to evoke such a reading, is shown in a contrastive example (15): C1 C2 (14) rikachpow wapoeri-ye riropa in. ri=kach-po-wo wapoeri-ye ri=er-pa in 3sgf=go-pfv-cop river-loc 3sgf=drink-intl water ‘She went to the river in order to drink water.’ (GP-N040801S.020) C1 C2 (15) ne’ nivia’ te notiapes pinik. ne’ ni=via’ te notiapes pi=nik here 1sg=take.away dem1m 1sg.calf 2sg=eat ‘Here, I take some of my calf (i.e. cut off) so that you eat.’ (RP-N030917S.307)


Swintha Danielsen and Lena Terhart

The suffix -sha ‘irr’ can mark conditional or hypothetical clauses, as shown in example (5) above. The departitive suffix -wana ‘dep’ marks an event that is done during or after leaving. It seems to have been particularly important for the marking of the temporal order of events, as illustrated in (16), but it is not used very frequently anymore: C1 C2 (16) ti rikachpow wapoeri-ye, riviawana tech porespa’. ti ri=kach-po-wo wapoeri-ye ri=via-wana tech porespa’ dem.f 3sgf=go-pfv-cop river-loc 3sgf=take-dep dem.m mate ‘She went to the river, taking a mate pumpkin on her way.’14 or ‘She went to the river after she had taken a mate.’ (GP-N040801S.019)

So, even though the strategies of clause embedding and (semantic) subordination are still existent in the Baure language today, we can note a decline from historical data and data collected by the SIL15linguists in the 1950s (unpublished), of the use of these morphological constructions. Some former strategies in Baure clause or phrase combinations have also been lost, such as the application of a coordinative or iterative suffix -piro ‘and, also’ in historical Baure, which could combine NPs in coordination (18a), but also two predicate clauses, as in (17). We do not have any evidence of further coordinative particles in early historical Baure, as we find them in the language today. C1 C2 (17) piyono pishimapiro pi=yono pi=shima-piro 2sg=go 2sg=arrive.irr-it ‘you go and come back’

[historical Baure] (Magio 1880 [1749]: 48)

Nowadays, Baure uses the connective ach ‘and’ for the coordination of two phrases or clauses, or for the general connection of an utterance to the preceding discourse. In (18), the contrast between the two phrase combining strategies is demonstrated, and in the following section, connectives will be addressed in more detail. (18) a. historical Baure in Magio (1880 [1749]: 28) b. Baure today: niti pitiapiro ndi’ ach piti’ niti piti-a-piro ndi’ ach piti’ 1sg 2sg-lk-it 1sg and 2sg ‘I and you’ ‘I and you’ 14 15

The “mate” is a kind of pumpkin used as a container for water. SIL = Summer Institute of Linguistics. SIL linguists are missionaries whose aim is to translate the Bible into all languages of the world. 

Borrowed Clause Combining Patterns in Two Arawakan Languages


The first evidence for the usage of ach ‘and’ in historical data comes from Cardús (1886: 319), where it is used to coordinate two NPs in a translation from Spanish. It seems that there also existed some clause marking particles or (bound) morphemes in historical Baure that were lost. Asis Coparcari (1880 [1767]: 68-70) mentions the two particles nichane ‘if, in case’ (16) and nenone ‘when, if’.16 co C1 C2 (19) nichane pamere yokise chase kamo renikapi kobe.17 [historical Baure] nichane pi=am=ro yokise chase kamo ro=inika=pi kobe if 2sg=take=3sgm stick time.ago neg 3sgm=bite.irr=2sg dog ‘If you had taken a stick then, the dog would not have bitten you.’ (Asis Coparcari 1880 [1767]: 68)

Summarizing the findings in this section, Baure used to combine clauses by embedding, marked subordination or juxtaposition, and the relation the clauses may have is marked by morphology on the embedded or subordinate predicate. Juxtaposition is much more frequent in the data collected by the SIL linguists in the 1950s, and in historical data, this is difficult to single out, because the Jesuits seem to have looked for direct translations of particular Spanish conjunctions, and they did not give us text with such complex contexts. So, in principle, Baure clause structure is not dependent on any connective, such as conjunctions. Nonetheless, and leading over to the next section, we showed that there is evidence of connectives in Baure, some of which may have been already lost. 3.2 Contact-induced clause combining constructions in Baure The construction focused on in this section is the combining of clauses by means of a connective. The argumentation is not that the construction itself is not native Baure, but that the extensive use of conjunction-like particles goes back to longterm contact with Spanish. As already shown in section 3.1, example (19), historical Baure also had a construction where particles occur clause initially. Some of these are coordinating, like ach ‘and’ (cf. example (18b), where it is used to combine two NPs), and others seem to mark a subordinated relation. Syntactically, however, the two clauses in combination are always independent clauses whose verbs remain unmarked. A semantic relation of subordination is caused only by the initial particle (cf. Danielsen 2007: 381-383). One example is given in (20):

There used to be more particles with an initial n- in Baure, as well as in Paunaka; the nasal presumably having a kind of deictic or pointing character. In Baure nowadays, only the local and temporal adverbials ne’ ‘here’, noiy ‘there’, naka ‘over there’, and nanan ‘later’ remain. 17 The abbreviation “co” refers to connective. 16


Swintha Danielsen and Lena Terhart

co C2 C1 (20) nar’inokow koech nowojik tech ndir eroserokchonev. ni=ar’inoko-wo koech no=wojik tech ndir eroserokoch-nev 1sg=be.sad-cop because 3pl=steal dem.m 1sg.poss lasso-pl ‘I am sad, because they stole my lassoes.’ (RP-030929S)

In the Jesuits’ data, there is no evidence of the connective koech ‘because’, and there are still many other ways to create a causal relation between two clauses. For example, a participle construction with or without introducing to ka ‘that which’ (cf. example (12) above); compare (20) to (21) with an introduced adverbial clause construction. C1 rel C2 (21) naskopi to ka paro’inokochow. ni=asko=pi to ka pi=aro’inok-cho-wo 1sg=help=2sg art ind 2sg=be.sad-ptcp-cop ‘I help you, because you are sad (lit. that which your being sad).’ (HC-040706S)

Baure has quite a number of these clause connectives, always in clause-initial position, and the majority of them are morphologically transparent. There is, for example, a group of particles derived from the generic verb root -a-: a/ach/achow ‘and, with’, aw ‘and not’, apo ‘or, if, that’, apon ‘or not’, avi ‘but’. Other connectives have been derived from the attributive ko- or privative mo-: koech ‘because’, koejkoe’ ‘so that’, moena’ ‘lest’18. Another productive particle is adversative tiwe’ ‘but’ (22), actually also decomposable as ti ‘dem’ + -wo ‘cop’ + -i’ ‘emph’). All of the connective particles can link two clauses or connect a clause to the complete preceding context, as discourse particles; to a lesser extent, this also holds true for Spanish conjunctions (cf. Stolz and Stolz 1996a:96). (22) and (23) are two more examples with clause connecting particles in Baure: C1 (22) kwe’ ndir mos chachachanev, kwe’ ndir mos exist 1sg.poss corn.cob

co C2 tiwe’ tischinev to choros. cho-a-cha~cha-nev big-clf:body-aug~aug-pl

tiwe’ ti-si-chi-nev to choros but small-clf:grain-dim-pl art corn.grain ‘I have big corn cobs, but their grains are small.’ (RP-081124L)


Lit. ‘no use’ < mo- ‘attr’ + -ina- ‘use’.

Borrowed Clause Combining Patterns in Two Arawakan Languages


co C2 C1 (23) piper ronik koejkoe’ rokamok. pi=pa=ro ro=nik koejkoe’ ro=kamok 2sg=give=3sgm 3sgm=eat so.that 3sgm=be.fat ‘Give him food, so that he gains weight.’ (RP-N081128L)

3.3 The influence of language contact in Baure How can we be sure that the connectives are related to language contact? In fact, it is not possible to prove that there is such a relation. However, taking the whole Baure language into consideration, we know that Spanish has had a great impact on the structure, much more than through actual lexical loans. The lexicalization and grammaticalization of constructions that are contact-induced in Baure have been studied, such as the preverbal particles for TMA marking, which have come to replace complex suffixing morphology on predicates. Three of the constructions are in fact similar to local Spanish constructions (cf. Danielsen 2015): a. Baure ver ‘already; prf’ and Spanish ya ‘already, prf’ b. Baure kach ‘go; intl’ and Spanish va ‘go to, fut’ c. Baure rom ‘recently, lately, soon; imm’ and Spanish recién ‘recently, soon, just, finally’ With the preverbal marking system of TMA in Baure, it is argued that the slot existed, but that the grammatical exploitation of the construction is a result of Spanish language contact, where analytic and more transparent constructions are preferred to the ones where the interpretation is dependent on the presence of some aspectual or modal suffixes on the predicates. The transparency of the connectives leads us to believe that they have evolved relatively recently. We also saw an example of a clause introducing particle in historical data in example (19), but this element is no longer in use19 − unfortunately we cannot present a complete list of formerly used particles, because a detailed analysis of historical Baure is still pending. Thus, even though the construction may also have a native model, the disappearance of connective constructions suggests a small degree of stability of (a) the particular particles with their semantics and functions, and of (b) the construction itself as a common means of clause combination. The stabilization of the construction in both cases − with preverbal particles as well as connective or subordinating particles − was controlled by the ubiquitous Spanish model construction. While preverbal particles reflect Spanish verb phrase structure, the clause connectives reflect Spanish clause structure. Baure also lost preverbal particles, such as the negative particle kamo, but this has not been studied in detail yet.



Swintha Danielsen and Lena Terhart

In the end, there are also a few examples of (ad hoc) borrowings of Spanish particles into Baure utterances, but these seem to be recent phenomena in the situation of language decay.20 One example is the particle ko ‘like’ in current Baure, which may have been interpreted as a truncated form of Spanish como ‘like’.21 The comparative construction modelled on Spanish is presented in example (24): C1 co C2 (24) ver nijinokopawori ko ti ponowaper. ver ni=jinoko-pa-wo=ri ko ti po-no-wapa=ro prf 1sg=see-intl-cop=3sgf like dem.f other-clf:generic-cos=3sgm ‘I already saw her like that other one.’ (AD_DC-C060401S)

The particle ko ‘what?’ has even been found to be used like Spanish que ‘that’ in a restricted relative clause construction, as in example (25). Two characteristics have probably facilitated the use of ko as a clause combining particle in Baure here: the phonological similarity of ko and que [kɘ] and the fact that ko has already been used as a particle in comparative constructions. co C [C1 co C2 ] (25) koech kwe’ ko pike. koech kwe’ ko pi=ke because exist what 2sg=do ‘because it is possible that something may happen to you (lit. because there is what you do)’ (LO_GP-P091220P-2)

Another example of a rather ad hoc borrowing of a Spanish particle into a Baure clause is the following. The interrogative clause in (26) would already be complete and grammatical without the introductory particle cuál ‘which?’ from Spanish.22 This is again a case of creating more syntactic transparency. The question for ‘which kind of’ is constructed by way of inserting the suffix -iyo ‘kind of’ into a nominal predicate.

It can be supposed that these borrowings go back to semi-speakers’ language, which has also been accepted partially by perfect Baure speakers. 21 Using ko ‘like’ in comparative constructions in Baure is a recent development, which is supported by the fact that in the closely related and generally mutually intelligible Carmelito dialect, the particle only serves the seemingly original function (ko ‘exist’, presumably related to ko- ‘attr’). 22 To be precise, the clause is complete as a subordinate indicative clause; in the interrogative clause it should get the nominalizer -no attached. However, it is not clear here if the speaker made a mistake by leaving the nominalizer out, or if the presence of cuál was the reason why it was dropped. 20

Borrowed Clause Combining Patterns in Two Arawakan Languages


co C (26) cuál ka’anoeyaw te? cuál ka’an-iyo-a-wo te dem.m which animal-kind-lk-cop ‘What kind of animal is this (lit. Which what kind of animal is this)?’ (RC-C-081012S-4)

Apart from the illustrated examples here, it was already mentioned in Stolz and Stolz (1996b: 286) that Baure has borrowed at least two common Spanish discourse particles, boen (< bueno, Sp.) ‘well’ and karaw/ karaj (< carajo, Sp.) ‘gosh’. While there is no similar native element to boen ‘well’, the exclamative interjection freely coexists with the native terms (y)ajpi and isher ‘gosh’. One last example shall demonstrate the use of a Spanish discourse particle in Baure: (27) Boen, … (0.96) … enevererapikoe’ rokachporeiyow royonpa. boen enevere-ro-a-piko-i’ ro=kach-poreiy-wo ro=yon-pa well 3sgm=go-rep-cop 3sgm=walk-intl ‘Well, … the next day he went again to hunt (lit. walk).’ (RP-N081126SL)

The particle boen ‘well’ is used to structure the discourse, and this means it can virtually occur at the beginning of every sentence, in particular with a slight hesitational pause before or after the particle, like in (27) − the number in parentheses indicates the time of the pause in seconds and milliseconds. To summarize the Baure situation, we can note that native constructions worked predominantly with clause embedding by nominalization and participles, the subordination relies on specific derivational suffixes (including the participle construction for adverbial clauses) and the juxtaposition of clauses, where certain morphemes could create a semantic subordination of a clause. The use of clause introducing connective particles was present also in historical Baure, but the exhaustive use of newly created connectives is rather to be regarded a recent phenomenon under the influence of Spanish. 4. Clause combining in Paunaka In Paunaka, there are three major strategies of clause combining. First, there are combinations that are not marked by a connective. Among them are completely unmarked combinations of two clauses, as well as cases where both clauses contain certain TMA affixes and clitics on the predicates that guide their interpretation. In addition, there is one affix that marks subordinate verbs. Second, there are some particles of Paunaka origin that connect clauses or chunks of discourse. These strategies, which we interpret as native Paunaka,


Swintha Danielsen and Lena Terhart

are described in 4.1. The third option is the combination of clauses by means of conjunctions and other connective particles borrowed from Spanish. Those connectives are used very frequently in Paunaka speech and are the topic of paragraph 4.2. The impact of Spanish on the Paunaka structures is reflected upon in 4.3. 4.1 Native clause combining constructions in Paunaka In Paunaka, two or more clauses may be juxtaposed without any devices marking the relation between the clauses. We interpret two juxtaposed clauses as belonging to one sentence, if they are not interrupted by a pause, and the intonation is that of a single sentence. The relation between the clauses is best translated as a coordinative one, like in example (28), but sometimes other interpretations are possible, too, like in example (29), which could either express that the state (of being afraid) and the action (of climbing up) happen simultaneously or that the action is the result of the state. C1 C2 (28) titupunubuji kimenukÿyae tisemaikuji echÿu kujubipi. ti-tupunu-bu-ji kimenukÿ-yae ti-semaiku-ji echÿu kujubipi 3i-arrive-refl-quot woods-loc 3i-search-quot dem liana.sp ‘He came to the woods, they say, (and) he looked for a liana, they say.’ (mox-n110920l) C1 C2 (29) tipikutu tipunu anÿke ti-piku-tu ti-punu anÿke 3i-be.afraid-iam 3i-go.up up ‘He was afraid and climbed up.’ or: ‘He was afraid, therefore he climbed up.’ (mxx-n101017s-1)

Some of the juxtaposed clauses most naturally translate into a sentence consisting of a main and a relative clause, as example (30) illustrates: C1 C2 (30) i kakuku naka chibutinekena eka aitubuchepÿi naka timuku i kaku-uku naka chi-buti-ne-kena eka aitubuche-pÿi naka ti-imuku and exist-also here 3-boot-poss-uncert dem here 3i-sleep ‘And here is also the boot of the young man, who sleeps here.’ (mox-a110920l-2)

In (30) aitubuchepuÿ ‘young man’ could then be interpreted as the head that is immediately followed by the relative clause naka timuku ‘he/who sleeps here’. Note that the verb does not carry any sign of subordination. Syntactically, the second clause is juxtaposed to the first one.

Borrowed Clause Combining Patterns in Two Arawakan Languages


Sometimes certain modal morphemes on the predicates indicate how the relation between the juxtaposed clauses is to be interpreted, similar to Baure. In (31) it is a combination of frustrative and irrealis that signal that this sentence is to be understood as a counterfactual conditional construction. C1 C2 (31) nÿkÿineini nakeini niyÿka nÿ-kÿi-ina-ini naka-ini ni-yÿka 1sg-gun-irr-frust here-frust 1sg-shoot.irr ‘If I had had my gun with me, I would have shot.’ (jxx-a120516l-a)

In (31) the subject of the clause expressing the condition, nÿkÿineini ‘my gun’, carries an irrealis suffix which states that the event is not real. The frustrative suffixes that mark the subject and the adverbial predicate nakeini ‘here (in vain)’ express that the event is not successful. Literally the clause means ‘if my gun had been here’. The verb of the matrix clause niyÿka ‘I (would) shoot’ is again marked for irrealis due to its proposition not having happened. The same clause C2 could also mean ‘I will shoot’ in isolation. Paunaka does not make use of nominalization to mark subordination23, which would be the typical strategy of a South Arawakan language (cf. Aikhenvald 1999: 100; Haude et al. 2011) and is also employed in Baure, see 3.1. Instead, verbs can be marked as subordinate by a suffix -i that is inserted after the root,24 i.e. between the last consonant of the verb base − in rapid speech the consonant is only palatalized − and the following vowel /u/ (default realis) or /a/ (irrealis). No other morpheme can break the sequence of the last root consonant and the following vowel. Among the clause types that mark the subordinate verb by this suffix are purposive clauses, complement clauses (32) and relative clauses (33). [C1 [[C2 ] [C3 ] ] ] (32) tanÿma tanÿmayu acheupu kuje kuinatu micha bichujikia amuke tanÿma tanÿma-yu acheupu kuje kuina-tu micha bi-chujik-i-a amuke now now-ints new moon neg-iam good 1pl-harvest-subord-irr corn ‘Now, because of the new moon, it is not good to harvest the corn.’ (nxx-a630101g-1)

The speakers do not generally make much use of nominalization. There are a few examples in older recordings of the 1960s (courtesy of Jürgen Riester – APCOB, Santa Cruz, Bolivia), but nowadays nominalization is falling out of use. 24 This suffix is certainly related to Baure -iy that had been shown in example (11) above, but in Paunaka it is only found on dependent verbs and not restricted to locative or instrumental notions. 23


Swintha Danielsen and Lena Terhart

The matrix predicate of the complement clause in (32) is kuinatu micha ‘it is not good’ and its complement is the verb bichujikia ‘that we harvest corn’ marked as subordinate by -i. Note that the clause expressing the cause – tanÿma tanÿmayu acheupu kuje ‘now there is the new moon’ – is completely unmarked. [C[C1 ] [C2 ] ] (33) niyunupunatu nauku nemukiu yuti ni-yunu-puna-tu nauku ne-imuk-i-u yuti 1sg-go-rep.irr-iam there 1sg-sleep-subord-real night ‘I will go back to where I slept at night.’ (jmx-e090727s)

In (33) the matrix clause with an adverbial head nauku ‘there’ is directly followed by the relativized verb nemukiu ‘that I slept’, which is marked subordinate by -i. More research on relative or adverbial clauses has to be undertaken in order to understand under which conditions a verb of a subordinate clause is marked with -i and which conditions trigger unmarked juxtaposed clauses as in example (30). The other strategy to signal certain combinations of clauses is by using particles. As in Baure, there is no indication on the predicates that would signal dependency. The connectives are the only means of establishing the relation between the clauses. On potential conditionals, like the one in example (34), the clause expressing the condition is introduced by a particle kue ‘if’. Both predicates encoding the condition and the consequence are marked for irrealis, because the realization of the event is hypothetical. co C1 C2 (34) kue piyuna tiyunauku echÿu kue pi-yuna ti-yuna-uku echÿu if 2sg-go.irr 3i-go.irr-also dem ‘If you go, he will go, too.’ (mux-c110810l)

Consecutive clauses are introduced by a particle nechikue ‘that’s why’. This particle may itself be complex, consisting of the deictic n- that is also found in Baure plus some other morphemes, but we are currently unable to break down its meaning.25 Most of the time, nechikue introduces a new utterance after a pause, like in (35), where the number in parentheses gives the duration of the pause in seconds and milliseconds. It is therefore rather to be classified as a discourse connective, which links a piece of discourse to a previous one, than as a clause connective, although instances of clause combining can be found, too, as in (36): 25

Other Paunaka particles that contain the deictic nasal are naka ‘here’, nauku ‘there’, nechÿu ‘over there’, nebu ‘loc.cleft (this is (to/from) where) and the deictic suffix -ni, which is most probably a loan from Bésɨro.

Borrowed Clause Combining Patterns in Two Arawakan Languages


co C2 C1 (35) tibeune chÿenu (1.23) nechikue bitiuku naka ti-beu-ne chÿ-enu nechikue biti-uku naka 3i-bring-1sg 3-mother that’s.why 1pl-loc? here ‘His mother brought me … that is why we are still here.’ (nxx-p630101g-1) C2 co C3 C1 (36) nipiji Maria bichujijikumÿne biyuna asaneti nechikue kuina nemejikupa ni-piji Maria bi-chuji~jiku-mÿne bi-yuna asaneti nechikue kuina 1sg-sibling Maria 1pl-talk~ints-dim 1pl-go.irr field that’s.why neg ne-mejikupa 1sg-forget.irr ‘Me and my sister Maria, we talked when we went to the field, that is why I haven’t forgotten [to speak the language].’ (jxx-x110916)

Another particle chejepuine ‘because’ may be used to introduce causal clauses, but this word is used only by one speaker and there are not many examples in the corpus. Again, chejepuine may rather be interpreted as a discourse particle ‘because of this fact’, since there is usually a pause between the two clauses that are connected to each other. 4.2 Borrowed clause combining constructions in Paunaka Paunaka has experienced much influence from Spanish. We do not have historical data to compare current Paunaka with older varieties, but if we compare Paunaka to the genetically closest related languages, we find that it is morphologically less complex. We are convinced that morphological complexity was lost in a process of language obsolescence favoured by close contact with the morphologically less complex Spanish. There are also many loans of Spanish origin and some semantic calques. One noticeable feature is the borrowing of conjunctions and other particles. Among them are the three prototypical conjunctions, coordinative i (y in Spanish orthography) ‘and’, disjunctive o ‘or’, and adversative pero ‘but’. In addition, we find consecutive entonces ‘so’ and contrastive más bien ‘rather’ being used quite frequently. At first sight, a Paunaka piece of oral discourse may look very Spanish-like due to all the loans, but when we examine the loans more closely, we also find differences with the Spanish model.26 The coordinative conjunction is the one with the highest number of occurrences in the corpus. In example (37), there are two clauses coordinated by i ‘and’: We are aware of it being problematic to compare spoken Paunaka to standard written Spanish, but, unfortunately, there are not any studies on the use of connectives in the local spoken variety of Spanish on which we could base our comparison.



Swintha Danielsen and Lena Terhart

co C2 C1 (37) kapunu naka i tikechune kapunu naka i ti-kechu-ne come here and 3i-say-1sg ‘He came here and (he) told me.’ (mxx-p110825l)

However, most of the time i occurs at the beginning of an utterance after a switch of turn or after a pause. It is therefore not as much related to clause combining as it functions as a discourse connective that links the utterance to the previous context, see (38), as also argued about Baure connectives ach ‘and’ and koech ‘because’ (Danielsen 2007: 385), compare example (25). C1 co C2 (38) tikubiakubuji (0.75) i kuina eka trabakuina ti-kubiaku-bu-ji i kuina eka trabaku-ina 3i-be.tired-refl-quot and neg dem work-irr ‘He was tired, they say, and he hadn’t worked.’ (mox-n110920l)

The conjunction also serves as a pause filler. It gives the speaker more time to organize the following utterance, as in example (39), where it appears between two pauses. C1 co C2 (39) naka tibebeikumÿne eka apimiya- ee aitubuchepÿimÿne (1.58) i (1.09) nechÿuku kakuku echÿu chipeumÿne kabe naka ti-bebeiku-mÿne eka apimiya ee aitubuche-pÿi-mÿne here 3i-lie-dim dem girl hes boy-body-dim i nechÿu-uku kaku-uku echÿu chi-peu-mÿne kabe and there-also exist-also dem 3-animal-dim dog ‘Here the girl – eh – the boy is lying [on the bed] … and … there is also his dog.’ (mox-a110920l-2)

Although we assume that in the local spoken variety of Spanish the conjunction is used as a discourse connective, it is probably not the most important function. We can conclude that the borrowed conjunction has restricted its function in the recipient language or at least given priority to a function that is not the main function in the donor language.27 The conjunction pero also frequently introduces utterances after a pause, as in example (40). In those cases, it has a similar discourse function as i, but adds an additional adversative flavor. 27

There are also a few NPs that are coordinated by i, but generally NPs are coordinated asyndetically by juxtaposition, a fact that fits the finding by Matras (1998:312) that coordination of constituents is not as vulnerable to conjunction borrowing as coordination of clauses.

Borrowed Clause Combining Patterns in Two Arawakan Languages


C1 co C2 (40) pijinepuÿ ja’a (1.14) pero tiyuikumÿnetu? pi-jinepuÿ ja’a pero ti-yuiku-mÿne-tu 2sg-daughter intj but 3i-walk-dim-iam ‘your daughter, aha … but does she walk already?’ (uxx-p110825l)

Nevertheless, pero is also frequently used for the combination of clauses, like in example (41), so we cannot speak of a restriction of its function. C1 co C2 (41) nisachuini niyuna pero nipiku ni-sachu-ini ni-yuna pero ni-piku 1sg-want-frust 1sg-go.irr but 1sg-be.afraid ‘I would like to go, but I’m afraid.’ (jxx-p110923l-1)

The third major conjunction borrowed from Spanish is disjunctive o ‘or’. We do not find many examples of o in our Paunaka corpus, but disjunction is also crosslinguistically less prominent than coordination (Haspelmath 2004:27; Ohori 2004: 62), so that this may be due to a universal tendency. This particle is found to introduce an added clause that either presents an alternative like in (42), or it is used for self-correction, like in (43), where it does not express disjunction in the strict sense. C1 co C2 co C3 (42) juchubu piyunia tukiu naka, Argentina? (0.86) o kuina o tÿpajÿka? juchubu pi-yun-i-a tukiu naka Argentina o kuina o tÿ-pajÿka where 2sg-go-subord-irr from here Argentina or neg or 3i-stay.irr ‘Where do you go from here, to Argentina? ... or not or does it stay? (the plane in the air, meaning a direct flight).’ (jxx-e120516l-1) C1 co C2 (43) naka echÿu chikamane kaku eka kabemÿne teukena (0.52) o tinikukena yÿtÿukukena naka echÿu chi-kama-ne kaku eka kabe-mÿne t-eu=kena o ti-niku=kena here dem 3-bed-poss exist dem dog-dim 3i-drink=uncert or 3i-eat=uncert C3 tiniku naka yÿtÿuku=kena ti-niku naka food=uncert 3i-eat here ‘Here in his bed there is a dog, he is drinking, it seems, ... or he is eating food, it seems, he is eating here.’ (mox-a110920l-2)

There are no verbs that are related by the disjunction particle in the Paunaka corpus, but that may be a coincidence. Nominal predicates can be combined by disjunctive o.


Swintha Danielsen and Lena Terhart

Another connective from Spanish frequently used in Paunaka is the consecutive marker entonces. It is employed as a discourse connective most of the time, but can also serve to connect two clauses as shown in (44), where it co-occurs with the subordination marker -i. co C1 co C2 (44) pero repentekena tamichaupunuyu entonces eyunia pero repente=kena ti-a-micha-upunu-yu entonces e-yun-i-a but maybe=uncert 3i-irr-good-rep-ints so 2pl-go-subord-irr ‘But maybe he will be well again so that you can go.’ (mux-c110810l)

The connective entonces competes with the native Paunaka consecutive marker nechikue (see (35) and (36)). Both markers appear in the same syntactic slot; the difference is a semantic one. Note that there are no examples of a borrowed connective porque ‘because’, which is frequently employed in Spanish. As has been argued before, the Paunaka causal connective chejepuine ‘because’ is also very rare. Speakers prefer consecutive structures over causal ones. It is therefore not surprising that two consecutive markers appear in Paunaka discourse. While entonces, like the Spanish particle, can be used consecutively or temporally (‘so, then’), nechikue always marks the clause as being a necessary consequence of some previously mentioned event (‘that’s why, therefore’). Its consecutive force is stronger than that of entonces, and this is probably the reason why it resisted being completely replaced by a Spanish loan. 4.3 How native constructions get replaced by Spanish ones Spanish conjunctions and connectives did not replace Paunaka strategies abruptly. They were added to the Paunaka strategy of combining clauses and utterances to support the link between them, and finally established as the primary means of marking these connections. Only then could they replace the Paunaka strategy completely. One strategy of clause combining that seems to currently undergo such a change is found with adverbial clauses containing a subordinate verb, which is interpreted as purposive, as in example (45). Nevertheless, sometimes the adposition tÿpi ‘ben’ occurs preceding the subordinate verb. In its core meaning tÿpi is marked for person and expresses beneficiaries of an event, but apart from this, tÿpi (with and without person-marking) is also developing into a semantic calque of the Spanish para ‘for’. This preposition marks benefactives in Spanish, but it is also used in many other contexts, for example to introduce purposive clauses. Example (45) shows a simple purposive clause with the purpose verb marked subordinate and (46) a purposive clause that is “supported” by tÿpi. One example was even found

Borrowed Clause Combining Patterns in Two Arawakan Languages


where there is no subordinate marking on the purpose verb and the relation is only marked by tÿpi (47). C1 C2 (45) mandau cinco mil dolares chiyunia. mandau cinco mil dolares chi-yun-i-a send five thousand dollars 3-go-subord-irr ‘She (one sister) sent 5.000 dollars so that she (the other sister) would go.’ (jxx-p110923l-1) C1 co C2 (46) echÿu isini echÿu kupisaÿrÿ chikutikubunube nauku kimenu chitÿpi chiretenaikianube yÿkÿkejaneyae. echÿu isini echÿu kupisaÿrÿ chi-kutiku-bu-nube nauku kimenu chi-tÿpi dem jaguar dem fox 3-run-refl-pl there woods 3-ben chi-retenaik-i-a-nube yÿkÿke-jane-yae 3-tie-subord-irr-pl tree-distr-loc ‘The jaguar and the fox ran there to the woods to tie themselves to the trees.’ (Villafañe 2007, glosses by the authors) C1 co C2 (47) chupunanube echÿu maquina tÿpi chananube echÿu epenue chi-upuna-nube echÿu maquina tÿpi chi-ana-nube echÿu epenue 3-bring.irr-pl dem machine ben 3-make.irr-pl dem hole ‘They would bring the machines for making the pit.’ (mxx-p110825l)

Examples (45) to (47) may well illustrate how native strategies are progressively lost: (45) can be taken as the more native Paunaka strategy, (46) as the already influenced, but still grammatically marked Paunaka strategy, while (47) is simply a loan translation of the Spanish strategy with tÿpi used as the connective. First, the more Spanish-like construction is added to the native one, like in (46), and then finally, the native strategy is dropped completely in favor of a solely Spanish-like construction, as in (47). 5. Comparison and explanations of the findings To summarize the findings, Baure native clause combining constructions include mainly embedding by deriving the predicates as arguments in the clauses, subordination marked by specific morphemes, or juxtaposition of clauses, where specific semantics can be interpreted from the markers attached to the predicates in the clauses. Baure also has a minor native strategy of combining two clauses by means of a connective, similar to conjunctions in Spanish. However, these connectives are not stable, so that particles from historical data are not present in contempo-


Swintha Danielsen and Lena Terhart

rary Baure data. On the other hand, a number of particles seem to have emerged relatively recently. Their use models Spanish clause combining constructions and it is argued that the exploitation of this strategy is a result of long-term language contact with Spanish. This coincides with other observations of the influence of Spanish structure on the Baure language. One of the dominant clause combining strategies in Paunaka is the juxtaposition of clauses without any subordination marker, with possible refinement of the interpretation of the relation between the clauses by verbal morphology. In addition, there is one verbal suffix that can mark subordination, employed for various types of relations. Furthermore, Paunaka has a number of native and borrowed clause combining connectives. Most of the conjunctions borrowed from Spanish do not seem to have any native counterpart. We can thus suspect that Spanish did not have influence only on the structure of Paunaka but also served as the lexical donor of various clause combining as well as discourse particles. In this section, we first discuss the general motivation for the borrowing of clause connecting strategies and then turn to the specific situations presented here. Since the two Arawakan languages are not only closely related genetically but have also undergone long-term contact with Spanish in similar settings, the question arises why the outcome diverges in the described ways. In 5.1 we explain the general motivations for borrowing structures from Spanish in both languages, and in 5.2 we contrast the two situations of Baure and Paunaka. 5.1 Borrowing of clause connectives and structure from Spanish Indigenous languages in contact with Spanish have long been reported to have borrowed conjunctions and other function words. Most information is available on these borrowings into Mesoamerican languages (cf. Stolz and Stolz 1996a; Suárez 1983: 135-136; Brody 1987 and numerous accounts of borrowings into individual languages, e.g. Zimmermann 1987 for Otomí; Muntzel 1988 for Ocuiltec), but there are some reports on other areas as well (South America and the Western Pacific cf. Stolz and Stolz 1996b; South America cf. Brody 1995). A detailed description of function words borrowed into Amazonian languages is given in Sakel and Matras 2008: 73-75. The explanation given in an influential paper by Stolz and Stolz (1996a) – see also (Zimmermann 1987) – is that the borrowing of function words gives the indigenous discourse a certain Spanish flavor, which is created deliberately by bilingual speakers to signal their competence in the more prestigious national language, and is also used by monolingual speakers (cf. Stolz and Stolz 1996a: 110-111, see also Brody 1987: 509-510). Matras (1998: 318) claims that the opposite is

Borrowed Clause Combining Patterns in Two Arawakan Languages


true: “[I]tems are not borrowed because their use is prestigious. Rather, speakers will ultimately give in to cognitive pressure to borrow them, unless this process entails a loss of prestige.” He showed that the borrowing of conjunctions, connectives and other particles is not restricted to situations with one language being socio-linguistically more prestigious. Hence, the explanation of borrowings being a marker of identity with a linguistically dominant group cannot fully account for the occurrence of such borrowings cross-linguistically. According to him, all of these words belong to a class of “utterance modifiers” (Matras 1998: 282) that “regulate interactional relations between speakers and hearers” (ibid: 308). That is, their main function is pragmatic, and they are used by the speaker to direct and monitor the hearer’s processing of the propositional content of the utterances and “maintain assertive authority” (ibid: 310). The use of such devices is highly automatized and gesture-like. The bilingual speaker is confronted with cognitive overload of keeping two systems apart and may reduce the pressure by making use of the devices that are provided by the language that is pragmatically dominant.28 The two systems fuse. This is both a synchronic and diachronic process, because the repeated spontaneous usage of devices of the pragmatically dominant language may give rise to a diachronic replacement of structures in the other language (Matras 1998: 291-321). Supporting the argument of high cognitive load is the fact that some of the above described native constructions to combine clauses involved the morphological marking of the subordinate predicate in a highly synthetic way, i.e. Baure has a number of embedding or subordinating suffixes, attached to a verb, and Paunaka has the subordinating suffix -i ‘subord’. This technique is considerably different from Spanish and seems to be more difficult to be maintained for that reason. The suffixal marking comes in a completely different syntactic and pragmatic position and is much less obvious when the whole clause structure is considered pragmatically, e.g. the suffixes are always completely unstressed. 5.2 Why do we have distinct outcomes from seemingly similar contact situations? Given a very similar scenario with Spanish being the pragmatically dominant language, why did Paunaka speakers borrow connectives from Spanish whereas Baure speakers did not or only borrowed structure? The answer may lie in the historically different social status of both languages and thus a different time depth of being confronted with a pragmatically dominant language. Pragmatically dominant is the language of a monolingual group that the bilingual speaker regularly has to speak with (Matras 1998: 285-323). It does not necessarily coincide with the language that is socially dominant. A language that a speaker is most exposed to at a given time span may also gain pragmatical dominance over another language.



Swintha Danielsen and Lena Terhart

Baure was established as a lengua general of some mission towns by the Jesuits (Saito 2009: 356-357) and fixed in a grammatical description. The Baure language of the missionaries (and thus of the Bible) has probably introduced the first changes towards a more Spanish-like structure into the language (cf. the argumentation of Saito 2009: 357 for Moxo). Until the 1920s, people moving to Baures − as a consequence of the rubber boom − have been learning Baure as a means of communication. It is possible that the increased communication with L2 learners, maybe with a Spanish linguistic background, led to the formation of particles that resemble Spanish connectives formally as well as semantically. Although the rubber boom had disastrous consequences for the demographic development of the town, Baure co-existed as an important means of communication until roughly the 1950s. It was then that the social dominance of Spanish was finally manifested with the educational reform, when education that was culturally and linguistically based on the urban majority society was introduced into the rural areas of Bolivia. The usage of indigenous languages inside the schools was forbidden and children who disobeyed were punished. As a result, many people switched to Spanish when raising their children, so that they would be able to get along in these surroundings. Nowadays, Spanish is socially and pragmatically dominant for all Baure speakers. Nevertheless, they may have received substantial Baure input during their childhood. In addition, Baure speakers seem to be linguistically more conservative and they are proud of their language that is considered difficult by outsiders. They are also very conscious about different proficiencies and do not fully accept semi-speakers as part of their linguistic community. A greater linguistic consciousness may be the result of the majority status in colonial times on the one hand, but also of the early interest that linguists paid to their language. Baure has already been investigated in the 1950s by two SIL linguists, who were highly appreciated by the community according to the present-day speakers. Today in Baures, although there are very few fluent speakers, those who speak the language strongly resist against loans from Spanish, and mostly only those words that have been more phonologically integrated (thus early loans), are accepted as part of their language at all. Concerning the area of clause combining, the presence of particles serving the very same functions as Spanish connectives may ease the prevention of integrating loans. The situation for Paunaka is quite different. It has been a minority language for more than three centuries. Speakers of Chiquitano dialects outnumbered speakers of other languages in all the mission towns in the Chiquitanía29, also in Concepción, where most of the Paunaka were settled (Tomichá 2002: 290). Paunaka was 29

The only exception is San Ignacio de Zamucos, where there were no speakers of Chiquitano (Tomichá 2002:290).

Borrowed Clause Combining Patterns in Two Arawakan Languages


probably already re-structured during that time, due to the close contact to Chiquitano.30 The insertion of Spanish connectives into Paunaka grammar is arguably more recent. In the approximately 20 minutes of recordings with one speaker of Paunaka dating from the 1960s31, we do find i ‘and’, and maybe o ‘or’ (only one occurrence in a context that is not entirely clear); however, there is no occurrence of pero ‘but’; instead another connective masa seems to function as an adversative. Masa may well be a loan from Spanish mas, which is sometimes used as an adversative, or Brazilian Portuguese mais, which also serves as an adversative conjunction. The Spanish consecutive connective entonces ‘so, then’ does not occur at all, but the Paunaka one nechikue ‘that’s why’ is found quite frequently. We can conclude that the replacement of nechikue by entonces is a very recent development, while the other connectives were already integrated into Paunaka in the 1960s. The current speakers of Paunaka experienced a very intense time of contact with other languages during their childhood, when they lived on the haciendas of a major landowner to whom they were subjugated through a system of debt bondage. Although they claim that during that time “everybody spoke Paunaka”, this may be over-estimated as there were also speakers of Bésɨro and other languages around, e.g. Napeka (Chapacuran). One speaker states that he learnt Bésɨro during that time to prevent other people from making fun of him in his presence. Comments like that let us assume that Bésɨro was socially dominant to Paunaka; the language of the major landowners, Spanish, was socially dominant to all the indigenous languages. We believe that during this time, the first Spanish connectives were integrated into Paunaka. Since there was also contact with the Bésɨro dialect, it is likewise possible that some loans and grammatical structures were adopted via this other indigenous language, but this still has to be proven by a closer comparative look at Chiquitano. In this language contact situation, which refers to two indigenous languages, on the one hand, and to the contact with Spanish, on the other hand, ALL people experienced a great cognitive pressure to keep languages apart. Speakers of Paunaka had to deal with two socially (and probably also pragmatically) dominant languages. Speakers of Bésɨro had to communicate with many L2 speakers, who may have had a very limited knowledge. All speakers were confronted with the necessity to learn the more prestigious language, Spanish. In this situation, the systems fused, and it is not only Paunaka that has

Unfortunately, there are no grammatical descriptions of Paunaka stemming from that time, so that it is not possible to compare colonial to modern Paunaka. 31 Courtesy of Jürgen Riester (APCOB) who gave us several old tapes of his field recordings with many different languages spoken in the Chiquitanía. 30


Swintha Danielsen and Lena Terhart

borrowed connectives from Spanish. In a field dictionary of Bésɨro32, we find exactly the same and some additional Spanish connectives. Thus, they were probably borrowed into both languages at the same time, a process in which the presence of Spanish connectives in Bésɨro additionally favored their introduction into Paunaka. During the past decades and with vanishing knowledge about the indigenous language(s), the use of Spanish connectives may well have increased. This has probably been accompanied by the loss of obligatoriness of verbal complexity. As has been shown above (see 2.1 examples (5), (14), and (16) for Baure and 3.1 example (31) for Paunaka), a combination of aspectual or modal affixes on verbs can signal their relationship to each other, but in Paunaka this is not a major strategy of encoding relations between clauses anymore. When the Bolivian agrarian reform of 1953 ended debt bondage, the Paunaka people emigrated from the haciendas and settled in the surroundings to build their own villages and make their own fields. Nevertheless, villages were re-settled a few times and people migrated from one place to another in search of water. Some went to the cities to make their living. The speakers of Paunaka remaining today were adolescents by that time, and it was probably the disruption of the community which started with the end of debt bondage that initiated the decay of the language. Nowadays, the speakers hardly ever speak Paunaka and therefore the Spanish system is even more cognitively anchored. Stolz and Stolz (1996b: 111-112) offer a model of function word borrowing with three layers: the discourse or text layer, the clause layer, and the word layer. The first function words are usually discourse particles. If we follow the explanation by Matras (1998), we could argue that the discourse layer is the first system that fuses, even before lexical or functional items are borrowed. As the borders between discourse particles and clause connecting devices are very fuzzy, a second logical step is the integration of borrowed clause connectives on the sentence level, which eventually may also spread to the word layer, if words or phrases are subsequently also coordinated by connectives. Baure has certainly taken over discourse organisation of a Spanish model, as was shown by the example of the omnipresent particle boen ‘well’ borrowed from Spanish. Furthermore, there are preverbal particles that show influence from the Spanish structure on the phrase level. The use of connectives, constructed of Baure lexical material, but with a similar distribution and semantics to corresponding Spanish connectives, presents yet another example of how Baure has adapted structurally to Spanish. Paunaka has experienced even more fusion with the Span32

We are very thankful to Pierric Sans, who provided us with his FLEx dictionary of Bésɨro for comparative matters.

Borrowed Clause Combining Patterns in Two Arawakan Languages


ish system, as it has lost much of its verbal complexity to show certain semantic connections between clauses, and makes extensive use of Spanish particles to mark discourse operations. It employs them to connect clauses, and sometimes even to connect phrases or words. In addition, the functions of native material, like the benefactive adposition, are broadened to match the whole array of functions of a similar Spanish item in Paunaka. This is a process that may well become more widespread in the future. The answer to the question of why Baure has resisted the integration of Spanish loans more strongly than Paunaka most convincingly lies in the different sociolinguistic development of the two languages over the centuries. Baure had been a dominant language during colonial times; it was even promoted to the general language of mission towns, and documented and analysed in Jesuit grammars. Its use was relatively widespread until the middle of the last century, and further attention to the language was paid by SIL linguists. Although its use has decreased since the 1950s, and usage of Baure was associated with a low social status since then, the former high prestige must have had an impact on its relative stability. The linguistic pride of the last speakers is reinforced by the ongoing endeavor of foreign people to document and analyze the language. Paunaka has been a minority language since the foundation of mission towns. It was never properly documented and described, except for a few short word lists stemming from the 19th century, and was considered socially inferior to Chiquitano, at least during the lifetime of the remaining speakers. In comparison to Baure, it shows a higher degree of obsolescence, of which the employment of Spanish connectives and the higher restructuring of the whole system of clause combination provides only one example. Abbreviations 1 = 1st person; 2 = 2nd person; 3 = 3rd person; all = allative; art = article; attr = attributive; aug = augmentative; ben = benefactive; C = clause; cert = certainty; cleft = cleft particle; clf = classifier; co = connective; cop = copula/imperfective; cos = change of state; dem = demonstrative; dep = departitive; dim = diminutive; distr = distributive; emph = emphatic; f = feminine; frust = frustrative; fut = future; hes = hesitation; i = intransitive; iam = iamative; imm = immediate; ind = indefinite pronoun; intj = interjection; intl = intentional; ints = intensifier; irr = irrealis; it = iterative; lk = linker; loc = locative; m = masculine; neg = negative; nmlz = nominalizer; pl = plural; poss = possessive; prf = perfect; pfv = perfective; ptcp = participle; rel = relativizer; quot = quotative; real = realis; refl = reflexive; rep = repetitive; sg = singular; subord = subordination; temp = temporal; uncert = uncertainty


Swintha Danielsen and Lena Terhart

REFERENCES Aikhenvald, A. Y. (1999). “The Arawak Language Family”. In R.M.W. Dixon, and A. Y. Aikhenvald (eds.), The Amazonian Languages. Cambridge: Cambridge University Press, pp. 65-106. Asis Coparcari, F. de (1880[1767]). “Gramática de la Lengua de los Indios Baure de la Provincia de Majos”. In L. Adam, and C. Leclerc (eds.), Arte de la Lengua de los Indios Baures de la Provincia Moxos, Conforme al Manuscrito Original del Padre Antonio Magio, Paris: Maissonneuve and Cia Libreros Editores, pp. 55-109. Brody, J. (1987). “Particles Borrowed from Spanish as Discourse Markers in Mayan Languages”. In Anthropological Linguistics, 29, 4, pp. 507-532. Brody, J. (1995). “Lending the ‘Unborrowable’: Spanish Discourse Markers in Indigenous American Languages”. In C. Silva-Corvalán (ed.), Spanish in Four Continents. Studies in Language Contact and Bilingualism. Washington: Georgetown University Press, pp. 132-147. Cardús, J. (1886). Las Misiones Franciscanas entre los infieles de Bolivia: situación del estado de ellas en 1883-1884. Barcelona: Librería de la Inmaculada Concepción. Crevels, M. (2002). “Why Speakers Shift and Languages Die: An Account of Language Death in Amazonian Bolivia”. In M. Crevels, S. van der Kerke, S. Meira, and H. van der Voort (eds.), Current Studies on South American Languages. Leiden: CNWS, pp. 9-30. Danielsen, S. (2007). Baure: An Arawak Language of Bolivia. Indigenous Languages of Latin America (ILLA) 6. Leiden: CNWS. Danielsen, S. (2011). “Clause Embedding Strategies in Baure (Arawak)”. In K. Haude, R. van Gijn, and P. Muysken, P. (eds.), Subordination in South American Indian Languages. Amsterdam/Philadelphia: John Benjamins, pp. 79-108. Danielsen, S. (2015). “Observing the Evolution of TMA particles − Replacing Agglutinating Strategies”. In E. Labeau, and Q. Zhang (eds.), Taming the TAME systems. Amsterdam/Philadelphia: Rodopi (Cahiers Chronos 27), pp. 1-23. Danielsen, S./Granadillo, T. (2008). “Agreement in Two Arawak Languages: Baure and Kurripako”. In M. Donohue, and S. Wichmann (eds.), The Typology of Semantic Alignment, Oxford: Oxford University Press, pp. 396-411. Danielsen, S./Terhart, L. (2014). “Paunaka”. In: M. Crevels, and P. Muysken (eds.), Las lenguas de Bolivia, Vol. 3. La Paz: Plural Editors, pp. 221-258. Haspelmath, M. (2004). “Coordinating Constructions: An Overview”. In M. Haspelmath (ed.), Coordinating Constructions. Amsterdam/Philadelphia: John Benjamins, pp. 3-39. Haude, K./van Gijn, R./Muysken, P. (eds.) (2011). Subordination in South American Indian Languages. Amsterdam: John Benjamins. Elliott, J. R. (2000). “Realis and Irrealis: Forms and Concepts of the Grammaticalisation of Reality”. In Linguistic Typology 4, pp. 55-90. Krauss, M. (2007). “Classification and Terminology for Degrees of Language Endangerment”. In M. Brenzinger (ed.), Language Diversity Endangered, Berlin: Mouton de Gruyter, pp. 1-8.

Borrowed Clause Combining Patterns in Two Arawakan Languages


Magio, A. P. (1880[1749]). “Gramática de la Lengua de los Indios Baures de la Provincia de Majos”. In L. Adam, and C. Leclerc (eds.), Arte de la lengua de los indios baures de la provincia Moxos, conforme al manuscrito original del padre Antonio Magio, Paris: Maissonneuve and Cia Libreros Editores, pp. 1-53. Matienzo, J./Tomichá, R./Combes, I./Page, C. (2011). Chiquitos en las Anuas de la Compañía de Jesús (1691- 1767). Cochabamba: Editorial Itinerarios. Matras, Y. (1998) “Utterance Modifiers and Universals of Grammatical Borrowing”. In Linguistics, 36, 2, pp. 281-331. Muntzel, M. (1988). “Spanish Loanwords in Ocuiltec”. In International Journal of American Linguistics 51, 4, pp. 515-518. Ohori, T. (2004). “Coordination in Mentalese”. In M. Haspelmath (ed.), Coordinating Constructions. Amsterdam/Philadelphia: John Benjamins, pp. 41-66. Rose, F. (2014). “Mojeño Trinitario”. In M. Crevels, and P. Muysken (eds.), Lenguas de Bolivia, Volume 3. La Paz: Plural editors, 59-97. Saito, A. (2009). “‘Fighting against a Hydra’: Jesuit Language Policy in Moxos”. In S. Kawamura, and C. Veliath (eds.), Beyond the Borders: Global Perspective of Jesuit Mission History. Tokyo: Sophia University Press, pp. 350-363. Sakel, J./Matras, Y. (2008). “Modelling Contact-Induced Change in Grammar”. In T. Stolz, D. Bakker, and R. Salas Palomo (eds.), Aspects of Language Contact. New Theoretical, Methodological and Empirical Findings with Special Focus on Romancisation Processes. Berlin: Mouton de Gruyter, pp. 63-87. Sans, P. (2010). Éléments de Sociolinguistique et de Phonologie du Bésɨro (Chiquitano). Langue en Danger des Basses Terres de Bolivie. Masters Thesis, Université Lumière Lyon2. Stolz, C./Stolz, T. (1996a). “Transpazifische Entlehnungsisoglossen. Hispanismen in Funktionswortinventaren beiderseits der Datumsgrenze”. In N. Boretzky, W. Enninger, and T. Stolz (eds.), Areale, Kontakte, Dialekte. Sprache und ihre Dynamik in mehrsprachigen Situationen. Bochum: Universitätsverlag Dr. N. Brockmeyer, pp. 262-291. Stolz, C./Stolz, T. (1996b). “Funktionswortentlehnung in Mesoamerika. Spanisch-Amerindischer Sprachkontakt (Hispanoindiana II)”. In Sprachtypologie und Universalienforschung, 49, 1, pp. 86-123. Suárez, J. A. (1983). The Mesoamerican Indian Languages. Cambridge: Cambridge University Press. Terhart, L. (2009). Klassifikatoren im Baure (Arawaksprache in Bolivien). Masters Thesis, Freie Universität Berlin. Tomichá Charupá, R. (2002). La primera evangelización en las reducciones de Chiquitos, Bolivia (1691-1767). Protagonistas y metodología misional. Cochabamba: Verbo Divino. Villafañe, L. (2007). Gramática Paunaka (Aruak). ms. Amsterdam. Zimmermann, K. (1987). “Préstamos gramaticales relevantes del español al otomí: una aportación a la teoría del contacto entre lenguas”. In Anuario de Lingüística Hispánica 3, 223-253.


CODESWITCHING AND BORROWING IN ARUBAN PAPIAMENTU: THE BLURRING OF CATEGORIES Yolanda Rivera and Patrick-André Mather University of Puerto Rico – Río Piedras

1. Socio-historical background Papiamentu, a Spanish-based creole primarily spoken in the Netherlands Antilles, shares features with other Atlantic Creoles that cannot be traced to their respective superstrata. For example, Papiamentu and Saramaccan, an Englishbased Creole, possess word-level tonal distinctions (Portilla 2009; Good 2004; Rivera-Castillo and Pickering 2004; Rivera-Castillo 1998; Remijsen and van Heuven 2005; Römer 1980, 1983, 1991; Rountree 1972; Voorhoeve 1961, 1968; Baum 1976; DeBose 1975). Papiamentu phonology significantly diverges from Spanish. Kouwenberg and Murray (1994: 9-11) describe the Papiamentu syllable structure as consisting of syllables with complex onsets (up to three consonants) and codas (up to two segments). Onsets include sC sequences, which violate general sonority restrictions (as in English and Dutch) (Clements 1990). Moreover, onsets include sequences with sN (smak, ‘flavor’), and sC+liquid (splica, ‘explain’). Birmingham (1976: 23) points out differences with Spanish syllable structure, which disallows such sequences in onset position. This creole is currently undergoing significant changes due to contact with other languages, such as Spanish, and is exhibiting signs of Hispanicization in the Aruban variety (Navarro Tomás 1953; Wood 1972a; Baum 1976). The foreignborn population has increased significantly in the last decades from a total of 6,117 immigrants between 1920 and 1979 to 9,448 between 1980 and 1991 (modified from p.74, Table P-C.7. of the Aruba’s “Censo 91”). Spanish, Dutch and English are widely spoken, with different degrees of fluency among speakers (Wattman Frank 1974; Lang 1997: 93). Maurer (1989: 143) provides data based on the 1981 census indicating that, in Aruba, 80.1% of the population speaks Papiamentu at home, while 5% speak Dutch, 10.6% speak English, and 4.3% prefer other languages. Among “other languages”, we can include Span-


Yolanda Rivera and Patrick-André Mather

ish. The percentages for this category are smaller in other islands: Curaçao, 3%; Bonaire 1.8%. Papiamentu only became an official language of the Netherlands Antilles in 2007. In Aruba, it has been the official language since 2003 (Migge, Léglise and Bartens 2010). Baum (1976: 84) indicated more than three decades ago that using Spanish in daily conversations became an indicator of higher socioeconomic status for Papiamentu speakers. Currently, the social prestige of Spanish is evident in Aruba, where newspapers such as the “Diario” use Spanish traditional c spelling in lieu of k for a voiceless velar stop /k/. As in many cases of diglossia, speakers tend to use Papiamentu mostly, though not exclusively, in informal settings, while Dutch, Spanish, and English have been the languages of work and schooling for centuries (Perl 1999; Maurer 1989: 143). A reaction to this situation, as well as a plan to turn the native language into the language of instruction (Èxtra 2000), has triggered the creation of an official orthography (1976) and a Committee for the Standardization of Papiamentu in 1984, on the island of Curaçao (Martinus 1997: 11; Perl 1999: 253). However, language planning and standardization can indicate a fear of language loss or the imposition of a Hispanicized variety of the language (Khubchandani 1984). The effects of Hispanicization (as well as the influence from other languages) have not been studied in detail. This chapter provides some groundwork in this area. Some studies assign an important role to Portuguese in the formation of postcolonial dialects and Atlantic creoles (Navarro Tomás 1953; Birmingham 1971; Wood 1972b; Voorhoeve 1973). According to these researchers, creoles in this area exhibit common traits that originated in a Portuguese-based Proto-creole that was the lingua franca in West Africa during the times of the slave trade (i.e., the monogenetic theory) (Alleyne 1980; Holm 1988; Martinus 1997). However, this theory does not describe the role that language contact has in the formation of creoles (Thomason and Kaufman 1988). Other approaches, like the substratist and superstratist hypotheses, assign a greater role to specific language groups in creole genesis. The superstratist model (e.g. Chaudenson 1992) proposes that most features of creoles can be derived from the superstrate (typically a European language), which would have been restructured by successive generations of L2 speakers, through second-language acquisition processes such as a reduction or regularization of inflectional morphology, greater analyticity and reanalysis of lexical morphemes as grammatical morphemes. One prominent substratist model, usually referred to as the Relexification Model (Lefebvre 1998), limits the role of language contact to one of combining lexical material from one language with grammatical features from another. Other recent approaches, such as the ecological approach, allow for a wider range of language-contact phenomena

Codeswitching and Borrowing in Aruban Papiamentu


and processes (Mufwene 2003), or emphasize the role of second-language acquisition and argue for a gradual creolization process, over several generations (Mather 2006, 2007). Due to these possible socio-historical scenarios, features in Papiamentu that resemble features in other languages might have originated during creole formation or be the result of recent contact. Therefore, we have found two types of data: (a) phonological features and lexical items resembling similar tokens in other languages but which presently constitute elements of the native system; (b) phonological features and lexical items resembling similar tokens in other languages, which, in fact, are non-native, and might have originated in situations of recent language contact. In this chapter, we analyze different types of lexical borrowing, codeswitching, and phonological adaptations. Regarding the latter, we describe the basic elements of Papiamentu and Spanish syllable structure, and phenomena related to differences between syllable types (Section 3). 2. Codeswitching versus borrowing Traditionally, codeswitching and borrowing are distinguished along several different criteria. First, codeswitching usually involves entire phrases, even clauses, whereas borrowings are normally single words or short phrases. Second, by definition only bilinguals can practice codeswitching, since this process involves a high degree of fluency in both languages. Third, while in codeswitching the phonology of each language is preserved, in borrowing, words are typically adapted to the phonetics and phonology of the host language. Poplack (1984: 55-56) explains that several criteria have been developed to distinguish between the two, but that they are not always reliable: “In seeking a way to identify full-fledged loanwords, a number of indices measuring various aspects of the linguistic and social integration of borrowed words were developed […]. These were abstracted from the types of criteria used implicitly or explicitly by scholars of bilingualism (e.g. Bloomfield 1933; Weinreich 1953) to characterize loanwords and included measures of frequency of use, native language synonym displacement, morphophonemic and/or syntactic integration, and acceptability to native speakers.” Usually, borrowings are distinguished from codeswitches based on monolingual standards within the speech community. For example, Gardener-Chloros (1995) looks at French words in Alsatian and shows that bilingual informants vary considerably in their opinion on whether specific words are integrated borrowings, or codeswitches. In our study, we found that some apparent codeswitches are in fact integrated loanwords since they appeared in Papiamentu dictionaries, even though they display various degrees of phonological adaptation.


Yolanda Rivera and Patrick-André Mather

Poplack (2004) distinguishes between established loanwords (phonologically adapted) and nonce borrowings (with partial phonological adaptation). She adds that borrowings are not only phonologically, but also morphologically and syntactically integrated in the recipient language. Mahootian (2006) suggests a diachronic process, beginning with isolated codeswitches, which become nonce borrowings (only used by bilinguals), and finally loanwords used recurrently by the entire community, including monolinguals, and listed in dictionaries. According to Muysken et al. (1996: 486), “codeswitching, a quite normal and widespread form of bilingual interaction, requires a high level of bilingual competence.” They add that balanced bilinguals produce intra-sentential codeswitching, whereas those with limited competence only do “emblematic” switching (like tag words: “okay?”). Muysken et al (1996: 488) compare Poplack’s definition of intra-sentential codeswitching as alternation between two systems, with Myers-Scotton’s model of insertion of embedded-language fragments into a matrix language. Muysken et al. (1996: 488) add that “insertion is frequent in neo- or excolonial settings and recent migrant communities, where there is a considerable asymmetry in the speakers’ proficiency of both languages and possibly also in the status of the languages involved.” This corresponds to the Aruban situation, where the prestige languages are the colonial or other European languages (Dutch, Spanish, English), while Papiamentu remains the main informal vernacular and home language. Muysken et al. (1996: 489) also argue that codeswitching is the first step to borrowing: “It is generally assumed, however, that lexical borrowing occurs through codeswitching: gradually, new words are introduced into the lexicon of the receiving language through repeated use by more and more speakers and by morphosyntactic and phonological integration.” This can also be seen in Aruba, where some apparent codeswitchings in our data are in fact well-integrated borrowings in Papiamentu, while others appear to be nonce borrowings by individual speakers. In his work on English loanwords in Irish, Hickey (2009: 672) states that “phonological integration is not always clear-cut.” This suggests that the traditional criterion of phonological adaptation, used to distinguish between borrowing and codeswitching, is not reliable since both phenomena include variable degrees of phonological adaptation. According to Bullock and Toribio (2009: 190), who studied English-Spanish codeswitching, it is generally assumed that borrowing typically involves phonological adaptation, whereas codeswitching does not. In codeswitching, it is assumed that there are only low-level phonetic effects. Bullock and Toribio (2009) show that several factors can condition these phonetic effects, including different proficiency levels in each language. Among low-level phonetic interference in codeswitching there are differences in voice-onset time (VOT) between, for example, Spanish (which has short VOT) and English (which

Codeswitching and Borrowing in Aruban Papiamentu


has longer VOT, which translates as aspiration of voiceless stops word-initially). Bullock and Toribio’s (2009) study included 288 test sentences in English, Spanish, and bilingual sentences with codeswitching. They found some bilateral influence in VOT for balanced bilinguals. The four outcomes were divergence (maintenance of contrast), convergence, interference (L1 on L2) and hypercorrection (L2 on L1) (2009: 202). Their study shows that, even in codeswitching, there can be considerable adaptation of codeswitched elements to phonological constraints of the host language. Our study on codeswitching between Papiamentu and other languages spoken in Aruba (Spanish, English and Dutch) also suggests, like Bullock and Toribio’s (2009) study, that phonological interference in codeswitching can occur in both directions: as we will see in the examples below, it is sometimes difficult to distinguish between single codeswitches (nonce borrowings) by bilinguals, and borrowings that are fully integrated into Papiamentu. While borrowings are, to some extent, adapted to the phonology of Papiamentu (for example in the devoicing of final consonants), in conversations displaying codeswitching, there is also some Spanish phonological influence on Papiamentu, which could lead to contact-induced language change (Thomason and Kaufman 1988). Our data also show that two of the three criteria traditionally used to distinguish borrowing from codeswitching are not always applicable to the Aruban data: first, codeswitching between Spanish and Papiamentu displays varying degrees of phonological adaptation; second, one cannot claim that a given word is an integrated borrowing based on the fact that it is used by monolinguals, since most of community is bilingual or even multilingual. 3. Papiamentu syllable structure Papiamentu syllable structure includes a large number of syllable types. Klein (2011: 182) recognizes six types in this language [following (Blevins 1995) description]: V, CV, CVC, VC, CCV, CCVC. Following Levelt and van der Vijver (2004), he includes Papiamentu in the Marked III class regarding its syllable structure. Levelt and van der Vijver’s (2004) scale ranges from a basic CV syllable structure (Marked I) to a Marked IV type for languages like Dutch. The main difference between Papiamentu and Dutch consists of the absence of complex codas in the creole. Nevertheless, besides words of Dutch origin, other words from different lexical sources allow complex codas: accent (English), ‘accent’; Diaweps, ‘Thursday’ (Spanish día jueves). Moreover, even words of Dutch origin, like heft, ‘to join, to stitch’ (from hechten), exhibit complex codas not attested in the source (Martinus 1997: 30). This indicates that this language allows a greater number of syllable types than Spanish, and should be included in the Marked IV class. There-


Yolanda Rivera and Patrick-André Mather

fore, the types of phonological adaptations to syllable structure we describe in this paper do not show a tendency to generate CV structures, as typically expected in creoles. On the contrary, adaptations show cases in which bilingual PapiamentuSpanish speakers eliminate complex onsets, or cases in which Papiamentu-English bilinguals devoice final codas. Some of the phenomena attested in the data show the dominant role of Spanish constraints on syllable structure (a Marked III type language), like the prosthesis of vowels in certain word initial groups. This is attested in the case of the bilingual Papiamentu-Spanish speaker (Speaker 3). Spanish can only have two consonants in onset position; and the second one is always a liquid (Quilis 1982: 139-140). Additionally, prosthesis is common in words borrowed from English: stress > estrés. Data collected from Speaker 3 provide evidence of prosthesis in sC wordinitial sequences (see Appendix C). Codas, in Papiamentu, include sequences of voiceless stops (kontrakt, ‘contract’), and stops followed by voiceless fricatives (Djaweps or Diahuebs, ‘Thursday’); as well as sonorants followed by voiceless stops (spalk, ‘splint’) (Kouwenberg and Murray 1994: 11). All these sequences are disallowed in Spanish, which can only have sequences of alveolar nasals followed by a voiceless fricative in coda position (Quilis 1982: 141). Additionally, historically, adaptations of lexical borrowings from Spanish show evidence of devoicing. Papiamentu consonants in syllable final position are devoiced (Birmingham 1971: 17): apsoluto, ‘absolute’ (Spanish: absoluto), berdat, ‘truth’ (Spanish: verdad). However, the cases of coda restructuring we found consist of devoicing of final consonants for speakers 1, 2, and 4, who constantly interact with English speakers (see Appendix A, B and D). Moreover, different studies describe a large number of diphthongs in Papiamentu, from twenty-three (Maurer 1989: 148) to nine (Kouwenberg and Murray 1994: 6). None of these include schwa (/ə/) among its components. However, Baum (1976: 86) indicates the presence of an allophonic schwa in Papiamentu. Speaker 4 produces a schwa segment in diphthongs for words of English origin. American English has four heterorganic diphthongs, as well as diphthongs with non-phonemic schwa. However, /ə/ cannot precede other vowels, and can occur in second or third position only (Hammond 1997: 7). In the cases attested in the data, /ə/ precedes the other vocoid. Additionally, according to Birmingham (1971: 38-39), Papiamentu presents historical rhotacism of intervocalic /d/ (/ð/): Spanish hígado ([iaðo]) > igra; mediodía > merdía, ‘noon’. Speaker 3 shows cases of rhotacism of intervocalic /d/ for words borrowed from Spanish (see Appendix C). In sections 5.1-5.3.2, we discuss instances in which speakers violate these general constraints on Papiamentu phonology. In other cases, Papiamentu phonological

Codeswitching and Borrowing in Aruban Papiamentu


constraints regulate the syllabic and melodic features of the items borrowed from Spanish and other languages. 4. Data sources The first author collected data for this analysis in 2000, with a Summer Research Grant awarded by the Office of Sponsored Programs at the University of Alabama-Tuscaloosa. Approximately five hours of spontaneous speech samples were recorded in conjunction with Rachel Shuttlesworth, currently at the University of Alabama-Tuscaloosa. Don Walicek, currently a professor at the University of Puerto Rico, completed a partial transcription of the data. We selected four speakers out of sixteen informants interviewed at the time. Speakers 1 and 2 are both Aruban and L1 speakers of Papiamentu. Speaker 3 is Venezuelan-born, and moved to Aruba at the age of 6. She is thus an L1 speaker of Spanish and an L2 speaker of Papiamentu. Speaker 4, her husband, is Aruban and an L1 speaker of Papiamentu. The materials include two separate conversations, between speakers 1 and 2 on the one hand, and speakers 3 and 4 (the married couple) on the other hand. Our analysis evaluates cases of codeswitching, borrowings, and phonological adaptations. Some of the adaptations apply to borrowings from other languages, like English, in contexts of codeswitching. Lexical borrowing and codeswitching are evaluated regarding their phonological shape, listing in local dictionaries, syntactic context, and the conversational situation. 5. Spanish and English borrowings and codeswitches in Papiamentu In everyday conversations in Papiamentu, the insertion of words and phrases from Dutch, Spanish and English is commonplace. In accordance with Muysken et al. (1996), it is assumed that single insertions of foreign words and phrases by native speakers of Papiamentu display various degrees of phonological integration, depending on whether these are one-time lexical insertions or longstanding borrowings in Papiamentu (e.g., listed in Papiamentu dictionaries). As we argued in section 2, the degrees of phonological integration alone are not always defining criteria to distinguish between borrowing and codeswitching, since even in codeswitching there is some degree of phonological interference between Papiamentu and other languages, as evidenced in other work on codeswitching (e.g. Bullock and Toribio (2009) on Spanish-English codeswitching). We first analyze sentences from the conversation between speakers 1 and 2, and subsequently between speakers 3 and 4.


Yolanda Rivera and Patrick-André Mather

5.1 Spanish borrowings and codeswitches (speakers 1 and 2) In a spontaneous conversation recorded between two native speakers of Papiamentu in Aruba (speakers 1 and 2), we found examples of both integrated loanwords from Spanish (that appear in printed dictionaries of the language) as well as one-word codeswitches (or “nonce borrowings”) from Spanish. As shown in section 2, it is not always straightforward to tell them apart. A clear example of the first category is the word fiesta, which appears in the following example from Speaker 1 (a male native speaker of Papiamentu): (1) bamo pone un fiesta, fiesta ki hor? ‘Let’s organize a party, a party at what time?’

By contrast, Speaker 2 (also an L1 speaker of Papiamentu) codeswitches into Spanish in the following utterance: (2) Awo, ki nan ta bezig […] de bida, […], de noche. ‘Now, when they are busy […] with their lives, […] at night time’

In Papiamentu, the phrase would be pa nochi. In the following sentence, Speaker 1 also codeswitches into Spanish, with some phonetic transference from Papiamentu: (3) so du muchu problema ‘besides many problems’

The standard form in Papiamentu is masha problema, whereas in Spanish it would be mucho(s) problema(s), which suggests that there is some phonetic adaptation into Papiamentu, specifically in the raising of the unstressed final vowel of mucho. Other examples of codeswitches from Spanish include the following sentences (Speaker 2): (4) antes ta de cink or seis ‘before it was five or six’ (5) ocho de mai ‘May eighth’

In (4), the form in italics is the Spanish word for ‘six’. Here, the similarity between Papiamentu and Spanish facilitates this transfer. In (5), it is possible to observe an intra-sentential codeswitch, or a Spanish ‘island’ (using Myers-Scotton’s terminology), with partial adaptation of ‘mai’ since the actual form in Spanish is ocho de mayo.

Codeswitching and Borrowing in Aruban Papiamentu


Other Spanish words and phrases (italicized in the examples below) inserted into Papiamentu by Speaker 1 are not phonetically adapted, and are also best described as intra-sentential codeswitches, as is shown in examples 6 to 9 below (segments in Spanish are italicized): (6) E ta una fiesta bastante bieu, di San Juan […] ‘It is quite an old celebration, dedicated to San Juan […] supuestamente, manera cu mi ta komprendé supposedly, according to what I know’ (7) […] no tin nada más ‘does not have anything else’ (8) Una usa más el holandés […] ‘One uses (speaks) Dutch much more’ (9) agua, súku, leche ‘water, sugar, milk’

The Papiamentu word for ‘milk’ is lechi. Example (9), and others, can be interpreted either as a single-word codeswitch by bilingual (or multilingual) speakers, or, alternately, as a low-level phonetic interference from Spanish, given that both languages (Spanish and Papiamentu) are lexically related. Similar phenomena are found in the speech of Speaker 2 (10-11), where Spanish words appear to be inserted into Papiamentu structure. (10) Entiende si sobra ya – en nos, nos idioma [...] ‘To understand enough already - in our, our language […]

ku, mi ‘understand’ which, which for me is to understand’

The standard Papiamentu word is kompronde (not entiende). (11) Sí, sí, es interesante, no? OK […] ke konos bisa? ‘Yes, yes, it is interesting, isn’t it? Ok, I want to know what (she/he/it) means.’

5.2 English borrowings and codeswitches (speakers 1 and 2) As we mentioned in the introduction, Aruba Papiamentu is in daily contact with Dutch, Spanish and also English since the island is a very important tourist destination. In the following examples, there are various types of borrowings and codeswitches that, like the examples from Spanish, also display varying degrees of phonological adaptation.


Yolanda Rivera and Patrick-André Mather

In example (12), Speaker 1 uses the English word ‘stop’, with some syllabic restructuring since there is an epenthetic [e] at the end of the word: (12) ku ta bisa di stope, no? ‘as they say, to just stop, right?’

Example (13) is more complex. On the surface, it looks like an English phrase within a Papiamentu sentence (or an embedded language ‘island’ according to Myers-Scotton’s terminology). However, the phrase “breakdown in alcohol” does not exist in English, nor does the preposition “en.” Syntactically, this is a Papiamentu phrase, with two English single-word codeswitches (“nonce borrowings”). However, in this case the word “breakdown” is partially adapted since the [r] is pronounced as an alveolar trill. The same applies to the nonce borrowing “groggy” (also realized with a trill) in example (14): (13) Sí, breakdown en alcohol, de hubentud, sí. ‘Yes, an alcoholic breakdown, for young people, yes.’ (14) Asi te aki groggy, te aki ta borrachi […] ‘Here until groggy, until (she/he) is drunk’

Example (15) is an English codeswitch with some phonological adaptation, namely the voicing of /t/ in “went up”, the hardening of /ð/in “other”, and voicing and hardening of /ð/in “with”: (15) wendup wid each udda ‘went up with each other’

Some English codeswitches are not adapted phonologically, as in examples (16) to (21) from Speaker 1. These appear to be mainly single-word codeswitches. (16) ken un dialekto bus (bos) ta high un Frances. ‘whose speech sounds are high for a French person’ (17) No e same ‘Not the same [...]’ (18) So what more?... yeah (codeswitch into English in addressing linguist) (19) It’s going good or? (English codeswitch in addressing linguist) (20) no sa […] awe ta bal e-e-e construction guys. ‘Do not know […] now expenses include the, the, the construction guys.

Codeswitching and Borrowing in Aruban Papiamentu


(21) then … a unas kaba. ‘then … some finished.’

5.3 Spanish and English borrowings and codeswitches in Papiamentu (speakers 3 and 4) The following data were recorded in a conversation between two married people (speakers 3 and 4). Speaker 3 is a woman whose family emigrated from Venezuela when she was 6 years old. Speaker 4, her husband, was born and raised in Aruba, and is an L1 Papiamentu speaker. First we examine examples of Spanish codeswitching and phonological interference from Spanish by Speaker 3. We then turn to data from Speaker 4. 5.3.1 Spanish codeswitches (Speaker 3) In example (22), Speaker 3 uses the Spanish phrase “tres hermanas”, with some adaptation to Papiamentu. “Rumán” is the Papiamentu word for brothers/sisters, whereas urmana sounds more like Spanish “hermanos.” Speaker 3 also does the Spanish “liaison” with the preceding /s/, as in Colombian Spanish. There is no audible plural marking, which could be linked to the aspiration/elision of the final –s in Caribbean Spanish. However, her final /s/ is clear in all other cases. The actual reason for the missing plural marking can be the fact that there is a number preceding the noun, and plural marking and preceding numbers are incompatible in Papiamentu. (22) tres urmana ‘three brothers’

Interestingly, later in the conversation, she talks about learning Dutch in school, and coming to Aruba to learn Papiamentu. She explains what each sibling has done in school and uses the correct Papiamentu word “rumán.” She frequently uses Spanish phrases / codeswitches, as in the following examples: (23) más o menos ‘more or less’ (24) Pero…bukinan… Eh…a mi lo gusta storia. ‘But…books… Umm… I would like stories.’

The italicized segment in (24) is in Spanish, with an inserted Papiamentu word (storia) with no prosthetic /e/. Spanish phonotactics do not show /st-/ clusters word-initially.


Yolanda Rivera and Patrick-André Mather

In (25), the Papiamentu complementizer sea introduces a Spanish sentence: (25) Sea a mí me gusta… ‘as I like it…’ (26) Bukinan e ta gusta… pero en español ‘Books he likes… but in Spanish’

In (26), Speaker 3 uses the Papiamentu form of a verb with the experiencer as subject (the husband), and with preverbal marker “ta.” However, in the second part of the sentence she switches to Spanish to explain that the stories he reads are in Spanish because there are few books written in Papiamentu. (27) Más inglés… pero mi ta comprende ma español ‘More English… but I understand Spanish better’

In (27), Speaker 3 is switching between Spanish and Papiamentu and explains that he prefers to read in English but that she prefers Spanish since she understands Spanish better. This preference is reflected in the fact that, during conversations in Papiamentu, he switches to English, whereas she switches to Spanish, her L1. Even though Speaker 3 is a native speaker of Spanish, there is some interference from Papiamentu in her Spanish codeswitches. In example 28, the Spanish NP lacks gender agreement (Papiamentu, and most creoles, do not show gender agreement on adjectives). In example 29, the speaker inserts a Papiamentu word (‘skol’) in a Spanish codeswitch: (28) Guitarra eléctrico (Spanish: “guitarra eléctrica”) ‘electric guitar’ (29) Pero…de vez en cuando una skol (Spanish: “escuela”) ‘But…from time to time some school’

5.3.2 Single-word codeswitches from Spanish and English (Speaker 4) Speaker 4, a native speaker of Papiamentu, is a musician, so his vocabulary in this area might be influenced by contact with other musicians, who are speakers of other languages. Even though he is Papiamentu-dominant, he sometimes uses recent borrowings from Spanish, alternating between the native form and the borrowed ones, e.g. ‘famia’ (traditional Papiamentu) versus ‘familiar’ (a recent borrowing from Spanish). Other nonce borrowings include ‘clásico’ (from Spanish) and ‘jeazz’ (from English, including a glide): (30) Tisha gusta clásico. ‘Tisha likes classical music.’

Codeswitching and Borrowing in Aruban Papiamentu


(31) Me gusta jeazz… Mi tata gusta jeazz. ‘I like jazz … my father likes jazz.’

Here, in addition to the English word ‘jazz’, Speaker 4 uses the Spanish personal pronoun ‘me’. It should be added that, in addition to codeswitches from and into English and Spanish, there are occasional single-word switches into Dutch, as in example (32) where Speaker 4 translates the Dutch word for the interviewer: (32) Huiswerk…ta bisa tarea ‘housework … it means ‘tarea’

There are even occasional multiple codeswitches between Papiamentu, Spanish and English, as in example (33): (33) Haci bo maximo, bo best ‘You do the utmost, your best’

In example (34), the English borrowing ‘fault’ is pronounced as [aw], instead of the English vowel [ɔ]. (34) Un na man bai “fault” ‘One signals a “fault” (with the hand)’

In example (35) below, the speaker switches between Papiamentu and English, but word-final voiced stops are devoiced (‘peet’ = ‘speed’) (35) peet reading… Lesa lo ke bo mester ‘speed reading, read whatever you have to read’

In example (36), the speaker begins the sentence in Spanish, switches to Papiamentu, then back to Spanish again. He is talking about the fact that she charges little as compared to others, who charge the same for poor services. (36) En el sentido que otra…peor dona mes su servicio. ‘Meaning that another one… would provide a less efficient service.’

In example (37), the speaker switches to Spanish interrupting his wife, who’s talking about local food. In the second part, after the interruption by children, they talk about food together, rather than in long monologues, like before. However, they talk about local food and there are few references to anything non-local. They use, for the most part, Papiamentu. Nevertheless, sometimes Spanish words show up in his speech when he elaborates on his wife’s descriptions.


Yolanda Rivera and Patrick-André Mather

(37) ora, ora ta pone cazuela. ‘Time, time to set up the cooking pan.’

In example (38), he explains that during San Nicolás (just before Christmas), people came to work at the refinery and they switched between Papiamentu and English, and that currently they call this area “the chocolate city.” (38) Parte ei, ta…e ta name the chocolate city ‘That section there, it is…it is called the chocolate city.’

6. Phonological adaptations The authors conducted detailed observations of recordings based on differences between speaker production and actual Papiamentu features. An analysis of phonetic features, in some cases, complements these observations. We measured formant height and duration in cases of prosthesis, including visual inspection of the spectrograms as well as drops in intensity measurements to determine vowel duration. To analyze devoicing phenomena, we conducted a further visual inspection of the spectrograms. As indicated in Section 5, the phonological adaptations generally include adjustments to fit the Papiamentu syllable structure. However, there are cases of changes to Papiamentu words in the case of the Papiamentu-Spanish bilingual: Speaker 3. This speaker produces sC onsets with prosthetics vowels in contexts of switching from the Spanish indefinite determiner una (there is no -a agreement in Papiamentu) to Papiamentu skol: (39) una əskol ‘a school’

Figure 1. Schwa insertion in una əskol, ‘a school’

Codeswitching and Borrowing in Aruban Papiamentu


The extra vowel is not a full vowel but a schwa (/ə/). Measurements for the first and second formants during the low vowel (/a/) mid portion are: 688.31 Hz and 1688.03 Hz. Measurements for the first and second formants during the prosthetic vowel (/ə/) portion are: 659.60 Hz and 1692.57 Hz. According to previous studies (Lindblom 1963), schwas show great variability, often adapting to surrounding vowels. Schwas are also typically shorter than other vowels. However, the most striking difference between cases of insertion and those that preserve sC sequences is the duration of vowel sequences (Duration of sequence in Figure 1: 229ms.). Compare this case with a similar case without prosthesis (Speaker 3): (40) pero mi ta comprende ma spañol. ‘but I understand Spanish much better.’

Figure 2. No prosthesis for ma spañol, ‘more Spanish’

Measurements for the first and second formants during the low vowel (/a/) portion are: 1122.90 Hz and 2754.68 Hz. The duration of the vowel /a/ in (40) is 58ms. There are other examples of vowel prosthesis (41-42): (41) bai əskol ‘attend school’ (42) hopi especial ‘so/very special’

In (42), there is a possible codeswitch (especial, ‘special’). However, insertion is not consistent throughout the conversation. Sometimes Speaker 3 avoids prosthesis in similar phonological contexts (43-45): (43) bai skol ‘attend school.’


Yolanda Rivera and Patrick-André Mather

(44) Eh… ami lo gusta storia. ‘Umm… I would like stories.’ (45) ba studia ‘is going to study’

On the contrary, cases of prosthesis are not attested for Speakers 1, 2, and 4 (see Appendix A, B and D). Devoicing of final consonants in Spanish and English words is the more salient feature for these speakers. (46) speet reading (English) ‘speed reading’ (47) berdat (from Spanish verdad) ‘truth’

There are cases of coda simplification: (48) exak (English) ‘exact’ (49) dialek (English) ‘dialect’

More phonological examples have been included in Appendix A-D. 7. Conclusions As we have seen in sections 5 and 6, even in a conversation held in Papiamentu, there are frequent borrowings and codeswitches from and into Spanish and English. Furthermore, foreign words and expressions are sometimes phonologically adapted to Papiamentu, but not always consistently, as we have seen in the use of competing forms such as famia/familiar (‘family, relative’). There are also considerable two-way phonetic interferences, from Spanish into Papiamentu and vice-versa. This suggests that in a multilingual society like Aruba, where there is contact in progress, there is no clear-cut distinction between single-word codeswitches and integrated borrowings. Following work by Bullock and Toribio (2009), fluent codeswitching does include overlap and mutual phonological interference between both languages, not only in voice-onset time, but also in the phonotactics of each language, such as the devoicing of final consonants, phonotactic constraints involving the use of prosthetic and epenthetic vowels, and the hardening of dental fricatives.

Codeswitching and Borrowing in Aruban Papiamentu


These results, although limited to four informants, suggest that in this contact situation codeswitching involves much phonological interpenetration of languages. Further work on Spanish/Papiamentu codeswitching is needed in order to cast more light on these phenomena. Nevertheless, this study provides a foundation on which future studies may be built. REFERENCES Alleyne, M. C. (1980). Comparative Afro-American. Ann Arbor: Karoma Publishers, Inc. Andersen, R. W. (1990). “Papiamentu Tense-Aspect, with Special Attention to Discourse”. In J.V. Singler (ed.), Pidgin and Creole Tense-Mood-Aspect. Amsterdam/ Philadelphia: John Benjamins, pp. 59-96. Baum, P. (1976). “The Question of Decreolization in Papiamentu Phonology”. In Linguistics: An International Review, 173, pp. 83-93. Birmingham, J. C., Jr. (1971). The Papiamentu Language of Curaçao. Unpublished Doctoral Dissertation, University of Virginia: Virginia. Blevins, J. (1995). “The Syllable in Phonological Theory”. In J. Goldsmith (ed.), Handbook of Phonological Theory, Oxford: Blackwell, pp. 204-244. Bloomfield, L. (1933). Language. New York: Henry Holt Bullock, B./Toribio, J. (2009). “Trying to Hit a Moving Target: on the Sociophonetics of Codeswitching”. In I. Ludmila, D. Winford, and K. de Bot (eds.), Multidisciplinary Approaches to Codeswitching, pp. 189–206. Chaudenson, R. (1992). Des Îles, des Hommes, des Langues: Essai sur la Créolisation Linguistique et Culturelle. Paris: L’Harmattan. Clements, G. N. (1990). “The Role of the Sonority Cycle in Core Syllabification”. In J. Kingston, and M. Beckman (eds.), Papers in Laboratory Phonology 1: Between the Grammar and Physics of Speech. New York: Cambridge University Press, pp. 283-333. DeBose, C. E. (1975). Papiamentu: A Spanish-based Creole. Unpublished Doctoral Dissertation, Stanford University, California. Èxtra. June 15, (2000). Papiamentu lo Bira Idioma di Instrukshon na Skol: Ta Bai Introdusí Akademia di Lenga i Facultat Genaral. Curaçao, Djaweps 15 Yüni, 2000, p. 40. Gardner-Chloros, P. (1995). “Code-Switching in the Community, Regional and National Repertoires: the Myth of the Discreteness of Linguistic Systems”. In L. Milroy, and P. Muysken (eds.), One Speaker, Two Languages. Cross-disciplinary Perspectives on Code-switching. Cambridge: Cambridge University Press, pp. 68-89. Good, J. (2004). “Tone and Accent in Saramaccan: Charting a Deep Split in the Phonology of a Language”. In Lingua 114, pp. 575-619. Hammond, M. (1997). “Vowel Quantity and Syllabification in English”. In Language 73, 1, pp. 1-17. Harris, C. C. (1951). Papiamentu Phonology. Unpublished Doctoral Dissertation, Cornell University, New York.


Yolanda Rivera and Patrick-André Mather

Hickey, T. (2009). “Codeswitching and Borrowing in Irish”. In Journal of Sociolinguistics 13, 5, pp. 670-688. Holm, J. A. (1988). Pidgins and Creoles. Vol. 1. Cambridge/New York, New Rochelle/ Melbourne/Sydney: Cambridge University Press. Khubchandani, L. M. (1984). “Language Planning Processes for Pluralistic Societies”. In C. Kennedy (ed.), Language Planning and Language Education. London: George Allen and Unwin, pp. 98-110. Klein, T. B. (2011). “Typology of Creole Phonology: Phoneme Inventories and Syllable Templates”. In Journal of Pidgin and Creole Languages 26, 1, pp.155-193. Kouwenberg, S./Murray, E. (1994). Papiamentu. Languages of the World/materials 83. München/Newcastle: Lincom Europa. Lang, G. (1997). “Papiamentu, the Caribbean Paradigm and Cultural Fractality”. In R. A. Young (ed.), Latin American Postmodernisms. Amsterdam: Rodopi, pp. 87-102. Lefebvre, C. (1998). Creole Genesis and the Acquisition of Grammar: The Case of Haitian Creole. Cambridge: Cambridge University Press. Levelt, C./van der Vijver, R. (2004). “Syllable Types in Cross-Linguistic and Developmental Grammars”. In R. Krager, J. Pater, and W. Zonnefeld (eds.), Constraints in Phonological Acquisition. Cambridge: Cambridge University Press, pp. 204-218. Lindblom, B. (1963). “Spectographic Study of Vowel Reduction.” In Journal of the Acoustical Society of America 35, pp.1773-1781. Mahootian, S. (2006). “Code Switching and Mixing”. In K. Brown (ed.), Encyclopedia of Language and Linguistics, Second Edition, Volume 2. Oxford: Elsevier, pp. 511-527. Martinus, E. F. (2004). The Kiss of a Slave: Papiamentu’s West African Connections. Curaçao: Fundasho Kas Di Kultura Kòrsou. Mather, P. A. (2006). “Second Language Acquisition and Creolization: Same I- processes, different E-results”. In Journal of Pidgin and Creole Languages 21, 2, pp. 231-273. Mather, P. A. (2007). “Creole Studies”. In D. Ayoun (ed.), Handbook of French Applied Linguistics, Amsterdam: John Benjamins, pp. 401-424. Maurer, P. (1989). “Les Réiterations et Reduplications Lexicalisés du Papiamento: Influence du Substrat Africain?”. In N. Boretzky, E. vom Werner, and T. Stolz (eds.), Vielfalt de Kontakte. Beiträge zum 5. Essener Kolloquium über “Grammatikalisierung: Natürlichkeit und Systemökonomie”, 18. Bochum: Brockmeyer, pp. 123-138. Migge, B./Léglise, I./Bartens, A. (2010). Creoles in Education: An Appraisal of Current Programs and Projects. Amsterdam/Philadelphia: John Benjamins. Mufwene, S. (2003). “Genetic Linguistics and Genetic Creolistics: a Response to Sarah G. Thomason’s Creoles and Genetic Relationships”. In Journal of Pidgin and Creole Languages 18, 2, pp. 273-288. Muysken, P./Kook, H./Vedder, P. (1996). “Papiamento/Dutch Codeswitching in Bilingual Parent–Child Reading”. In Applied Psycholinguistics 17, 4, pp. 485-505. Myers-Scotton, C. (1992). “Comparing Codeswitching and Borrowing”. In Journal of Multilingual and Multicultural Development 13, 1-2, pp. 19-39. Navarro Tomás, T. (1953). “Observaciones sobre el Papiamento”. In Nueva Revista de Filología Española 22, 7, pp. 184-189.

Codeswitching and Borrowing in Aruban Papiamentu


Perl, M. (1999). “Problemas actuales de la estandarización del papiamento”. In K. Zimmermann (ed.), Lenguas criollas de base lexical española y portuguesa. Madrid/ Frankfurt: Iberoamericana/Vervuert, pp. 251-60. Poplack, S. (1984). “Contrasting Patterns of Codeswitching in Two Communities”. In E. Wande et al. (eds.), Aspects of Multilingualism. Berlin: Mouton de Gruyter, pp. 51-77. Poplack, S. (2004). “Code-switching”. In U. Ammon, N. Dittmar, K.J. Mattheier, and P. Trudgill (eds.), Sociolinguistics. An International Handbook of the Science of Language and Society, Second Edition, Berlin: Mouton de Gruyter, pp. 589-596. Portilla, M. (2009). “Tono y Acento en el Pidgin Afroportugués Americano”. In Filología y Lingüística 35, 1, pp. 139-177. Prince, A./Smolenski, P. (1993). Optimality Theory: Constraint Interaction in Generative Grammar, New Brunswick, NJ: Rutgers University Center for Cognitive Science. Quilis, A. (1982). Curso de Fonética y Fonología Españolas. Madrid: Consejo Superior de Investigaciones Científicas. Remijsen, B./van Heuven, V. (2005). “Stress, Tone, and Discourse Prominence in the Curaçao Dialect of Papiamentu”. In Phonology 22, pp. 205-235. Rivera-Castillo, Y./Pickering, L. (2004). “Phonetic Correlates of Tone and Stress in a Mixed System”. In Journal of Pidgin and Creole Languages 19, 2, pp. 261-284. Rivera-Castillo, Y. (1998). “Tone and Stress in Papiamentu: the Contribution of a Constraint-Based Analysis to the Problem of Creole Genesis”. In Journal of Pidgin and Creole Languages 13, 2, pp.1-38. Römer, R. G. (1980). “Proclisis y Enclisis en una Lengua Tonal”. In Diálogos Hispánicos de Amsterdam 1, pp. 113-123. Römer, R. G. (1983). “Papiamentu Tones”. In E. Muller (ed.), Papiamentu: Problems and Possibilities, Papers Presented at the Conference on Papiamentu on the Occasion of the 65th Anniversary of Madero and Curiels Bank. Zutphen: De Walburg Pers, pp. 85-96. Römer, R. G. (1991). Studies in Papiamentu Phonology (Caribbean Culture Studies). Amsterdam and Kingston: The University of Amsterdam. Rountree, C. S. (1972). “Saramaccan Tone in Relation to Intonation and Grammar”. In Lingua 29, pp. 308-325. Thomason, S. G./Kaufman, T. S. (1988). Language Contact, Creolization and Genetic Linguistics. Berkeley: University of California Press. Voorhoeve, J. (1961). “Le Ton et la Grammaire dans le Saramaccan”. In Word 17, pp. 146-163. Voorhoeve, J. (1968). “Towards a Typology of Tone Systems”. In Linguistics 46, pp. 99-114. Wattman Frank, F. (1974). “Language and Education in the Leeward Netherland Antilles”. In Caribbean Studies 13, 4, pp. 111-117. Weinreich, U. (1953). Languages in Contact: Findings and Problems. New York: Mouton. Wood, R. (1972a). “Hispanization of a Creole Language: Papiamentu”. In Hispania 55, 4. pp. 857-64. Wood, R. (1972b). “New Light on the Origins of Papiamentu: an Eighteenth-Century Letter”. In Neophilologus 56, pp. 18-30.


Yolanda Rivera and Patrick-André Mather


Discourse Context

Linguistic Context




They talk about parties and differences with older people

Word final devoicing

“hubentut” (‘youth’) instead of “hubentud”

Voicing after nasal


They talk about language

Following nasal consonant

diferende, no?, instead of “diferente”



Complete section in English about English speakers’ lack of knowledge of the Dutch language

Word initial

In America, they don’t know a ting about Dutch

Coda simplification


They talk about language

Word final

“dialek”, instead of “dialecto”

No-rhotic schwa but flap after vowel


They talk about language

Before final consonant

dark (/˄/ followed by flap)

Coda simplification


They talk about language

Word final

exak, instead of “exact”

Voicing after nasal


They talk about language

Following nasal consonant

ta interesande, no?



They talk about construction work

Word final

“berdat”, instead of “berdad



Speaker #

Discourse Context

Linguistic Context




They talk about parties and differences with older people

Word final devoicing



They talk about parties

Word final

“tiki borach” (‘a little drunk’) instead of Papiamentu “borache”

Loss final weak vowel

Codeswitching and Borrowing in Aruban Papiamentu


Coda simplification


They talk about language

Word final


Voicing after nasal


They talk about language

Following nasal consonant


They talk about construction work

Change in stress position due to the fact that Papiamentu verbs carry stress in penultimate position. English pronunciation preserved with schwa and retroflex [l] in final position.

“ba ‘rapel” (‘rapelling’)

Change in stress position



Speaker #

Discourse Context

Linguistic Context


No Prosthetic vowel


They talk about family


ba studia

No Prosthetic vowel


They talk about family

Liaison with preceding /i/

bai skol kinan

Prosthetic vowel


They talk about family

Preceded by a front vowel (/i/)

bai (e)skol

Prosthetic vowel


They talk about family

Preceded by a front vowel (/i/)

bai eskol

No Prosthetic vowel


They describe being able to read only in Spanish because very few books are written in Papiamentu

Sentence in Spanish: inserted Papiamentu word with no prosthetic /e/)


No Prosthetic vowel


They describe being more fluent in Spanish than English

Sentence in Papiamentu with insertion of Spanish words

ma spañol

Prosthetic vowel


They start talking about food in Aruba

Papiamentu sentence and Spanish word inserted after final /i/

hopi especial


Yolanda Rivera and Patrick-André Mather

Prosthetic vowel


They describe going to school.

After Spanish definite article ‘una’

una eskol

No Prosthetic vowel


They talk about life in Aruba

No prosthesis, although there is a preceding /e/.

den…, eh, eh… stat

Alternation /d/ and /ɾ/


They talk about life in Aruba

Intervocalic context

un tiki de cara cosa


Speaker #

Discourse Context Linguistic Context


Long vowelschwa


They talk about his wife’s musical taste

Use of Spanish “Me gusta” before switching to English

No Prosthetic vowel


They talk about life in Aruba

Sentence in Papiamentu


No Prosthetic vowel


They talk about education in general

Sentence in Papiamentu


No Prosthetic vowel and devoicing


They talk about reading

English phrase

speet reading

Prosthetic vowel


They talk about reading

After Spanish determiner, on a Spanish noun

otro escritor

Liaison with the initial /s/


They talk about reading

Sentence in Papiamentu

bai mas paspaño

Me gusta jəazz… Mi tata gusta jəazz.


Sandro Sessarego1 and Javier Gutiérrez-Rexach2 University of Texas at Austin and 2The Ohio State University

1. Introduction This article develops an analysis of nominal ellipses in Choteño Spanish (CS) and compares this grammatical phenomenon to its respective counterpart in standard Spanish (stSp) to provide a unified account of N-drop in these two language varieties. CS is an Afro-Hispanic vernacular spoken in Chota Valley, Department of Imbabura (Ecuador), by the descendants of the slaves taken to this region during colonial time to work on plantations (Sessarego 2013). The generative literature on Spanish N-drop and related ellipsis patterns is particularly rich. Several models have been proposed to account for the processes explaining their distribution in the standard variety. Certain authors have identified the richness of Spanish nominal morphology as the property responsible for the licensing of nominal empty categories (Torrego 1988); other accounts have explained the nature of Spanish N-drop by recurring to N movement to DP-internal agreement projections to check nominal phi-features (Kester and Sleeman 2002; Ticio 2003, 2005); finally, others have ascribed nominal properties to the preposition de ‘of’ to account for the cliticization of the definite article to this element (1) (Brucart and Gràcia 1986; Brucart 1987; 1999; Torrego 1988). (1) stSp a. El papel blanco y el [e] de manchas negras the paper white and the [e] of spots black b. *El papel blanco y el [e] con manchas negras the paper white and the [e] with spots black ‘The white paper and the black-spotted one’

Nevertheless, if we take a closer look at N-drop phenomena in CS, we can immediately realize that the aforementioned assumptions on the nature of ellipsis cannot be applied to this Afro-Ecuadorian variety. In fact, ellipsis takes place in CS even though this language is not inflectionally rich (Sessarego 2013). Additionally, in


Sandro Sessarego and Javier Gutiérrez-Rexach

this Afro-Hispanic dialect, N-drop can also occur when the elided category is followed by a preposition other than de ‘of’, namely con ‘with’ (2). (2) CS a. El papel blanco y el [e] de mancha negro the paper white and the [e] of spots black b. El papel blanco y el [e] con mancha negro the paper white and the [e] with spots black ‘The white paper and the black-spotted one’

In the present article, we propose a unified account to explain the nature of nominal ellipsis in these closely related dialects. Sections 2 and 3 present the data on Ndrop and preposition use in CS and stSp. Section 4 summarizes the main proposals that have been put forward in the literature to account for ellipsis phenomena in stSp. Section 5 provides a new, unified explanation to account for the data. Finally, section 6 presents our conclusions. 2. The data Both stSp and CS allow nominal ellipsis across clauses with ‘sloppy identity’ for number features, but not for gender features. As a result, a change in number features allows N-drop, while a change in gender features blocks the elliptical process (cf. Depiante and Masullo 2001; Saab 2004) (3, 4)1. (3) stSp a. El perro negro the-M.SG dog-M.SG black-M.SG ‘The black dog and the white ones’ b. * El perro negro the-M.SG dog-M.SG black-M.SG (4) CS a. El perro negro the-M.SG dog-M.SG black-M.SG ‘The black dog and the white ones’ b. * El perro negro the-M.SG dog-M.SG black-M.SG

y los [e] blancos and the-M.PL [e] white-M.PL y la [e] blanca and the-F.SG [e] white-F.SG y los [e] blanco and the-M.PL [e] white-M.SG y la [e] blanco and the-F.SG [e] white-M.SG

Demonstratives, cardinals, and quantifiers can stay by themselves in their D position (5-6) in the case of nominal ellipsis, while definite articles need to be followed by an AP, a PP or a CP (7).


Note that in traditional CS, in contrast with stSp, gender agreement does not affect adjectives and strong quantifiers. Additionally, number is only marked on determiners (cf. Sessarego 2013: ch.4).

Nominal Ellipses in an Afro-Hispanic Language of Ecuador


(5) stSp a. No compro dos chocolates, compro tres no buy two chocolates buy three ‘I do not buy two chocolates, I buy three’ b. No compro estos chocolates, compro aquellos no buy these chocolates buy those ‘I do not buy these chocolates, I buy those’ (6) CS a. No compro dos chocolate, compro tres. two chocolate buy three no buy ‘I do not buy two chocolate, I buy three’ b. No compro estos chocolate, compro aquellos. no buy these chocolates buy those ‘I do not buy these chocolates, I buy those’ (7) stSp/CS a. El perro negro y el [e] blanco the dog black and the [e] white b. El perro negro y el [e] de color blanco the dog black and the [e] of color white c. El perro negro y el [e] que es de color blanco the dog black and the [e] that is of color white ‘The black cat and the white one’ d. *El perro negro y el [e] the dog black and the [e]

Nevertheless, not all adjectives behave in the same fashion; while post-nominal adjectives allow nominal ellipses, pre-nominal ones block ellipsis operations (8). (8) stSp/CS a. *El verdadero amigo alto y el supuesto [e] bajo the true friend tall and the supposed [e] short ‘The true tall friend and the supposed short one’ b. El amigo alto y el [e] bajo the friend tall and the [e] short ‘The tall friend and the short one’

It must be pointed out that D + [e] + PP constructions (see 7b) are subject to certain grammatical constraints since not all Ps can be used in these sentences. In stSp, the only P capable of following [e] in such a configuration is de ‘of’ (9); while in CS, N-drop can occur when the elided N is followed by de ‘of’ and con ‘with’ (10). (9) stSp a. El bocadillo de jamón y el [e] de chocolate the sandwich of ham and the [e] of chocolate ‘The ham sandwich and the chocolate one’


Sandro Sessarego and Javier Gutiérrez-Rexach

b. * El bocadillo con jamón y el [e] con chocolate the sandwich with ham and the [e] with chocolate ‘The ham sandwich and the chocolate one’ c. *El bocadillo para mí y el [e] para ti the sandwich for me and the [e] for you ‘The sandwich for me and the one for you’ (10) CS a. El bocadillo de jamón y el [e] de chocolate the sandwich of ham and the [e] of chocolate ‘The ham sandwich and the chocolate one’ b. El bocadillo con jamón y el [e] con chocolate the sandwich with ham and the [e] with chocolate ‘The ham sandwich and the chocolate one’ c. *El bocadillo para mí y el [e] para ti the sandwich for me and the [e] for you ‘The sandwich for me and the one for you’

De is a multifunctional preposition in stSp and CS. It heads a wide range of PPs, which are assigned a variety of theta-roles: possessors (11a), agents (11b), and objects (11c). It also takes as complements a variety of non-thematic PPs (12). A restriction applying to stSp and CS N ellipses is the need for a thematic-role correspondence between the coordinated clauses (compare: 11a, b, c vs. 11d, e). (11) stSp/CS a. El libro suyoposs y el [e] tuyoposs the book his and the [e] your ‘His book and your book’ b. El libro de Cervantesagent y el [e] de Juanagent the book of Cervantes and the [e] of Juan ‘Cervantes’ book and Juan’s one’ c. El libro de físicaobject y el [e] de sintaxisobject the book of physics and the [e] of syntax ‘The physics book and syntax one’ d. *El libro de Maríaagent/poss y el [e] de físicatheme the book of María and the [e] of physics ‘María’ book and the physics one’ e. *El libro de fisicaobject y el [e] de Maríaagent/poss the book of physics and the [e] of María ‘The physics book and María’s one’

The thematic constraints conditioning the grammaticality of the examples presented in (11) only apply to theta-marked PPs. On the other hand, PPs free from theta-assignment –i.e. those without thematic or argumental relation with the head noun– may occur more freely (12).

Nominal Ellipses in an Afro-Hispanic Language of Ecuador


(12) stSp/CS a. Dame el libro de Juanagent/poss y el [e] de color rojo give-me the book of Juan and the [e] of color red ‘Give me Juan’s book and the red one’ b. Dame el libro de Petrarcaagent y el [e] de color rojo give-me the book of Petrarca and the [e] of color red ‘Give me Petrarca’s book and the red one’ c. Dame el libro de físicatheme y el [e] de color rojo give-me the book of physics and the [e] of color red ‘Give me the physics book and the red one’

Finally, only CPs introduced by que ‘that’ are allowed as relative clause remnants; other Cs result in ungrammatical constructions (13). (13) stSp/CS a. El hombre que es bueno y el [e] que es malo the man who is good and the [e] who is evil ‘The man who is good and the one who is evil’ b. *El bar donde ella toma y el [e] donde ella trabaja the bar where she drinks and the [e] where she works ‘The bar where she drinks and the one where she works’

In conclusion, the data presented thus far show that N-drop phenomena in CS and stSp obey the same constraints on ellipsis. The only difference that we could identify concerns DPs headed by a definite article and including nouns followed by a con-PP. In fact, in CS such constructions allow N-drop, while in stSp the elision of N would yield an ungrammatical structure. One of the grammatical aspects of CS that diverges quite significantly from stSp is the use of prepositions. Section 3 will provide a brief summary of the main attested differences between these languages with respect to this property. 3. Preposition uses in CS and stSp As shown in Sessarego (2013: ch.4), prepositions in CS are often omitted (14-15). (14) a. CS: Yo vivo lejos las casita I live far away the little houses b. stSp: Yo vivo lejos de las casitas I live far away from the little houses ‘I live far away from the little houses’ (15) a. CS: Yo soy abajo I am down ‘I am from down there’ b. stSp: Yo soy de abajo I am from down ‘I am from down there’ (cf. Lipski 1987: 163)


Sandro Sessarego and Javier Gutiérrez-Rexach

Moreover, CS a ‘to’ is often used as en ‘in’ in stSp, with a locative function (16). (16) a. CS: Mi novio vive a Guayaquil My boyfriend live to Guayaquil b. stSp: Mi novio vive en Guayaquil My boyfriend live in Guayaquil ‘My boyfriend lives in Guayaquil’

The preposition con is also used in a differential fashion; it can replace stSp y ‘and’ (17). (17) CS Yuca con arroz, eso se come mucho. Yuca with rice this reflex. eat much ‘Yuca and rice, this is a common dish’

Most interestingly, CS con is used in many constructions in which de would be the preferred preposition in stSp (18). (18) a. CS: Hombre con esta edad no tiene que trabajá man with this age not have that work b. stSp: Un hombre de esta edad no tiene que trabajar a man of this age not have that work ‘A man that old should not work’ c. CS: La mujer con ojos verde tomó mucho the woman with eyes green drunk much d. stSp: La mujer de ojos verdes tomó mucho the woman of eyes green drunk much ‘The green-eyed woman drank a lot’

It may be of interest to point out that in all these cases CS parallels Afro-Bolivian Spanish (ABS) perfectly (19) (cf. Gutierrez-Rexach & Sessarego forthcoming for a similar analysis of N-drop phenomena in ABS). (19) ABS a. Huahua cun eje edad pesa veinte kilo Kid with this age weighs twenty kilo ‘A kid at this age weighs twenty kilos’ (Sessarego 2011: 54) b. Mururata cun Chijchipa, nojoto siempre fue uno nomá Mururata with Chijcipa we always were one no more ‘Mururata and Chijcipa we have always been just one’ (Lipski 2008: 132) c. Juan nació a La Paz Juan was born to La Paz ‘Juan was born in La Paz’ (Sessarego 2011: 53)

Nominal Ellipses in an Afro-Hispanic Language of Ecuador


In summary, this section has shown that the use of prepositions in stSp and CS diverges in several ways. An interesting difference, which may explain why con-PPs allow N-drop to obtain in CS in constructions like ‘definite article + [e] + PP,’ has to do with the fact that con ‘with’ oftentimes occurs in constructions which would use de ‘of’ in stSp. In all these examples, con appears to head a reduced relative clause similar to que tiene ‘that has’ (cf. Kayne 1994). In other words, con works as a C/P, thus it should be analyzed as a prepositional complementizer and consequently as a functional head, while other prepositions such as para ‘for’ would be lexical heads (cf. §4.2 below). 4. Literature review on stSp N-drop This section provides an overview of the main models that have been proposed to account for stSp N-drop. The hypotheses presented here will be tested on the data provided in the previous section to assess their validity. 4.1. Brucart (1987) Brucart (1987) offers an account of N-drop in Spanish. The author contrasts the grammatical examples shown in (20), where the definite article is followed by an AP (20a), a de-PP (20b), or a CP (20c), with the ungrammatical structures lacking an overt determiner (21a), lacking a remnant complement (21b), or presenting a PP complement headed by a preposition other than de (21c). (20) stSp a. Mi cuñado utiliza e coche antiguo para ir a trabajar y el [e] nuevo para trasladarselos fines de semana a su casa de campo ‘My brother in law uses the old car to go to work and the new one to go to his countryside house during the weekend’ b. El hijo de Luis y el [e] de Antonio se han hecho muy amigos ‘Luis’s son and Antonio’s one became good friends’ c. La casa que visitaste ayer y la [e] que has visto esta mañana pertenecieron a un mismo dueño ‘The house that you saw yesterday and the one that you saw this morning belonged to the same owner’ (21) stSp a. *El hijo de Luis y [e] se han hecho muy amigos ‘Luis’s son and Antonio’s became good friends’ b. *El hijo de Luis y el [e] se han hecho muy amigos ‘Luis’s son and Antonio’s became good friends’ c. *El tren a Barcelona y el [e] a Madrid ha salido con retraso ‘The train for Barcelona and the one for Madrid left late’


Sandro Sessarego and Javier Gutiérrez-Rexach

One of the main points made by Brucart is that an empty N preceded by a definite article should be postulated to account for the aforementioned grammatical examples (21). This would go against previous models, which claimed that in such cases definite articles would act as pronouns (cf. Bello 1847). Brucart also postulates the existence of an empty N in constructions like (22), in contrast with Alarcos-Llorach (1973), who suggests that in such constructions the head would be the preposition following the definite article. (22) stSp El de filosofía ‘The one of philosophy’

Brucart supports these claims by pointing out that postulating the existence of unstressed articles acting as pronouns in (20) would be identical to doubling the number of personal pronouns, a counterintuitive and non-economical proposal. Likewise, claiming that de is the head in (22) would go against the basic assumptions of X’-Theory, which indicate that heads and their projections must belong to the same category; thus, an empty N must be postulated to account for such a construction. The author also highlights the difference between definite articles and other determiners. Definite articles, in fact, due to their unstressed nature, must cliticize on some DP elements in order to lead to a grammatical elliptical construction (23). (23) stSp a. *Buscaba el [e] sought the b. Buscaba uno [e] sought a ‘I was seeking one’ c. Buscaba éste [e] sought this ‘I was looking for this’ d. Buscaba alguno [e] sought some ‘I was seeking some.’ e. Buscaba tres [e] sought three ‘I was looking for three’

Brucart also points out that the only PPs capable of following a definite article in elliptical constructions are de-PPs. Such a restriction does not apply to other D elements (24).

Nominal Ellipses in an Afro-Hispanic Language of Ecuador


(24) stSp a. Compré el/éste/alguno/uno/tres [e] de matemáticas bought the/ this/ some/one/three [e] of mathematics ‘I bought the/ this/ some/one/three of mathematics’ b. Compré *el/éste/alguno/uno/tres [e] con lazo bought the/ this/ some/one/three [e] with lace ‘I bought the/ this/ some/one/three with lace’

The grammaticality of all the constructions in (24a), in contrast with the ungrammaticality of the structure with el ‘the’ in (24b), would be due to the fact that all the determiners (except for the definite articles) allow the recovery of their antecedent and subsequently enable [e] to participate in the assignment of thematic roles across ‘true’ (con) and ‘false’ (de) prepositions. On the other hand, the definite article does not trigger such an operation; hence, the following constituent would be able to receive thematic assignment only if preceded by a ‘false’ preposition like de (Brucart 1987: 245). Brucart’s account is valuable in that it recognizes the existence of empty Ns as heads of nominal phrases and provides a framework to account for differences between definite articles and other determiners. Nevertheless, it leaves certain phenomena unexplained: it does not acknowledge the difference in grammaticality due to variation in gender and number features and it does not provide an account for pre-nominal and post-nominal adjectives. Moreover, it fails to provide an explanation for thematic correspondences between pronounced and elided Ns. 4.2. Kester and Sleeman (2002) Kester and Sleeman (2002) refine Torrego’s (1988) analysis, which accounted for the grammaticality and ungrammaticality of N-drop constructions in Spanish. According to Torrego, definite articles, differently from other D categories, lack person features and therefore are not rich enough to appear by themselves in D position. Rather, they must have such a feature supplied by a [+N] category, which may be a noun, an AP, a que-CP or a de-PP. After presenting Torrego’s model, Kester and Sleeman (2002) offer a significantly different analysis, which adopts Kayne’s (1994) Antisymmetry framework. Kester and Sleeman do not explain the peculiar behavior of definite articles in terms of [person] features. Rather, they claim that the Spanish definite article cannot appear in isolation because it is weaker in a semantic sense. Therefore, in order to yield a semantically interpretable construction, it has to co-occur with a predicate (2002: 111). In their view, the empty noun must be considered as part of a CP acting as the complement of the definite article. The authors claim that


Sandro Sessarego and Javier Gutiérrez-Rexach

“within this clausal constituent the null noun has to enter in a checking relation, at some point of the derivation, with the highest overt functional head in order to be licensed” (2002: 116). Thus, the differences in grammaticality between dePPs, que-CPs and APs versus other instances of PPs and CPs would be formally represented as the contrasts highlighted in (25) and (26). (25) a. el [D/PP proi [D/P° de [IP Juan [I [ei] … ]]]] b. el [CP proi [C° que [IP nos regaló [ei] tu padre]]] c. la [CP proi [C° [IP [ei] [I° [amarilla]]]]] (26) a. *el [CP [proi] [C° [IP [ei] [I° [PP ei para Jaime]]]]] b. *el [CP proi [C° donde [IP fuimos [ei] ]]]

Kester and Sleeman account for the data by postulating, in line with Kayne (1994), that the definite article is followed by a clausal complement corresponding to either a CP or a D/PP (see 25). Within this model, an empty N would thus move to the specifier position of such a clause and instantiate a Spec-Head relation. The differences in grammaticality between de-PP/que-CP (28 a, b), on one hand, and para-PP/donde-CP (29 a, b), on the other, come from the fact that de and que are analyzed as functional heads, while para and donde are analyzed as lexical ones. Para- and donde-CPs block N-drop because feature checking only takes place in functional projections (Chomsky 1995). Example (25c) yields a grammatical construction as well. In this case, the empty N would be able to enter in a checking relation with a functional head, which consists of the head contained in the functional projection Agreement AP; thus, (25c) may be reformulated as (27). (27) la [CP proi [IP ei [AgrAP ei [AgrA amarillaj [AP ei ej ]]]]]

Kester and Sleeman’s model provides an interesting hypothesis for analyzing N-drop in stSp. These authors, in fact, are able to account for the differences in grammaticality between DPs headed by definite articles and other determiners by applying Kayne’s (1994) Antisymmetry framework. Nevertheless, their model leaves several issues unresolved. First of all, this account does not take the theta-marking constraints exemplified in examples (11-12) into consideration. Secondly, the postulation of an AgrP to which N should move appears counterintuitive since we know that sloppy identity across clauses is allowed for number but not for gender, thus suggesting that these two features should not be checked in the same syntactic position. Finally, as in this study, there is no mention of the ungrammaticality of N-drop with prenominal As.

Nominal Ellipses in an Afro-Hispanic Language of Ecuador


4.3. Ticio (2003, 2005) Ticio (2003, 2005) elaborates on Grohmann and Haegeman’s (2002) analysis of a layered CP and postulates the parallel presence of three different syntactic domains within the DP. In her view, a domain would concern theta-role assignment; another would be devoted to phi-agreement and case relations; while the last one would encode discourse information. Within Ticio’s model, adjectives would occupy different positions in the syntactic structure depending on their semantic properties. For this reason, prenominal As would be located in Spec-NP, postnominal As are adjoined to NP, and R(elational) As are generated in Spec-nP. (28) DP




Agr’ POSS Agr



n NP

NP ADJUNCTD/Pronominal As

Prenominal As



By postulating that the ellipsis operation targets only the lower NP node, Ticio provides an explanation for the different behavior of pronominal and postnominal As in elliptical constructions (see 29, cf. Ticio 2005: 136). nP

(29) (…)




NP ADJUNCTD/Postnominal As

Prenominal As



Ellipsis site


Sandro Sessarego and Javier Gutiérrez-Rexach

Ticio (2005: 136-137) shows how her account can explain the (un)grammaticality of a variety of elliptical constructions involving Adjuncts, Postnominal As, Prenominal As, Possessors, Agents, etc. Nevertheless, she acknowledges that, given the structure in (29), an example like (30) should not be grammatical since objects should be targeted by elliptical processes. (30) stSp Compramos varios libros de Matemáticas y alguno [e] [de Física]obj (we)bought several books of Math and some [e] of Physics ‘We bought several Math books and a Physics one’

Ticio accounts for this case by postulating a stylistic rule that would be able to move the object from its original site to a higher position in the derivation before the ellipsis operation applies, as shown in (31). (31) stSp a. Varios [e] de Físicaobj de Juanposs several [e] of Physics of Juan b.



nP of Juan NP


of Physics

n NP

NP ADJUNCTD/Postnominal As

Prenominal As



of Physics

Ticio provides a model to explain why only PPs headed by de can be found as remnants when the empty N is preceded by a definite article, which acts as a clitic on [+N] elements (adjectives and nouns) (cf. also Brucart and Gràcia 1986, and Raposo 1999). According to this author, cliticization cannot occur across phase boundaries because they represent spell out units. In her view, two different types of PPs are found in Spanish: full PPs, which constitute phases (cf. Chomsky 2001, Gallego 2010) headed by Ps such as con ‘with’, and ‘false’ PPs (which do not constitute phases), headed by de ‘of’. This would explain the ungrammaticality of (32) and the grammaticality of (33). In fact, in (32) the presence of con gives rise

Nominal Ellipses in an Afro-Hispanic Language of Ecuador


to a phase boundary at Spell-Out, which prevents article cliticization. On the other hand, in (33), the article can freely cliticize, since de is inserted after Spell-Out and does not introduce a new phase unit. (32) stSp a. [DP La [[NP chica] || [PP con gafas]]] Spell Out b. [DP La [[NP e]] || [PP con gafas]]] NP-Ellipsis c. *[DP La [[NP e]] || [PP con gafas]]] Cliticization (33) stSp a. [DP La [[NP chica] [NP gafas]]] Spell Out b. [DP La [[NP e] [NP gafas]]] NP-ellipsis c. [DP La [[NP e] de-[NP gafas]]] DE(of)-insertion d. [DP La[[NP e] + de-[NP gafas]]] Cliticization

Ticio’s account has the advantage of acknowledging that not all adjectives behave in the same way in elliptical constructions and it points out the existence of thematic-roles within the DP in relation to N-drop phenomena. Nevertheless, this model also leaves several issues unresolved. First, postulating pre-nominal and post-nominal adjective base-generated positions is problematic in that it goes against hypotheses assuming that all adjectives are originally pre-nominal and the overt distribution is just a result of N movement (Cinque 1993, 2005, 2007)2. Second, as with the model proposed by Kester and Sleeman (2002), the postulation of an Agreement Projection cannot account for sloppy identity phenomena related to gender and number features. Third, the movements of objects before ellipsis and de insertion at PF seem to be somewhat ad-hoc solutions and are not independently motivated operations from a theoretical viewpoint. In fact, PF should not be able to read narrow-syntax features, but only phonological matrixes. Moreover, Ticio’s model assumes a right-branching structure with right-sided specifiers and rightward movement operations that appear to contradict standard assumptions on left branching and dislocation; it also violates the Linear Correspondence Axiom (Kayne 1994). 5. Toward a unified account of N-drop Our account combines certain aspects of the literature on N-drop with several consolidated assumptions on the nature of DP, without relying on ad-hoc insertion or movement operations. In line with Kester and Sleeman (2002) and with our recent account of Afro-Bolivian Spanish (ABS) N-drop (cf. Gutiérrez-Rexach & Sessar


For a different account of adjective position and movement within the DP see also Gutiérrez-Rexach and Mallén (2001)


Sandro Sessarego and Javier Gutiérrez-Rexach

ego 2014), we claim that the semantic and syntactic differences between definite articles and other determiners play a crucial role in the patterns attested in CS and stSp. In fact, from a semantic point of view, definite articles are not truly quantificational: they only contribute a variable to the semantic representation (cf. Heim 1982 and many others). Syntactically, it can be said that other determiners are encoded with a [+referential] feature, while definite articles lack such an encoding. As pointed out by Kester and Sleeman (2002), definite articles are restricted elements in a semantic sense, which require the presence of a predicate to yield to semantically interpretable constructions. Conversely, demonstratives, cardinals, universal determiners, etc. are base-generated under their respective projections or possibly move to Spec DP to check their referential feature. They are capable of “freezing” the DP, thus they act as pronoun-like elements, regardless of the existence of an overt N3. See Cardinaletti and Giusti (1991); Zamparelli (2000); Ishane (2008); Gutiérrez-Rodríguez (2009) for an account of base-generation and movement of these elements within the DP. Thematic layers project syntactically in the DP (Ticio 2005), instantiating a hierarchy such as the one in (34), proposed by Giorgi and Longobardi (1990) and Valois (1991) among others. (34) Possessor>>Agent>>Object

The preposition de is a multifunctional preposition, capable of introducing a variety of non-thematic and thematic PPs, which are assigned under nP (Carstens 2000). In line with Kayne (1994), we assume that all PPs and APs are adjoined to the left and that the surface order is the result of leftward N movement and constituent dislocation. Adjectives and de-phrases carrying thematic roles are basegenerated within the nP shells and can move to higher projections in the course of the derivation4. On the other hand, the rest of the post-nominal adjectives (e.i. el libro rojo), as well as descriptive de-phrases free from thematic assignment (e.i. el libro de color rojo) are generated higher in the structure. The fact that a strict correspondence between thematic roles is required between the pronounced and the elided constituents suggests that the thematic specification generated under nP is frozen in place, and the associated elements cannot be displaced from this constituent. Thus, there is a matching requirement according to which the thematic configuration of the elided N and the source nominal have to be identical, as shown in (11-13). Given these data, our proposal claims that As and PPs (when free from theta-assignment) can be claimed to be base-generated higher in the

3 4

See Leonetti (1999) for a more detailed account on this issue. The order is obtained by phrasal rising operations (cf. Cinque 2010; Picallo 2010).

Nominal Ellipses in an Afro-Hispanic Language of Ecuador


syntactic structure, in a location different from that of the theta–marked elements, which is immune to ellipsis operations. An advantage of this account over previous models is that it does not rely on ad hoc object movement or procrastinated de insertion at PF (Ticio 2003; 2005), a desirable aspect since principles such as Greed or Procrastination, although embraced in the initial versions of minimalism, have several ‘look-ahead’ characteristics which are undesirable from a computational point of view. Moreover, assigning special [+N] features to de and que (Torrego 1988) is not required either. N-drop operations appear to involve the recovery of a nominal element and the information contained in the lower nP shells, from the lower thematic nP (object) up to the higher nP (possessor). As a result, the information contained in higher projections (e.i. NumP, non-thematic-APs/de-PPs) does not have to match and can be variable. This is in line with models viewing ellipsis resolution as a semantic process similar to anaphora resolution (cf. Merchant 2001 among others). Thus, there is a parallelism in the ellipsis behavior of CS and stSp despite the fact that the former has impoverished number specifications. Since ellipsis applies at the lower nP level, it is not affected by differences at the higher functional projections. Additionally, this lends support to the claim that ellipsis does not involve movement to, or deletion of, agreement projections, since this would predict deeper differences between CS and stSp. For example, ellipsis with asymmetric Gender specifications should be possible in CS, contrary to fact, as shown in (3, 4). Furthermore, since nouns are not overtly specified for number, ellipsis of plural nominals with determiner remnants should not be possible either, again contrary to fact (5, 6). To account for the differences in grammaticality between stSp and CS con, we consider con in CS to share many of its structural features with stSp de, as suggested by the cross-linguistic data presented in section 3. In other words, con in CS is a functional preposition with C features, i.e. a prepositional complementizer (C/P) (Cf. Kayne 1994). As such, it has a functional layer where the null pronominal to be resolved by the ellipsis process can check its relevant functional features (category, etc.), as proposed by Kester and Sleeman (2002). This layer is different from the one in the higher embedding DP, which, as we have proposed, is not targeted by the ellipsis resolution process and consequently cannot support feature checking of the empty pronominal. 6. Conclusions This article has provided a unified account of N-drop phenomena in CS and stSp. The study of CS is of significance because it is a dialect with several important peculiarities (lack of or impoverished overt agreement, etc.), which should make


Sandro Sessarego and Javier Gutiérrez-Rexach

ellipsis processes markedly different, according to existing theories. Nevertheless, this is not so and the available arrangements can be explained by general principles on ellipsis resolution and antisymmetric derivations. The actual differences between stSp and CS can be explained as a by-product of the differential specification of syntactic resources while processes and operations apply uniformly. REFERENCES Alarcos-Llorach, E. (1973). Estudios de gramática funcional del español. (BRH, Estudios y ensayos, 147). Madrid: Gredos. Bello, A. (1847). Gramática de la lengua castellana destinada al uso de los americanos. Santiago de Chile: Imprenta de Progreso Brucart, J. M./Gràcia M. L. (1986). “I Sintagmi Nominali Senza Testa”. In Rivista di Grammatica Generativa 11, pp. 3-32. Brucart, J. M. (1987). La elisión sintática en español. Doctoral Dissertation. Universitat Autònoma de Barcelona. Brucart, J. M. (1999). “La Elipsis”. I. In Bosque, and V. Demonte (eds.), Gramática descriptiva de la lengua española. Vol. II. Madrid: Espasa, pp. 2787-2866. Cardinaletti, A./Giusti, G. (1991). “Partitive ne and the QP-Hypothesis. A Case Study”. In University of Venice Working Papers in Linguistics, Venice, pp. 1-19. Carstens, V. (2000). “Concord in Minimalist Theory”. In Linguistic Inquiry 31, 2, pp. 319-355. Chomsky, N. (1995). The Minimalist Program. Cambridge, Mass.: MIT Press. Chomsky, N. (2001). “Derivation by Phase”. In M Kenstowicz (ed.), Ken Hale: A Life in Language. Cambridge, Mass: MIT Press, pp. 1-52. Cinque, G. (1993). “On the Evidence for Partial N-Movement in the Romance DP”. In Venice Working Papers in Linguistics 3, 2, pp. 21-40. Cinque, G. (2005). “Deriving Greenberg’s Universal 20 and Its Exceptions”. In Linguistic Inquiry 36, pp. 315-332.  Cinque, G. (2007). “The Fundamental Left-Right Asymmetry of Natural Languages”. In University of Venice Working Papers in Linguistics 17, pp. 77-107. Cinque, G. (2010). The Syntax of Adjectives. Cambridge, Mass: MIT Press. Depiante, M./Masullo, P. (2001). “Género y número en la elipsis nominal: consecuencia para la hipótesis lexicalista”. Paper presented at I Encuentro de Gramática Generativa, Gral Roca, 22-24 November 2001. Gallego, Á. (2010). Phase Theory. Amsterdam/Philadelphia: John Benjamins. Giorgi, A/Longobardi, G. (1990). The Syntax of Noun Phrases, Cambridge: Cambridge University Press. Grohmann, K. K./Haegeman, L. (2002). “Resuming Reflexives”. In Proceedings of the 19th Scandinavian Conference in Linguistics, Tromsø, Norway, pp. 46-62. Gutiérrez-Rexach, J./Mallén, E. (2001). “NP Movement and Adjective Position in the DP Phases”. In J. Herschenshon, E. Mallén, and K. Zagona (eds.), Features and In-

Nominal Ellipses in an Afro-Hispanic Language of Ecuador


terfaces in Romance: Essays in Honor of Heles Contreras. Amsterdam/Philadelphia: John Benjamins, pp. 107-132. Gutiérrez-Rexach, J./Sessarego, S. (2014). N-drop Parallelisms in Afro-Bolivian Spanish and Standard Spanish: A Microparametric Account. In Orozco R. (ed.). New Directions in Hispanic Linguistics. Newcastle: Cambridge Scholars Publishing, pp. 188-216. Gutiérrez-Rodríguez, E. (2009). Rasgos gramaticales de los cuantificadores débiles. Madrid: Universidad Complutense de Madrid. Doctoral Dissertation. Heim, I. (1982). The Semantics of Definite and Indefinite Noun Phrases. Doctoral Dissertation. Amherst: University of Massachusetts at Amherst. Ishane, T. (2008). The Layered DP: Form and Meaning of French Indefinites. Linguistik Aktuell / Linguistics Today, Vol. 124, Amsterdam/Philadelphia: John Benjamins. Kayne, R. (1994). The Antisymmetry of Syntax. Cambridge, Mass: MIT Press. Kester, E. P./Sleeman, P. (2002). “N-Ellipsis in Spanish”. In Linguistics in the Netherlands 2002, pp. 107-116. Amsterdam/Philadelphia: John Benjamins. Leonetti, M. (1990). El artículo y la referencia. Madrid: Taurus Universitaria. Lipski, J. (1987). The Chota Valley: Afro-Hispanic language in highland Ecuador. Latin American Research Review, 22, pp.155-170. Lipski, J. M. (2008). Afro-Bolivian Spanish. Madrid/Frankfurt: Iberoamericana/Vervuert. Merchant, J. (2001). The Syntax of Silence. Oxford: Oxford University Press. Picallo, C. (2010). “Structure of the Noun Phrase”. In Handbook of Hispanic Linguistics. Oxford: Blackwell. Raposo, E. (1999). “Towards a Minimalist Account of Nominal Anaphora in Spanish and English”. In Saab, A. (2004). El dominio de la elipsis nominal en español: identidad estricta e inserción tardía. Masters Thesis. Universidad Nacional del Comahue. Sessarego, S. (2011). Chota Valley Spanish. Madrid/Frankfurt: Iberoamericana/Vervuert. Sessarego, S. (2013). Introducción al Idioma Afroboliviano: Una Conversación con el Awicho Manuel Barra. Cochabamba/La Paz: Plural Editores. Ticio, M. E. (2003). On the Structure of DPs. Doctoral Dissertation, Storrs: University of Connecticut. Ticio, M. E. (2005). “NP-Ellipsis in Spanish”. In Proceedings of the 7th Hispanic Linguistics Symposium 7, pp. 128-141. Torrego, E. (1988). On Empty Categories in Nominals. ms., Boston: University of Massachusets. Valois, D. (1991). The Internal Structure of DP. Doctoral Dissertation, Los Angeles: UCLA. Zamparelli, R. (2000). Layers in the Determiner Phrase. New York: Garland.


1. Introduction A multitude of previous studies have examined the possible existence of a Spanish-based creole in the formation of Caribbean Spanish (e.g., Álvarez and Obediente 1998; Díaz-Campos 1998; Díaz-Campos and Clements 2005, 2008; Laurence 1974; Lipski 1993, 2000, 2005; McWhorter 1995, 2000; Megenney 1985, 1989, 1999; Otheguy 1973; Schwegler 1996). Díaz-Campos and Clements (2008) previously presented sociohistorical and linguistic evidence refuting McWhorter’s view of colonial Venezuela as a place with the ideal conditions for the development of a creole language. In particular, they argue that the disproportion between Africans and Europeans fostered by the plantation system did not exist in colonial Venezuela, specifically in the coastal area of Barlovento. What remains to be examined in elaborating an explanation of the potential (or not) for creole language development in colonial Venezuela is the available sociohistorical and linguistic evidence from areas of predominantly African influence. The potential for creole genesis in the Barlovento area of Venezuela may have been feasible in cimarroneras and cumbes where enclaves of Africans who had revolted and/or fled from slave masters could have fostered the appropriate conditions for the development of a creolized Spanish. The present study provides an analysis of available demographic, sociohistorical, and linguistic evidence to contribute to the ongoing discussion of the potential formation of Afro-Caribbean Spanish. 2. Overview As pointed out by Díaz-Campos and Clements (2005), creoles based on English, French, Portuguese, and Dutch have raised questions about the origin of Latin American Spanish and the contribution of African populations to its formation.


Avizia Yim Long and Manuel Díaz-Campos

With regard to the latter, there are considerably fewer accounts of how diverse populations in the Spanish Caribbean played a role in the formation of the spoken varieties encountered in those regions. These accounts analyzing the African contribution generally fall into one of the three following perspectives (DíazCampos and Clements 2005): scholars who argue for a possible creole or semicreole origin (e.g., Álvarez and Obediente 1998; Megenney 2000), scholars who hypothesize that Africans living in the Caribbean acquired Spanish as a second language (e.g., Laurence 1974; Lispki 2005), and scholars who propose that a new variety emerged as a result of the languages in contact (Spanish, African, indigenous), that is, through a process of koineization (e.g., De Granda 2001; Parodi 2001). As stated in the introduction, the present study will examine the potential for creole genesis in the Barlovento area of Venezuela, specifically in cimarroneras or enclaves of runaway African slaves living outside of Spanish rule. To best situate the analysis in the existing literature, the following subsections provide a brief review of previous analyses of Barlovento Spanish as well as examinations of a possible creole origin for Barlovento Spanish. 2.1 Barlovento Spanish The Barlovento region of Venezuela is located along the coast of north central Venezuela in the state of Miranda. It is located approximately 130 kilometers southeast of the capital city of Caracas, also located near the coast in the Capital District. Geographically, Barlovento is located between two parallel mountain ranges – Serranía del Litoral in the north and Serranía del Interior in the south – that form part of the coastal range of Venezuela (Guerra Cedeño 1984). This geographic placement makes the Barlovento region susceptible to heavy annual rainfall triggered by coastal winds, which in turn gives way to fertile valleys and dense vegetation (Guerra Cedeño 1984). In addition, temperatures often reach 40 degrees Celsius (104 degrees Fahrenheit). During colonial times, the Barlovento region remained largely unpopulated due to the aforementioned geographic and climatic conditions. It was not until the beginning of the 16th century that the Barlovento region was exploited for the economic motivations presented by the Spanish crown, specifically the cultivation of cacao and coffee for which thousands of African slaves were imported to work the land (Guerra Cedeño 1984). Nowadays, the Barlovento region contrasts strikingly with the capital, given its rural surroundings and its underdeveloped infrastructure. The region also differs in terms of its population since African heritage is prevalent, while in others areas of the country mixed race communities are the norm.

Cimarroneras in Venezuela: The Role of Isolated Communities


Figure 1. Map of Miranda State highlighting Barlovento area1

The work of Megenney (1985, 1989, 1999) provides the most comprehensive analysis of potential linguistic contributions of African languages to the development of the variety of Spanish spoken in the Barlovento area. He focuses on five phonological processes to argue for the existence of a pidgin Spanish in the Caribbean more generally: fortition of [ð] > [d] and [ɣ] > [ɡ] after a flap ([ɾ]), lenition of [f] > [h] in intervocalic position, aspiration and deletion of syllablefinal /s/, nasalization of vowels adjacent to nasal segments, and velarization of syllable-final nasals (see Díaz-Campos and Clements 2005, 2008, for a concise review and critical analysis). He also identifies several items at the level of the lexicon that point more clearly to African influence (cf. Megenney 1985). In addition to phonological and lexical features, analyses of several syntactic features have been considered to determine the potential existence of a Spanishbased creole. These features (also summarized in the work of Díaz-Campos and Clements 2005: 50-51, see also Díaz-Campos and Clements 2008) include omission of copula verbs, omission of a with personal direct objects and indirect objects, SVO word order in interrogative phrases (e.g., ¿Qué tú quieres? ‘What do you want?’), and double negation (e.g., no quiero no ‘I don’t want no’), among others. The only syntactic feature argued to clearly demonstrate African influence is double negation, which Díaz-Campos and Clements (2005) explain as an example of language transfer. In sum, previous research examining these claims in more detail has provided sufficient evidence to support the hypothesis that a semi-creole or creole language did not exist in Venezuela 1

Map obtained from


Avizia Yim Long and Manuel Díaz-Campos

(Díaz-Campos and Clements 2005, 2008). These studies are briefly reviewed in the following subsection. 2.2 A creole origin for Barlovento Spanish? To date, Díaz-Campos and Clements (2005, 2008) have presented the most detailed examinations of sociohistorical and linguistic evidence to explain the lack of a creole language in the Barlovento area of Venezuela. Through a detailed analysis of historical evidence, specifically the Spanish Crown’s role in trade with its colonies and the proportion of African to non-African peoples, Díaz-Campos and Clements (2005) argued that the appropriate conditions did not exist for the development of a creole language. An analysis of linguistic evidence further revealed that those traits previously attributed to pidginization and/or creolization could also be encountered in dialectal areas without African influence (Díaz-Campos and Clements 2005: 51). Díaz-Campos and Clements (2008) went on to present demographic evidence from colonial Venezuela as well as evidence from a variety of historical sources (e.g., the role of the Spanish Crown in the slave trade system, African population, and economic activity) to argue that “McWhorter’s view of colonial Venezuela as a place where a creole language could have developed is untenable” (2005: 378). Given that mixed-race groups with Spanish as a native language constituted a majority in colonial Venezuela and there was a predominance of small plantations that allowed for more direct contact between Africans and Europeans, Díaz-Campos and Clements (2008) effectively ruled out the possible creolization of Caribbean Spanish in the area of Barlovento, Venezuela. Díaz-Campos and Clements (2008) did point out that the evidence presented does not rule out the possibility of a creole existing in isolated regions of predominantly African influence (2008: 378). According to Brito Figueroa (1961) and Acosta Saignes (1967), the African population of colonial Venezuela consisted of slaves, indentured servants, and a smaller group called cimarrones. Cimarrones (lit. ‘runaways, fugitives’) was the name given to Africans that had fled from plantations and/or their slave masters to take refuge away from the coastal area where slave trading and plantations were prevalent (Belrose 1988). Cimarrones took refuge in cimarroneras or cumbes, generally located in the mountains, which served as a safe haven for runaway slaves as well as meeting places for planning raids of nearby farms and plantations (Belrose 1988: 15). Of particular importance for the present study, cimarroneras also served as sites for the reconstruction of traditional, African social structure (Belrose 1988: 15). These sites lie at the heart of Megenney’s (1999) assumption that there existed the potential for the development of a semi-creole language among Africans living in communities founded by runaway slaves. It is also the case

Cimarroneras in Venezuela: The Role of Isolated Communities


that there exists the assumption in some of the previous work that slaves spoke a Portuguese-based pidgin variety acquired in West Africa and used it as the primary means of communication. Thus, the present study aims to contribute to this discussion by focusing on available information for cimarroneras in the Barlovento area of Venezuela. Following a brief review of slave insurrections in colonial Venezuela, the next section details relevant demographic, sociohistorical, and linguistic evidence related to cimarroneras in colonial Venezuela to explore potential African influences on Barlovento Spanish. 3. The present study This section deals with several aspects that are important for understanding the sociohistorical conditions that contributed to the formation of isolated communities of Africans in coastal Venezuela. An important aspect that is examined in this section is related to the demographic configuration of cimarroneras and cumbes. The study of the social configuration of these isolated communities provides us with important information about their social structure and related cultural and linguistic practices of its inhabitants. A description of the origins of cimarroneras and cumbes is presented to situate these communities historically. 3.1 Insurrections documented in colonial Venezuela and the emergence of cimarrones, cimarroneras, and cumbes Guerra Cedeño (1984: 25) astutely points out that, historically, African slaves’ struggle for freedom began at the moment they were captured to be sold. The first documented uprisings in Venezuelan colonial territories were led by Miguel, an African slave described by colonizers as being un negro muy ladino en lengua castellana y resabido en toda suerte de maldad ‘an African very sly in speaking Spanish and well known for all sorts of evil’ (Brito Figueroa 1973). The insurrections led by Miguel in the mid-16th century gave way to numerous slave uprisings, first in the region of Yaracuy and later throughout colonial Venezuela (Guerra Cedeño 1984). As can be seen in Table 1, the most prominent slave insurrections marked a period of instability and unrest for the Spanish Crown that lasted well over two centuries. A natural consequence of these uprisings was a considerable increase in the number of runaway slaves or cimarrones throughout Venezuela. Guerra Cedeño (1984: 26) notes that by the end of the 16th century, hundreds of cimarrones could be found in the Tuy Valleys, Pariaguán, Cumaná, Tucacas, Calabozo, La Guaira, Yare, and Maracaibo. Guerra Cedeño (1984: 26) cites Acosta Saignes (1967), who makes reference to groups of armed cimarrones in 1655 that would ambush and rob cows and slaves from ranchers residing in the plains of Barcelona.


Avizia Yim Long and Manuel Díaz-Campos

Another notable string of insurrections occurred between Panaquire and the Tuy Valleys, where a large group of cimarrones led by the African Guillermo Ribas overthrew several plantations to free slaves and captured runaways over a period of three years in the late 18th century. Year(s)


Slave group



African slaves


Yaracuy (Buría)

African miners



African pearl divers


Central Valleys

Cimarrones (runaway slaves)


Puerto Cabello, Tucacas

African slaves


Yare (Tuy Valleys)

African slaves


Panaquire, Tuy Valleys

African slaves


Caucagua, Capaya

African slaves


Sierra de Coro

African slaves


Eastern areas (Cumaná, Carúpano, and Cariaco)

African slaves



Mulatto slaves

Table 1. Chronology of key slave uprisings in Venezuela (adapted from Guerra Cedeño 1984)

Olavarriaga (1943) estimated that by 1720 as many as 20,000 cimarrones were living in Venezuela. This would have represented approximately one third of the African slave population, estimated to be nearly 60,000 (Humboldt 1941). Acosta Saignes (1967) pointed out that cimarrones lived together in structured communities, rather than dispersing, most likely in order to protect each other from colonial authorities and other fugitive groups inhabiting colonial Venezuela at that time. In Venezuela, those communities were first called cumbes, derived from quimbo, a word from a Mandingo language of Africa meaning “remote/isolated place” (Vila 1954). In fact, the first slave revolts led by Miguel initiated the establishment of one of the first cumbes along the San Pedro River in Venezuela. Cimarroneras is a generic term given to those communities consisting of fled African slaves, though as we will describe, the term cimarroneras was reserved for temporary sites, whereas cumbes were more sedentary and farther removed from neighboring plantations or colonial towns (Guerra Cedeño 1984). This distinction seems to be crucial for understanding social and cultural practices among Africans, as cimarroneras would have tended to have more contact with mainstream society, whereas cumbes would have been more unique and isolated.

Cimarroneras in Venezuela: The Role of Isolated Communities


It is estimated that some 400 cimarroneras existed in the Barlovento region of Venezuela between the last decades of the 17th century and the end of the 18th century (Guerra Cedeño 1984)2. Given the temporary and mobile nature of cimarroneras, attempts to situate them physically at any point in time presents a considerable challenge. However, using historical reports and references to the locations of a handful of cumbes, Guerra Cedeño (1984) presents a potential layout of cimarroneras and cumbes in the state of Miranda. Cumbes were fewer in number compared to cimarroneras; however, by the end of the 18th century their numbers grew and several notable cumbes could be found in Morocopo, Caucagua, Panaquire, El Guapo, Taguaza, Aragüita, Capaya, Chuspa, Curiepe, Mamporal, Tacarigua, Tapipa, Guatire, and Guarenas (Guerra Cedeño 1984). This increase in cumbes was due in large part to the Crown’s establishment of an annual tax that had to be paid by freed slaves. Thus, to avoid payment of this tax, freed slaves organized themselves into more permanent groups and settlements, that is, cumbes, in neighboring areas (Acosta Saignes 1967). This section has presented information about cimarroneras and cumbes as isolated communities where African populations lived outside of mainstream Spanish influence. The existence of a large number of these locations indicates the possible formation of varieties of language with heavy African influence, particularly in those locations where there was the impact of one ethnic group with a majority of members. It is also the case that cimarroneras and cumbes would have fostered contexts favoring a return to original sociocultural practices, including social organization, religious celebrations, and language. Following a detailed overview of the demographic configuration of these communities (Section 3.2), a description of the sociohistorical conditions of cimarroneras and cumbes is outlined (Section 3.3). 3.2 Demographic configuration of cimarroneras and cumbes Through extensive analysis of historical data, Guerra Cedeño (1984) proposes that cimarroneras were of two main types: those serving as sites of vigilance and meeting places for carrying out raids of nearby plantations, and those formed by fugitive slaves and native Indians who were able to survive by attacking nearby ranches and landowners. The two types are distinguishable in several ways. The former generally consisted of a smaller number of armed slaves, generally around 20 men, whose main function was to protect other Africans and indigenous per2

These numbers are further supported by reports sent to the Governor of the Caracas Province between 1780 and 1784, where specific reference to cimarrones and cimarroneras was made (Acosta Saignes 1967).


Avizia Yim Long and Manuel Díaz-Campos

sons in route to removed areas of the interior (Guerra Cedeño 1984). These cimarroneras were always changing location to avoid colonial authorities and other groups that inhabited the Venezuelan interior (Guerra Cedeño 1984). In addition, although the sites for these cimarroneras were always changing, its inhabitants resided temporarily in two or three huts until advancing further into the interior to live permanently in hiding (Guerra Cedeño 1984). As previously mentioned, the latter group was comprised of a larger number of fugitives from a wider variety of ethnic backgrounds. Inhabitants of the second type of cimarroneras were also constantly changing from one location to the next, although they generally slept in trees or outside rather than in small huts (Guerra Cedeño 1984). Considering the demographic configuration of the first type of cimarroneras in more detail, the first observation to highlight is their multi-ethnic character. Recall that African slaves were imported legally via Portuguese or French slave traders from different regions of Africa, initially through Cape Verde and the Gulf of Guinea (Guerra Cedeño 1984). Soon after, Africans from a variety of indigenous groups were captured and trafficked to the Americas from areas such as Congo, Angola, Mozambique, and Madagascar (Guerra Cedeño 1984). The first quarter of the 16th century marks the beginning of slave importation to Venezuela, specifically around the year 1530 (Guerra Cedeño 1984: 15). African slaves were first introduced to the Cabo de la Vela region, located along the northern coast of Venezuela close to Colombia. From there, African slaves were distributed to other regions of the country throughout the 1500s, with the majority being located in the province of Caracas (Guerra Cedeño 1984). It is estimated that by 1840 approximately 50,000 slaves resided in Venezuela, with the grand majority (nearly 35,000) being concentrated in the fertile areas of Barlovento and the Aragua and Tuy Valleys (Guerra Cedeño 1984). It is important to point out that these workers were used in small farms with an average of 15 Africans per farm, whereas in the city the tendency was one or two per household. Regarding the origins of Africans imported to Venezuela, the majority appeared to have originated from Zaire, Congo, or Angola and were of Bantu origin (Acosta Saignes 1967). This Bantu heritage implies a rich linguistic group that comprises the languages classified as part of the Niger-Congo family. This configuration does not take into account refugees from other islands and the large number of smuggled slaves from Dutch traders (Acosta Saignes 1967; Pollak-Eltz 1977). Cumbes are generally grouped together with cimarroneras as a generic term referring to communities principally consisting of fled African slaves. However, Guerra Cedeño (1984) points out that the term cimarroneras was reserved for temporary sites whereas cumbes were more sedentary and farther removed from neighboring plantations or colonial towns. The same distinction will be maintained in our anal-

Cimarroneras in Venezuela: The Role of Isolated Communities


ysis to highlight key demographic and sociohistorical differences between the two groups. Recall that the establishment of one of the first cumbes of Venezuela was motivated by the revolts led by the African slave Miguel (Acosta Saignes 1967). These insurrections represented the collective effort of African ex-slaves as well as Jirajara Indians, an indigenous group inhabiting the Nirgua area that had not yet been “pacified” by Spanish missionaries (Pollak-Eltz 1977). Both the presence and contributions of indigenous groups are important to consider in terms of the ethnic configuration of cumbes. Pollak-Eltz (1977: 439) points out that the cumbe founded by Miguel along the San Pedro River was inhabited by Africans and Indians who “lived peacefully together.” Between Maracaibo and Rio Hacha, south of the Guajira Peninsula in western Venezuela and coastal Colombia, fled Africans took refuge among the Guajiro Indians who took part in raiding plantations and attacking travelers (Pollak-Eltz 1977). In the Barlovento region, runaway slaves received considerable help from the Tomusa Indians – an indigenous group well known for the ability to traverse and navigate the densely wooded area – and often took refuge with them, particularly in the province of Caracas (Guerra Cedeño 1984). Thus, the ethnic configuration of cimarroneras and cumbes was very much heterogeneous, an observation that has implications for the maintenance of individual linguistic and cultural practices. For the purposes of the present study, the distinction between the two types of cimarroneras as well as the description of cumbes outlined previously is significant. As proposed by Acosta Saignes (1967) and Guerra Cedeño (1984), we maintain that cumbes served as more intact communities or institutions of African slaves than cimarroneras given that members of cumbes generally sought refuge as far as possible from colonial towns and plantations. Therefore, in cumbes it was possible for inhabitants to rebuild the social structure of their native homelands. It is also the case that cumbes formed by fled slaves were more likely to favor African culture than cimarroneras because the latter tended to have a more diverse ethnic structure. In summary, it is more tenable that a creole language developed in cumbes and perhaps in the more homogeneous cimarroneras, thus the remaining analysis will focus on these sites. We now turn our attention to the available sociohistorical data specific to these sites in the Barlovento region of Venezuela. 3.3 Sociohistorical conditions surrounding cimarroneras and cumbes In this section we provide an analysis of relevant sociohistorical data to better understand the context in which cimarroneras and cumbes emerged and developed. The following three areas will be addressed to highlight key aspects of these communities relevant to our discussion of the potential for the formation of a cre-


Avizia Yim Long and Manuel Díaz-Campos

ole language in Venezuelan Spanish: (1) cultural-spiritual practices of Africans in Venezuela, (2) social structure and organization of isolated communities, and (3) acculturation of isolated Africans into rural, peasant communities. First, to address the importance of cultural-spiritual practices, recall that the African slave population of Venezuela was multi-ethnic: Slaves of African origin were brought from a variety of tribal groups where many different western and central African languages and dialects of those languages were spoken. Therefore, it should come as no surprise that Africans also differed culturally. Differences were often so great that, while working under slave masters, Africans originating from distinct tribes did not get along and distrusted each other (Pollak-Eltz 1994). Despite these differences, colonizers of European descent insisted on Africans’ assimilation to European religious and cultural practices, which in turn required all slaves to be baptized, attend mass, and participate in other Catholic rites (PollakEltz 1994). The extent to which Venezuelan colonizers sought to replace Africans’ native spiritual and cultural practices was considerable, although there was some tolerance for Africans to carry out their native rites and celebrations after baptism (Pollak-Eltz 1994). In fact, it is well known that Catholic and African traditions were mixed and even remain today as cultural traditions (e.g., the Celebration of Saint John the Baptist, the Dancing Devils of Yare). Historically, it may be the case that the conservation of several religious elements, some of which were shared amongst Africans, may have played a role in uniting Africans. This situation of meeting and sharing may have contributed to the organization of slave revolts and the establishment of isolated communities by cimarrones. Though originating from distinct cultures, shared spiritual beliefs may have contributed to unification and possibly uprising against a collective struggle as explained previously. African slaves shared a similar worldview: There existed a Supreme Being and connections between the living world and supernatural worlds were mediated by ancestral spirits (Pollak-Eltz 1994: 30). These diverse groups also shared beliefs regarding the payment of homage to ancestors, often through ritualistic practices accompanied by dance and music (particularly drums), and the central role of the village’s medicine man or healer (Pollak-Eltz 1994: 30). In their native cultures, entire groups and villages were led by a single leader, generally a sorcerer who could serve as a liaison between living humans and sacred spirits inhabiting the supernatural world (Pollak-Eltz 1994). Collectively, these shared beliefs and “magicoreligious” elements are particularly important, considering that the establishments of cumbes were often led by sorcerers (Pollak-Eltz 1994). For example, the origins of Birongo – a cumbe located in the mountainous area between Curiepe and Capaya in the Barlovento area – are attributed to revolts by African slaves and Indians whose leaders were brujos, chamanes o curanderos

Cimarroneras in Venezuela: The Role of Isolated Communities


‘witches, shamans, or healers’ (Guerra Cedeño 1984). In addition, many slave revolts were led by such sorcerers, as in the case of a slave woman on Margarita Island who gathered followers and “tried to work magic” against authorities to free pearl divers (Pollak-Eltz 1977). This particular example illustrates the potential importance of shared spiritual beliefs and worldviews in the unification of distinct cultural groups under a single leader. A = ranches

C = hilltop covering cumbe E = trees

B = watch posts

D = meeting place

G = escape path

F = entrance to cumbe

Figure 2. Visual representation of the organization of cumbes (Guerra Cedeño 1984: 52)

Another aspect of cumbes to consider is the nature of their social structure and organization. These concepts go hand-in-hand given that, according to Pollak-Eltz (1977), the organization and vigilance present in these settlements was necessary not only for the survival of the inhabitants but also for “keeping discipline” (1977: 444). In the Barlovento region, cumbes were tightly organized and strategically constructed to serve as a stable site of social structure and defense from outsiders (Guerra Cedeño 1984). A visual representation of the organization of cumbes is available in Figure 2. Recall that the Barlovento region of Venezuela is mountainous and densely vegetated, two topographical features that were often exploited by settlers of cumbes. In fact, many towns across the coastal area between Barlovento and the State of Vargas remain to this day difficult to access due to unpaved roads and the lack of reliable public transportation. These populations consisted of small towns along the central coast with a topography combining mountains and dense vegetation. In Figure 2, we can observe that cumbes were generally blocked from direct horizontal view by a large hilltop or peak. Ranches, where inhabitants slept, were strategically constructed between rows of trees for additional shelter and natural protection. Constant vigilance was maintained by establishing watch posts along the main entrances to the cumbe as well as along the perimeter behind ranches. Regarding the social structure of cumbes, Figure 2 is useful for highlighting the


Avizia Yim Long and Manuel Díaz-Campos

existence of one central meeting place, also located in the center of the settlement. It can be argued that this arrangement was particularly useful for “keeping discipline,” as noted by Pollak-Eltz (1977), and it is further suggestive of the power that the leader of the community had as a cohesive force. Because of the leader’s status as a spiritually-endowed individual, maintaining order and establishing cultural traditions was a priority, a priority that we argue was reflected in the layout of these communities. One final aspect this section will address is the acculturation of isolated communities of Africans into rural, peasant life. To facilitate the demonstration of the outcomes of this process, we highlight key aspects of a work of sociohistorical fiction, Peonía, written by Manuel Vicente Romero García (1920) and critically analyzed in depth by Belrose (1984). Peonía reflects the beginnings of Venezuelan literature, given that it is regarded as the first novel to be written by a Venezuelan writer. Peonía is described by Belrose (1984) as una aproximación costumbrista al tema del Negro ‘a local-customs approach to the issue of Blacks’ (1984: 33) that attempts to provide a snapshot of the social conditions of Africans living in rural Venezuela during the post-dictatorship era of Guzmán Blanco (Belrose 1984: 33). The story is situated in and around a sugarcane plantation located in the Tuy Valley of Venezuela (a rural area located near the capital city, Caracas, approximately 70 kilometers south with a notable population of African ancestry), and paints a picture of the world of peones, or Africans, from the curious and sardonic perspective of a Caracas agricultural engineer (Belrose 1984: 34). First, the novel is populated with descriptions of “poetic duels” in rustic settings that are suggestive of ritualistic practices. These duels are accompanied by music and dance, alluding to the perception of Africans as natural-born poets and musicians. Also suggestive of ritualistic practices is the presence of certain folklore or popular myths described in the story. For example, Romero García elaborates the beliefs surrounding the death of a child in the community, highlighting not only the influence of Catholic traditions (e.g., the child becomes an angel since he did not have time to sin) but also traditions of African origin (e.g., dancing the joropo, a ‘happy and sensual’ dance) on peasant folklore (Belrose 1984: 40). Second, these events were carried out in the context of an ethnically diverse group of people: whites, Indians, blacks, mulattoes, and mestizos who had been living together since early colonial times. Belrose (1984: 40) points out that use of the term negro ‘black’ referred not to the race of the individual but was used in an endearing manner as in Venezuela and other Caribbean countries today. Although the novel attempts to contrast African from mainstream Venezuelan (i.e., European-based) cultural practices, it is clear that what is perceived as “African” is really no different than “rural” or “peasant” (Belrose 1984: 40). The novel paints a picture of Black peasant culture, not Black African culture, suggesting that by the end

Cimarroneras in Venezuela: The Role of Isolated Communities


of the 1800s, isolated communities populated by Africans had already been assimilated to the rural lifestyle of peasants in Venezuela. Further evidence for this assertion can be found in Pollak-Eltz (1977), who maintains that the “…assimilation of the maroons into the Venezuelan peasant culture was accomplished long ago” (Pollak-Eltz 1977: 444). The question that remains is why this process of acculturation took place. One piece of evidence is highlighted in the novel itself and argued in detail by DíazCampos and Clements (2005, 2008). Not only were rural communities multiethnic, but also the predominant group was of mixed-raced descent (i.e., pardos or mixed African-European and African-Indian ancestry). Díaz-Campos and Clements (2008) report that 38.22% of the population of colonial Venezuela was considered to be constituted by pardos. Another piece of evidence for this acculturation was the widespread destruction of cumbes that took place as colonial authorities sought to exterminate these communities and return fled slaves to their owners (Guerra Cedeño 1984). Thus, the gradual disappearance of cumbes went hand-in-hand with re-integration into (rural) plantation life or further removal into farther regions and subsequent integration with other indigenous or peasant (rural) communities. One last piece of evidence of integration of Africans to Venezuelan culture comes from the research of Pollak-Eltz (1977), who argues that cimarrones went to isolated areas and enlisted with military forces fighting against the Spanish Crown. Particularly during the wars of independence in the early 1800s, many slaves fled plantations and fought alongside “patriotic forces” (Pollak-Eltz 1977: 444), further facilitating the integration of Africans into Venezuelan culture. In this section we have discussed three components of the sociohistorical situation surrounding mainly cumbes to shed light on the discussion of the potential for a creole language in Venezuela. The African presence in these communities seems favorable for the emergence and development of a creole language. However, it is clear that cumbes were multi-ethnic and multi-cultural settlements. The lack of a stable, homogeneous group and the surrounding environment of unrest appear to have contributed to the development of a more dynamic community that existed to join groups of diverse ethnic descent that shared the desire to simply survive and avoid a life of slavery. The latter is echoed by Pollak-Eltz (1977: 444), who states that: With regard to Venezuelan maroons we have to make it clear that no specific maroon culture ever existed in this country. In the early days the ‘cumbes’ might well have been organized according to an African pattern, but they were in constant contact with free blacks in the ‘criollo’ villages. Therefore we cannot speak of a maroon culture in the sense that it existed in Jamaica or Surinam.


Avizia Yim Long and Manuel Díaz-Campos

Cimarroneras in principle seem to be an ideal place for the survival of African traditions including language. However, their changing dynamic and their temporary nature did not contribute to the establishment of strong cultural patterns that remain today. For example, there is no clear indication of the existence of a creole language that was spoken in those communities. The following section will highlight key linguistic features of African influence in Barlovento Spanish that further contribute to the discussion of the lack of a creole language variety in Venezuela, particularly in Barlovento. 3.4 Linguistic aspects of African influence in Barlovento Spanish As summarized in the background literature for Barlovento Spanish, several phonetic phenomena present in Afro-Venezuelan Spanish previously attributed to the existence of a creole variety have been put into question given the presence of the same phenomena in dialectal areas without African influence (DíazCampos and Clements 2008, 2005). Some of those same features appear to be representative of black speech in sociohistorical works of fiction such as Peonía3, as exemplified in the following excerpt: “No le negamos el vicio a los músicos llaneros en el Tuy toos semos negros pero semos caballeros” (from Belrose 1984: 38)

The first, toos [todos] ‘all’ reflects a common, ongoing phonological process of many varieties of Spanish: deletion of intervocalic voiced fricatives. Intervocalic /d/ deletion has been documented not only in the Caribbean but also in Spain (northern and southern areas), Chile, Colombia, and coastal Peru. The second, semos [somos] ‘(we) are’, exemplifies a general characteristic of rural speech, which is typical alongside other lexical items such as asigún for según ‘according to’ and naiden for nadie ‘no one.’ These examples are representative of rural varieties of Spanish in Latin America (Parodi 2001). However, as discussed in the previous section, this novel is one about peasant culture, not African culture, and the same observations can be applied to phonetic features of Black speech in the novel. Peonía is considered to be the first Venezuelan novel, thus it would be the first document in which Venezuelans, including African Venezuelans, are represented in fiction.


Because there are no available sources of recordings of colonial times, one way to study possible patterns is the use of literary representations of African speech. However, we understand that these representations may be stereotypical and can be used only when systematic phenomena are observed.

Cimarroneras in Venezuela: The Role of Isolated Communities


More generally, a consideration of relevance is the extent to which Africans in Venezuela were able to utilize and maintain their native languages, as explained by Pollak-Eltz (1994: 22): To facilitate acculturation of “bozales” (recent arrivals from Africa) and to avoid riots, the newcomers were quickly dispersed over large areas and had no chance to get together. Their languages were soon lost and they were taught Spanish. Some already knew a lingua franca based on Portuguese, that was widely spoken on the West African coast. Indians and blacks often worked side by side and their proximity fostered miscegenation and cultural amalgamation, which culminated in the development of the criollo folk culture of today.

Again we observe the creation of multi-racial societies in Spanish colonies as well as the lack of assuming a more active role in slave trading by Spaniards (Díaz-Campos and Clements 2005, 2008) as two important factors influencing not only the sociohistorical situation of colonial Venezuela, but also the potential for integration of African linguistic features in Barlovento Spanish. As pointed out by Megenney (1985), a pidgin Portuguese was used as a lingua franca in most West African trading posts. Furthermore, before the use of Spanish was enforced by colonial authorities, it is believed that many African slaves arriving to the Americas in the 17th and 18th centuries were already familiar with this variety and were using this language to communicate with their slave masters (Megenney 1985). It appears that Africans, who spoke this variety, did not develop this pidgin into a creole language, but instead acquired Spanish due to the process of acculturation that they were forced to adopt. Linguistic evidence clearly demonstrating African influence in Barlovento Spanish is largely constrained to lexical items (Megenney 1985; Pollak-Eltz 1994; Sojo 1947). Words of possible African origin reported by Megenney (1985) are found in chants honoring saints, though their meanings are unknown to people today:

cocotí, sambalá, samborombino, congo del cangalí, tango guachicamé malabí maticú lambí etc. Alá alá láay sámbalaa balá, balá, bumbé, el hombre no má, ay sambalá balá bumbé (Pollak-Eltz, 1994: 34)

Sojo (1947) presented more words of possible African origin that are still in use in Barlovento Spanish today. Unlike the evidence provided by Megenney (1985), their meanings are known and some have been linked to words of an African language. Table 2 displays those words and their corresponding meanings.


Avizia Yim Long and Manuel Díaz-Campos

Words in use today


Linked to… (original language)


a sweet






Black magic



a chant








little seashell (Bantu)


Maroon village

group; to be united (Fang)

Table 2. Words of possible African origin encountered in Barlovento Spanish (Sojo 1947)

Lipski (2007) provides an analysis of one phonetic feature (/ɾ/ > [d]) preserved by members of Afro- Hispanic speech communities in certain lexical items traced to the Kikongo language of Bantu origin. The Barlovento community is one such community examined, where the phrase hablar loango ‘speak loango’ is used to refer to having poor pronunciation and more generally the lack of education that is consequently associated with the inability to speak well (Lipski 2007: 103). The feature /ɾ/ > [d] is the “most stereotypical” characteristic of speaking loango, as demonstrated in the example Caracas > Cadaca (Lipski, 2007: 104). Lipski further illustrates that this change is only present in Kikongo and neighboring languages of the Congo River basin, providing clearer evidence of African influence on Barlovento Spanish spoken in Venezuela. The existing linguistic evidence clearly demonstrating African influence is minimal for thorough consideration of potential creole genesis in isolated communities. These observations alongside the discussion provided for ethnic configuration and the sociohistorical situation of these settlements overall point to unfavorable conditions for creole genesis in Barlovento communities and more generally in areas of predominantly African influence in Venezuela. 4. Concluding remarks and future directions This paper has provided additional evidence for non-creole bases of Afro-Caribbean Spanish. Through consideration and discussion of the ethnic configuration and sociohistorical situation of cimarroneras and cumbes, as well as existing linguistic evidence, we provide additional insight into why it is unlikely that a

4 5

ND indicates no data available. Pinga is also used to make reference to something cool in the expression de pinga or it is also used to refer to male genitalia.

Cimarroneras in Venezuela: The Role of Isolated Communities


creole language did not develop in the Barlovento area of Venezuela, a region of predominantly African influence since its settlement during the colonial era. The arrival of African slaves who potentially spoke a pidginized variety of Portuguese coupled with a diverse, multi-ethnic demographic and widespread acculturation did not produce favorable conditions for the development of a creole language in Venezuela. The study of cimarroneras and cumbes suggests that the complex cultural dynamic of multi-ethnic and mobile communities did not facilitate the maintenance of African traditions, including language. Instead, it appears that over time these communities were forced to integrate into the rural life of Venezuela. If anything, constant contact between different groups over time appears to have been influenced more by similarities than differences in terms of cultural and linguistic practices to maintain communication and ensure survival in these isolated communities. This study focused on northern coastal areas but future investigations should examine origins of other Afro-Venezuelan groups residing in other regions, especially those located farther in the interior and whose contact with outsiders may have been more limited. Additionally, future analyses of popular traditions and culture of Afro-Venezuelans may unveil more linguistic sources of data necessary to address in further detail the extent to which African languages played a role in the development of the variety of Spanish spoken in Venezuela. REFERENCES Acosta Saignes, M. (1967). Vida de los esclavos negros en Venezuela. Caracas: Hesperides. Álvarez, A./Obediente, E. (1998). “El español caribeño: antecedentes sociohistóricos y lingüísticos”. In P. Matthias, and A. Schwegler (eds.), América negra: panorámica actual de los estudios lingüísticos sobre variedades hispanas, portuguesas y criollas. Madrid/Frankfurt: Iberoamericana/Vervuert, pp. 40-61. Belrose, M. (1988). África en el corazón de Venezuela. Maracaibo: Universidad del Zulia. Brito Figueroa, F. (1961). Las insurrecciones de los esclavos negros en la sociedad colonial venezolana. Caracas: Editorial Cantaclaro. Brito Figueroa, F. (1973). El problema de la tierra y esclavos en la historia de venezuela. Caracas: Teoría y Praxis. de Granda, G. (2001). “Procesos de estandarización revertida en la configuración histórica del español americano: el caso del espacio surandino”. In International Journal of the Sociology of Language, 149, pp. 95-118. Díaz-Campos, M. (1998). The Trace of African Languages in Latin American Spanish: A Study of the Spanish Variety Spoken in Barlovento, Venezuela. ms., Columbus, OH: Ohio State University.


Avizia Yim Long and Manuel Díaz-Campos

Díaz-Campos, M./Clements, J. C. (2005). “Mainland Spanish Colonies and Creole Genesis: The Afro-Venezuelan Area Revisited”. In L. Sayahi, and M. Westmoreland (eds.), Selected Proceedings of the Second Workshop on Spanish Sociolinguistics. Somerville, MA: Cascadilla, pp. 41-53. Díaz-Campos, M./ Clements, J. C. (2008). A Creole Origin for Barlovento Spanish? A Linguistic and Sociohistorical Inquiry. In Language in Society, 37, 3, pp. 351-383. Guerra Cedeño, F. (1984). Esclavos negros, cimarroneras y cumbes de barlovento. Lagoven. Humboldt, A. de (1941) [1826]. Viaje a las regiones equinocciales del Nuevo Continente. Viajes y Naturaleza. Caracas: Biblioteca Venezolana de Cultura. Laurence, K. (1974). “Is Caribbean Spanish a Case of Decreolization?”. In Orbis 23, pp. 484-99. Lipski, J. (1993). On the Non-Creole Basis for Afro-Caribbean Spanish. Albuquerque, NM: University of New Mexico Press. Lipski, J. (2000). “Bozal Spanish: Restructuring or Creolization?”. In E. W. Schneider, and I. Neumann-Holzschuh (eds.), Degrees of Restructuring in Creole Languages. Amsterdam/Philadelphia: John Benjamins, pp. 55-83. Lipski, J. (2005). A History of Afro-Hispanic Language. Five Centuries, Five Continents. Cambridge: Cambridge University Press. Lipski, J. (2007). “El cambio /R/ > [D] en el habla afrohispánica: ¿un rasgo fonético “Congo”?”. In Boletín de lingüística XIX 27, pp. 94-114. McWhorter, J. (1995). “The Scarcity of Spanish-Based Creoles Explained”. In Language in Society 24, pp. 213-244. McWhorter, J. (2000). The Missing Spanish Creoles: Recovering the Birth of Plantation Contact Languages. Berkeley, CA: University of California Press. Megenney, W. (1985). “África en Venezuela: Su Herencia Lingüística y Cultura Literaria”. In Montalbán 15, pp. 3-56. Megenney, W. (1989). “Black Rural Speech in Venezuela”. In Neophilologus 73, pp. 52-61. Megenney, W. (1999). Aspectos del Lenguaje Afronegroide en Venezuela. Madrid/Frankfurt: Iberoamericana/Vervuert. Megenney, W. (2000). “The Fate of Some Sub-Saharan Deities in Afro-Latin American Cult Groups.” Paper delivered at the XXII International Congress of the Latin American Studies Association. March 16-18. Miami, Florida. Olavarriaga, J. de (1943). Descripción Histórico-Geográfica de la Antigua Provincia de Venezuela. Caracas: Academia Nacional de la Historia. Otheguy, R. (1973). “The Spanish Caribbean: A Creole Perspective”. In C.-J. Baley, and R. Shuy (eds.), New Ways of Analyzing Variation in English. Washington, DC: Georgetown University Press, pp. 323-339. Parodi, C. (2001). “Contacto de Dialectos y Lenguas en el Nuevo Mundo: La Vernacularización del Español de América”. In International Journal of the Sociology of Language, pp. 33-53. Pollak-Eltz, A. (1977). “Slave Revolts in Venezuela”. In Annals of the New York Academy of Sciences 292, pp. 439-445.

Cimarroneras in Venezuela: The Role of Isolated Communities


Pollak-Eltz, A. (1994). Black Culture and Society in Venezuela (La negritud en Venezuela). Caracas: Lagoven Booklets. Romero García, M. V. (1920). Peonía: novela de costumbres venezolanas. EditorialAmérica: Sociedad Española de Librería. Schwegler, A. (1996). “La Doble Negación Dominicana y la Génesis del Español Caribeño”. In Hispanic Linguistics 8, pp. 247-315. Sojo, J. P. (1947). “El Negro y la Brujería”. In Revista Venezolana de Folklore 1, p. 1. Vila, M. A. (1954). El Estado Miranda, Su Tierra y Sus Hombres. Caracas: Edición del Banco Miranda.


1. Introduction The present chapter is concerned with language contact and change in the bilingual West Indian speech community in the Republic of Panama where Spanish and a variety of Creole English are spoken. I focus on the production of the voiceless dental plosive in terms of an acoustic property known as voice onset time (VOT). VOT is defined in acoustic phonetics as the time interval between the occlusion of a consonant stop and the onset of vocal cord vibration (Lisker and Abramson 1964: 387). In Spanish, VOT is considered to be short lag because vocal cord vibration, or voicing, occurs shortly after the release of the stop. During this time interval very little aspiration can be heard. However, in English, VOT is categorized as long lag because voicing occurs well after the release of the stop and there is a great deal of aspiration. Therefore, this acoustic property provides what is known as a conflict site, which is an area where the two languages do not line up (Poplack and Meechan 1998: 132). The conflict site allows us to analyze potential effects of contact between Spanish and Creole English, which is one of the goals of the present study. The second goal of the analysis is to focus on the individual as the locus of variation and change in the context of linguistic contact and to combine quantitative results with qualitative data to explain contact-induced change occurring in this speech community. In this light, the analysis contributes not only to the extensive work carried out in situations of language contact and change, but also to the growing literature on variationist sociolinguistics, especially in terms of methodology. 2. Spanish and Creole English in Panama Besides being the national language of Panama, Spanish has been the official language of this country since the early 1940s with the promulgation of the Consti-


Delano S. Lamy

tution of 1941 (see Constitution of 1941, Title 1, Article 10).1 However, Creole English has been the second most widely spoken language in Panama due to migration from the English-speaking Caribbean during three important projects: the construction of the Panama Railroad and the first and second phases of the construction of the Panama Canal. The construction of the Panama Railroad began during the California Gold Rush in the mid-1800s. Many were searching for the quickest route from the east coast of the United States to the west, and because of its shape as an isthmus, Panama was considered the ideal location for this journey. In 1838 slavery had been abolished in the British West Indies, and many freedmen who were unable to find jobs in their homeland decided to seek work abroad, particularly in Panama. This initiated the first wave of migration of about 5,000 workers from Jamaica to the Isthmus to labor in the railroad construction (Connif 1985: 17). A second wave of migration of West Indians was triggered in the late-1800s when the French began the first phase of the construction of the Panama Canal. The majority of the workers migrated from the French-speaking islands of Martinique and Guadaloupe (Diez Castillo 1981; de Banville 2005). About 50,000 workers went to Panama, some of whom returned to their islands after the construction of the canal failed; however, many decided to stay and raise families while they waited for another project (Connif 1985:18). The second phase of the construction of the Panama Canal was initiated by the United States after Panama won its independence from Colombia on November 3, 1903, thus triggering the third wave of migration mainly from English-speaking islands. It is estimated that 150,000 laborers and their families arrived in Panama between 1904 and 1914. Many remained after the canal was completed because of the many opportunities available in this country. According to Connif (1985: 4), the West Indians, or antillanos, became the largest group of immigrants in a country that had a total population of 400,000 people. The presence of West Indians in Panama brought about extreme racism as Panamanians of mestizo origin felt they had to compete with West Indian immigrants for well sought after Canal jobs (Connif 1985: 4). Bad race relations were also noticeable in the Canal Zone where the Americans implemented Jim Crow segregation disguised under a gold-silver payroll system (Connif 1985: 4-5). The racism that West Indians had to endure in Panama caused this group to unite despite their island differences. They began to establish their own schools, churches and businesses, which brought about the emergence of a new West Indian subculture that


A copy of Constitution 1941 can be found at Biblioteca Digital Panameña: http://www.binal.

The Individual as the Locus of Variation and Change in a Contact Situation


incorporated traditions borrowed from England, the Caribbean, North America, and Panama (Connif 1985: 12). Their establishments were normally run in English, which helped to maintain the language in later generations, especially in the cases where their children had limited access to education in Spanish (Connif 1985: 6). These establishments were located in areas where the majority of West Indians had settled, such as El Chorrillo, San Miguel, Calidonia, and El Marañon, also known as Guachapalí (Thomas-Brereton 1992: 49). Although race relations have improved tremendously, many West Indians continue to live in these areas and speak both Spanish and English daily.2 It is also important to note that what West Indians refer to as “English” in Panama has been labeled as “Creole English” by linguists in creolistics literature. This term is used to describe this variety for two reasons: 1) it is a dialect that has evolved from the language spoken by the immigrants who had come from islands such as Jamaica, Barbados, Saint Vincent, etc., and 2) the variety maintains many characteristics of the Creole English dialects spoken in the islands mentioned. The term “English” has been used in the present paper up to this point because the concept of (Creole) English in Panama is not as clear as it may be in Jamaica. However, in order to keep with the tradition of creolistics, “Creole English” will be used henceforth. West Indian speech, particularly in Spanish, is not often commented on in Panama; however, in the rare occasion that a person points out any differences, they highlight phonetic characteristics. Some refer to their speech as being slow paced, using the phrase “hablar con el tiempo”. Another trait that is highlighted when imitating a West Indian is the aspiration of consonants. It seems that this aspiration is a salient characteristic of West Indian speech, and thus, could obtain some type of sociolinguistic meaning that could emerge when analyzing effects of contact in this particular speech community. Thus, the linguistic variable chosen for this analysis is VOT duration of the voiceless dental plosive because this acoustic cue is correlated with consonant aspiration. 3. Voice onset time (VOT) According to Lisker and Abramson (1964: 387), VOT is the time interval between the release of a consonant stop and the onset of glottal vibration. The voiceless dental plosive in Spanish exhibits a VOT value of no more than 25 msec. In Eng2

There is also an enclave of West Indians in the city of Colón, which can be considered the mecca for West Indians. Also, West Indians can be found in the province of Bocas del Toro. They have a slightly different history from those living in the metropolitan areas of Panama, but they also speak a variety of Creole English.


Delano S. Lamy

lish, the voiceless alveolar plosive features a VOT value of no less than 30 msec. (Lisker and Abramson 1964). According to these measurements, Spanish stops are considered short lag because voicing occurs shortly after the release burst. English stops, however, are long lag because voicing occurs well after the release. Although these are well-established values in the literature for monolingual speakers of Spanish and English, researchers have found that bilinguals can show different patterns in their production of stops when compared to their monolingual counterparts in both languages. In his Speech Learning Model, Flege (1995a) affirms that the age of second language acquisition has a significant effect on VOT duration in bilingual speech. He claims that as the age of acquisition increases, the ability to discern phonetic differences between two languages becomes less likely, and thus, brings about a merger in the language. Flege’s (1987) merger hypothesis is based on evidence of compromise VOT values found in the speech of late bilinguals. ‘Compromise’ refers to the fact that the mean VOT values for both languages fall between the prototypical ranges produced by monolinguals. This process is known as crosslanguage assimilation and occurs because bilinguals make an equivalence classification of similar sounds from their two languages (Flege 1987; Flege 1995b: 101). Flege (1995b: 101) also finds evidence of what is known as cross-language dissimilation, which is more common in early bilinguals. In this process, bilinguals separate their two languages by producing extra-long VOT values in one language to compensate for long VOTs in the other. This behavior is common in early bilinguals since the first language (L1) is not completely developed at a young age, the second language (L2) is established independently of the L1 (Flege 1987). Therefore, it can be concluded based on the literature concerned with VOT duration in both monolingual and bilingual speech, that this acoustic property is ideal for examining the effects of contact between Spanish and Creole English such as convergence and divergence. With these ideas in mind, in Section 4, I discuss certain social factors that have been considered in language variation and change literature in order to explain contact-induced changes in bilingual speech communities. 4. Language contact, variation, and change In the framework of language variation and change (LVC), the evolution of language has been attributed to the geography or the social setting in which language changes. Such external factors, along with internal ones, have been considered to be the “ecologies” that bear on language evolution (Mufwene 2001). There can be many explanations for why certain forms die out, and why others enter or become

The Individual as the Locus of Variation and Change in a Contact Situation


more frequent in a language. Many researchers concerned with LVC take into account social factors, such as intensity of contact (e.g. Thomason 2001), language attitudes (e.g. Appel and Muysken 2005; Hill and Hill 1980), language loyalty and maintenance (e.g. Fishman 1964; Grosjean 1982) and identity (e.g. Eckert 2000, 2008; Hall-Lew 2009) to explain variation, change, and language shift in contact situations. One of the most widely discussed social factors is intensity of contact, which Thomason (2001) quantifies using three criteria: duration of the contact period, relative sizes of the populations, and socioeconomic dominance. If two groups of speakers have been in the same region for a long period of time, this could lead to more opportunities for language change due to contact (Thomason 2001: 66). Second, the group of speakers with the smaller population more than likely will adopt features from the other group’s language because the larger one has the dominant culture. Third, if this dominant group is also socioeconomically more successful than the smaller one, it is even more likely that the smaller group will adopt features from the larger one (Thomason 2001: 66). When considering these conditions, other factors, such as language attitudes, loyalty and maintenance, and identity could work together to bring about change in a speech community. These factors can also be at work when observing the behavior of the individual because change is not only noticeable in the general speech community, but also at an individual level. Therefore, it becomes necessary to examine contact-induced change in individual speakers, since such analysis could give an additional explanation of the emergence of new varieties in contact situations due to language change. 4.1 The individual as the locus of change Literature concerned with language contact, as previously mentioned, normally focuses on the speech community as a whole when analyzing the effects of contact. In recent years, many researchers (e.g. Blondeau and Nagy 2008; Meyerhoff 2009; Poplack and Levey 2010; Torres Cacoullos and Aaron 2003) have taken on the task of determining the source of change in a contact situation using the comparative variationist method established by Shana Poplack and her colleagues. This method helps to determine whether change occurring in a bilingual speech community is internally motivated or if it is contact-induced by performing a series of comparisons of statistical results between monolingual and bilingual varieties. The comparative variationist method has been considered one of the most adequate ways of examining possible contact-induced change. However, in spite of this, one finds it necessary to hone in on the individual to have a better understanding of the particular contact situation since speakers tend to vary their language according to style (see Eckert 2008; Podesva 2007; Zhang 2005). If bi-


Delano S. Lamy

linguals have the ability to access different speech styles, this means that they could produce different forms in a single context either from their native language or from their second language according to certain social factors. One method of determining the effect of social factors on individual speech style is incorporating mixed effects into regression models (Baayen, Davidson and Bates 2008; Drager and Hay 2012; Johnson 2009). Mixed-effects modeling refers to the inclusion of two types of factors. There are factors that are considered fixed effects, which can be verb type, syllable stress, phonetic environment, or gender. They have a small number of levels and can be replicated in further studies. There are also factors that are random effects, that is, they are drawn from a larger population such as lexical item or individual speaker, and cannot be replicated in further studies (Johnson 2009: 364-365). Including speakers as a random effect in a regression model allows for individuals to vary significantly from the norm without underestimating or overestimating the significance of factors treated as fixed effects (Drager and Hay 2012; Johnson 2009). Most importantly for this study, the results of speaker as a random effect represents stylistic variation, which can complement the interpretation of the fixed effects results when drawing from qualitative data found in sociolinguistic interviews (see Section 5 for further discussion). This type of interpretation can be used to explain why certain linguistic changes may be taking place in a bilingual speech community. The goal of the present study is to use this representation of speaker style to understand why change is occurring in the speech of Creole English-Spanish bilinguals living in Panama. In Section 5, I discuss the methodology used to obtain this goal. 5. Methodology 5.1 Speech community and data collection This study originates from a larger analysis of language change in Panama in which samples of natural, spontaneous speech were extracted from sociolinguistic interviews conducted with 5 monolingual speakers of Spanish, 5 monolingual speakers of Creole English and 5 Creole English-Spanish bilinguals. In this paper, I focus on the production of Spanish and Creole English by the 5 bilingual speakers. These individuals are of West Indian origin and have lived in Panama all their lives. They also claim to be early bilinguals; that is, they learned Creole English in the home from birth and Spanish in their community at least by the age of five. They use both languages daily, as they have integrated into the general Panamanian population and also belong to a social network of Creole English-speaking West Indians.

The Individual as the Locus of Variation and Change in a Contact Situation


The speakers were recorded using a Sony Digital Voice Recorder ICD-SX750 and a Shure Microphone SM10A. This equipment was selected given the need for high-quality audio files to perform acoustic measurements of voice onset time of the voiceless dental plosive. 5.2 Analysis The dependent variable, as previously mentioned, is the voiceless dental /t/ occurring in three phonetic environments: absolute word-initial, prevocalic (#_V); postconsonantal, prevocalic (C_V); and intervocalic (V_V). In an attempt to follow the principle of accountability (Labov 1972: 72), all words containing a voiceless dental were extracted from 10-minute segments beginning after the 10th minute of each interview. The total number of tokens obtained for the larger analysis was 2,128. 1,040 of these tokens were produced by the bilinguals, with 528 being extracted from Spanish and 512, from Creole English. VOT duration was measured in milliseconds using Praat (Boersma and Weenink 2012), a free software program for analyzing speech sounds. The point of release of the /t/ was detected on the spectrograms as a short periodic pulse. The measurement was taken from this point up to the onset of voicing, which is indicated by regularly spaced vertical striations. The following figures are examples of spectrograms of voiceless dentals in Spanish and Creole English.

Figure 1. Spectrogram and waveform of /t/ in ‘terreno’ in Spanish

Figure 2. Spectrogram and waveform of /t/ in ‘text’ in Creole English


Delano S. Lamy

In order to pinpoint variation and potential change in bilingual speech, linear regression models fitted with mixed effects were included using Rbrul (Johnson 2009). This package, which is loaded in the free statistical environment R, provides the statistical significance level for factor groups. It also estimates the effect of factors on mean values and presents the results in coefficients that can be positive or negative. In the case of VOT duration, positive coefficients indicate that a longer VOT is favored with a certain factor and negative coefficients indicate that a shorter VOT is favored. In the LVC framework, the order of effect of the conditioning factors, also known as the constraint hierarchy, is considered the underlying system of the language and offers a detailed view of each factor being tested (Poplack and Tagliamonte 2001). Therefore, certain linguistic factors were included in the linear regressions based on previous literature concerned with VOT duration. These factors were preceding segment, following vowel height, position of /t/ within the word, syllable stress, rate of speech, word class, word frequency and lexical item.3 All of these factors, except lexical item, were considered fixed effects. Also, since I am concerned with how speaker style plays a role in language change in this contact situation, “individual speaker” was included in the linear regression as a random effect. Each speaker is assigned a coefficient in the regression model, which allows us to observe how they vary from one another (Baayen, Davidson & Bates 2008; Drager & Hay 2012; Johnson 2009). Therefore, in terms of VOT duration, I am able to determine the likelihood that a speaker will produce a longer VOT as opposed to a shorter one. These quantitative results are then explained by exploring the qualitative data drawn from the sociolinguistic interviews to elucidate the linguistic behavior of the speakers. In Section 6, I present and discuss the results of the analysis, focusing mainly on the bilingual speaker results; however, I mention the findings from the monolingual data where they are relevant. 6. Results 6.1 Mean VOT values Spanish

Creole English