sign in sign up
<<
Home , Belizeans, Belizean Creole, Belizean society, Belizean Spanish, Bolivian Spanish, Central American Spanish

LANGUAGE USE, LANGUAGE CHANGE AND INNOVATION IN NORTHERN CONTACT SPANISH

By

OSMER EDER BALAM

A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2016

1 ACKNOWLEDGMENTS

This dissertation would not have been possible without the guidance and support from many people, who have been instrumental since the inception of this seminal project on contact

Spanish outcomes in Northern Belize.

First and foremost, I am thankful to Dr. Mary Montavon and Prof. Usha Lakshmanan, who were of great inspiration to me at Southern Illinois University-Carbondale. Thank you for always believing in me and motivating me to pursue a PhD. This achievement is in many ways also yours, as your educational ideologies have profoundly influenced me as a researcher and educator.

I am indebted to my committee members, whose guidance and feedback were integral to this project. In particular, I am thankful to my adviser Dr. Gillian Lord, whose energy and investment in my and research were vital for the completion of this dissertation. I am also grateful to Dr. Ana de Prada Pérez, whose assistance in the statistical analyses was invaluable to this project. I am thankful to my other committee members, Dr. Benjamin

Hebblethwaite, Dr. Ratree Wayland, and Dr. Brent Henderson, for their valuable and insighful comments and suggestions. I am also grateful to scholars who have directly or indirectly contributed to or inspired my work in Northern Belize. These researchers include: Usha

Lakshmanan, Ad Backus, Jacqueline Toribio, Mark Sebba, Pieter Muysken, Penelope Gardner-

Chloros, and Naomi Lapidus Shin. A special thanks to Christine Kray, who generously shared interview transcripts from interviews she conducted with third and fourth generation descendants from San José Yalbac in 2005.

I am grateful to Dámaris Mayans, my friend and colleague, whose presence has been a blessing and inspiration to me. I am forever indebted to your friendship and your support in my research endeavors. My humble gratitude also goes out to Dr. Michael A. Claudio, whose

2 constant support and words of advice during the past five years were crucial to my scholarly accomplishments. ¡Mil gracias Claudio!

Next, I would like to thank my parents for their faith in my dreams and for their prayers. I am especially thankful to my sister Ruby who, in the process of data collection, became my research assistant. I am grateful to Mr. Alejandro Pérez, from the , whose assistance in the data collection process was fundamental to this dissertation as well.

Last but not least, I am indebted to the consultants of the present dissertation, the protagonists of this project, who gave me an opportunity to read and learn more not only about linguistic aspects of Northern and Northern Belize bi/multilingual code- switching, but about my people, my past and their struggles during the last century. Without their willingness to participate and their willingness to help me recruit other participants, this dissertation would not have been possible. ¡Muchas Gracias! Dios bo’otik!

3 TABLE OF CONTENTS

page

ACKNOWLEDGMENTS ...... 2

LIST OF TABLES ...... 8

LIST OF FIGURES ...... 10

ABSTRACT ...... 12

CHAPTER

1 INTRODUCTION ...... 14

1.1 Overview ...... 14 1.2 The Sociolinguistic Context: Northern Belize ...... 16 1.2.1 Northern Belize: Demographic and Linguistic Profile ...... 16 1.2.2 History of Contact Spanish and Bi/ in Northern Belize ...... 22 1.3 Language Attitudes and Identity in Northern Belize ...... 36

2 THEORETICAL BACKGROUND...... 45

2.1 Defining Code-Switching ...... 45 2.2 Opposing Views of Code-Switching ...... 47 2.2.1 The Dominant View of Code-Switching ...... 47 2.2.2 Criticisms of CS as the Combination of Grammars or Lexicons ...... 52 2.2.3 Code-Switching and Convergence ...... 56 2.3 Approaches in the Grammatical Analysis of Code-Switching ...... 59 2.3.1 Bilingual Speech Typology ...... 63 2.3.2 Myers-Scotton’ The Matrix Language Frame (MLF) Model ...... 69 2.3.3 Minimalist Approach to Code-Switching ...... 76 2.4 The Debate on Social versus Linguistic Factors...... 81 2.5 Situating the Northern Belize Context ...... 89 2.5.1 Research Questions ...... 94 2.5.2 Structure of the Dissertation ...... 95

3 THE INTERVOCALIC RHOTIC IN NORTHERN BELIZEAN SPANISH ...... 97

3.1 Introduction ...... 97 3.2 Intervocalic Rhotic Contrast(s) and Neutralization in Spanish Varieties ...... 99 3.3 Previous Work ...... 106 3.3.1 Rhotic Distribution in Spanish Bidialectal Contexts ...... 106 3.3.2 Studies on the Intervocalic Tap/Trill Contrast ...... 109 3.4 Methodology ...... 113 3.4.1 Participants ...... 113

4 3.4.2 Tasks ...... 115 3.4.3 Procedure ...... 117 3.5 Results...... 119 3.5.1 Rhotics in Northern Belizean Spanish ...... 119 3.5.2 Neutralization of the NBS Intervocalic Rhotic Contrast ...... 125 3.5.2.1 Elicited oral production...... 125 3.5.2.2 Read-aloud ...... 129 3.6 Discussion ...... 138 3.6.1 A Closer Look at Task Type and Phonetic Outcomes ...... 138 3.6.2 Social-indexical Factors in Northern Belize ...... 142 3.7 Concluding Remarks ...... 147 3.8 Limitations ...... 147

4 BILINGUAL LIGHT VERB CONSTRUCTIONS ...... 149

4.1 Introduction ...... 149 4.2 Bilingual Light Verb Constructions...... 150 4.3 Bilingual Light Verb Constructions in Spanish Contact Situations ...... 158 4.3.1 Verb Frequency and Stativity ...... 158 4.3.2 Passivization ...... 163 4.4 Study 1 Methodology: Verb Frequency and Stativity ...... 166 4.4.1 Consultants ...... 167 4.4.2 Data ...... 169 4.5 Results...... 173 4.5.1 Speakers’ Intuitions: Verb Frequency and Stativity in BLVCs ...... 173 4.5.2 BLVCs in Control Structures ...... 176 4.5.3 BLVCs in Oral Production ...... 178 4.5.4 Restriction of Frequency in Oral Production ...... 181 4.5.5 Restriction of Stativity in Oral Production ...... 185 4.5.6 Beyond Verb Frequency and Stativity: Hybridity and Innovation in BLVCs .....188 4.6 Study 2 Methodology: Passivization ...... 193 4.6.1 Participants ...... 193 4.6.2 Data ...... 195 4.7 Results...... 199 4.8 Discussion ...... 206 4.8.1 Speakers’ Attitudes to their Language Varieties and Bi/Multilingualism ...... 210 4.8.2 Speakers’ Identification with Spanish/English CS ...... 212 4.8.3 BLVCs, Creolization, Convergence and Language Change ...... 213 4.9 Concluding Remarks ...... 217

5 SEMANTIC CATEGORIES AND GENDER ASSIGNMENT ...... 218

5.1 Introduction ...... 218 5.2 Semantic Category Incorporation and Gender Assignment in Determiner Phrases ...... 219 5.3 Previous Studies...... 224 5.3.1 Semantic Domains in Spanish/English Code-Switching ...... 224 5.3.2 Gender Assignment in Spanish/English Code-Switching ...... 227

5 5.4 Methodology ...... 236 5.4.1 Participants ...... 236 5.4.2 Data ...... 239 5.4.2.1 Semantic categories ...... 239 5.4.2.2 Gender assignment ...... 242 5.5 Results...... 244 5.5.1 in Northern Belizean Spanish ...... 245 5.5.2 Non-switched versus Switched DPs across Age-groups ...... 248 5.5.3 Choice of Determiner in Mixed DPs ...... 251 5.5.4 Semantic Category Incorporation ...... 255 5.5.5 Gender Assignment in Non-Switched and Switched DPs ...... 265 5.5.6 Analogical Gender ...... 265 5.5.7 Biological Gender and Animacy ...... 267 5.5.8 Feminine Gender Assignment ...... 269 5.5.9 Overgeneralization of the Masculine Gender? ...... 272 5.6 Discussion ...... 275 5.6.1 Semantic Domains ...... 276 5.6.2 What Determines Gender Assignment in Code-Switched Discourse? ...... 277 5.6.3 What Accounts for Feminine-Marked Switched DPs? ...... 281 5.6.4 The Importance of Type of Code-Switching ...... 282 5.7 Concluding Remarks ...... 285

6 IMPLICATIONS OF FINDINGS ...... 287

6.1 Conclusion ...... 287 6.2 Summary of Main Findings ...... 287 6.3 Language Practices among Maya/ in Northern Belize ...... 288 6.4 Toward a Reconceptualization of Code-Switching ...... 298 6.4.1 Convergence and Creolization ...... 299 6.4.2 Re-examining the Nature of Code-Switching ...... 307 6.5 Weighing in on Social and Linguistic Factors ...... 311 6.6 Final Remarks ...... 319 6.7 Limitations and Areas for Future Research ...... 320

APPENDIX

A READING PASSAGE...... 324

B EL COCODRILO ROSADO ...... 325

C LANGUAGE BACKGROUND QUESTIONNAIRE ...... 326

D ACCEPTABILITY JUDGMENT TASK ...... 329

E LANGUAGE BACKGROUND QUESTIONNAIRE ...... 338

F ACCEPTABILITY JUDGMENT TASK ...... 340

6 G LANGUAGE BACKGROUND QUESTIONNAIRE ...... 346

REFERENCES ...... 349

BIOGRAPHICAL SKETCH ...... 383

7 LIST OF TABLES

Table page

1-1 Population by for Northern Belize districts* ...... 18

1-2 Population by Language, Districts and Sub-divisions (1970 Census)* ...... 20

1-3 Percentage of population four years and older by languages spoken and district in Northern Belize* ...... 21

2-1 Factors, strategies, and outcomes in CS ...... 68

2-2 Main topics examined in the present study ...... 95

3-1 Rhotic distribution in normative varieties of Spanish...... 102

3-2 Divergent versus normative patterns of rhotic distribution ...... 107

3-3 Participant information on self-reported proficiencies and language use ...... 114

3-4 Rhotic production in elicited oral production ...... 127

3-5 Phonetic distribution of the intervocalic tap in reading ...... 130

3-6 Phonetic distribution of the intervocalic rhotic in reading ...... 131

3-7 Mean durations of taps across tasks ...... 133

4-1 Linguistic background of consultants for Study 1 ...... 168

4-2 Categories used for Verb Type ...... 171

4-3 Descriptive statistics for stativity and frequency in BLVCs with transitive predicates ..173

4-4 Stativity in BLVCs ...... 174

4-5 Frequency in BLVCs ...... 176

4-6 Descriptive statistics for control structures ...... 177

4-7 ANOVA results for sentence type in control structures ...... 178

4-8 BLVC production across two speaker groups ...... 179

4-9 Lexical verb type and level of diffusion ...... 183

4-10 Lexical verb distribution in canonical BLVCs ...... 187

8 4-11 Linguistic background of consultants for Study 2 ...... 194

4-12 Mean group ratings of BLVCs across two distance levels ...... 200

4-13 Consultants’ ratings of feminine- versus masculine-marked passive BLVCs ...... 205

5-1 Semantic categories and examples from Northern Belize bi/multilingual speech ...... 240

5-2 Gender dis(agreement) in Northern Belizean Spanish...... 246

5-3 Production of determiner phrases across age groups ...... 248

5-4 Openness of semantic domains to in different types of DPs ...... 256

5-5 Logistic regression analysis of factors that favor non-switched or switched Spanish/English DPs ...... 261

5-6 Cross-tabulation of DP production by speaker type and semantic domains...... 263

5-7 Distribution of masculine- and feminine-marked DPs ...... 265

5-8 Translation equivalents for English-origin nouns ...... 266

5-9 Gender assignment according to biological gender and animacy ...... 267

5-10 Feminine gender assignment...... 270

5-11 Gender assignment incongruences across three generations ...... 273

9 LIST OF FIGURES

Figure page

1-1 Map of the Yucatan Peninsula and Belize. © OpenStreetMap contributors, CC-BY- SA ...... 17

1-2 Third generation Mayan/ descendants from San José Yalbac...... 24

1-3 Second generation Mayan/Mestizo descendants from San José Yalbac...... 25

2-1 Muysken’s (2000, p.9) representation of the different types of switching ...... 65

2-2 Bilingual speech strategies for code-switching...... 67

2-3 Morphemes types in the 4-M model (Myers-Scotton & Jake, 2000, p.1064) ...... 72

2-4 A minimalist approach to code-switching (MacSwan, 2005, p.7) ...... 78

3-1 Apico-alveolar trill produced by speaker in tonces agarra ‘then he grabs’ (Willis & Bradley, 2008, p.89) ...... 100

3-2 Tap produced by Veracruz Mexican Spanish speaker in parec(e) ‘seems’ (Bradley & Willis 2012:52) ...... 101

3-3 Picture Story ‘The Flood’ (Timm & Eccott, 1972, p.64) ...... 116

3-4 Canonical tap in corazón ‘heart’ produced by speaker S2 ...... 120

3-5 Retroflex approximant in agarró ‘caught’ produced by speaker F1 ...... 121

3-6 Bunched approximant in cigarro ‘cigarette’ produced by speaker S4 ...... 122

3-7 Two-closure trill in rana ‘frog’ produced by speaker S2 ...... 123

3-8 Null rhotic in uninflected light verb hacer ‘do’ in bilingual light verb construction asé come down ‘to come down’ ...... 124

3-9 Lenited tap in miró ‘saw’ produced by speaker F1 ...... 126

3-10 Overgeneralized tap in cerrado ‘closed’ produced by speaker S2 ...... 132

3-11 Speaker S2’s neutralized production of pero ‘but’ and perro ‘dog’ ...... 135

3-12 Speaker S3’s production of cerrado ‘closed’ and ‘seashell’ ...... 136

3-13 Overgeneralized tap in la rata ‘the rat’ produced by S1 ...... 137

4-1 German root lauf- cannot value Spanish little v’s conjugation feature ...... 165

10 4-2 The light verb hacer values the [uConj] feature of vsp ...... 165

4-3 Root cant- values the [uConj] feature in little v and then move up ...... 165

4-4 Sample items from the AJT ...... 169

4-5 Frequency of lexical verbs in BLVCs ...... 181

4-6 Distribution of lexical verbs following Fuller Medina’s classification ...... 186

4-7 Male consultants’ ratings of passive BLCVs for distance levels 1 and 2 ...... 202

4-8 Female consultants’ ratings of passive BLCVs for distance levels 1 and 2 ...... 203

4-9 Ratings for stative passive BLVCs at distance level 1 ...... 204

5-1 Mean production of determiner phrases by age group and type of determiner phrase ....249

5-2 Mean proportions of switched DPs across different age groups...... 250

11 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

LANGUAGE USE, LANGUAGE CHANGE AND INNOVATION IN NORTHERN BELIZE CONTACT SPANISH

By

Osmer Eder Balam

August 2016

Chair: Gillian Lord Major:

The present mixed-methods dissertation examines the language contact situation in

Northern Belize, where Spanish is in intense contact with English and Belizean Kriol. I specifically analyze three linguistic phenomena – intervocalic rhotics, bilingual light verb constructions and mixed determiner phrases – while attending to concomitant sociolinguistic factors in an effort to unravel patterns of language use, language change and innovation in

Northern Belize.

In light of previous contentions regarding linguistic convergence in Northern Belizean

Spanish, a crucial question which arises is whether phonological and/or morphosyntactic structures in this multilingual context show an approximation toward or movement away from normative Spanish forms. To address this overarching question, quantitative and qualitative data from a total of 161 native speakers of Northern Belizean Spanish were examined.

For rhotics, intervocalic rhotic neutralization was found to be a variable rather than a global phenomenon; thus, no evidence was found of a merged retroflex approximant category.

For bilingual light verb constructions, results showed that stativity, verb frequency and passivization do not restrict the incorporation of ‘hacer + V’ in code-switched discourse,

12 revealing that these syntactic innovations are not constrained by these linguistic factors. For the examination of semantic categories, results revealed that several domains were favorable contexts for English-origin nouns. The openness of semantic domains to English-origin nouns was particularly attested in the speech of translanguagers. For gender assignment, the analysis showed that the Spanish gender distinction was only neutralized in mixed discourse; hence, revealing a composite mechanism specifically induced by code-switching. I argue that sociolinguistic factors such as speakers’ positive attitudes toward CS, their identification with bilingual language practices, and a low degree of normativity are factors that have contributed to the innovative use of Spanish/English CS in Northern Belize.

Overall, results revealed that contact Spanish in Northern Belize exhibits cross- generational stability (i.e., rhotics and gender assignment do not reveal convergence), linguistic innovation (bilingual light verb constructions), and synchronic convergence (gender assignment in mixed determiner phrases). The instantiation of convergence and/or elements of creolization in the CS data support recent work that advocates for a re-evaluation of monolingual notions and perspectives of code-switching and bilingualism.

13 CHAPTER 1 INTRODUCTION

1.1 Overview

Almost four decades ago, Hagerty (1979) aptly observed that the study of Belizean

Spanish was a void in dialectology that needed to be filled, as scant research had been conducted on the contact Spanish situation in Belize. While extensive research on

Spanish/English bilingualism in the U.S. flourished in the 1980s and thereafter, the case of

Spanish/English contact in Belize remained practically unnoticed. To date, scant empirical work has been carried out to understand the contact Spanish situation in Belize, where Spanish as the first language of the majority is in intense contact with English as an official language and

Belizean Kriol (henceforth BK) as the . Endeavoring to contribute to the understanding of linguistic outcomes in this borderlands Central American/ context, I examine phonological and morphosyntactic phenomena to explore the nature of both contact

Spanish and code-switching (henceforth CS) outcomes in Northern Belize.

In this dissertation, I examine a phonological phenomenon in Northern Belizean Spanish

(henceforth NBS) and two morphological/morphosyntactic CS phenomena in an effort to provide a broader insight into contact outcomes and linguistic variation in Northern Belize. Central to my analyses is the concept of convergence, which in the present dissertation, refers to the enhancement (Bullock & Toribio, 2004) or exploitation (Gardner-Chloros & Edwards, 2004) of structural similarities between linguistic systems in bi/multilingual speech.

The examination of convergence in CS outcomes is particularly instrumental to our understanding of the synchronic dimension of CS, as the relationship between CS and convergence in bilingual speech has been contentiously debated in previous work (see Toribio

2004; Torres Caucoullos and Travis 2010). Whereas some scholars argue that the alternation of

14 languages serves as a ‘reflex’ (Toribio, 2004, p.172) that triggers convergence in bi/multilingual speech, others reject this view (Torres Cacoullos & Travis, 2010). From a diachronic perspective, the study of convergence provides an insight into the potential alignment or merging between two similar structures or forms from different languages in situations of intense language contact; hence, allowing us to understand cross-generational patterns of linguistic evolution.

Endeavoring to elucidate contact outcomes in the Northern Belize context, specifically as it relates to NBS and Spanish/English CS, this dissertation examines the following three linguistic phenomena:

1. The neutralization of the intervocalic rhotic contrast in Northern Belizean Spanish. In Spanish, the tap/trill contrast, e.g., /caro/ > [ˈkaɾo] ‘expensive’ versus /carro/ > [ˈkaro] ‘car’, is alternatively maintained via segmental duration measured in milliseconds rather than the number of lingual contacts. In some contact varieties, the normative contrast co- exists with an innovative one, where the intervocalic rhotic contrast is maintained via the systematic use of a tap/retroflex or assibilated rhotic contrast, e.g., /caro/ > [ˈkaɾo] ‘expensive’ versus /carro/ > [ˈkaɻo] ‘car’. In Chapter 3, I examine whether in elicited oral production and reading-aloud, adolescent or post-adolescent speakers produce a converged rhotic category or not.

2. Restrictions, namely (i) verb frequency, (ii) stativity and (iii) passivization, that have been proposed to constrain the incorporation of bilingual light verb constructions, e.g., ‘hacer + V’ as in Hacemos communicate en ‘We communicate in Spanglish’, in bilingual discourse. These three restrictions have been posited to either disfavor or ban the incorporation of ‘hacer + V’ in switched discourse. In Chapter 4, I test whether these restrictions are operative in the speech of adolescent, post-adolescent and adult bi/multilinguals.

3. Mixed determiner phrases (DPs), in particular (i) semantic category incorporation and (ii) gender assignment in non-switched versus switched DPs. The former semantic phenomenon allows us to examine whether certain domains such as education and technology favor other-language items whereas others do not (e.g., technology-related terms such as ‘software’ versus abstract concepts like ‘life’ and ‘happiness’). The latter phenomenon allows us to further understand how gender is assigned to other-language nouns (e.g., unasfem cancionesfem versus unosmasc songs ‘some songs’ or unasfem songs). In Chapter 5, I examine these two phenomena in Northern Belize bi/multilingual CS.

15 The present chapter, which consists of two main sections, provides a detailed background of the sociolinguistic context under investigation. In section 1.2, I first describe Northern

Belize’s demographic and linguistic profile, and highlight important sociohistorical information regarding the presence of Spanish and bi/multilingualism in this region since the 1840s. In section 1.3, I provide an overview of previous studies that provide an insight into the language attitudes and identities of Spanish-speaking bi/multilinguals in Northern Belize.

1.2 The Sociolinguistic Context: Northern Belize

Given that language varieties are forms of social practice, which we cannot understand and/or investigate in isolation of the cultural, political and social contexts of their production and use (Del Valle, 2014; García, 2009; García & Kleifgen, 2010; García & Wei, 2014; Leeman,

2005; Pennycook, 2010; Zentella 1997), I provide here a detailed overview of the sociohistorical presence of bi/multilingualism in Northern Belize, which will elucidate how the context under study differs from other Spanish contact situations in the Spanish-speaking world that have been examined thus far. To this end, I first present information on demographic and linguistic statistics, which provide an insight into Northern Belize’s speakers and their language varieties.

Secondly, I highlight important sociohistorical developments, in particular since the 1840s, which relate to the presence of Spanish and bi/multilingualism in Northern Belize.

Understanding the sociohistorical context of NBS is integral to the present dissertation and its findings, as I argue that some of the attested patterns, especially those that relate to intra- sentential CS, can only be attributed to social factors which have had a profound influence on language contact outcomes in Northern Belize.

1.2.1 Northern Belize: Demographic and Linguistic Profile

Northern Belize comprises two districts: Corozal, which borders the Mexican state of

16 , and Walk (or la ciudad azucarera ‘Sugar City’), which borders both

Mexico to the north and to the west (see Figure 1-1).

Yucatán

Corozal

Orange Walk

Figure 1-1. Map of the Yucatan Peninsula and Belize. © OpenStreetMap contributors, CC-BY- SA

Today, although Northern Belize is a culturally diverse region, it continues to be predominantly populated by Spanish-speaking Maya/Mestizos1. This is not surprising given that during the Caste War of Yucatan (1847-1901), Yucatec Maya and Mestizo migrant families primarily settled in this region of Belize. The 2010 Belize Population and Housing Census reveals that in both Corozal and Orange Walk, Mestizos comprise more than 75% of the total

1 I use the term Maya/Mestizo as some speakers in Belize identify as Maya/Mestizos, claiming their Yucatec Maya roots despite their lack of proficiency in Yucatec Maya. The use of the began to rapidly decline in Northern Belize after the 1930s and 1940s (Koenig, 1975). This was partly triggered, according to Brockmann (1979), by the social stigma associated with the Maya people and language. This time period corresponds with the dates of birth of some of the first generation consultants in the present dissertation (see Chapter 5), who are the only speakers with some listening and/or speaking proficiency in Maya.

17 population (see Table 1-1). In contrast, Creoles (i.e. individuals of mixed

African/European/Amerindian ancestry) in both districts comprise less than 6% of the population. Importantly, this has been the case since the nineteenth century. In 1861, for example, “the Maya and Mestizo ethnic groups accounted for 56% of Belize’s population and

86% of the population of northern Belize” (cf. Camille, 1996, p.54). Thus, although migration from other ethnic groups has increased in the last century, leading to some demographic changes,

Maya/Mestizos remain the ethnic majority in Corozal and Orange Walk.

Table 1-1. Population by ethnic group for Northern Belize districts* Corozal % Orange % Walk Asian (Japanese, Chinese, Taiwanese) 297 0.7 360 0.8 Black/African 70 0.2 40 0.1 White/Caucasian 359 0.9 100 0.2 Creole 2,279 5.6 2,415 5.3 East Indian 1,215 3.0 164 0.4 244 0.6 268 0.6 Hindu 113 0.3 94 0.2 Lebanese 58 0.1 - - Maya Ketchi 349 0.8 215 0.5 Maya Mopan 89 0.2 238 0.5 Maya Yucatec 303 0.7 167 0.4 Mennonite2 2,732 6.7 5,036 11.0 Mestizo/Spanish/ 31,029 75.6 35,513 77.3 Other 32 0.1 46 0.1 Mixed 1,891 4.6 1,290 2.8 Not reported 1 0.0 - - Total 41,061 100.0 45,946 100.0 *Percentages added. For original table, cf. Belize Housing and Population Census (2010, p.78)

In terms of the linguistic landscape of Northern Belize, demographic data from colonial times provide limited insight into the prevalence of bi/multilingualism (i.e., Spanish/Mayan,

2 Mennonites speak a variety of German. They are part of “a Protestant sect that migrated first to Northern and Southern Russia, to Pennsylvannia in the late 1700s, and to Canada a century later” (Barry & Vernon, 1995, p.72). Mennonites in Belize began arriving in 1958, mainly from Manitoba, Canada, and , (Dobson, 1973).

18 Spanish/English, Spanish/Kriol) in Belize, a fact which Hagerty (1979) describes as “unfortunate

[given that] the census of a country where an overwhelming large proportion of the population is bilingual gives no data concerning bilingualism (p. 25)3.”

Census data prior to Belize’s independence in 1981 does support, however, the fact that

Northern Belize has largely remained a Spanish-speaking region. As a point of reference, consider census data from 1970 for Corozal and Orange Walk regarding language use and sub- divisions (i.e., town versus rural areas)4. Table 1-2 below reveals that overall, the most widely spoken languages in 1970 were Spanish and English. In particular, Spanish was spoken primarily in rural areas, both in Corozal (72.2%) and Orange Walk (63.6%). In terms of English, greater competence was reported among speakers who lived in (70.7%) than in rural areas (29.3%), in line with Brockmann’s (1979) observation that “the breadth and extent of

English use in [Orange Walk] villages is less (p.174).”

In Corozal, the difference in English proficiency between speakers from the town and rural areas was less marked in 1970. Notably, English in the 1970 Population Census was inclusive of BK, a fact consonant with the widespread notion at the time that Kriol was simply a of English and not a “language” in its own right. Based on this data, therefore, it is difficult to distinguish how many speakers were proficient in NBS versus BK. Table 1-2 also

3 Le Page & Tabouret-Keller (1985, p.67) provide Census data from 1960, which provides some insight into bilingualism in Belize. From a total population of 90, 505 inhabitants in British , the Census revealed that approximately 51% of inhabitants reported speaking ‘English only’. This was followed by 20% of the population who reported speaking ‘English and Spanish’ and 10% who reported speaking ‘Spanish only’. In this case again, English refers to both English and Belizean Kriol. The higher number of speakers reporting proficiency in ‘English only’ reflects the demographic make-up in Belize for most of the eighteenth and nineteenth centuries, when Creoles comprised the largest ethnic group. Census data from the last five decades reveal, however, that by 1991, Mestizos became the largest population group in Belize (for Census data from 1981, 1991 and 2000, see Gabbert, 2007, p.54; Moberg, 1997, p.85).

4 In his seminal work on Belizean Spanish, Hagerty (1979) employs data from the 1970 Census in order to illustrate language use in Belize. Given that the present dissertation further investigates some of the observations Hagerty made in his work, I have chosen the 1970 Census data as a point of reference to compare language use in Belize during the colonial era versus more recent times.

19 shows that by the 1970s, only a minority of speakers spoke Yucatec Maya. Most speakers of the

Yucatec Maya language were from rural areas, where Spanish/English bilingualism was less prevalent.

Table 1-2. Population by Language, Districts and Sub-divisions (1970 Census)* Districts and Population % English % Spanish % Maya/ % Sub-divisions Ketchi 4724 30.4 1805 54.1 2811 24.8 - - Corozal Rural 10827 69.6 1534 45.9 8528 72.2 592 100 Total 15551 100 3339 100 11339 592 100

Orange Walk Town 5698 33.4 1736 70.7 3831 36.4 22 7.8 Orange Walk Rural 11343 66.6 720 29.3 6691 63.6 261 92.2 Total 17041 100 2456 100 10522 283 100 *An adapted table from the 1970 Population Census data, provided by Hagerty (1979). Percentages added. I report here only data from Northern Belize. Data excluded include Garifuna (appears as Carib in original table), spoken by a minority of speakers in Northern Belize, and Low German, spoken only in Orange Walk, (for table with full details, see Hagerty, 1979, p.27).

The most recent Belize Population and Housing Census (2010) confirms that following the decline of the Yucatec Maya language in the 1930s and 1940s, Spanish has remained the most widely spoken language in Corozal and Orange Walk, spoken by more than 80% of speakers in both Corozal and Orange Walk (see Table 1-3). Thus, in terms of linguistic make-up,

Northern Belize has remained rather stable during the past decades, with Spanish being the predominant language in this area of Belize. The two other languages that are most widely spoken in Corozal and Orange Walk are English and Kriol, a pattern that also bears out for the country. Note that in the most recent Census, English and Kriol are presented as separate language categories, reflecting local status planning efforts to recognize ‘Bileez Kriol’ as a legitimate language.

20 Table 1-3. Percentage of population four years and older by languages spoken and district in Northern Belize* Country Corozal % Orange % Total Walk Chinese 2,600 272 0.7 304 0.7 Kriol 130,467 6,995 18.9 6,974 16.8 English 183,903 20,161 54.4 25,794 62.2 Garifuna 8,442 138 0.4 188 0.5 German 9,364 2,151 5.8 4,147 10.0 Maya (Ketchi/Mopan/Yucatec) 30,748 920 2.5 939 2.3 Spanish 165,296 31,424 84.7 35,522 85.6 Other 2,729 202 0.6 229 0.6 * Note that percentages do not sum to 100, as speak multiple languages. I report here only the data that relate to Northern Belize. For details on the entire country, refer to original tables in Belize Population and Housing Census (2010, p. 21, p. 82).

Although the 2010 Census provides a more detailed overview of the languages spoken in contemporary Belize, it still does not provide information on the prevalence of bi/multilingualism. English appears as the second most widely spoken language (see Table 1-3), reflecting developments in Northern Belize’s educational system (see section 1.2.2). The possibility still remains, however, that some speakers who reported speaking English may have in fact referred to Kriol (for relevant discussion, see Belize Census, 2010, p.21; also see Le Page,

1992, p.85). It must be underscored that although the 2010 Census gives the impression that

English is widely spoken in Belize, the use of English is marked in informal contexts (for relevant discussion, see Balam, 2014, 2016a; Balam & Prada Pérez, 2016; Koenig, 1975; Elliott,

1995, p.143). In Belize, English is learned as a second language in schools, with native English speakers representing a very small minority. As will be discussed later, this has been a historical trend from the time Belize was a British colony.

In the present dissertation, I on Northern Belize as a contact Spanish situation, examining Spanish/English contact outcomes in particular. Sociohistorical data suggest that intense Spanish/Kriol contact is a more recent phenomenon, which likely became more prevalent

21 within the last 40-50 years (for bi/multilingual CS data that reveal the limited use of

Spanish/Kriol switching, see Balam, 2014). Although competence in Kriol among Northern

Belize Maya/Mestizos has increased, recent census data suggest that Spanish is still the dominant language (see Table 1-3). Speakers from villages are typically Spanish-dominant bi/multilinguals, and there are many high school adolescents who have limited or no competence in BK (see, for instance, speakers F3 and F5 in the analysis of the intervocalic rhotic contrast in

Chapter 3). In contrast, Spanish/English contact is prevalent in both urban and rural settings given that English is the official language of school instruction. It is essential to keep in mind, however, that BK is Belize’s lingua franca (Balam, 2013, 2014; Escure, 1983; Salmon &

Gómez-Menjívar, 2014; Le Page & Tabouret-Keller, 1985; Ravindranath, 2009; Seitz, 2005), and as such, its historical and sociolinguistic influence is taken into consideration in the present work, especially as it relates to speakers’ language attitudes, linguistic identities and their CS practices.

1.2.2 History of Contact Spanish and Bi/Multilingualism in Northern Belize

Although sociopolitical influence from Spanish seems very limited after the 1531 defeat of Alonso Dávila’s attempt to establish a Spanish town in (close to present-day Corozal town), there is evidence that the Spanish culture and language may have started to seep into the area we know today as Belize by the early and mid-1600s. A census conducted by Captain Francisco Pérez in 1655 shows that there were Yucatec Mayans with

Spanish names living in Tipú, an early settlement close to the Belize/Guatemala border (Scholes

& Thompson, 1977). The use of Christian first names by Mayan chiefs are also attested in the accounts written by Fr. José Delgado, a Dominican priest who traveled from South Belize to

Mérida in Mexico in 1677 (Thompson, 1972, p.18). It remains under speculation, as Hagerty

22 (1979) aptly points out, however, whether Spanish existed during this period as a or as a second or even first language among the native population of Belize.

Today, researchers concur that the presence of Spanish in Belize only became prominent after 1847, when the Caste War of Yucatan broke out in Mexico (Barry & Vernon, 1995;

Camille, 1996; Church, Yaegar, & Dornan, 2011; Dobson, 1973; Reed, 1964). During this time, more than 7,000 Mestizo and Maya refugees fled their homeland, searching for a new beginning.

They settled in the Northern districts of Corozal and Orange Walk (see Figure 1-1), part of a frontier that at the time was claimed by Great Britain, Guatemala and Mexico (Church et al.

2011). By 1850, Reed estimates that more than 10,000 refugees from the Yucatan and Quintana

Roo had settled in what today constitutes Northern Belize (cited in Hagerty, 1979, p.21). By

1858, Corozal and in the were the largest towns in Northern

Belize (Bolland, 2003, p.114).

In the census of 1861, more than 50% of the colony’s 25,365 people in Northern Belize were inhabitants who had been born either in Yucatan or (Bolland, 1977;

Camille, 1996; Dobson, 1973; Woods, Perry, & Steagall, 1997). Thus, by the time Belize was declared the colony of in 1862, these Mayan- and Spanish-speaking migrants had already established communities such as Patchakán and Xaibé in Corozal (Bolland, 2003, p.114; Koenig, 1975, p.32) and San José Yalbac and San Pedro Sirís in Orange Walk (Church et al., 2011, p.178; Kray, Church & Yaeger, in press). This wave of migration in the mid-19th century established the ethnic make-up of present-day Northern Belize, where Maya/Mestizos have remained as the predominant ethnic group for at least four to five generations (Balam,

2015).

23 In the present corpus, the eldest consultant was a second generation descendant who was originally from San José Yalbac, a village that was among the earliest settlements established by

Maya and Mestizo refugees following the Caste War of Yucatan. First, second and third generation descendants were usually Maya/Spanish bilinguals. The third and fourth generations were the transient generations between Maya/Spanish and Spanish/English bilingualism.

Oral production data in the present dissertation are primarily based on sociolinguistic interviews with fifth and sixth generation Northern Belize Maya/Mestizos, who are

Spanish/English bilinguals or Spanish/English/Kriol multilinguals. The following pictures depict

Northern Belize Maya/Mestizos from different generations.

Figure 1-2. Third generation Mayan/Mestizo descendants from San José Yalbac.

Photo by Eusebio Balam.

24

Figure 1-3. Second generation Mayan/Mestizo descendants from San José Yalbac.

Photo by Carmen Pérez.

It is important to note that there was another major wave of migration that helped to solidify the current status of Spanish speakers as the majority in Belize. In the 1980s, between

30,000 to 40,000 Maya and Mestizo refugees escaping civil wars in Guatemala, Honduras and El

Salvador settled in different parts of Belize (Gabbert, 2007, p.53), particularly in the . Today, there are some villages that are known to be “Salvadoran” communities (e.g., Las

Flores ‘The Flowers’, a village close to the capital city of ). Some of these Central

American refugees did settle in Northern Belize, but in considerably smaller numbers. As a consequence of these major migration waves in the , there is a dialect boundary between Northern and Central Belize (Hagerty, 1979, p.119). It is the former region, however, that is known for several phonological (e.g., use of retroflex approximant) and morphosyntactic innovations (e.g., bilingual light verb constructions). The latter context is generally known for

25 linguistic patterns that are typical of other Central American varieties of Spanish such as

(Hagerty, 1979).

Although Spanish has been in contact with English for more than a century in Northern

Belize, the intensity of this contact has surged only recently. Up to the 1930s and early 1940s, main towns and smaller communities in other parts of Belize were isolated from , which has historically been populated by Creoles (i.e., speakers of mixed African and/or British ancestry). A road connecting Belize City to Orange Walk Town was not completed until 1939.

Prior to 1939, the only way to Belize City was via a 30-hour boat trip from Corozal (Brockmann,

1979, p.163). Thus, given the lack of highways connecting main municipalities in the 1930s, communication and mobility between towns and the city was rather limited (Hagerty, 1979, p.9;

Koenig 1975, p.39; Le Page & Tabouret-Keller, 1985), and Maya/Mestizo people’s exposure to native speakers of English and/or BK was not commonplace.

Le Page & Tabouret-Keller (1985, p.183) observe that “contact with systems of Standard

English was through school books and through a handful of administrative expatriate

Government officers and their families”. A relevant fact is that during the colonial era, L1

English speakers from Great Britain or the U.S. always comprised a tiny minority (for full details, see Dobson, 1973, p.251). Between 1861 and 1946, for example, census data reveal that the number of inhabitants who were born in the consistently accounted for less than 1% of the total population. In 1946, Britain-born inhabitants only accounted for a meager

0.2% (107/59,220). Thus, exposure to English via schooling was very limited.

A pertinent fact is that there was restricted access to secondary school education during this time. Archaeological evidence suggests that as early as the 1930s and perhaps even in prior decades, there were primary schools in remote Maya/Mestizo communities in Northwestern

26 Belize (Church et al., 2011, p.190). Church et al. reported finding inkwell fragments and remnants of toys and porcelain dolls, associated with nineteenth century educational discourse practices, in San Pedro Sirís, an early Maya/Mestizo settlement established in Northwestern

Belize following the Caste War. Access to , however, was not as accessible as primary education. It is not surprising, therefore, that speakers from this generation developed only limited proficiency in English. In Orange Walk Town, it was only after 1953 that urban adolescents were able to obtain a secondary education when Muffles High School was established (Brockmann, 1979, p.169-170). Thus, contact between Spanish and English in academic contexts began to increasingly take place during the 1950s and thereafter, when several high schools opened in Northern Belize (for further discussion, see Brockmann, 1979, p.168).

We must also bear in mind that up to the 1950s, exposure to media via radio and/or television was also limited in Northern Belize, especially in rural areas. One of the consultants interviewed in the present study vividly retold his childhood memory of the first time villagers saw a black and white television with animation but no sound. This was in the early 1950s.

Brockmann (1979, p.166) reports that radios began to appear in Orange Walk in the 1950s. This suggests that during this time Northern Belizeans were almost exclusively exposed to NBS

(which may have sounded at the time more like Yucatan Spanish, the language of first generation migrants) and Yucatec Maya.

Sociohistorical data, therefore, strongly suggest that prior to the 1950s, Spanish in

Northern Belize was primarily in contact with Maya and to a lesser extent with English and BK.

Our eldest consultant reported that BK was not used by any of the villagers in her native village of San José Yalbac in the 1920s and 1930s. It must be noted that up to the mid-1940s, the timber industry dominated the colonial economy, with forest products accounting for more than 90% of

27 the country’s exports (Dobson, 1973, p.265). A salient aspect of the forestry industry was that the men would usually work as permanent or seasonal laborers in logging camps felling or extracting chicle, while the women stayed in the villages (Camille, 1996;

Thompson, 1963).5 It is in the social context of these camps that we can conjecture some contact between NBS and BK may have begun among the crews of Maya, Mestizo and Creole men.

An interesting phenomenon is that in the early 20th century it seems Creole men were more proficient in Spanish rather than Spanish-speaking men more proficient in Kriol. Often times, Creole men would switch to Spanish when speaking to foreigners (see Thompson, 1963, p.230), as this would be more intelligible to them. Brockmann (1979, p.163) observes that there was a small number of Creoles living in Northern Belize, who “spoke Spanish, sometimes to the exclusion of English or Creole.” These speakers were locally known as “Black Spanish.” During this time, Spanish seems to have had a higher social status than BK.

Even though some younger Maya/Mestizos spoke BK in the 1970s, the Spanish variety spoken in Northern Belize was more positively perceived than BK (Koenig, 1975, p.110), and this may account for some Spanish speakers’ initial reluctance to embrace and use BK, which archaeologists such as Thompson (1963, p.230) described as “extremely difficult to understand.”

But BK did not stay for long as a stigmatized variety. In fact, there is evidence that by the 1970s, it had covert prestige and was already displacing the in some areas of Western

Belize (Le Page & Tabouret-Keller, 1985, p.172, 220-221). Prior to this, Allsopp (1965) had also suggested that BK was the language of the British colony, so the covert prestige that BK had in certain areas of Belize may have been present as early as the 1960s or 1950s. More recently, status planning efforts launched by the National Kriol Council (established in 1995) and the

5 Chicle was used in the manufacturing of chewing gum. The latex, obtained by bleeding the sapodilla tree, was boiled until hard blocks were formed (Dobson, 1973, p.265).

28 Belize Kriol Project have helped not only to change people’s attitudes toward BK, but they have also solidified ‘Bileez Kriol’ as the national language of Belize. The overt prestige that ‘Bileez’ or ‘Belizean’ Kriol has gained is especially evident today in the country’s local media (i.e., radio and TV advertisements) and the younger generation’s preferential use of BK over their native languages (Balam, 2013; Balam & Prada Pérez, in press; Ravindranath, 2009).6

The prestige afforded to BK in Belize today is unprecedented, both regionally (i.e.,

Creoles are stigmatized in other Central American contexts, see Balam & López Alonso, 2015) and historically. Even before Belize was an independent country, BK was already symbolic of

Belizean identity and Belizean unity (Ravindranath, 2009, p.130), a position that has never been held by English and/or NBS. As Lawrence (2001, p.126) rightly observes, “The fascinating aspect of this great phenomenon is that it was never encouraged as a language either by the

British or in the schools. Yet Maya children in Big Falls, Garifuna children in , and

Mestizo children in Orange Walk and Corozal can speak it.” Although English was the language of power in colonial Belize, in the 1970s, English was clearly not a popular second language

(Koenig, 1975, p.85). In fact, Maya/Mestizos avoided speaking English even if they were proficient, given that attempts at speaking Standard English were frowned upon in cases of inter- ethnic communication (Koenig, 1975, p.111).

These negative attitudes among adolescents toward standard varieties still seem to be commonplace in Northern Belize (see Balam, 2013; Balam & Prada Pérez, in press). Recent research suggests that BK continues to gain prestige in Northern Belize, particularly now that

6 Salmon and Gómez-Menjívar (2014), who examined attitudes toward Kriol among speakers from Belize City and Gorda Town, argue that Kriol has covert prestige with overt status in Belize. In colonial Belize, Northern Belize Mestizos were also exposed to Kriol via Radio Belize (for relevant discussion on media in Belize prior to 1981, see Elliott, 1995). In contemporary Belize, the use of Belizean Kriol and/or English/Kriol switching is especially prevalent in radio and television advertisements. See, for instance, the following advertisements from Belize Telemedia Limited (https://www.youtube.com/watch?v=PGc9lS0p5uQ; https://www.youtube.com/watch?v=oPxVOHGZVcc), a local company that offers telecommunication services to Belizeans.

29 there are more official efforts to promote the legitimacy of BK as a language (Balam, 2013).

Particularly among some young urban Maya/Mestizos in Orange Walk, a kind of ethnic creolization is attested, where some adolescents identify not only with BK but perceive themselves as Creoles as well (Balam & Prada Pérez, in press). Thus, the covert prestige that BK had in the 70s has certainly transitioned to a more overt manifestation today.

During the last four to five decades, Spanish/English CS has also become increasingly frequent (Balam, 2013, 2014, 2015), adding yet another important dimension to Northern

Belize’s contemporary linguistic landscape. Northern Belize is a borderlands context where

Spanish/English CS co-occurs with BK and English, a contact phenomenon that is rather unusual in the American mainland given that English has official status only in Belize and .

In light of the fact that CS and BK have thrived among Northern Belize Maya/Mestizos whereas standard varieties have remained marked among younger generations (Balam, 2013; Balam &

Prada Pérez, in press), Northern Belize offers fertile ground to examine language contact outcomes in a context that has not been characterized by a strong, monoglossic tradition.

Post-independent Belize is well known for its celebration of ethnic and linguistic diversity. This is especially evident in the national language policy which recognizes the status of English as the official language of instruction, but which also promotes the use of first languages as “important vehicles for [children’s] transition from home to school” (Government of Belize Language Policy, 2000, p.183). The national language policy:

respects the multicultural and multi-lingual nature of the country and the acknowledged wishes of members of the community who belong to a variety of cultural groups and will

i) encourage the use of native/home languages, where necessary, to facilitate learning, and

30 ii) support school and community efforts in fostering cultural preservation through the provision of time and space to teach native/home languages other than English and Spanish.

(Government of Belize Language Policy, 2000, 183)

In this regard, it is notable that Belize does not have a subtractive language policy, but an additive one, which promotes the development of speakers’ rich, linguistic repertoire. The country’s national language policy does not alienate or ban the strategic use of bi/multililingual language practices in the classroom. This partly explains why among Belizean teachers, first languages and/or CS are generally perceived as pedagogical resources (for further discussion, see

Balam and Prada Pérez, in press). In fact, the use of first languages alongside English is a practice that seems to have been prevalent even in colonial times (Hagerty, 1979, p.10; Koenig,

1975, p.82-85). Therefore, it is not surprising that native bi/multilinguals show a positive predisposition to this practice, which has historically comprised a crucial element of classrooms in Belize where English is generally learned as a second language.

The use of first languages or bi/multilingual language practices is not only present in social and academic domains but other public spaces as well. For example, linguistic variation has always played a key role in the country’s local media. In his examination of media systems in Belize, Elliott (1995, p.31) noted the following about Belizean radio announcers’ language practices:

Radio announcers often [show] off a fluency in English, Spanish and Creole, mixing Mexican music with U.S. hits and a variety of song styles from the . Radio Belize in 1995 moved seamlessly from English to Spanish to Creole, from music to talk shows, from rap to religion, including weather, obituaries, and local announcements in a potpourri of styles.

Elliott’s keen observation vividly illustrates how the dynamic nature of bi/multilingualism in Belize pervades the media, and in doing so, mirrors Belize’s acceptance of

31 linguistic variation and its rejection of a monolingualist tradition where it is only one language that is overwhelmingly associated with the society’s everyday functioning.

In Northern Belize, the value that Spanish/English CS and BK attained as identity markers may have been intensified by the people’s search for a national identity, which was an important driving force during the recent struggle for independence (Barry & Vernon, 1995;

Bolland, 2003; Elliott, 1995; Le Page & Tabouret-Keller, 1985; Shoman, 2010). This search for a Belizean identity took precedence in nationalistic discourse not only in the years leading to

Belize’s independence, but after 1981 as well. A call for unity was a recurrent theme in both political and educational discourse.

In the 1960s, for instance, the anti-colonial movement in pre-independent Belize was strong, with the United Black Association for Development (UBAD) activist Evan X Hyde championing the notion that black comprised of “four parts: the black, the brown, the red and the yellow”. Hyde made a call for the unity of all non-white minorities in Belize because in his view, they were all oppressed by the white British colonial master (Shoman, 2010, p.15). A statement about “Culture and Sovereignty” in the textbook A History of Belize: Nation in the Making

(1983, p.73) candidly illustrates the kind of advocacy for national unity that was prevalent in the

1980s:

For much of our history, the natural interaction of cultures which co-exist within one community was inhibited by the colonial policy of divide and rule, which ensured that our various cultures remained largely isolated from, and suspicious of each other, and that the colonizer’s culture remained dominant. An essential part of the decolonization process must therefore be the elimination of all colonially inherited prejudices about each other’s cultures.

The historical origins of our people and the more recent influences upon our culture have produced diversity. Out of this diversity we must seek unity, while recognizing the value of our different customs and traditions.

(cited in Bolland, 2003, p. 217)

32 Given that Standard English was associated with the colonial master, it is not surprising that although English had prestige, it never became the language of the people. Among Northern

Belize Maya/Mestizos, Spanish/English CS developed as the distinctive marker of their national and ethnic identities (Balam 2015; also see Chapter 6), at the same time that BK solidified its status as the lingua franca of Belize. This is a unique sociolinguistic phenomenon, as we can see that Spanish/English CS and BK almost concurrently developed as important identity markers.

Notably, both CS and Creoles have historically been perceived as corrupt or structurally imperfect or impoverished off-shoots of standard languages (for perceptions in relation to

Creoles, see DeGraff, 2005), yet in the Belize context, these are precisely the contact varieties that have been embraced by the people.

Importantly, the 1980s were not only about nationalism, but about other key social developments such as the access to technology. Prior to 1981, Belize had no national television service, but the same year that Belize attained its independence was the same year it got access to

American programming via satellite technology (Elliott, 1995, p.5; Lent, 1989; Oliveira, 1986).

Thus, speakers became increasingly more exposed to (Oliveira, 1990). Some scholars, such as Elliott (1995), questioned whether the sudden advent of American media and the “the television voice of the 1980’s [with] an American accent” would somehow negatively affect or weaken Belizeans’ emerging sense of national identity. At least at a linguistic level, it seems that Belizeans’ fascination with American television did not translate to a desire to ‘sound

American’ or to use more Standard English in informal contexts. As the present data will show

(Chapter 5), the “television generation” (Elliott, 1995), which had greater access to secondary and post-secondary education, chose to employ Spanish/English CS rather than monolingual

English and/or monolingual Spanish to assert their identities.

33 In sum, language contact between Spanish and English became increasingly intense in

Northern Belize during the last six to seven decades, catalyzed by important developments in

Northern Belize such as the establishment of secondary schools in the 1950s and thereafter, and the exposure to English via American mass media in the 1980s. Linguistically, whereas the abandonment of the Yucatec Maya language for NBS became prevalent among many

Maya/Mestizo families in the 1940s (Brockmann, 1979, p.167; Koenig, 1975, p.82), the last four decades have been characterized by an increase in Spanish/English/Kriol bi/multilingualism

(Balam, 2015; Brockmann, 1979) and a decrease in the use of NBS among younger

Maya/Mestizos in urban areas of Orange Walk (Balam, 2013; Balam & Prada Pérez, in press).

The last two decades have also opened way for a more regularized or ‘focused’ (in the sense of Le Page & Tabouret-Keller (1985)) linguistic situation vis-à-vis BK. While previous scholars (e.g., Kernan et al., 1976) described the Creole spoken in Belize as a continuum (i.e.,

Standard Belizean English, Non-standard Belizean English, City Creole and Broad Creole; also see Escure, 1982), we can now consider English and BK as separate linguistic systems with their own regional variations (for work on BK varieties, see Salmon & Gómez-Menjívar, 2014;

Salmon, 2015). This conceptualization has been made possible thanks to the increased access to higher , coupled with an increased exposure to both BK and Standard English input in the mass media; and the birth of ‘Bileez Kriol’ as a national ‘language’ with its distinctive grammar (Decker, 2005; Greene, 1999; Young, 1973). The development of BK as a language of prestige marks an important turning point for the future of Spanish in Belize, as the contact between Spanish, English and BK will become increasingly more intense. Given that the younger Maya/Mestizo generation in Northern Belize is more positively predisposed to Kriol

34 than previous generations, Spanish/English/Kriol multiligualism is bound to become more commonplace (for relevant discussion, see Balam, 2014, p.88-89).

The intensity of language contact is clearly evident in present-day Orange Walk, where the exposure to different varieties is not only constant but also pervasive. On any given morning, speakers can listen to Despierta Belice, a local television show where the host speaks NBS and casually switches back and forth between NBS, English and BK. During the day, speakers can listen to the local radio station Estereo Amor, where announcers speak a variety closer to standard (Mexican) Spanish7, or they can tune in to radio stations from other parts of the country, such as Krem FM, where radio announcers speak either BK or English. In the evenings,

Orangewalkeños can watch the national news where the reporters speak English and the interviewees typically respond in BK, or they can watch the local news where the news anchor speaks English, but the interviewees usually speak Spanish or switch between languages.

Furthermore, on cable television, which has now become accessible even in rural areas, speakers can watch programs on CNN, NBC and HBO, but also on Telemundo, El Canal de las Estrellas, and Univisión. Thus, to speakers in Northern Belize, language variation is the norm as they are constantly exposed not only to local and regional varieties of Spanish and English, but also to

BK and other languages (e.g., , German, Hindi) as well. This multilingualism pervades both everyday life and the media as well.

7 By , I mean the variety that is taught in academic contexts. Following Yiakoumetti and Esch (2010), standard languages are associated with educational systems. In contact situations, there is significant variation vis-à-vis the similarities and differences between standard and non-standard varieties. Thus, the variation between standard and non-standard varieties is better conceived as existing on a continuum. Historically, as a result of family ties to the Yucatan region and exposure to Mexican media, Norteños in Belize have primarily been exposed to Mexican Spanish.

35 In order to further contextualize bi/multilingualism and the status of language varieties in

Northern Belize, in the following section, I elaborate on Maya/Mestizos’ attitudes toward and identification with the language practices in their multilingual repertoire.

1.3 Language Attitudes and Identity in Northern Belize

In previous work, several authors have underscored the important role that language attitudes have on speakers' use of CS in bi/multilingual contexts (Backus, 2005; Coulmas, 2013;

Field, 2005; Jacobson, 1998; Muysken, 2000, 2013; Sebba, 1998; Toribio, 2004, among others).

For instance, both Muysken (2013, p.714) and Toribio (2004, p.172) suggest that dense CS practices are most likely to be found in contexts with low levels of linguistic prescriptivism, where CS is valued and positively perceived. In terms of identity, Gardner-Chloros (2009, p.27) suggests that frequent CS is especially prevalent in communities where this language practice assumes an identity function.

CS, therefore, can be analyzed as the linguistic manifestation or enactment (in the sense of Le Page & Tabouret-Keller, 1985) of a group (Gardner-Chloros, 2009, p.5; Siegel 2005) or bicultural (Mahootian, 2003; Zentella, 1997) identity. In contrast, lack of identification with this practice, due to larger sociolinguistic or sociopolitical factors, may be indexed by the limited use of CS among fluent bilinguals (Toribio, 2002). In light of Belize’s history of bi/multilingualism and important developments that have taken place in the last few decades (e.g., incorporation of

Spanish in school curricula, status-planning efforts to promote BK as a language, etc.), an important question raised is whether present-day younger generations of Northern Belize

Maya/Mestizos are more positively predisposed to CS or to monolingual varieties.

Language attitudes, or the perceptions and value judgments that people ascribe to varieties (Hidalgo, 1986) and the speakers of these varieties (Anderson & Toribio, 2007; Fasold,

1984; Galindo, 1995; Romaine, 1995), are integral to our understanding of multilingual contexts,

36 as they allow us to have insight into more general societal attitudes toward linguistic variation.

This, in turn, can serve as a barometer of the nature of linguistic purism or prescriptivism in a community.

To date, while extensive research has examined Spanish/English bilinguals’ attitudes toward U.S. Spanish varieties (Anderson, 2010; Carranza & Ryan, 1975, Flores & Hopper, 1975;

Ryan & Carranza, 1977; Sawyer, 1975; Solé, 1977, among many others), scant work has been carried out on language attitudes among bi/multilinguals in Northern Belize. There are some sociolinguistic studies, however, that do provide some insight into the language attitudes and identities of Maya/Mestizos in Belize. Overall, these studies reveal that prior to Belize’s independence in 1981, NBS was associated with a Mestizo identity, and NBS had a higher status than BK among Northern Belize Mayan/Mestizos. In contrast, more recent studies point to (i) the ascension of BK as a prestige variety associated with a strong pan-Afro Belizean linguistic identity and (ii) the use of Spanish/English CS as a marker of Northern Belize Maya/Mestizo identity.

Early sociolinguistic work conducted by Koenig (1975) emphasized the close association between the use of NBS and a strong Mestizo identity. Koenig, who collected data from surveys, participant observation, and interviews, examined the patterns of inter-group communication among speakers from different ethnic backgrounds in Patchakán village and Corozal Town.

Among other findings, Koenig (1975, p.188) found that “language was an important marker of ethnic identity.” Thus, being Mestizo equated with speaking NBS. Interestingly, Koenig noted that villagers closely identified with “the Mexican people, their culture and their language”

(p.82).

37 Koenig’s findings also revealed that there was a general negative predisposition toward

BK. She reported that of the 50% of Mestizos in her study who expressed negative attitudes towards the first language of other ethnic groups, 80% had negative attitudes towards BK. In contrast, only 27% of Creoles stated that they disliked NBS, whereas 50% of native BK speakers had negative attitudes toward their first language. Thus, NBS clearly had a higher status than

BK. Although competence in Kriol was attested among speakers who were younger than 30 years, it was not prevalent.

Brockmann (1979), who descriptively analyzed language use in Orange Walk Town via surveys and participant observation, made similar assertions. Consonant with Koenig’s findings,

Brockmann noted that ethnicity was closely linked to language choice, as Mestizos spoke

Spanish amongst themselves. Brockmann also found that BK had a minority status in Orange

Walk. He reported that there was “only a limited amount of Creole speaking competence among

Mestizos” (p.170). Importantly, however, Brockmann pointed out that some Mestizo parents were concerned about the “Creolization” of the younger generation in terms of clothes, dance and language, particularly among males. Thus, in the late 1970s, the use of BK was already noticeable among some youth in Orange Walk Town, in line with Le Page and Tabouret-Keller’s

(1985) observation that prior to Belize’s independence in 1981, BK was already starting to gain ground as the country’s lingua franca.

According to Brockmann, the low incidence of intermarriage between Mestizos and

Creoles and the use of NBS within ethnic boundaries contributed to a diglossic speech community. Whereas NBS was used within social and private domains, English was used in official contexts such as the workplace and government offices. Importantly, Maya/Mestizo men used more English than women, given that men typically dealt with matters that were not related

38 to the home domain. In line with Koenig’s observation of the marked status of English in informal speech, Brockmann observed that “[the] inappropriate use of English was regarded as pretentious and ethnically disloyal” (p. 177). He observed that the trend at the time was toward increased competence in NBS, English and BK.

In their longitudinal study of language use and identity in the Cayo District, Western

Belize, Le Page and Tabouret-Keller (1985) also noted a process of “Creolization”, which seemed more advanced in this region of Belize. Although their investigation8 focused on Belize as a Creole contact situation, pertinent insights into the status of Spanish and BK surfaced in their interview data. For example, GM, a speaker Le Page had originally recorded in 1970, described Spanish as “Creole Spanish” and “broken-up Spanish” (1985, p.167), suggestive of the fact that in colonial Belize, both Belizean Spanish and Belizean Kriol (BK) were stigmatized.

GM also admitted to mixing both Spanish and BK, “balanc[ing] half of each (1985, p.167)” revealing that CS was also practiced, in line with Brockmann’s (1979) observation in Orange

Walk.

In a follow-up interview in 1978, Le Page reports that GM commented that people in San

Ignacio, a town in Western Belize, were “becoming Creole speakers” (1985, p.169). In a case study of a Spanish/BK bilingual family in Western Belize, Tabouret-Keller (1980) examined this shift to BK, and suggested that younger speakers’ conception of themselves as being “mixed” allowed them to embrace their Mestizo ethnic identity and their Creole linguistic identity. Thus, data from Le Page and Tabouret-Keller suggest that the shift to BK in some municipalities of the

Cayo District occurred at a strikingly rapid rate, a process Le Page (1992, p.90) later described as the double shift in three generations (i.e., from Maya to NBS to BK). Le Page and Tabouret-

8 Le Page and Tabouret-Keller based their longitudinal work on two main datasets. In their 1970 study, they examined language use from interviews with 280 children (between ages 10 – 16) from the Cayo District. In their 1978 follow-up study, they interviewed forty consultants from their original sample.

39 Keller’s observations are important, as they reveal the complexity of identity affiliations (ethnic, linguistic, national) that were emerging in Belize at a time when the country was still struggling for its independence from Great Britain.

Seitz (2005) more cautiously delved into the aspect of identity among Belizeans. Seitz analyzed data from a questionnaire (with 113 adult consultants), interviews (with 21 consultants) and participant observation in an effort to investigate speakers’ attitudes in relation to their cultural and ethnic identities. Among other findings, Seitz found that Belizeans from different ethnic groups generally had a negative attitude toward immigrants from Central American countries. Although he does not state that this negative predisposition extended to Central

American varieties of monolingual Spanish, we can surmise this was the case. Le Page (1992, p.87) suggests that one of the main reasons why Spanish may have become perceived with a sense of antagonism in Belize was that it was associated with Guatemala, which has long claimed Belize as part of its territory9.

Notably, Seitz noted that among the different ethnic groups (i.e., Mestizo, Creole,

Garifuna, Maya, East Indian, Other), Mestizos had the highest percentage of negative attitudes

(41%), and younger Belizeans (especially those between the ages of 25 – 34) were more negatively predisposed to immigrants than older Belizeans. In his analysis, Seitz (2005) reports that the most hostile attitudes toward Spanish-speaking immigrants were found among Northern

Belize Mestizos “of ‘long-term’ Belizean heritage of Mexican ancestry” (p.101). Based on his findings, Seitz concluded that Northern Belize Mestizos use BK as a means of asserting their authentic “Belizean” identity and to avoid being associated with Spanish-speaking Central

American immigrants by Creoles (for an alternative view, see Chapter 6).

9 Before Belize obtained its independence from Great Britain in 1981, “Guatemala, which claimed Belize as part of its own territory which should have been inherited from had it not been illegally occupied by the British, threatened to annex the whole country as soon as it became independent” (Le Page, 1992, p.87).

40 More recently, Balam (2013) provided further insight into the identities of Northern

Belize Maya/Mestizos. Balam examined speakers’ language attitudes (toward NBS, standard

Spanish and bi/trilingual CS) and linguistic identities through the analysis of survey and interview data from 42 native bi/multilinguals (between ages 15 – 25) from Orange Walk,

Belize. The study’s findings revealed that most speakers did not have strong pejorative attitudes toward their native Spanish variety. Consultants did not agree with the assertion that NBS was incorrect. Furthermore, none of the four groups (adolescent males; adolescent females; post- adolescent males; post-adolescent females) agreed that Standard Spanish sounded better than

NBS. All groups concurred that they were proud of their unique Spanish accent. They also agreed that when they travel abroad, they feel proud of their ability to speak different languages; thus, revealing participants’ overall positive attitude toward their bi/multilingualism. Notably, the post-adolescent group showed a more positive attitude to bi/trilingual CS than adolescents.

Results also showed that the use of monolingual varieties was marked among adolescents, especially males, consonant with Koenig’s (1975) observations that the use of standard varieties was disparaged among adolescents.

An important finding was that all speaker groups generally concurred that bi/trilingual CS reflects who they are, in line with speakers’ frequent use of bilingual language practices in naturalistic speech. It was evident in the interviews, however, that some speakers also had a BK linguistic identity. Specifically, adolescent consultants did engage in bilingual CS but showed a preference to use monolingual BK. Some adolescents embraced a strong pan-Afro Belizean

Kriol linguistic identity and spoke more BK than either Spanish or Spanish/English CS. A subsequent study on the perceptions of 32 Belizean teachers of Spanish regarding students’ attitudes toward Spanish instruction further revealed that students demonstrate the most negative

41 attitudes toward Spanish varieties, yet hold the most positive attitudes toward BK (Balam &

Prada Pérez, in press).

Speakers’ positive attitudes toward CS and their identification with this sociocultural practice was replicated with survey data from a second group of 52 native consultants (34 females, 18 males; ages between 14 - 40) from Northern Belize (Chapter 5). Crucially, there were two revealing pieces of information in the second data set. First, speakers gave higher ratings to the idea that bilingual Spanish/English CS rather than trilingual Spanish/English/Kriol

CS reflects who they are. There was a tendency for post-adolescents (Mean = 6.00) and adults

(Mean = 6.04) to give higher ratings (on a scale where 1 indicates ‘Totally Disagree’ and 7 indicates ‘Totally Agree’) to Spanish/English CS as an identity marker than the adolescent group

(Mean = 5.44). The value attributed to Spanish/English CS is particularly important, as it differs from previous findings, and it points to the crystallization of a Northern Belizean Mayan/Mestizo identity. Recall that Koenig’s and Brockmann’s observation was that there was a direct relationship between ethnic and linguistic identity. In other words, being Mestizo was strongly associated with the use of NBS and not BK and/or CS. On the other hand, Le Page and Tabouret-

Keller suggested that Mestizos in Western Belize embraced a Mestizo ethnic identity but a Kriol linguistic identity. In this case, however, we see that speakers closely identified with

Spanish/English CS.

The other important finding was that speakers did not identify with BK. Although speakers’ overall group rating for the idea that Belizean Kriol is not a language (overall Mean =

3.22) reveals that they in fact recognize the legitimacy of Kriol as a language, they do not identify with it. Although BK is associated with a national identity (i.e., Belizean), the data suggest that contra Seitz’s (2005) view, Northern Belize Maya/Mestizos actually employ

42 bilingual Spanish/English CS in order to ‘project’ (in the sense of Le Page and Tabouret-Keller

(1985)) their national and ethnic identities (for further discussion on this issue, see Chapter 6).

This is supported by the fact that when consultants were asked to write down in terms of level of importance (where 1 indicated ‘most important’, 2 indicated ‘important’ and 3 indicated

‘somewhat important’) the three labels (i.e., Latino, Hispanic, Mestizo, Creole, Maya-Mestizo,

Central American, Orangewalkeño, Belizean, Other) that best identified who they are, speakers gave precedence to their ‘Belizean’ (i.e., 65%, N = 34) and ‘Mestizo/Mayan-Mestizo’10 (i.e.

27%, N = 12) identities.

In summation, the insights and findings from the previous studies reveal several crucial trends. The data strongly suggest that Northern Belize Maya/Mestizos do not have ultra- normative attitudes toward their varieties. This was evinced in their neutral responses or disagreement with the notion that Standard Spanish sounds better than NBS and in their positive attitudes toward CS. In terms of identity, recent work shows that Northern Belize Maya/Mestizos identify with bilingual discourse practices, and perceive Spanish/English CS as representative of who they are. Thus, unlike previous generations who had limited competence in English, and who identified more with the Mexican culture and language (Koenig, 1975, p.85), the younger generations of Maya/Mestizos mostly associate their identity with Spanish/English CS (Balam,

2013, 2015). The emergence of CS as an identity marker in the last four to five decades, alongside the rise in popularity and prestige of BK, are important driving forces that seem to have paved the way for the extensive use of bilingual language practices in this multilingual

10 I combined the identity labels ‘Mestizo’ and ‘Maya-Mestizo’ as they are sometimes used interchangeably in Northern Belize. The reason why some speakers may choose one label over another may have to do with their proficiency or lack thereof in Yucatec Maya, their familial and network ties (i.e., rural vs. urban) and/or a desire to want to assert their Mayan roots. Although Yucatec Mayan is a dying language in Belize, there are local efforts to revitalize the Mayan language. A well-known local activist in the Orange Walk District is Mrs. Felicita Cantún, who promotes Mestizo/Mayan culture through dance in national events and through language (i.e., Yucatec Mayan) in high schools.

43 context. In this dissertation, I further explore how these factors may have contributed to stability in the use of rhotics (Chapter 3), but convergence and/or innovation in the case of bilingual light verb constructions (Chapter 4) and mixed determiner phrases (Chapter 5).

Before examining these linguistic phenomena, in the following chapter, I provide an overview of the theoretical concepts, frameworks and debates that are central to this dissertation.

I also elaborate on how this large-scale study contributes to previous work on language contact and intra-sentential CS.

44 CHAPTER 2 THEORETICAL BACKGROUND

In this chapter, I elaborate on the theoretical concepts, frameworks and discussions that are central to the present work. In section 2.1, I operationalize the concept of CS to provide a working definition for the remainder of the dissertation. In section 2.2, I problematize the notion of CS itself, specifically by (i) outlining competing views in the literature on what CS constitutes, and (ii) elaborating on the concomitant debate on the relationship between

CS and convergence. In section 2.3, given my examination of both variationist and generative restrictions on the use of bilingual light verb constructions, I provide a brief overview of variationist and generative constraints that were influential in the early study of CS. I subsequently summarize the main tenets of three theoretical approaches to CS, namely

Muysken’s typological approach to CS, the Matrix Language Framework model (Myers-Scotton and colleagues) and the Minimalist approach to CS (MacSwan and colleagues). Although I do not provide a comparative analysis of these approaches in relation to the Northern Belize CS data, each model is fundamental to our current understanding of the study of CS, both from typological and formal linguistic perspectives. In section 2.4, I provide an overview on the debate vis-à-vis the deterministic effect of social versus linguistic factors in language contact outcomes. Lastly, in section 2.5, I situate the Northern Belize context within the extant literature on contact linguistics and CS. I also provide the structure and overarching research questions of this dissertation.

2.1 Defining Code-Switching

CS has been defined in a variety of ways1. Heller (1988, p.1) defines CS as “the use of more than one language in the course of a single communicative episode”. McCormick (2002,

1 For relevant discussion on the history of the study of CS, see Álvarez-Caccamo (1998) and Benson (2001).

45 p.216) asserts that CS can be perceived as a superordinate term, referring to “the juxtaposition or alternation of material from two (or more) languages or .” More recently, Bullock and

Toribio (2009, p.1) highlight the notion of bilingual competence, defining CS as “the ability on the part of bilinguals to alternate effortlessly between their two languages.”

Examples (1) and (2) from Spanish/English CS below exemplify inter- and intra-sententential

CS respectively, a distinction that is uncontroversial in CS research. Whereas (1) illustrates switching that is characterized by alternation at sentential boundaries, in (2), switching occurs at phrasal boundaries within the sentence. Importantly, it is the latter type of CS, associated with a high degree of bilingual competence (Anderson & Toribio, 2007; Bullock & Toribio, 2009;

Poplack, 1980; Toribio, 2001, 2004; Zentella, 1981, 1997), which has been at the core of CS research for the past few decades.

(1) ¿Qué pasa? Can’t you come over today?

What happen can’t you come over today

‘What’s happening? Can’t you come over today?’

Pfaff (1979, p.316)

(2) Yo no comprendo cómo un gobierno can allow una cosa asi to happen

I.1SG not understand how a government can allow a thing like that to happen

‘I don’t understand how a government can allow something like that to happen.’

Moyer (1992, p.194)

Although scholars today agree on the inter- and intra-sentential CS distinction, there is much less concensus on what CS constitutes (Rodríguez-González & Parafita-Couto, 2012). In the following section, I elaborate on this debate, as the data from Northern Belize allow us to examine the nature of CS in a context where this language practice is not stigmatized; hence,

46 allowing us to investigate the dynamicity of CS. As Bhatia & Ritchie (2016) rightly highlight, the study of CS oftentimes faces methodological problems, especially when it is examined in contexts where this practice is regarded as deviant. Therefore, in order to have a better understanding of the dynamic nature of CS, it is of paramount importance that we study CS in contexts where language mixing is not stigmatized but is rather a a norm that is tacitly accepted and embraced by the community.

2.2 Opposing Views of Code-Switching

In this section, I present an overview of the two opposing views of what CS entails, and I further elaborate on the debate of CS as a mechanism that induces convergence in bilingual speech. These are two controversial issues that relate not only to theoretical discussions about the nature of CS but also to the overall conceptualization of bilingualism.

2.2.1 The Dominant View of Code-Switching

Even after decades of extensive research on CS, scholars still disagree on what exactly

CS entails (Bullock, Hinrichs & Toribio, 2014; Gardner-Chloros & Edwards, 2004; Rodríguez-

González & Parafita-Couto, 2012; Winford, 2003). The view of CS as involving the combination of two discrete linguistic systems, however, remains the prevailing view. For example, the

Matrix Language Frame (MLF) model (Myers-Scotton and colleagues) and the Minimalist approach to CS (MacSwan and colleagues) – formal models that I describe in more detail later – are grounded on this notion.

The MLF asserts that classic CS occurs in cases where a clear distinction between the

Matrix Language (ML) and the Embedded Language (EL) is strictly maintained (Myers-Scotton,

1993, 2002). As it relates to the Minimalist approach to CS, Herring et al. (2010, p.557) note that

MacSwan’s approach also suggests that “bilinguals have discrete and separate lexicons for the languages they speak, each with its own internal principles of word formation, as well as

47 separate phonological components.” MacSwan (2005, p.5) highlights that CS is the union of two lexically-encoded grammars or “the simple consequence of mixing two lexicons in the course of a derivation.” MacSwan and Colina (2014, p.190) further argue that bilinguals have “a separately encapsulated phonological system for each language”. Thus, although the MLF and the

Minimalist approach to CS differ on conceptual grounds, they are both based on the idea that CS entails the alternation of two discrete languages.

This dominant view may relate to the fact that in the last three decades, there has been a particular focus on quantitatively distinguishing borrowing from CS (e.g., Aaron, 2014; Poplack

& Meechan, 1998; Poplack, 2012). This two-way distinction has been a fundamental concern because singly occurring donor-language items (e.g., English-origin nouns in Spanish discourse, see Chapter 5) are among the most common kinds of other-language items in bilingual speech

(Gardner-Chloros, 1995; Muysken, 2000; Myers-Scotton & Jake, 2013; Poplack & Meechan,

1998; Weston, 2013). Therefore, analyzing bilingual discourse data typically requires drawing a line between both phenomena if one is to study them. It is unsurprising, therefore, that the borrowing/CS dichotomy has continued to play a key role in CS research.

Earlier work on contact linguistics distinguished between different types of borrowing.

For instance, Haugen (1950, 1972) distinguished between three main types of borrowings: loanwords, loanblends and loanshifts. Loanwords refer to cases when morphemes and their meanings are imported from another language with little or some phonological integration (e.g.,

French word déjà vu in English discourse). Loanblends refers to cases where one part of the word is borrowed from the donor language, whereas the other part reflects the native language

(e.g., bonchote ‘bunch’ in Belizean Spanish: Hagerty, 1996, p.136). In contrast, loanshifts refer to cases of semantic extension, which occurs when a word from the base or host language is

48 extended in meaning so that it corresponds to a word in the donor language (e.g., grosseria ‘rude remark’ is extended to ‘grocery’ in North American Portuguese: Haugen, 1950, p.84). Poplack and Sankoff (1984) identified four basic criteria for the characterization of loanwords, which include: frequency of use, displacement of synonym in the host language, morphophonemic and/or syntactic integration, and acceptability of donor-language item in the host-language lexicon.

To date, the main criterion that has been used to distinguish between borrowing and CS is the degree of morphophonemic integration. In terms of integration, established borrowings infiltrate deeply into the host lexicon so that in time they are commonly used by monolingual speakers who may lack knowledge or competence in the donor language (Lipski, 2005; Winford,

2003). Thus, established borrowings are fully integrated in the borrowing language, and speakers often treat them as native items. Poplack and colleagues differentiate established from ‘nonce’ borrowings. According to the Nonce Borrowing Hypothesis (Poplack & Meechan, 1998, p.137;

Poplack, 2012; Sankoff, Poplack, & Vanniarajan, 1990), if lone other-language items are nonce borrowings, then they should pattern with the equivalent structure in the recipient language and not the donor language. Thus, what distinguishes these borrowings is that in terms of morphosyntactic structure, they align with the base language (i.e., the ML in Myers-Scotton’s terms). This, however, is not the case of CS, as switches are perceived to be “transitory phenomena” (Winford, 2003, p.107), and hence entails the maintenance of patterns that clearly evince two separate language systems. Sankoff, Poplack and Vanniarajan (1990) highlight that

CS is attested only when there is a clear change of and syntax within an utterance.

Thus, whereas borrowing is generally believed to entail a certain degree of integration in the host or base language, CS is assumed to entail the alternate use of two discrete, monolingual

49 grammars that remain autonomous in bilingual discourse (Gardner-Chloros, 2010, p.30). Today, most researchers concur that borrowings involve the partial or full syntactic, morphological and/or phonological integration of a foreign item into the host lexicon, whereas CS entails the maintenance of the monolingual norms that govern the structure of the input languages (Bullock

& Toribio, 2009; Bullock, Hinrichs, & Toribio, 2014; Herring, et al., 2010; MacSwan, 2000,

2005, 2014; Poplack, 1980, 1988, 2012; Poplack & Sankoff, 1984; Rodriguez-González &

Parafita-Couto, 2012; Torres Cacoullos & Travis, 2010; Toribio, 2001, 2004, among many others). From this perspective, therefore, CS involves the combination of elements from two discrete linguistic systems.

According to Bullock and Toribio (2009, p.2), an “incontrovertible” example of intra- sentential CS would be the title of Poplack’s (1980) , as in (3), which illustrates the clear distinction between English and Spanish in the switched utterance. On the other hand, (4) is ungrammatical, as it violates CS constraints.

(3) Sometimes I’ start a sentence in Spanish [sic] termino en español

“…and I finish in Spanish.”

(4) *Sometimes yo will empezar a oración in inglés and termino in español

It is important to highlight that convergence is generally not viewed as constituting a key element or strategy of CS (but see Backus, 2004; Bullock & Toribio, 2004; Gardner-Chloros &

Edwards, 2004; Toribio, 2004). Myers-Scotton (2002), for instance, maintains the strict distinction between ‘classic’ and ‘composite’ CS. While the former type of CS (akin to

Poplack’s notion of ‘intra-sentential CS’) is based on the assumption that bilinguals “fully maintain both language systems” (Bullock & Toribio, 2009, p.4) as in (3), the latter occurs when the language or structure in question contains elements from both languages (i.e., there is

50 linguistic convergence). Thus, classic CS is not envisioned to co-occur with composite CS. They are taken to be distinct phenomena.

Previously, researchers have taken different approaches to explain the distinction between borrowing and CS. For instance, in her analysis of Welsh/English CS, Deuchar (2005) considers only English-origin items in the Welsh dictionary as borrowings, following Myers-

Scotton’s (2002, p.41) assertion that the status of lexical item as a borrowing is unquestionable if it appears as a dictionary entry. Thus, items absent in the Welsh dictionary are considered switches. Aaron (2014), in her analysis of English-only nouns in , distinguishes between ‘singletons’ and ‘diffuse’ items. Whereas the former refer to nouns that are used only once by one speaker, the latter refers to nouns that are used at least three times by three different speakers. Aaron assumes that ‘diffuse’ items are established borrowings, whereas

‘singletons’ may be single-word switches. Hebblethwaite (2007, p.76), on the other hand, in his analysis of left-periphery items in Haitian Creole/English CS, distinguishes between ‘borrowing- in-progress’ (i.e., lexical item that is ‘being borrowed into a recipient language but that is still linked to the donor language through codeswitching’) and CS by coding the surrounding language indexation.

Crucially, not all scholars support the categorical distinction between borrowing and CS.

Several researchers such as Deuchar and Stammers (2016), Gardner-Chloros (1987, 1995),

Hebblethwaite (2007, 2010), Myers-Scotton (2002), Thomason (2001), Treffers-Daller (1994),

Samar and Meechan (1998) and Winford (2010, p.182) have suggested that there is no clear-cut categorical distinction between CS and borrowing. Instead, borrowing and CS exist on a continuum. In particular, Gardner-Chloros (1995) concurs with Haust (1995) in that borrowings begin their historical trajectory as code-switches that later become increasingly used by speakers

51 of the borrowing language. Thus, borrowings and switches fall within a continuum and the status of an item can change depending on how speakers adopt and integrate these forms into the host lexicon. It must be emphasized, however, that there is the possibility for borrowings to be adopted by monolinguals who do not engage in CS (e.g., Japanese word sudoku by English monolinguals).

Gardner-Chloros (2010, p.59) underscores that “in the present state of knowledge, it has not been demonstrated that the difference between CS and other language contact phenomena are categorical differences as opposed to differences of degree.” Gardner-Chloros (2009, p.51) further adds that “there is a grey area where borrowing, code-switching, convergence and innovation cannot easily be distinguished.” For Gardner Chloros, therefore, there is no categorical distinction between borrowing and CS. In this regard, her view of CS is reminiscent of Muysken’s (for details, see section 2.3.1), in that she suggests CS co-exists and interacts with other contact phenomena (i.e., ‘strategies’ in Muysken’s terms).

2.2.2 Criticisms of CS as the Combination of Grammars or Lexicons

The prevailing view of CS, as the combination of two discrete grammars or lexicons that maintain their separate identities in bilingual discourse, has been previously criticized as an oversimplification (Gardner-Chloros & Edwards, 2004, p.126; Gardner-Chloros, 2010) and misrepresentation (García 2009, 2010) of the dynamic nature of CS. Gardner-Chloros (1995) describes this belief as the new orthodoxy, which has replaced the notion of the ideal speaker/listener in homogenous communities.

Álvarez-Cáccamo (1998, p.36) adds that there is no concrete evidence to substantiate the widely held assumption that “speakers who code-switch possess two (or more) identifiable linguistic systems or languages”. This notion of two discrete systems is especially questionable in the case of bi/multilingual speakers’ lexical inventories, where lexical gaps are known to

52 comprise a natural aspect of code-switchers’ linguistic repertoires (Clegg, 2010; Balam & Prada

Pérez, 2016). For scholars such as Gardner-Chloros who oppose a monoglossic view of bilingualism, CS is envisioned as a dynamic phenomenon that is inclusive of “a range of interlingual phenomena within which strict alternation between two discrete systems is the exception rather than the rule” (Gardner-Chloros, 1995, p.68). Thus, CS is viewed as potentially involving the instantiation of linguistic creativity and the exploitation of similarities between languages.

From this perspective, bi/multilinguals play an active role in CS. Muysken (2000, p.32) highlights that if the necessity arises, code-switchers break or ‘bend’ rules if structural equivalence is not present. Thus, bi/multilinguals have the creative capacity of taking an active role in how CS forms are used or manipulated in switched discourse. Sebba (2009) concurs with this view, positing that code-switchers employ different strategies (e.g., harmonization, neutralization, compromise) to facilitate bilingual discourse and build switch sites in cases of interlinguistic incongruence. More recently, even scholars who examine CS data through the

Minimalist framework question the view that intra-sentential CS mainly consists of features found in the input languages (as contended by MacSwan (1999, 2005)). For instance, González-

Vilbazo and López (2011, p.846) point out that such a view “does not fully consider the role of

UG in language acquisition” (for details on these authors’ view, see Chapter 4). In their view, child speakers have access to the universal pool of features that may allow the formation of innovative CS structures not attested in any of the component languages.

In CS corpora, there are several innovations that fall within what Gardner-Chloros (2009, p.51) describes as a “grey area” and which reveal bilingual speakers’ creative use of the linguistic resources that are available to them. For example, compromise forms are often

53 employed in CS, as in the French/Alsatian example in (5), where the French verb cuellir is used.

In both French and Alsatian, the verb should be conjugated in the third person to be grammatical, but here it is not. Gardner-Chloros points out that the Alsatian ending – iere (e.g., marschiere ‘to march’) is also the third person plural ending. Thus, cuellir is a compromise form, as the French infinitive ending –ir sounds like an Alsatian conjugated third person plural (for other examples, see Sebba, 2009, p.49).

(5) Ah voila, nitt dass se do cueillir, un gehn dann uf d’ander Sit

“Yes there you are, they shouldn’t pick, and then go to the other side.”

(Gardner-Chloros, 1991, p.159)

In some contexts, the two varieties in contact may share phonemes and many words, as in the case of French/Dutch CS in Brussels (Treffers-Daller, 1994). Particularly in cases where a certain variety is in contact with a typologically related language, the boundaries in code- switched speech become less distinct, as there are many lexical and phonological overlaps. For instance, although Belizean English and BK differ in several aspects of grammar and phonology

(for overview, see Decker, 2013), many words in Belizean English and BK, for instance, could belong to either variety. Such contact situations problematize models of CS that emphasize clearly delineated lines between two languages in bilingual discourse2.

The prevailing conceptualization of CS, as the combination of two discrete grammars of lexicons, becomes particularly problematic in cases where CS structures cannot be attributed to either of the component languages (Gardner-Chloros, 2009; Gardner-Chloros & Edwards, 2004; but see González-Vilbazo & López, 2011, 2012) or where the switched structures evince convergence. Recall that phenomena such as convergence are generally not considered to be an

2 It is important to note that for some scholars, these contact situations where two related varieties are alternated may not necessarily be illustrative of CS but style shifting instead (for relevant discussion, see Bullock & Toribio, 2009).

54 inherent characteristic of CS. As Edwards & Gardner-Chloros (2004, p.1448) underscore, “there is no place for the variation which precedes and underlies the refocusing of norms”. This

‘refocusing of norms’ refers to code-switchers’ bending or breaking of ‘rules’ and/or restrictions that ‘disallow’ certain switches in order to build CS sites.

Noteworthy is that the opposing views of what CS constitutes bring to the fore not only a terminological debate, but they also raise larger theoretical issues (Bullock, Hinrichs, & Toribio,

2014) and questions about views of what bilingualism is. A recurrent theme in previous work that supports a more dynamic view of bilingual language practices is that the conceptualization of CS as the combination of two discrete systems downplays the hybridity, complexity and dynamicity that characterize bi/multilingualism across different parts of the world (García and colleagues; Gardner-Chloros, 1995).

Thus, instead of fostering an appreciation and an understanding of linguistic variation and the translanguaging practices of bi/multilinguals, it paradoxically promulgates the old-fashioned concept that a ‘true’ bilingual equals a native speaker of Language X and a native speaker of

Language Y, as Zentella (1997, p.270) aptly illustrates it, “as if a bilingual were two monolinguals joined at the neck.” In the mainstream view of intra-sentential CS, CS is in fact viewed as a ‘1 + 1 = 2’ phenomenon given that the inherent assumption is not only that CS is the combination of two separate grammars and or lexicons, but that these systems retain their identities in bilingual discourse (for relevant discussion, see García, 2014; Gardner-Chloros &

Edwards, 2004; Otheguy, García, & Reid, 2015). Thus, convergence and/or other interlingual phenomena are not viewed as a natural or key component of intra-sentential CS.

In this dissertation, the borrowing/CS distinction does not constitute a central question in my examination of the two CS phenomena examined. In the case of bilingual light verb

55 constructions (Chapter 4), I focus on restrictions that have been proposed to constrain the incorporation of these structures. In the case of mixed determiner phrases (Chapter 5), I focus on how gender is assigned to donor-language nouns, regardless of their status as borrowings or switches.

In the following section, I provide a brief overview of the highly contested issue on whether CS triggers convergence in bi/multilingual speech.

2.2.3 Code-Switching and Convergence

In discussions on the nature of CS, a highly contested issue is whether CS is a causative mechanism of convergence and/or contact-induced change. Whereas some researchers espouse the view that CS and linguistic convergence go hand in hand (Backus, 2004; Edwards &

Gardner-Chloros, 2004; Gardner-Chloros, 2009; Gumperz & Wilson, 1971; Myers-Scotton,

2002; Savić, 1995; Sebba, 2009; Clyne, 1987), others disagree (Field, 2005; Torres Cacoullos &

Travis, 2010).

Convergence can be attested in the presence or absence of CS (Myers-Scotton, 2002). In composite CS, there is “convergence in regard to the source of some frame-building procedures, as well as in the features of the abstract grammatical structure in some ” (p.8). In the case of monolingual utterances, convergence is attested when all surface morphemes in a clause come from one language, whereas “the abstract lexical structure projecting these morphemes no longer comes from one language, but includes some abstract structure from another language”

(p.164). Thus, in both cases, the utterances will show a composite morphosyntactic frame.

In the literature, it is well established that over time languages influence each other in such indelible ways that convergence can be attested at the phonetic/phonological, morphophonemic, prosodic and syntactic levels (Clyne, 2003, p.104). For researchers such as

Backus (2004), Bullock and Toribio (2004), Davies and Deuchar (2010), Johanson (1999, p.53),

56 and Myers-Scotton (2002), convergence is an outcome but also a process of language contact; thus, evincing both diachronic and synchronic dimensions (for relevant work, see Bullock &

Gerfen, 2004; Fuller & Lehnert, 2000; Simonet 2010). Consequently, it can occur in both monolingual and bilingual speech.

Bullock and Toribio (2004, p.91) highlight the synchronic dimension of convergence, which they define as “the enhancement of inherent structural similarities found between two linguistic systems”. Note that this notion of convergence as the ‘enhancement’ of structural similarities had previously been suggested by other researchers such as Edwards & Gardner-

Chloros (2004), Muysken (2000), Otheguy and Lapidus (2003), Sebba (1998), Silva-Corvalán

(2004) and Thomason (2001). In Bullock and Toribio’s (2004) and Toribio’s (2004) view, convergence is specifically induced by CS.

Thus, in CS, a composite structure may be attested which does not necessarily indicate a

‘turnover’ of the matrix language (in the sense of Myers-Scotton (1998)), but which may reveal a

“reflex of bilingual usage” (Toribio 2004, p.172). It is particularly when bilinguals are in a bi/multilingual mode that they exploit the structural similarities between their languages. From this perspective, convergence is skillfully used as an optimization strategy (in the sense of

Muysken, 2000, 2013) in bilingual discourse to reduce the complexity and cognitive cost of processing two linguistic systems simultaneously (Bullock & Toribio, 2004; Otheguy & Lapidus,

2003; Silva-Corvalán, 1995), and most importantly, to facilitate CS.

Bullock et al. (2014) observe that this process of structural enhancement is a relative construct. Therefore, convergence is instantiated differently in different communities. In Sebba’s

(1998, 2009) terms, congruence will not be created the same way. Bullock, Hinrichs and Toribio point out that some speakers may identify similarities between their languages in a different way

57 than other bilinguals; hence, resulting in different patterns of convergence at different grammatical levels.

In addition, frequency of CS may also have an effect on convergence. When a close-knit community of bi/multilinguals frequently engages in dense CS practices, then they are more likely to capitalize on convergence than other bilinguals who engage in less dense CS practices

(e.g., intersentential CS). Bullock, Hinrichs and Toribio’s view is in line with Backus’ (2004, p.179) suggestion that convergence may be more pronounced among bilinguals who more frequently engage in CS. Thus, frequency of CS is an important factor when considering the prevalence or extent of convergence attested in a group of code-switchers. As it relates to the present dissertation, this hypothesis predicts that the presence of convergence in CS should be pronounced in the Northern Belize context, especially because of the frequent use of bilingual

CS (Brockmann, 1979; Balam, 2013) and the low levels of linguistic prescriptivism (Balam &

Prada Pérez, in press).

In line with Gardner-Chloros and Edwards’ (2004) conceptualization of CS, the view I adopt in the present dissertation is that CS is inclusive but not limited to borrowing and alternation. CS, therefore, includes interlingual phenomena such as borrowing, convergence, and linguistic creativity (Gardner-Chloros, 1995, p.68). Given that Northern Belize Maya/Mestizos frequently engage in bilingual language practices in a context where CS is sanctioned as a norm,

I set out to examine whether the data support Toribio’s (2004, p.172) contention that the simultaneous activation of languages in CS indeed “favors the searching of parallels between them, and hence promulgates the striving towards convergence.” As previously highlighted, particularly because CS constitutes an everyday norm in Belize, language mixing data from this context offers fertile ground to examine the dynamic nature of intra-sentential CS.

58 In the following section, I shift our focus to the formal study of CS. I first recapitulate earlier work on CS by providing an overview of early constraints in variationist and generative approaches to CS, which undoubtedly laid the foundation for more current scholarly work on CS.

Subsequenly, I provide an overview of both typological (Muysken, 2000, 2013) and formalist approaches in the study of intra-sentential CS (Myers-Scotton, 1993; MacSwan, 1999, 2005).

2.3 Approaches in the Grammatical Analysis of Code-Switching

Gardner-Chloros (2009, p.95) notes that there have been three major approaches to the grammatical analysis of CS: variationist (e.g., Timm, 1975; Poplack, 1980), generative (e.g.,

Joshi, 1985; Belazi, Rubin, & Toribio, 1994) and production-based (e.g., Myers-Scotton, 1993).

In particular, variationist approaches thrived during the 1970s and 1980s when the formal study of CS first garnered notable attention after the pioneering works of researchers such as Pfaff

(1979) and Poplack (1980). During this time, researchers’ main focus was on formulating and testing universal constraints that could explain where intra-sentential CS was possible and where it was not (Gumperz, 1976, 1982; Kachru, 1977; Lance, 1975; Lipski, 1977; McClure & Wentz,

1976; Naseh, 1977; Pfaff, 1979; Poplack, 1980; Sankoff & Poplack, 1981; Timm, 1975).

Primarily basing her work on Spanish/English data, Poplack (1980) later proposed the free morpheme constraint and the equivalence constraint, which Poplack argued were universal intra-sentential CS constraints. The former constraint bans word switches such as *eat-iendo

‘eating’, given that the Spanish morpheme –iendo that marks progressive aspect is morphologically bound and not free. The only way that such mixing is permissible is if the lexical form is phonologically integrated into the language of the bound morpheme as a nonce borrowing. The latter constraint predicts that CS is facilitated at junctures where the surface structures of the two languages are similar. Thus, switching cannot occur between two sentence elements if they are not equivalent in the way that they are linearly ordered. Like the

59 constraint, the equivalence constraint bans switches between Spanish and English verbs, given that the order of Spanish and English are different.

Generative approaches to CS, grounded on Government and Binding syntactic theory

(Chomsky, 1981), became prominent in the 1980s and 1990s, and during this time, more constraints were proposed to account for intra-sentential CS. Central to this wave of research was that it sought to determine how abstract, syntactic principles that hold true across natural languages could account for CS as well, without having to resort to CS-specific constraints.

Earlier constraints such as Poplack’s (1980) free morpheme constraint were highly criticized due to the fact that counterexamples surfaced in many subsequent CS studies involving other language pairs. For instance, counterevidence to the free morpheme constraint was found in CS pairs that involve agglutinative (e.g., Bantu, Swahili) and non-agglutinative (e.g., Hindi) languages (Bhatia & Ritchie, 1996, p.640; Gardner-Chloros, 2009, p.96).

Constraints proposed by generative researchers also became vulnerable to this scrutiny.

Such is the case of the government constraint, for instance, proposed by Di Sciullo, Muysken, and Singh (1986). According to this constraint, there could not be any switches between a governing syntactic element such as a verb and a governed element such as an object, as in (6).

These kinds of switches, however, were found to be rather frequent in Spanish/English CS

(Toribio, 2001, p.208) and other language pairs (Pandit, 1990; Myers-Scotton, 1993).

(6) Los niños pidieron pillows and blankets.

The children request.3PL.PRET pillows and blankets

‘The children requested pillows and blankets.’

Toribio (2001, p.208)

60 Another example is the functional head constaint (Belazi et al., 1994), which bans switching between a functional head (e.g., complementizer, determiner) and its lexical complement (e.g., phrase, verb phrase) as a result of a mismatch between language features. Belazi et al. postulate that in light of the fact that functional heads select the features of their complement, the language of the functional head must correspond with the language of its complement. Alternatively, the derivation crashes, resulting in an ungrammatical utterance.

The functional head constraint (FHC) correctly bans switches such as (7), where a switched Spanish verb after the auxiliary verb ‘have’ is ungrammatical. At the same time, however, it wrongfully precludes switches with the auxiliary verb estar ‘to be’, as in (8), which are well-attested in Spanish/English CS (Pfaff, 1979; Reyes, 1982; Zentella, 1997). Futhermore,

Mahootian and Santorini (1996, p.465-466) provide many counterexamples from different language pairs, which reveal the empirical inadequacy of the FHC.

(7) *The students had visto la película italiana

The students had see.PASTPART the movie italian

‘The students had seen the Italian movie.’

MacSwan (2014, p.7)

(8) Estaba training para pelear.

Be.3SG training.PROG to fight

‘He was training to fight.’

Pfaff (1979, p.296)

In their work, Mahootian and Santorini highlight that there is no necessity to resort to any constraints that are specific to intra-sentential CS. Instead, they assert that general principles of phrase structure within Universal Grammar should be able to account for CS data. In their

61 analysis of noun-modification structures, for instance, they rely on the distinction between complements and adjuncts to explain why there are unusual cases when the position of the violates the Spanish requirement for it to appear post-nominally rather than pre- nominally (e.g., I got a lotta blanquito friends ‘I got a lot of whitey friends’, from Poplack, 1980, p.600). In their view, it is only in head-complement configurations that heads determine the syntactic properties of their complements. Since noun-modification have the looser relations of adjunctions and not subjected to tight-knit sisterhood relations, the adjunct (i.e., adjective) is able to appear in a word order that reflects the structure of either Spanish or English.

Although earlier research on CS constraints received ample criticism due to the extensive counterevidence in different language pairs (for counterexamples to CS constraints, see Belazi et al., 1994; Berk-Seligson, 1986; Chan, 1998; Di Sciullo et al., 1986; Edwards & Gardner-Chloros,

2004; Jacobson, 1998, p.56-64; Mahootian & Santorini, 1996; Moyer, 1992; Myers-Scotton,

1993; Muysken, 2000; Pandit, 1990; Romaine, 1989; Toribio, 2001), it must be pointed out that that these seminal works were instrumental in demonstrating that intra-sentential CS is not “a grammarless mixture of two languages” (Grosjean, 1982, p.146), as some had believed (e.g.,

Labov, 1972, p.457; Lance, 1975, p.143), but rather a rule-governed phenomenon. The general concensus among scholars today is that CS among bi/multilinguals is systematic (Ritchie &

Bhatia, 2004, p.339; Toribio, 2004). The notion of ‘universal’ CS constraints, however, is still contested.

More recent scholarly works have continued to contribute to our understanding of the typological and grammatical nature of CS. In relation to the former, Muysken’s typological approach to CS, which is more descriptive rather than predictive in nature, has contributed to our understanding of CS as optimization strategies. Regarding the theoretical debate on how to best

62 account for the grammaticality of intra-sentential CS, there has been emphasis on two main approaches, namely, the Matrix Language Frame (MLF) model (as proposed by Myers-Scotton and colleagues) and the Minimalist approach to CS (as proposed by MacSwan and colleagues).

While the latter approach is a recent development of earlier attempts to account for CS data within a generative framework (e.g., Mahootian & Santorini, 1996), the former is a production- based, abstract model that was specifically proposed to account for intra-sentential CS data.

In the ensuing sections, I discuss these approaches in more detail. First, I provide an overiew of Muysken’s (2000, 2013) bilingual speech typology, which provides a grammatical typology of CS. Subsequently, I outline the main tenets of the Matrix Language Framework model (Myers-Scotton, 1997; Myers-Scotton & Jake, 1995; Jake, Myers-Scotton & Gross, 2002;

Myers-Scotton & Jake, 2013) and the Minimalist approach to CS (as proposed by MacSwan,

1999, 2000, 2005, 2014).

2.3.1 Bilingual Speech Typology

Muysken’s (2000) original typology of bilingual speech included three main distinctions: insertion, alternation, and congruent lexicalization (see Figure 2-1). In insertional switching or

‘code-mixing’, the first language serves as the base language and foreign elements or chunks from another language are inserted into it. In the Quechua/Spanish example in (9) from Urioste

(1966, p.7), the expression las cuatro is inserted into a Quechua clause.

(9) Q’aya suya-wa-nki las cuatro-ta Qo-yku-sqa-sun-ña bukis

tomorrow wait.1OB.2SG at four.ACC give.ASP.ASP.1PL.con box

“Tomorrow you wait for me at four. We’ll have a go at boxing.”

(cited in Muysken, 2013, p.712)

In alternational switching or ‘code-switching’, stretches of discourse from both languages alternate, and this may involve phrases or sentences. Importantly, universal combinatory

63 principles are used to combine items or phrases from the two independent grammars. In the

Spanish/English example in (7) from Gumperz & Hernández Chavez (1971, p.118), the Spanish expression ándale pues is juxtaposed to the English expression do come again.

(10) Ándale pues, and do come again.

‘That’s all right then, and do come again.’

(cited in Muysken, 2013, p.713)

On the other hand, congruent lexicalization occurs in cases where elements from two different languages are used in a grammatical structure that is partly or wholly shared by the base language and the donor language. Thus, in this type of CS, linear and categorical equivalence facilitate switching. In the Spanish/English CS example in (8), fluid switching between English and Spanish is facilitated by the structural similarities between the two languages. According to

Muysken (2000, 2013), typological proximity facilitates congruent lexicalization3.

(11) Bueno, in other words, el flight que sale de around three o’clock.

‘Good, in other words, the flight that leaves Chicago around three o’clock.’

(cited in Muysken, 1997, p.362)

Muysken’s most recent (2013) typology includes back-flagging, which occurs when heritage language discourse markers are inserted in second language (L2) discourse. These markers, as the Spanish/English CS example from Specker (2008, p.111) in (12) illustrates, are typically single items, clause-peripheral and have an ethnic connotation (i.e., used to signal ethnic identity).

(12) Gracias. You won’t regret this.

‘Thank you. You won’t regret this.’

3 For relevant discussion on criticisms regarding the nature and existence of congruent lexicalization, see Muysken, 2014.

64 (cited in Muysken, 2013, p.713)

It is important to note that insertion, alternation, congruent lexicalization, and back- flagging are not mutually exclusive; thus, a bilingual community can evince all four types of switching. In this regard, Muysken’s classification takes into consideration the variation that is sometimes present even within a single extended family of different generations; hence, capturing the dynamic nature of CS patterns attested not only at the level of the community, but at the level of individual speakers as well.

In Muysken’s (2000, 2013) view, there are several factors (i.e., linguistic, cognitve and social) that determine what type of CS is most prevalent in a bi/multilingual community. Thus, whereas speakers’ bilingual proficiency is important, so are other social factors such as power relations and the prestige of language varieties. The triangle in Figure 2-1 illustrates the co- existence and interaction of the different instantiations of switching in a bilingual community, as originally proposed by Muysken (2000). Note that the different CS types appear in the same triangular space, indicating that CS is a complex yet unified phenomenon that consists of different types of CS. In Muysken’s view, one type of CS does not exist to the exclusion of the other(s).

insertion

alternation congruent lexicalization

Figure 2-1. Muysken’s (2000, p.9) representation of the different types of switching

65 In his latest framework of language contact outcomes, Muysken (2013) provides a more detailed framework that accounts for different processes and outcomes of language contact (i.e., code-switching, creoles, , mixed languages and ethnolects). In this framework, Muysken proposes a four-dimensional taxonomy of CS, formulated in terms of speaker optimization strategies that are operant in different types of bilingual contexts.

Thus, the four CS types (i.e., insertation, alternation, congruent lexicalization and back- flagging) are envisioned as possible strategies that are at the disposal of bilingual speakers, but the way these strategies are used largely depend on both social and linguistic factors, as

“speakers are influenced by different constraints on language behavior, given the different circumstances in which they find themselves and the languages involved.” Crucially as well, the role of universal principles (UP)4 is also taken into consideration. Thus, different combinations of factors favor particular strategies, but not to the exclusion of other strategies.

As the schema in Figure 2-2 shows, there are four strategies that are operant in CS. The

L1- and the L2-focussed strategies are in opposite corners. In the former contact scenario, social factors such as unequal power dynamics in the community and low proficiency in the L2 will contribute to the more prevalent use of insertions. At the opposite end of the quadrangle, we see the inverse. In this case, heritage speakers typically become dominant in their L2; hence, resulting in the prevalent use of back-flagging.

4 Muysken does not adopt Chomskyan UG-principles (i.e., universal properties of the language faculty). He envisions these language-independent UPs as more general combinatory principles, whose key elements include discourse sequencying, iconicity, and paratactic adjunction (for further details, see Muysken, 2013, p.716). These UPs hold for all languages and language combinations.

66

Figure 2-2. Bilingual speech strategies for code-switching

Importantly, CS patterns or structures could appear in different positions within the schema, reflecting the combination of different strategies. The more a CS variety reflects a mixture of the four strategies, the closer its position will be near the center of the quadrangle.

Alternatively, other factors may favor the extensive use of only one type of CS in a certain bilingual community. Muysken makes general predictions about certain societal conditions and the resulting CS outcomes.

For instance, low levels of L2 proficiency and unequal power relations between two groups of people result in the use of insertional CS (see Table 2-1). In contrast, low levels of normativity, closely-knit networks and high bilingual proficiency contribute to the development of congruent lexicalization. In particular, Muysken’s inclusion of degree of normativity is crucial, as degree of normativity can be a powerful deterministic factor of how CS evolves in a bilingual community, as degree of normativity I argue can be a gatekeeper of CS-driven innovation. Thus, linguistic innovation and convergence is least likely to be attested in language contact situations where monolingualism and a purist tradition are privileged over bi/multilingualism.

67 Table 2-1. Factors, strategies, and outcomes in CS Factors Strategies Outcomes

Unequal power, (post)colonial L1 Insertion

settings/immigrant bilingualism, low L2

proficiency, considerable typological

distance

Relaxed language norms, closely-knit L1/L2 Congruent Lexicalization

networks, long contact, high bilingual proficiency, little typological and/or lexical

distance

Political competition between languages, UP Alternation

high bilingual proficiency, considerable

typological distance

Shift in second or third generation, L2 L2 Back-flagging

prestige, high L2 proficiency

*Slightly modified table showing combinations of linguistic, social and cognitive factors (for original, see Muysken,

2013, p.720).

In summation, Muysken views CS as strategies that remain at the disposal of bilinguals, which are differentially incorporated depending on the influence of concomitant social, linguistic and cognitive factors. Importantly, Muysken’s framework does not preclude convergent (i.e., congruent lexicalization) and/or innovative structures (i.e., attributed to Universal Processes) from emerging and/or co-occurring with insertion and/or alternation. This element of natural interaction between the different strategies is a crucial element of Muysken’s (2000, 2013) conceptualization of CS, one which captures not only variation patterns in CS, but the creative

68 language manipulation and linguistic indexicalization that bilinguals often engage in to ‘project’ different aspects of their social identities (Bhatia & Ritchie, 2008; Bucholtz & Hall, 2005;

Bullock et al., 2014; Edwards, 1983; Le Page & Tabouret-Keller, 1985; Myers-Scotton, 2002;

Pérez Casas, 2008; Tabouret-Keller, 1980; Zentella, 1997). In the following section, I provide a brief overview of two formal models that have been influential in research on the grammatical aspects of CS (Gardner-Chloros, 2009).

2.3.2 Myers-Scotton’s The Matrix Language Frame (MLF) Model

The MLF is an abstract theoretical model that was proposed to specifically account for intra-sentential CS. Its fundamental assumption is that in classic CS, there is an identifiable

Matrix Language (ML) and an Embedded Language (EL). In classic CS, the critical grammatical elements in bilingual constituents, such as system morphemes (e.g., grammatical gender and number markers), come from the ML (Myers-Scotton & Jake, 2013). In contrast, the EL is the language whose sole function is to contribute content morphemes (e.g., nouns, verbs, and some prepositions) and/or EL islands (e.g., EL noun phrases that follow EL grammar in otherwise ML discourse) to the ML in switched discourse.

In the MLF, the dominant influence of the ML is crucial to a bilingual CP (i.e., Projection of Complementizer: the minimal unit of analysis consisting of mixed constituents) in classic CS.

This contrasts cases of ‘composite’ CS (Myers-Scotton, 1997, 1998), where the morphosyntactic frame or bilingual constituent exhibits elements from both languages, and where in essence, a certain level of linguistic convergence has occurred. It is important to highlight that although

Myers-Scotton views convergence as a potential contact outcome, classic CS and composite CS are conceptualized as distinct phenomena (Myers-Scotton & Jake, 2009, p.339).

In classic CS (Myers-Scotton, 1993; Myers-Scotton & Jake, 2009, 2013), the ML is defined in terms of two principles: the Morpheme Order Principle (MOP) and the System

69 Morpheme Principle (SMP). In simple terms, the former requires that morpheme order in a bilingual constituent adheres to the ML morpheme order, whereas the latter stipulates that all relevant system morphemes must be derived from the ML.

Another important element of the MLF is the existence of ‘EL islands,’ which refer to constituents or maximal projections (e.g., DP, PP) containing EL morphemes (internally governed by EL grammar) rather than ML system morphemes, but which nonetheless abide by the MOP. In the example below from Haitian Creole/English CS (Hebblethwaite, 2007, p.298), the prepositional phrase EL island nonetheless abides by the MOP.

(13) L ap gaspiye kòb for nothing

'She is wasting money for nothing.'

(cited in Jake & Myers-Scotton 2009, p.221)

Also crucial to the MLF are the distinction between content versus system morphemes and the notion of lemmas. System and content morphemes are stored in the mental lexicon. What distinguishes these morphemes is their status as receivers or assigners of a thematic role. Items such as verbs and prepositions which assign thematic roles are content morphemes, whereas system morphemes such as markers of gender, number and case do not assign or receive thematic roles. Morphemes that express quantification over individuals or events, such as adjectives and tense/aspect markers, are also considered system morphemes. Lemmas are the abstract addresses of lexical items in the lexicon. These lemmas, containing an item’s non-phonological subcategorial information, are used to build switched utterances. Lemmas also have abstract features that activate the selection of the ML (Myers-Scotton & Jake, 2000, p.286).

The 4-M model (Myers-Scotton & Jake, 2000) is an elaboration of the content/system morpheme distinction that was initially proposed in Myers-Scotton (1993). In this expanded

70 version of the MLF, there are four morpheme types (see Figure 2-3); namely, content, early system morphemes, and two types of late system morphemes (i.e., bridges and outsiders). What distinguishes early system morphemes is that they do not assign or receive theta roles, and they are accessed early on in the production process (e.g., plural ‘s’, determiners, some prepositions).

They also depend on content morphemes for their form, given that content morphemes are heads of the maximal projection of the early system morphemes. Thus, when a phrasal verb like ‘fly away’ is accessed, the preposition ‘away’ is accessed along with the verb ‘fly’, hence making it an early system morpheme. Both content and early system morphemes are ‘conceptually activated’. This means that speakers’ intentions activate language-specific semantic/pragmatic feature bundles that become lemmas in the mental lexicon. Content morphemes such as nouns and verbs are directly selected to convey meaning, and early system morphemes such as plurality markers are indirectly selected early on in the production process by the speaker.

Bridge late morphemes are similar to early morphemes in that they rely on information within the maximal projections in which they occur for their form. The possessive ‘of’ in English would be an example of a bridge morpheme that connects two content morphemes. In contrast, outsider late morphemes refer to grammatical information markers (e.g., case and gender) that are added the last in the formulation of a mixed utterance (see Figure 2-3). Importantly, all outsider late morphemes must come from the ML, in line with the SMP.

71

Figure 2-3. Morphemes types in the 4-M model (Myers-Scotton & Jake, 2000, p.1064)

The Abstract Level Model is an important addition that was also made to the MLF. The

Abstract Level Model was proposed to determine whether there was sufficient congruence for constructions to be possible in different language pairs. It also helps to explain “the nature of the abstract morphosyntactic frame that structures bilingual clauses” (Myers-Scotton, 2002, p.19).

According to this model, all lemmas in the mental lexicon consist of three levels: the lexical- conceptual structure, the predicate-argument structure and the morphological realization pattern.

At the first level, the conveyance of the speaker’s intended message is what matters the most.

Here intentions activate semantic and pragmatic features that are bundled together, and a specific lemma that best conveys the speaker’s message is chosen. At the second level, thematic structure is mapped onto grammatical relations (e.g., , beneficiary, etc.). At the last level,

72 grammatical relations are realized on the surface (e.g., number agreement morphology). Myers-

Scotton and Jake (2013) posit that that the reason why EL nouns are primarily switched is because they only need to be checked at this first level. EL verbs, on the other hand, are checked for congruence at the three levels (Myers-Scotton & Jake, 1995) to determine if they are congruent with ML equivalents (for discussion on recent analysis, see Chapter 4).

Thus, an EL content morpheme cannot be used if it is incongruent with its equivalent,

(i.e., the ML equivalent is a system morpheme, the morphemes assign thematic roles differently, or morphemes differ in pragmatic function). According the Blocking Hypothesis, there is a filter that blocks any EL content morpheme which is not congruent with the ML (Myers-Scotton,

1993). If there is only sufficient congruence, a compromise strategy may be employed (e.g., bare nouns, ‘do-constructions’, EL islands). Overall, the 4-M model predicts that whereas the lexical- conceptual structure is easily transferred inter-linguistically, the predicate-argument structure and the morphological realization pattern should be influenced the least by language contact.

Therefore, outsider system morphemes such as tense and agreement should not be prime candidates for cross-linguistic transference.

Of particular relevance to this dissertation is the Bilingual NP Hypothesis (see Chapter

4), proposed by Jake, Myers-Scotton and Gross (2002, p.78-79), which makes the following predictions:

The system morphemes in mixed NPs come from only one language, called the ML. An asymmetry between mixed NPs and full NPs from the EL obtains: full EL NPs are dispreferred because their system morphemes (and their uninterpretable features) do not match other system morphemes and their uninterpretable features elsewhere in the bilingual CP.

The Bilingual NP Hypothesis, which was specifically proposed to account for mixed DPs such as unmasc house ‘a house’, predicts that when a speaker’s intention motivates the incorporation of a noun from the EL, the ML frames it. Importantly, when the speaker selects an

73 EL noun, the relevant feature matching takes place between the EL form and the feature of an

ML (real or hypothetical) counterpart, and not between the EL noun and the ML system morpheme. In relation to Spanish/English CS, this hypothesis predicts that when English nouns are incorporated, they should occur with . This preference occurs because system morphemes from the ML “maximize uniformity in the bilingual CP…they are the glue that holds together the bilingual CP” (p.79). In contrast, the internal structure of EL islands is under the control of the EL.

The ML Feature Hypothesis further requires that the gender feature be realized as in

Spanish, but this is not necessarily dependent on satisfying the formal gender feature of the translation equivalent which bears gender features. Jake et al. posit that multiple factors (e.g., gender of the translation equivalent, masculine gender default strategy) affect how an English- noun is incorporated in mixed DPs. Jake et al. suggest, however, that in classic CS, when English nouns occur with Spanish determiners, the determiner should match the gender of the translation equivalent (for further details, see Chapter 5). Such prediction, of course, where the translation equivalent plays an important role in bilingual constituents is in line with the idea that in classic

CS, “only one [language] is the source of the ML frame; in other types of bilingual speech (e.g., attrition, convergence, or mixed languages), the principle may be violated, with grammatical features appearing from more than one language” (p.72). It must be emphasized that the MLF also predicts that are possible in bilingual clauses where the ML is English (e.g.,

He bought the casa ‘He bought the house’). In Myers-Scotton's (1993) view, either language can serve as the ML in switched discourse, but whether this occurs or not depends on other factors such as the prestige of language varieties.

74 The MLF, therefore, makes specific predictions regarding intra-sentential CS. In line with the USP (Myers-Scotton, 2002) and the Bilingual NP Hypothesis (Jake et al., 2002), which emphasizes the preference given to ML structures over EL structures in bilingual speech, it is predicted that outsider late system morphemes or grammatical information markers must come from the ML. This restriction applies to both classic and composite CS (Jake & Myers-Scotton,

2009). On the other hand, conceptually-activated content morphemes such as nouns and verbs are most likely to be supplied by the EL in mixed constituents.

Although the MLF has been an influential model, some of its constructs have been criticized. For instance, the notion of the ML has been a controversial issue (Cantone &

MacSwan, 2009; Gardner-Chloros & Edwards, 2004; Muysken, 2000; Muysken & de Rooij,

1995). In earlier work, the ML was the language that provided the greater number of morphemes to the mixed discourse, a criteria that has been criticized as vague (Backus & Boeschoten, 1996;

Bentahila, 1995). The ML was subsequently proposed as being the language which provides the majority of system morphemes, a problematic notion given that the classification of content and system morphemes may differ across languages (Muysken, 2000; Romaine, 1995).

More recently, MacSwan (2005, p.20) contends that “the MLF model should be rejected on grounds of scientific parsimony” given that CS can be explained using existing empirical and theoretical mechanisms without having to resort to the distinction between the ML and the EL and/or the constructs of the MLF. The details of the theoretical debate between MacSwan (2005) and Jake et al. (2002) are not of central importance to the present investigation, as my main concern relates to variationist and generative restrictions that have been proposed in relation to bilingual light verb constructions and mixed determiner phrases in Spanish/English CS.

75 Therefore, although it is presented here because it is fundamental to our understanding of CS and different accounts of intra-sentential CS, it will not be addressed again.

2.3.3 Minimalist Approach to Code-Switching

The Minimalist approach to CS, as proposed by MacSwan (1999, 2005, 2014), is based on the Minimalist Program (Chomsky, 1991, 1995, 2000), whose ultimate aim is to minimize the necessary apparatus used to model language. Minimalism is a theory of linguistic competence, concerned specifically with the abstract nature of language and attempts to explain both universal and idiosyncratic aspects of it. Under the Minimalist Program, syntax comprises two central components: namely, a computational system for human language (CHL), which is invariant across languages, and a lexicon to which differences across languages is attributed.

Thus, linguistic variation is attributed to feature checking of morphological properties (abstract and concrete) of the lexicon (MacSwan, 2005). The lexicon is also crucial for the construction of phrase structure, which is achieved through several operations.

In the derivation of an utterance, the operation SELECT picks lexical items from the lexicon and places them in a LEXICAL ARRAY, which is a subset of the lexicon employed in the construction of the derivation. The operation MERGE takes items from the LEXICAL ARRAY and arranges them in a hierarchical manner. Elements within the phrase structure tree are then rearranged through the operation MOVE, which is triggered by feature checking.

Importantly, movement may entail either HEAD MOVEMENT, where a head undergoes movement and adjoins to another head. Alterntively, XP MOVEMENT occurs, where a maximal projection moves to the Specifier position of a head. In either case, movement occurs in order for lexically encoded features to be checked. The operation AGREE establishes a relation (such as number or gender agreement) between a lexical item and a formal feature. Uninterpretable

76 features (e.g., case, gender) must be checked and deleted. If features are not checked, the derivation crashes (i.e., ill-formed constructions).

As it relates to intra-sentential CS, MacSwan (2005, p.5) describes the Minimalist approach to CS as constraint-free in that “Nothing constrains codeswitching apart from the requirements of the mixed grammars.” Grounded in the Chomskyan (1995, 2000, and 2001) tradition, the Minimalist approach to CS asserts that no CS-specific constructs, principles or mechanisms are necessary to account for CS phenomena (contra the MLF, which is a CS- specific model). Thus, CS data are explained through the principles and requirements of the specific grammars in question and embedded principles of Universal Grammar, a built-in biological endowment all human beings have (Chomsky, 1965).

In this approach, CS is explained through the interaction of an invariant computational system and a variable lexicon, which accounts for the idiosyncratic differences across different languages. Also integral are operations and the checking of uninterpretable features. As Figure 2-

4 illustrates, in the derivation of a switched utterance, the operation SELECT chooses lexical items from each of the lexicons and puts them into the LEXICAL ARRAY or a subset of both lexicons that contribute to the bilingual construction. In MacSwan’s (2005) view, each lexicon contains a collection of morphological and phonological rules that applies to a specific set of lexical items (see Figure 2-4). Importantly, grammatical requirements are carried along with lexical items from the two systems as they enter a derivation (MacSwan, 2005, p.5). The operation MERGE ensures that items from the array are formed into a hierarchically arranged sequence. Operations of movement, driven by feature checking, then take place so that elements in the syntactic tree are rearranged.

77

Figure 2-4. A minimalist approach to code-switching (MacSwan, 2005, p.7)

As is the case in monolingual utterances, a head or a maximal projection may undergo movement to value features such as number and gender. For example, uninterpretable phi- features such as gender in determiners must be checked and deleted. If a feature is not valued, the derivation subsequently crashes, and if there is a mismatch, it is canceled. When a derivation successfully converges, the phonological component which is a subsystem of the computational system ensures that the utterance proceeds to SPELL-OUT so that it is phonetically realized.

Moro’s (2001) analysis of mixed DPs (cited in MacSwan, 2005, p.18) illustrates how the feature checking or valuation process takes place and how this process bans switches such as

*the casa ‘the house’, which some researchers have reported Spanish/English bilinguals do not

78 produce or judge as ungrammatical (Lipski, 1978). In the derivation of a switched DP such as un(a) house ‘a house’, phi-features of the determiner, but not the noun enter the derivation with unspecified values. The set of phi-features PERSON, NUMBER and GENDER is a probe that seeks to match features between the determiner and the noun in order to establish agreement. In the computational system, the operation AGREE ensures that checking and deletion of features takes place. The determiner, however, can only value and delete its own features if it has a full set of phi-features that matches with the corresponding phi-features in the noun. The deletion must take place as “a ‘one fell swoop’ operation, dealing with the [phi]-set as a unit” (Chomsky,

2000, p.124).

According to Moro, mixed constituents such as *the casa ‘the house’ crash and result as ill-formed switches given that there is no corresponding feature for grammatical gender in the

English determiner. Thus, the Spanish noun’s phi-features PERSON, NUMBER and GENDER attempt to value and delete the English determiner’s features PERSON, NUMBER and GENDER in one swoop, but the GENDER feature fails to do so (cited in MacSwan, 2005, p.18). Given that the determiner’s uninterpretable features are not deleted, the derivation fails to converge and crashes. Notably, feature valuation and subsequent deletion does take place in the monolingual derivations (a & b). It is also possible in the switched case, when the noun phi-features comprise a subset of the corresponding features in the determiner (c). This is not the case, however, in *the casa ‘the house’ in (d).

(a.) for D, phi-features = {person, number, gender};

for N, phi-features = {person, number, gender}

(Spanish D, Spanish N)

e.g., unafem casafem ‘a house’

79 (b.) for D, phi-features = {person, number};

for N, phi-features = {person, number}

(English D, English N)

e.g., a house

(c.) for D, phi-features = {person, number, gender};

for N, phi-features = {person, number}

(Spanish D, English N)

e.g., unmasc house ‘a house’

(d.) for D, phi-features = {person, number};

for N, phi-features = {person, number, gender}

(English D, Spanish N)

e.g., *the casa ‘the house’

It must be highlighted that in Minimalist analyses of CS data (González-Vilbazo &

López, 2011; MacSwan, 1999, 2005), certain switches are categorically excluded. As

Hebblethwaite (2007, p.368) points out, MacSwan's (1999) approach to CS necessitates unidirectional CS, as a clash in phi-features or lack of feature valuation implies no CS at all. For instance, in González-Vilbazo and López’s (2011) view, monolingual and bilingual data suggest that light verbs cannot co-occur with the passive voice due to “a restriction imposed on the computational system rather than a language-specific – or code-switching specific – type of restriction” (p.843). Given that the restriction is computational in nature, the implication is not that it should be ‘disfavored’ but that it should be categorically disallowed (for details on this restriction, see Chapter 4). Furthermore, in Spanish/English CS, a Minimalist approach to CS

80 predicts that switched DPs such as *the casa (MacSwan, 2005; Moro, 2001) should not be attested.

2.4 The Debate on Social versus Linguistic Factors

The last issue that warrants elaboration is the ongoing debate regarding the role of social versus linguistic factors in CS, which has become especially relevant to theoretical discussions on the most adequate way to account for CS outcomes. A fundamental issue that has been debated since the 1980s is whether the primary determinant of contact outcomes are linguistic or social factors (see Thomason & Kaufman, 1988) or a combination of social, linguistic and cognitive factors (Muysken, 2000, 2013). Thomason and Kaufman’s (1988) well-known contention is that it is “the sociolinguistic history of the speakers, and not the structure of their language, that is the primary determinant of the linguistic outcome of language contact (p.35).”

Thomason and Kaufmann further contend that the direction and degree of interference are determined not by structural aspects of languages but by social factors such as intensity and length of contact and prestige of the varieties (for similar views, see García Tesoro, 2010;

Gardner-Chloros, 1995, 2009; Gardner-Chloros & Edwards, 2004; Hebblethwaite, 2007, 2010;

Sebba, 1998, 2009; Verschik, 2008; Winford, 2013).

This perspective has been highly criticized (e.g., Field, 2002, 2005; King, 2002; Sánchez,

2005; Sankoff, 2008), especially because Thomason and Kaufman’s assertion has been understood as suggesting that under the right social conditions, linguistically “anything goes”. In particular, Sankoff (2008, p. 641) highlights that “[t]he cumulative weight of sociolinguistic research on language contact suggests that although it may be true that ‘anything can happen’ given enough social pressure, [Thomason & Kaufman] are very far from the truth in their blanket rejection of internal constraints”.

81 It is worth noting, though, that many researchers seem to have misunderstood Thomason and Kaufman’s view, as their contention has never been that linguistic factors or constraints are not relevant or important (for work on Thomason’s examination of linguistic factors and contact- induced language change, see Thomason (2001)). Thomason (2008) notes that in their original work (Thomason & Kaufman, 1988, p.4, emphasis in original), they contend that “[they] do not deny the importance of purely linguistic factors such as pattern pressure and markedness considerations for a theory of language change, but the evidence from language contact shows that they are easily overridden when social factors push in another direction”. Thus, Thomason and Kaufman’s primary contention has been that (i) there are no absolute linguistic constraints on contact-induced change and that (ii) in cases where social and linguistic factors are competing forces, social factors override linguistic ones.

In their work, Thomason and Kaufman (1988) provide counterexamples to linguistic constraints and proposals on typology5, interference and borrowing (e.g., phonological rules cannot be borrowed/transferred from one language to another), revealing that “all efforts to come up with successful linguistic constraints have failed” (Thomason, 2008, p.44). Although the debate sparked by Thomason and Kaufman was initially about general linguistic outcomes and constraints, the notion of ‘universal’ CS constraints is now contested as well.

Gardner-Chloros and Edwards (2004, p.104) point out that the endeavor to find universal

CS constraints has failed (also see Rodríguez-González & Parafita-Couto, 2012, p.473). As previously pointed out, the main problem with previously proposed constraints is that while they are applicable to one language pair, they are violated by switches in other language pairs; therefore, having a relative rather than a universal value (Gardner-Chloros, 2009, p.96; Muysken,

5 Given my interest on CS constraints in a particular contact situation, I do not focus here on attempting to define these linguistic constraints. The reader is advised to refer to Thomason and Kaufman’s work for a more thorough examination of these constraints and counterexamples.

82 2000; Romaine, 1995). The view held by scholars such as Thomason and Kaufmann (1988) and

Gardner-Chloros and Edwards (2004) is that constraints exist, but the universal and/or deterministic aspect of it is what they do not subscribe to.

The debate between linguistic versus social factors is also intricately connected to the study and prediction of CS outcomes, as typological and grammatical approaches vary in the degree to which social factors are considered. For instance, in the work of certain generative researchers, social factors are simply not considered, given that they do not constitute a central focus of their research. Hebblethwaite (2007, p.262) notes that “MacSwan (1999) attempts to build a theory of code-switching divorced from the influence of social structure.” Although

MacSwan (1999) is not concerned with sociolinguistic aspects of CS, he does state that “while social motivations for codeswitching can be catalogued and described in interesting ways, theoretical constructs of general or universal use may be quite beyond our reach [...]” (p. 39).

Hebblethwaite (2007) and others question the validity of such a stance, especially since the organization and nature of CS is not derived merely from formal aspects of grammar (e.g., phi features, syntactic operations, etc.) alone. CS is after all a sociocultural phenomenon that is profoundly connected to and reflective of speakers’ experiencies, histories and identities. Thus, factors such as the degree of linguistic prescriptivism and the prestige of language varieties are extremely important in determining how CS is grammatically manifested in a community. In

Gardner-Chloros and Edwards’ (2004, p.126) view, “although syntax plays an important role in

CS, it cannot be assumed a priori that the constructs of syntacticians are the best means for characterising the processes of performance data such as CS.”

Scholars that argue for the deterministic effect of structural/linguistic factors often contend that patterns in bi/multilingual speech which often seem to be contact-induced are in

83 reality pre-existing or ‘language internal’. Thus, contact per se does not result in any change, but in the acceleration of changes or extension of phenomena that were already present in the pre- contact language. For example, while many authors analyze the overuse of Spanish subject pronouns as an effect from contact with English in the U.S. context (e.g., Montrul,

2004; Otheguy & Zentella, 2007; Toribio, 2004, among many others), Torres Cacoullos & Travis

(2010) argue that CS does not radically affect subject expression in their New Mexican Spanish data.

Instead, they argue that in both monolingual and bilingual discourse, syntactic priming plays an important role. In the case of first person singular (yo ‘I’) subject expression, the higher the proportion of ‘I’ or ‘yo’ in the environment of a preceding expressed subject, then the higher rate of expressed subjects is attested. Thus, a speaker’s utterance is influenced by previous discourse and not by CS per se. Torres Cacoullos and Travis affirm that “priming may play a role in ostensible language change in contact situations by modestly raising the rate of a parallel structure without involving change in the structure of linguistic variation” (p. 260).

Another example is the application of a default gender, which is often attested in Spanish contact situations (for details, see Chapter 5), where other-language nouns are overwhelmingly assigned the masculine gender. In her analysis of gender assignment in New Mexican Spanish,

Aaron (2014) argues that the overwhelming use of the masculine gender has nothing to do with

CS per se or the existence of a default gender. Instead, she argues that gender assignment patterns may simply follow patterns and preferences that are internal to Spanish.

Noteworthy is that some researchers who follow a Minimalist approach to CS do recognize the important role of social factors, but their work still suggests that linguistic factors ultimately determine the grammaticality of CS. For example, although González-Vilbazo and

84 López (2011, p.846) propose a Minimalist analysis of bilingual light verb constructions in

German/Spanish CS (for further details, see Chapter 5), they differ with MacSwan in important respects. In particular, they assert that linguistic competence is the product not only of comprehensible input but the social environment as well. They argue that this is especially evident in cases of Creole genesis and the formation of sign languages, where we see the emergence of features that are absent from the input languages child speakers are exposed to.

Thus, speakers are not necessarily restricted to the available features in the mixed grammars and, accordingly, bi/multilingual children are able to resort to a more universal pool of features, which they are able to use for the creation of novel structures. At the same time, however, González-Vilbazo and López (2011) argue that light verbs are incompatible with passives due to a restriction from Universal Grammar (for details, see Chapter 4). In this regard, we see that González-Vilbazo and López ultimately give precedence to syntactic restrictions imposed by Universal Grammar. They do not suggest in their work that social factors can override restrictions imposed by Universal Grammar.

In terms of the dominant influence that social factors have on contact outcomes, scholars cite cases of contact varieties such as Creoles and/or mixed languages, where social aspects of the contact situation clearly have a deterministic effect in the resulting linguistic outcomes.

Winford (2013, p.365), for instance, discusses how demographic differences in colonial slave societies led to different contact varieties. In societies where the population grew through natural increase (i.e., locally born slaves and free slaves) and people of mixed descent had greater contact with settlers and their languages, second language varieties that were closer to the language of the Europeans emerged (e.g., Barbados). On the contrary, in contexts like

85 where the population grew through continuous, large scale importation of slaves, Creoles that are more typologically distant from their superstrate (i.e., European) languages emerged6.

Thomason (1995, 2008) adds that the most striking evidence of the influential effects of social factors is provided in the rather abrupt emergence of mixed languages such as Michif

(spoken in Northern Dakota), Ma’a (spoken in Tanzania) and (spoken in

Ecuador). She notes that these contact varieties did not evolve out of necessity, but as a deliberate attempt to create a language used to symbolize an in-group identity (also see Winford,

2013). Bilingual mixed languages, thus, reveal that their emergence is driven by “social reasons” (Thomason, 2008, p.53).

Other scholars highlight the deterministic role that sociohistorical and political changes in a certain context can have on language contact outcomes. Verschik (2008), for example, who examines Estonian/Russian contact in Estonia, observes that prior to the restoration of Estonia’s independence in 1991, Russian-speaking newcomers remained monolingual, and Estonian had very little cross-linguistic effect on Russian in the Soviet era. In light of the radical sociocultural and sociolinguistic changes, however, that began taking place after the 1990s, there have now emerged varieties of Russian, particularly among the younger generation, with different degrees of influence from Estonian. This has also been accompanied with the emergence of new identities (i.e., Estonian Russians) and inter- and intra-ethnic patterns of communication among

Russian speakers. Verschik aptly notes that “structural properties of the languages in question are not relevant for the fact that starting from a certain point, Estonian has a unidirectional impact on

Russian” (p.1965). Instead, the contemporary linguistic effects that Estonian has had on Russian

6 For discussion on how access to the superstrate language may account for differences in grammatical gender marking in Creoles, see Liceras, Martínez, Pérez-Tattam, Perales and Fernández-Fuertes (2006).

86 was brought about by an important series of historical and sociocultural events in the post-Soviet era.

The powerful effect of language prestige has also been investigated. Recently, Stell

(2015) examined whether language prestige can in fact predict CS outcomes, as suggested by

Muysken (2000). Muysken predicts that insertions are more common in contexts of colonialism and recent immigration, where there is a socially dominant language and a subordinate language.

In contrast, contexts where languages enjoy similar prestige favor alternation. In terms of linguistic factors, typological distance between languages favors insertions and alternations, whereas typological proximity leads to insertions and congruent lexicalization. In his examination of naturalistic oral production data from approximately 186 South African multilinguals (high school and university students), Stell investigated whether symmetry/asymmetry in language prestige induces specific conversational and grammatical types of CS to co-occur.

Among other findings, Stell found that congruent lexicalization is sociolinguistically rather than linguistically determined, as mixed morphology in finite verbs which is a primary feature of congruent lexicalization was either marginally or absent in the Coloured and White samples. In contrast, English finite verbs (e.g., Di formed mokgahlo ke Cope ‘They formed that association called Cope’) were attested in the Sesotho sample. Given that Sesotho (i.e., Bantu) and English (i.e., Germanic) are typologically distant (in contrast to Afrikaans and English which are both Germanic), this outcome was not expected to occur. His data also revealed a ‘White’ grammatical type of CS characterized by the certain features, and a ‘Nonwhite’ CS type characterized by the opposite features. This finding was consistent with the sociolinguistic histories of the different ethnic groups. Switching into English was higher in non-white samples

87 who afford prestige to English and lower in the white Afrikaner sample that identified more with

‘pure’ Afrikaans and hence, have a lower predisposition to use English. His analysis, therefore, revealed that symmetry/asymmetry in language prestige can determine certain grammatical and conversational CS patterns.

It is important to underscore that some researchers argue for the view that both linguistic and social factors shape CS outcomes (e.g. Muysken, 2013; Myers-Scotton, 2002). At the same time, they question how useful and predictive a purely abstract model of grammar is in the analysis of CS. For example, in his in-depth analysis of switched adverbs in Haitain

Creole/English CS, Hebblethwaite (2010) criticizes MacSwan’s (1999) approach to analyzing

CS phenomena (i.e., in isolation of social, historical, and sociopolitical factors), which focuses on abstract syntactic operations and feature valuation. Among other findings, in his analysis of

CS data from second generation Haitian Creole/English bilinguals in , Hebblethwaite found that switched English adverbs in Haitian Creole speech (90.5%) were overwhelmingly preferred over switched Haitian Creole adverbs in English discourse (9.5%), revealing the co- occurrence of asymmetry and bidirectionality. In other words, the fact that switched Haitian

Creole adverbs were attested suggests that this switch is not only possible but available for more ubiquitous use given the right social context (i.e., reversed sociolinguistic conditions).

Thus, Hebblethwaite asserts that “[i]t is not a conflict of grammatical features that causes asymmetry; it is instead the prevailing sociolinguistic and psycholinguistic conditions where

English has a ‘high’ status and Haitian Creole has a ‘low’ status” that explains the patterns observed (for a similar view in relation to Dutch/French CS, see Treffers-Daller, 1994). Thus,

Hebblethwaite rejects the claim that morphological feature-valuation can be used for predicting

CS outcomes. Note that MacSwan’s approach to CS predicts asymmetry with no bidirectionality,

88 as in the case of mixed determiner phrases where examples such as ‘the casa’ are banned/considered ungrammatical CS. Along the lines of Myers-Scotton (1993, 2002) and Chan

(2008), Hebblethwaite asserts that “the lexical and functional properties of words and the sociolinguistic and psycholinguistic realities of the community provide a better framework for predicting outcomes in code-switching” (p. 410).

In sum, the powerful influence that social factors have on language contact outcomes cannot be underestimated, as it has been previously shown that social factors can indeed affect the grammatical outcomes of CS. Endeavoring to contribute to our understanding of language contact outcomes in contact communities where multilingualism is the norm, I examined three linguistic phenomena in light of important historical and sociolinguistic aspects that are unique to the Northern Belize contact situation.

In the following section, I take a closer look at how the Northern Belize Spanish contact situation contributes to previous work on contact linguistics and Spanish/English CS.

2.5 Situating the Northern Belize Context

In the past decades, several influential works (e.g., Pfaff, 1979; Poplack, 1980; Silva-

Corvalán, 1993, 1994, 1995, 2004; Zentella, 1997) have greatly contributed to our understanding of the outcomes that result from Spanish/English bilingualism and concomitant issues such as language attrition and . In particular, extensive research has been conducted on

Spanish/English CS in the U.S. (e.g., Anderson & Toribio, 2007; Jenkins, 2003; Otheguy &

Lapidus, 2003; Pfaff, 1979; Poplack, 1980; Reyes, 1982; Sankoff & Poplack, 1981; Timm, 1975;

Toribio, 2001, 2002, 2004; Wilson Vergara, 2013, among many others). It is surprising, therefore, that Spanish/English CS in Northern Belize has remained grossly understudied. The examination of language contact outcomes in this borderlands context, however, warrants scholarly attention for several reasons.

89 Belize stands out, as it is the only Central American country where English is the official language. Belize is also the youngest nation in Central America. Given its recent colonial history

(see Chapter 1), the search for a national identity still remains an important endeavor for

Belizeans; thus, raising questions of how multiple identities are linguistically indexed in this culturally diverse nation. Although Belize is generally considered an Anglophone Caribbean country due to the official status of English and to its historical ties with the Caribbean (Balam,

2013; Shoman, 2010), English is rarely used in informal settings.

In line with Waddell’s (1961) prediction that Spanish would eventually rise as the primary language of Belize, Spanish has today become the most widely spoken first language in

Belize (Balam, 2013; Gabbert, 2007). It is BK, however, as previous work indicated (Allsopp,

1965; Le Page & Tabouret-Keller, 1985), that occupies the unrivaled position of lingua franca since colonial times. These sociolinguistic conditions, coupled with the absence of ultra- normative language attitudes, render the Northern Belize context one that is singular in nature within the Spanish-speaking world.

When comparing the Northern Belize and the U.S. Hispanophone contexts, it is particularly in the degree of linguistic purism and general attitudes toward bi/multilingualism that the two contact situations starkly differ. This sociolinguistic difference is important to contact outcomes. We know that high levels of purism “hinder extensive and intimate code- switching practices” (Muysken, 2013, p.714); thus, having an effect on the nature, use, and status of CS and bi/multilingualism in a community’s private and public spaces.

In the U.S. Spanish/English context, there are high levels of linguistic purism (García,

2014; Valdés, González, López García & Márquez, 2003; Villa, 2002; Zentella, 1997, among others), and this is evinced not only in programs (for relevant discussion, see

90 García 2009, 2014), but also in the general attitudes toward CS. As an example, consider Odón

Betanzos Palacios’ description of Spanglish, a term often used to refer to CS in the contact

Spanish literature (Montes-Alcalá, 2009; Palmer, Martínez, Mateus & Henderson, 2014;

Rodríguez-González & Parafita-Cuoto, 2012; Zentella, 1997). Palacios, former president of the

North American Academy of the Spanish Language, described Spanglish as “an unnecessary and ignorant creation that constitutes a temporary problem” (as cited in Rodríguez-González &

Parafita-Cuoto, 2012, p.463; cf. Olszanki, 2007). The idea that Spanglish, non-standard varieties and/or bilingualism are the root of bilinguals’ academic and/or economic problems is one that has historically been very widespread in the U.S. context (for relevant discussion, see Escamilla,

2006; Zentella, 1997, p.285).

Border communities in the Southwest U.S. have a painful history of code-segregation, which in Guerra’s (2015) words, refers to an ideology that was founded on the basis of monolingual and monocultural values, beliefs and practices, which demanded complete assimilation from children who spoke minority or underprivileged varieties (for further details on author’s personal narrative and on code-segregation, see Guerra, 2015). Given the presence of code-segregation in the U.S., it is not surprising that today linguistic hegemony is rigidly associated with Standard English and Standard Spanish varieties (Escamilla, 2006; Leeman,

2005; García, 2009, 2014; García & Kleifgen, 2010; García & Wei, 2014; Martínez, 2010; Mrak,

2011; Palmer & Martínez, 2013; Palmer et al., 2014; Ramirez & Milk, 1986; Ramirez, Milk &

Sapiens, 1983; Villa, 2002; Zentella, 1997), whereas bi/multilingual language practices are typically disparaged and pejoratively perceived.

It must be taken into consideration that the use and frequency of bi/multilingual language practices can be intricately related to the degree of linguistic prescriptivism in a sociolinguistic

91 context, as communities where monolingual language practices are valued over bilingual ones may not foster practices such as the frequent use of intra-sentential CS in everyday discourse

(Muysken, 2013). We could surmise that there would be “little code-switching at all in communities where prescriptivist attitudes about maintaining the ‘purity’ of the language are part of the culture” (Myers-Scotton, 1998, p.100).

Communication patterns between linguistic groups may also affect the possibility of CS assuming an identity function. In bilingual communities that are polarized (e.g., the two linguistic groups have little communication due to historical tensions), for example, CS would not play an important identity function as allegiance to a linguistic group would generally be indexed through the use of either one language or the other rather than a mixture of the two (for relevant discussion, see Jacobson, 1998, p.100). In Northern Belize, however, bilingual CS does not only serve as an identity marker, but as an important linguistic resource employed in a variety of contexts (Balam, 2013; also see Chapter 5).

In this regard, Northern Belize allows us to study Spanish/English CS in a context where this practice is largely accepted and valued. Hence, we are able to investigate the complex and dynamic nature of CS phenomena, either via intuitional or naturalistic speech data, in a contact situation where speakers’ linguistic behaviors are not as heavily influenced or affected by the overt stigmatization of CS.

Northern Belize is similar to other contexts of migrant bilingualism, or bilingualism which results from the mass movement of a group of people from one place to another

(Agnihotri, 1987; Backus, 1996; Bailey, 2002; Cheshire & Gardner-Chloros, 1997; Gardner-

Chloros & Finnis, 2006; Meuwiss & Blommaert, 1998; Toribio, 2006; Zentella, 1997).7

7 I use the term migrant rather than immigrant bilingualism given that the latter term may be restrictive in meaning, as waves of migration may not necessarily imply movement from one country to another. In the case of Belize,

92 Crucially, a distinctive characteristic of CS in cases of migrant bilingualism is that “[it] is often not only very frequent but very intricate at a grammatical level” (Gardner-Chloros & Edwards,

2004, p.123).

In the case of Belize, previous work has highlighted the prevalence of CS in Belize. For example, Hagerty (1979, p.134) reported “the common practice of code-switching” in Belize, and Brockmann (1979, p.175) noted the “quite impressive” and “quick and frequent code switching” among bi/multilingual speakers from Orange Walk, but neither Hagerty nor

Brockmann provided quantitative or qualitative insight into this phenomenon. More recently,

Balam (2013) only examined speakers’ language attitudes. The linguistic aspects of NBS and

Northern Belize Spanish/English CS, however, remain poorly understood. In an effort to fill this gap, in the present dissertation, I provide in-depth insight into the grammatical nature and unmarked status8 of Spanish/English CS in Northern Belize.

More specifically, the present dissertation provides quantitative analyses of phonological

(i.e., rhotic neutralization) and morphosyntactic (i.e., bilingual light verb constructions and determiner phrases) structures, while also exploring how sociohistorical factors may have helped to shape present-day contact Spanish outcomes in Northern Belize. As the foregoing discussion has revealed, three distinctive sociolinguistic characteristics of the Spanish contact situation include: (i) positive attitudes toward Spanish/English CS, (ii) CS as a marker of Northern Belize

Maya/Mestizo identity, and (iii) a low degree of normativity. Given the singular nature of its recall that the waves of migration of Yucatec Maya and Mestizo families became prominent in the mid- and late- 1840’s, when Belize was not still not a British colony. Belize did not officially become a British colony until 1962 (Bolland, 2003; Dobson, 1973; Shoman, 2010). In this regard, these Maya and Mestizos were not necessarily immigrants.

8 In line with Myers-Scotton’s (1993) ‘markedness’ theory of CS, this means that in bilingual communities, the practice of CS itself becomes the ‘unmarked choice’ or the preferred mode of communication simply because it is the norm and is part of speakers’ social knowledge of language use. Previous work suggests that in Northern Belize, the use of CS is unmarked (Balam, 2013; Balam & Prada Pérez, 2016).

93 history and sociolinguistic make-up, Northern Belize allows us to examine the nature of contact outcomes in relation to those that have been more extensively studied.

Note that I analyze novel data and also provide preliminary comparisons between contact outcomes in this context and the U.S. Hispanophone context, particularly the Southwest U.S.

While a thorough between the Northern Belize and the Southwest U.S. context (as two regions where a Mexican Spanish-derived variety of CS is spoken) is clearly beyond the purview of this dissertation, it is nonetheless useful to make connections and identify potential avenues for future research.

2.5.1 Research Questions

The two overarching questions and hypotheses guiding the present dissertation are as follows:

1. What role does convergence (diachronic for the analysis of rhotics and synchronic in the

case of bilingual light verb constructions and mixed determiner phrases) play in the

language practices of Northern Belize bi/multilinguals?

Hypothesis: If CS is a causal mechanism of convergence, then convergence should be

attested in the production of ‘hacer + V’ and mixed determiner phrases. If CS also leads

to diachronic convergence, at least some evidence should be found of a merger of rhotic

categories in NBS.

2. What do findings in the Northern Belize context reveal about the roles of linguistic and

social factors in language contact outcomes?

Hypothesis: If social factors are deterministic (Gardner-Chloros & Edwards, 2004), we

should find bilingual patterns that demonstrate insertion and congruent lexicalization

(Muysken 2013), given the low degree of normativity in Northern Belize and closely knit

94 networks in a contact situation where bi/multilingualism has remained stable since the

1840s.

2.5.2 Structure of the Dissertation

This dissertation consists of three inter-related projects that collectively provide an insight into the patterns of language use, language change and innovation in Northern Belize contact Spanish. Whereas the first project explores a phonetic/phonological feature of Northern

Belize contact Spanish, the second and third projects deal with two phenomena that specifically elucidate the morphosyntactic nature of Spanish/English CS in Northern Belize.

Table 2-2. Main topics examined in the present study Topic Data Participants Project 1 Neutralization of the intervocalic elicited oral 10 speakers rhotic contrast production data, read-aloud data Project 2 Bilingual light verb constructions acceptability 89 speakers judgments, oral production data Project 3 Determiner phrases oral production 62 speakers data

As Table 2-2 illustrates, data from the present study come from a total of 161 native consultants from Orange Walk, Northern Belize.

In the study on rhotics, I examine the intervocalic rhotic contrast. Specifically, I investigate whether a rhotic merger is attested among adolescents, especially now that bilingual CS and the use of BK have become more frequent. The project on bilingual light verb constructions examines restrictions that have been proposed in the antecedent variationist and generative literature (González-Vilbazo & López, 2011). Lastly, for the project on non-switched and switched determiner phrases, I examine the incorporation of semantic categories and gender assignment in mono- versus bi/multilingual discourse. In the concluding chapter, I discuss how

95 the overall findings from the present large-scale study inform our current understanding of language practices in Northern Belize and the nature of intra-sentential CS.

The present investigation contributes to our current understanding of phonological and syntactic microvariation in a Central American/Caribbean context where bi/multilingual CS occurs in a context with low levels of linguistic purism. It also provides foundational work on phonological and grammatical outcomes in Northern Belize; thus, filling a gap in the contact

Spanish literature on a Spanish variety that has been given little scholarly attention (Balam,

2013; Brockmann, 1979; Fuller Medina, 2005; Koenig, 1975, 1980). Lastly, it contributes to theoretical discussions on (i) the prevailing conceptualization of CS as the combination of two grammars or codes that maintain their identities in bilingual speech and (ii) the role that CS plays as a causative mechanism of convergence.

96 CHAPTER 3 THE INTERVOCALIC RHOTIC CONTRAST IN NORTHERN BELIZEAN SPANISH

3.1 Introduction

In his seminal work on Belizean Spanish, Hagerty (1979, p.91) noted that a phonemic merger of intervocalic Spanish rhotics was taking place in the younger generation, some of whom produced a merged rhotic category in minimal pairs such as caro ‘expensive’ and carro

‘automobile’. A question that arises as a result of Hagerty’s decades-old observation is whether in today’s younger generation, the same pattern is present or whether the merger has progressed to a more advanced stage of convergence. In previous sociolinguistic work, it has been argued that particularly adolescents employ vernacular linguistic features and advance existing patterns of sound change (Labov, 2001; Eckert, 1997b, 2004). In Eckert’s (1988) view, “[t]he relatively high degree of phonological innovation in the adolescent age group with relation to other age groups is an indication that the development of adolescent social structure provides a major impetus for phonological change” (p.197). There is the possibility, therefore, that among contemporary Northern Belizean adolescents, who are more proficient in English and BK than previous generations, we may find evidence of further intervocalic rhotic neutralization.

The Northern Belize context offers fertile ground to test whether the intervocalic rhotic contrast has been lost. To begin with, NBS has been noted for its distinctive non-standard phonological and grammatical features (Balam, 2014; Hagerty 1996). Crucially, within the last four to five decades, there has been an increase in both English and BK proficiency in Northern

Belize, bringing into question whether the increase in the intensity of language contact and changes in bi/multilingual proficiency and language use at a societal level have resulted in the complete merger of intervocalic rhotics. We know from previous work that it only takes a few decades before contact-induced, phonological innovations are manifested in a community (for

97 work on phonetic convergence involving rhotics, see Bullock & Gerfen, 2004; Ravindranath,

2009, p.92-93).

Furthermore, given that there has also been a cross-generational increase in the frequency of CS (see Balam, 2015, 2016a, 2016b), the study of the intervocalic rhotic contrast provides some insight into the proposed causal relationship between CS and convergence (Backus, 2004;

Bullock & Toribio, 2004; Gardner-Chloros & Edwards, 2004; Myers-Scotton, 2002). We could posit that a drastic increase in the frequency and density of CS in the younger generations, coupled with relaxed language norms, has favored the neutralization of the intervocalic rhotic contrast in NBS. Recall that in Muyken’s (2000, 2013) grammatical typology, relaxed language norms stimulate congruent lexicalization. At the phonological level, strong normative pressure may lead to hypercorrection in elicited speech, but we can surmise that weak normative pressure may favor convergence1 across time. If indeed NBS is moving in the direction of English, as

Hagerty (1996) claims, convergence could be manifested not only at the morphosyntactic level but at the phonological level as well, especially in the case of rhotics, known to be susceptible to cross-dialectal variation (Bradley, in press; Willis & Bradley, 2008) and diachronic change

(Jaworski & Gillian, 2011)2.

Endeavoring to determine whether adolescent speakers of NBS produce a merged rhotic category in intervocalic position, I examined oral production data drawn from elicited, semi- spontaneous speech and reading. These two modes of production were examined to investigate whether a difference emerged depending on the modality.

1 Convergence may reflect different degrees of approximation toward another phonological or morphosyntactic structure in the contact variety or varieties. Here I use the terms ‘convergence’, ‘merger’ and ‘neutralization’ interchangeably. Following Bullock and Gerfen (2004), I take phonological neutralization to be an instantiation of linguistic convergence between two or more contact varieties.

2 In linguistic contexts involving other languages, the trill has been noted for its diachronic vulnerability to phonetic weakening and non-usage (see, for example, the case of the intervocalic trill in Polish: Jaworski & Gillian, 2011).

98 This chapter is organized as follows. In section 3.2, I briefly describe the phonological phenomenon in question and situate the present study within the more general aims of the overall dissertation. In section 3.3, I provide an overview of previous studies on rhotic production and the intervocalic rhotic contrast in Spanish. Section 3.4 provides details on the methodology that was employed in the present acoustic study. Section 3.5 outlines the results of the study. Lastly, in sections 3.6 and 3.7, I provide a discussion of the study’s findings and offer concluding remarks.

3.2 Intervocalic Rhotic Contrast(s) and Neutralization in Spanish Varieties

In this section, I summarize the work that has been conducted on Spanish rhotics. I also spectrographically differentiate the canonical tap from the apico-alveolar trill. Although I describe the general differences between normative and divergent patterns of rhotic distribution,

I provide a more detailed description in section 3.3.

To date, most of the extant literature on the production of Spanish rhotics investigates dialectal variation in the realization of rhotics (e.g., Blecua, 2001; Colantoni, 2006; Díaz-

Campos, 2008; Henriksen & Willis, 2010; Lewis, 2004; Quilis, 1993; Sánchez Corrales, 1986;

Vásquez Carranza, 2006; Willis, 2006, 2007), or the L2 acquisition (e.g., Face, 2006; Hurtado &

Estrada, 2010; Johnson, 2008; Rose, 2010; Olsen, 2012; Reeder, 1998) of the tap and/or the trill.

Particularly the trill, which presents difficulty to both native speakers and second language learners of Spanish, has been widely studied and noted for its cross-dialectal allophonic variation

(Blecua, 2001; Colantoni, 2006; Díaz-Campos, 2008; Face, 2006; Jaworski & Gillian, 2011;

Lewis, 2004; Olsen, 2012; Solé, 2002; Widdison, 1998, among others).

While the tap involves the production of only one closure with an average duration of 20 ms (Quilis, 1993), the trill, lasting approximately 85 ms (Quilis, 1993) involves a series of lingual contacts, typically more than two, which can only be produced under specific

99 aerodynamic conditions (Bradley & Willis, 2012; Face, 2006; Ladefoged & Maddieson, 1996;

Solé, 2002; Willis, 2007). The apico-alveolar trill can only be produced when an airstream passes through an aperture produced by high air pressure, and the pressure behind each closure dips dramatically, consequently triggering the apex of the tongue to produce another occlusion

(Jaworksi & Gillian, 2011). Thus, voiced apico-alveolar trills are constrained in their aerodynamic requirements, and slight changes in oropharyngeal and subglottal pressure lead to rhotic productions that are non-trilled (Solé, 2002).

Acoustically, the canonical trill is distinguished by the presence of clear breaks that correspond to brief occlusions or lingual contacts between the apex of the tongue and the alveolar ridge (Hualde, 2005, p.181). The representative spectrogram in Figure 3-1 illustrates a canonical apico-alveolar trill with three lingual contacts, produced by a native speaker of

Mexican Spanish. Spectrographically, the trill appears as consecutive blank breaks.

Figure 3-1. Apico-alveolar trill produced by Mexican Spanish speaker in tonces agarra ‘then he grabs’ (Willis & Bradley, 2008, p.89)

100 In contrast, the production of a tap does not entail strict articulatory requirements, as it involves only one closure or occlusion. As the example from Veracruz Mexican Spanish in

Figure 3-2 illustrates, the closure has minimal formant structure. Importantly, the tap may include a release burst (appears immediately after closure), reflective of their status as non- (Bradley & Willis, 2012, p.51).

Figure 3-2. Tap produced by Veracruz Mexican Spanish speaker in parec(e) ‘seems’ (Bradley & Willis, 2012, p.52)

The study of the tap and the trill has been of interest to Hispanic linguists not only because of their variation, but also because they have an intervocalic phonemic contrast in

Spanish. Table 3-1 below presents the general distribution of the tap and trill in normative

Spanish3 (from Hualde, 2005, p.183, cited in Willis & Bradley, 2008, p.88).

3 Following Bradley & Willis (2012), the patterns of rhotic distribution in normative Spanish are those that conform to Hualde’s (2005, p.183) description.

101 Table 3-1. Rhotic distribution in normative varieties of Spanish Rhotic (contrast) Phonetic context and example

Contrast tap /ɾ/ 1a) V _V Intervocalic versus trill // /kaɾo/ ‘expensive’ vs. /karo/ ‘cart;car’

Only trill /r/ 2a) #_ Word-initial /roka/ ‘rock’

2b) C_ After a heterosyllabic consonant /alrededor/ ‘around’, /enredo/ ‘mess’ /israelita/ ‘Israeli’

Only tap /ɾ/ 3a) C_ After tautosyllabic consonant /bɾoma/ ‘joke’, /gɾamo/ ‘gram’

3b) V_#V Word-final before a vowel /ser amigos/ ‘to be friends’

Variable rhotic 4a) V_C Before a consonant (most commonly [ɾ]) /paɾte/ [paɾte] ~ [parte]

4b) V_#C Word-final before a consonant /seɾ poeta/ ‘to be a poet’

4c) V_ ## Word-final before a pause /seɾ o no seɾ/ ‘to be or not to be’

As Table 3-1 illustrates, in normative Spanish, trills appear intervocalically (1a), word- initially (2a), and syllable-initially, after a heterosyllabic consonant (2b). It is only in intervocalic position, however, that there is a phonemic contrast between the tap and the trill. On the other hand, taps appear in the second position of a complex onset cluster (3a), and in word-final position, when resyllabified into the onset of a following vowel-initial word (3b). Lastly, the tap variably occurs in syllable-final position before a consonant (4a), and in word-final position before a consonant (4b) or pause (4c).

102 Of particular interest in recent work on Spanish rhotics has been the study of the maintenance of the intervocalic tap/trill contrast (1a). Evidenced in fewer than 30 minimal pairs such as pero ‘but’ and perro ‘dog’, and caro ‘expensive’ and carro ‘car/automobile’ (Bradley &

Willis, 2012), this contrast is one of the most distinctive features of normative Spanish varieties, and is also orthographically differentiated in Spanish, with the “r” representing the tap and the

“rr” the trill.

Despite its saliency in both spelling and pronunciation, however, it has been previously argued that in many Spanish varieties, the intervocalic tap/trill contrast is non-existent. In particular, Hammond (1999, p.147) has argued that neutralization of the tap/trill contrast has occurred in many Spanish varieties. Basing his contention on recordings of a text read aloud by

229 speakers of different Latin American and varieties, Hammond noted that the three-contact trill in intervocalic position, as prescribed by the Real Academia Española, does not occur in the naturalistic discourse of a vast majority of native Spanish speakers. In his work, he found that only 16 of the 1603 trill occurrences were produced as canonical trills.

Contra Hammond’s claim, however, in more recent studies, scholars have found that the intervocalic tap/trill contrast in Spanish may be maintained via segmental duration (in milliseconds) of trill variants. I provide details on these studies in section 3.3.2.

To date, only a handful of acoustic studies have specifically examined the Spanish intervocalic tap/trill contrast. There is no acoustic study, however, that explores this phenomenon in a bi/multilingual Central American context where normative patterns of rhotic distribution co- exist with divergent ones. By divergent, I mean that the distribution of rhotics in the local

Spanish variety systematically differs from the patterns attested in normative Spanish. A phonological feature that individualizes these Spanish varieties is the existence of an assibilated

103 rhotic (e.g., for : Adams, 2002; Vásquez Carranza, 2006; for Highland

Ecuadorian Spanish: Bradley, 1999) or retroflex approximant (e.g., for NBS: Hagerty, 1979), systematically used in place of the canonical apico-alveolar trill.

Of relevance to this chapter is the retroflex approximant variant, which Hagerty (1979,

1996) identified as a distinctive phonological feature of the Spanish variety spoken in Northern

Belize4. Previously, it has been shown that increasing contact with varieties that have a retroflex category (e.g., English and Kriol) can lead to phonological innovation and/or change in languages that lack retroflex variants (e.g., for Spanish in contact with English: Ramos-Pellicia,

2007; for Garifuna in contact with English and BK: Ravindranath, 2009).

Diachronically, Latin American Spanish varieties that make variable or no use of the canonical apico-alveolar trill stand out because they have evolved in such a way that their phonological systems have undergone a restructuring in their patterns of Spanish rhotic distribution. Thus, these varieties have seemingly gone one step further in the process of phonological evolution, developing an alternative way to maintain the intervocalic rhotic contrast in Spanish. Oftentimes, the innovative rhotic substitutes the place of the trill in several contexts, as will be discussed further below. Thus, instead of the canonical tap/trill intervocalic contrast, speakers may adopt a tap/assibilated rhotic contrast (e.g., for Central Valley Costa Rican

Spanish: Vásquez Carranza, 2006) or a tap/approximant intervocalic contrast, as the present analysis shows.

It is worth noting that in Spanish contexts with both normative and divergent patterns of rhotic distribution, even if the divergent patterns largely exclude the use of the canonical trill in

4 In previous work, the use of the retroflex approximant has been attested in other varieties, including: New Mexican Spanish (Cassano, 1977); Spanish (Sánchez, 1973); Yucatan Spanish (Lope Blanch, 1975); and Lorain (Ramos-Pellicia, 2007). Notably, the retroflex variant is often associated with influence from the American English approximant rhotic (Canfield, 1940; Face, 2006; Major, 1986).

104 informal speech, the trill often comprises an integral phoneme of academic Spanish in classroom contexts, as is the case in Belize. A key question that arises, therefore, is how the interaction of normative and divergent patterns of rhotic distribution manifests itself in the speech of bi/multilinguals that live in a community where CS is the norm. More specifically, do CS and the presence of divergent patterns of rhotic distribution (in informal discourse) alongside normative ones (in the media, classroom discourse, etc.) contribute to loss or maintenance of the intervocalic rhotic distinction in Spanish? There are two potential scenarios. There is the possibility that the intervocalic rhotic contrast has simply been lost, as it has occurred in other contact situations (e.g., Judeo-Spanish dialects in the Balkan Peninsula: Bradley & Willis, 2012;

Quintana, 2006, p.84; : Granda, 1984; Lipski, 1985; O’Brien, 2013;

Palenquero and Papiamentu: O’Brien, 2013, p.26).

It must be considered, however, that younger generations in Northern Belize have had greater access to formal education, and hence, more classroom instruction on normative Spanish in post-independent Belize, especially after the incorporation of Spanish into the high school curriculum in 1998. This development, therefore, could have potentially contributed not to convergence but to divergence (i.e., the presence of discrete normative and divergent intervocalic rhotic contrasts that are employed in different social domains). Thus, the incorporation of

Spanish as a subject in the high school curriculum could have led to greater use of the trill. We know that changes in schooling can have an impact on the phonological systems of speakers

(e.g., Guion, 2003), and across time, varieties with non-standard phonological forms may actually undergo a rapid process of standardization in which younger speakers gradually abandon

‘regional’ forms and employ more standard, pan-Hispanic phonological variants (for example of this phenomenon in Yucatan Spanish, see Michnowicz, 2006, 2012).

105 In light of Hagerty’s (1979, p.81) observation that rhotic neutralization was in its final stage among some adolescent Maya/Mestizos in the 1970s, one can anticipate that if this was indeed the case more than three decades ago, the present data should reveal at least some indication of this. Endeavoring to determine whether the intervocalic rhotic contrast has neutralized in NBS, I analyzed the segment duration of a total of 580 intervocalic rhotics

(elicited production = 330 tokens; reading = 250 tokens) produced by adolescents from Orange

Walk, Northern Belize.

3.3 Previous Work

In this section, I provide an overview of studies that have specifically shed light on (i) divergent patterns of rhotic distribution and (ii) the neutralization of the tap/trill phonemic contrast in Spanish.

3.3.1 Rhotic Distribution in Spanish Bidialectal Contexts

To my knowledge, no acoustic study has specifically examined the intervocalic rhotic contrast while also taking into consideration the coexistence of normative and divergent patterns of intervocalic rhotic distribution. We must bear in mind, however, that an important aspect of many communities in and the U.S. Hispanophone context is that there are varying degrees of Spanish bidialectalism. Bidialectalism is present when “two varieties of the same language are used alongside each other” (Yiakoumetti & Esch, 2010, p.294). Typically, the non- standard variety is the language spoken at home (e.g., NBS) whereas the standard variety is the second dialect students learn in the classroom context (i.e., standard Spanish).

An important phonological consequence in these contexts is that speakers are exposed to and may use both divergent and normative patterns of rhotic distribution in different social domains. Thus, while in informal speech speakers may employ a tap/assibilated rhotic intervocalic contrast, they may employ the normative tap/trill contrast in more formal situations.

106 This raises the question of whether the co-existence of two rhotic distributions leads to phonological convergence.

Divergent patterns of rhotic distribution are relatively different from those attested in

Spanish varieties with normative patterns of rhotic distribution. In order to further illustrate these differences, let us consider previous descriptive work conducted on Highland Ecuadorian

Spanish and Central Valley Costa Rican Spanish (see Table 3-2), two varieties that have been noted for their similarities in their divergent patterns of rhotic distribution (Vásquez Carranza,

2006, p.297).

Table 3-2. Divergent versus normative patterns of rhotic distribution Ecuadorian Costa Rican Standard 1 a. Rosa [ř] [ř] [r] ‘rose’ b. Enrique [ř] [ř] [r] ‘Enrique’ c. perro [ř] [ř] [r] ‘dog’ d. pero [ɾ] [ɾ] [ɾ] ‘but’ 2 a. tres [ř] [ř] [ɾ] ‘three’ b. vendrá [ř] [ř] [ɾ] ‘s/he will come’ c. premio [ɾ] [ɾ] [ɾ] ‘prize’ d. cruz [ɾ] [ɾ] [ɾ] ‘cross’ 3 a. cuerpo [ɾ] [ɾ] [ɾ ~ r] ‘body’ b. largo [ɾ] [ɾ] [ɾ ~ r] ‘long’ c. carne [ř] [ř] [ɾ ~ r] ‘meat’ d. persona [ř] [ř] [ɾ ~ r] ‘person’ 4 a. ir ahora [ř] [ɾ] [ɾ] ‘to go now’ b. mayor gusto [ř] [ɾ] [ɾ ~ r] ‘greatest pleasure’ c. la flor [ř] [ɾ~ř] [ɾ ~ r] ‘the flower’

Citing the works of Argüello (1978, 1980) and other scholars, Bradley (1999) reports that in Highland the assibilated [ř] shares the phonetic distribution of the trill in syllable-initial contexts. Thus, as Table 3-2 shows, the assibilated [ř] occurs in word-initial position (1a), in post-consonantal position (1b), and in contrastive distribution with the tap in

107 intervocalic position (1c, d). A similar pattern in attested in Central Valley Costa Rican Spanish

(Vásquez Carranza, 2006, and references therein)5.

The distribution of the tap in varieties with divergent patterns of rhotic distribution aligns with normative patterns not only in the intervocalic tap context but in other phonetic environments as well. For instance, in certain post-consonantal positions (2c, d), the tap is attested across both divergent and normative Spanish. Unlike the tap and/or the trill, however, the assibilated rhotic occurs in complex syllable onsets after non- coronals such as [t] and [d] as in tres ‘three’ (2a) and vendrá ‘will come’ (2b), yielding an alveolar [tʃ] sound as in the English word ‘tree’ (for relevant discussion, see Vásquez Carranza, 2006, p.299).

In addition, the assibilated rhotic occurs in word-internal coda position where a coronal follows the rhotic, as in carne ‘meat’ (3c) and persona ‘person’ (3d).

There are minor differences between Highland Ecuadorian Spanish and Central Valley

Costa Rican Spanish. For instance, in Highland Ecuadorian Spanish, the assibilated rhotic appears word-finally, before a vowel (4a), consonant (4b), or a phrase boundary (4c). On the contrary, in Central Valley Costa Rican Spanish, the tap primarily occurs in these contexts. In addition, in Central Valley Costa Rican Spanish, elision of the tap occurs only in /rC/ clusters at the morpheme boundary when the rhotic is followed by a consonant-initial clitic (e.g., [po.

ˈnɛø.lɛ], ponerle ‘to put something on it’). Overall, however, patterns of rhotic distribution in these two varieties are strikingly similar (Vásquez Carranza, 2006).

Although divergent patterns of rhotic distribution are documented in the literature, it is a phenomenon that is not specifically addressed and/or discussed. The distinction and presence of both normative and divergent patterns of rhotic distribution, however, are important factors to

5 Vásquez Carranza’s (2006) impressionistic analysis was based on the read-aloud data from 6 consultants who were native speakers of Central Valley Costa Rican Spanish.

108 consider, as emergent bidialectal speakers of Spanish often have an acute awareness of rhotic variants and their pronunciation. In the case of Northern Belize, for example, speakers are quite aware that their home variety differs markedly from the standard variety vis-à-vis rhotic sounds

(see Balam 2013, p.264; Balam & Prada Pérez, in press).

This is to be expected, especially since the trill is often emphasized in the classroom. In some Spanish textbooks, the trill is highlighted and presented as a distinctive sound of the

Spanish phonological system. In the textbook Español para la vida 2 Workbook, which some high schools in Belize still use, authors Moore and Abrikian (1995) provide explicit instruction as to how to pronounce the trilled rhotic. They also provide practice exercises that focus on the trill. Regardless of the textbook in use, explicit instruction on the use of the trill in normative

Spanish is an aspect of pronunciation that Spanish teachers in Belize give special attention to

(Balam & Prada Pérez, in press).

In the next section, I summarize studies that have specifically examined the intervocalic tap/trill contrast in Spanish.

3.3.2 Studies on the Intervocalic Tap/Trill Contrast

In recent research on rhotics, some scholars have focused on examining the maintenance of the tap/trill contrast in Spanish. This research has revealed that contra Hammond’s claim, in some Spanish varieties, the tap/trill phonological contrast in maintained via segmental duration rather than the number of closures or lingual contacts.

Willis and Bradley (2008), for example, examined the intervocalic tap/trill contrast in

Dominican Spanish (DS) and found that the tap/trill contrast in DS has not been neutralized.

They compared the semi-spontaneous production of 154 taps (produced by 12 adult consultants) with previously recorded data on trills extracted from a recorded narrative (Willis, 2006, 2007).

Results revealed that both the tap and the trill appeared in a continuum of variants (i.e., the trill

109 appears as [r], [ɦr], [ɦɾ] and [ɦ]), consistent with the findings of other researchers. Most importantly, their findings revealed that the rhotic contrast is maintained in DS. Even though trills and taps seemed auditorily similar, duration differentiated them. In terms of duration, the mean trill duration was at least three times as long as the mean tap duration (e.g., for Cibao speakers, 89 ms versus 20 ms). In their analysis, Willis & Bradley suggest that consonant reduction and elision in taps, and pre-breathy voicing in DS trills occur in order to help maintain the tap/trill contrast.

Bradley and Willis (2012) found further evidence in their study on Veracruz Mexican

Spanish (VMS), where the intervocalic tap/trill contrast is also maintained via segmental duration rather than the number of lingual contacts of the trill. Their study was based on data collected in 2005 from ten university students who are native speakers of VMS. In the guided, semi-spontaneous speech task, speakers narrated a story using Mayer’s (1969) Frog Where are

You? picture story. Their analysis revealed that taps were typically lenited or elided, whereas trills typically had voiced and voiceless and approximant variants; and post- approximantized [rɹ]6. Their acoustic analysis of 216 phonemic taps and 339 phonemic trills also showed that there was no overlap in the segmental duration of normative taps and trills in intervocalic position. Trills were significantly longer than taps (i.e.36 ms versus

+60 ms for nine out of ten speakers).

Henriksen (2015) also found supporting evidence for the maintenance of the tap/trill contrast via segmental duration in speakers of Mexican Spanish speakers living in the

6 Acoustically, post-approximantized allophones of the trill appear as a series of breaks, typically two or three, followed by continuous formant structure (i.e., the presence of an approximant rhotic that appears as a vocalic segment) and possible frication (see Bradley & Willis, 2012, p.59). It involves, therefore, the articulation of a trill immediately followed by an approximant rhotic. According to Bradley and Willis, while these allophones have been documented in Italian, Russian, Peninsular Spanish and other languages, they have only been attested in VMS, but not in other varieties of Latin American Spanish.

110 Chicagoland area. Henriksen’s study was based on semi-spontaneous speech data from a total of

16 Spanish/English bilinguals (8 first generation and 8 second generation) who also narrated

Mayer’s (1969) Frog Where are You? picture story. Henriksen’s analysis of 514 trills and 216 taps revealed that 15 of the 16 speakers maintained the tap/trill contrast via segmental duration, with the trill being more than twice as long as the tap (Generation 1: trill = 74.17 ms, tap = 31.99 ms; Generation 2: trill = 70.49 ms, tap = 30.61 ms). Although speakers most commonly produced trills with a single occlusion, extensive intra-speaker and inter-group variation was attested in the number of occlusions in trills. Based on his findings, Henriksen concluded that phonetic variation does not result in phonological instability in the grammar of heritage speakers of Spanish.

Unlike findings in these previous studies, O’Brien (2013) found supporting evidence for intervocalic rhotic neutralization in Equatoguinean Spanish. Her study was based on read-aloud data from five multilingual speakers who spoke Equatoguinean Spanish (2 females, 3 males).

O’Brien examined rhotics in different phonetic contexts (i.e., onset clusters with voiceless stops, word initial, word final, and word medial positions). As it relates to neutralization, O’Brien examined word medial intervocalic rhotics to determine whether the tap/trill contrast is maintained by type of realization (tap, trill, fricative, etc.) or by segment duration. Results showed that speakers did not maintain a contrast via type of realization, as both word medial trills and tap showed similar patterns of distribution in terms of variants, with the fricative being the most frequent realization. Furthermore, the analysis of 44 trills and 29 taps revealed that speakers neutralized the tap/trill contrast, producing a mean duration of 27 ms for trills and 22 ms for taps; these durations were not found to be statistically different. Overall, only one

111 canonical trill was produced in the data. Thus, O’Brien’s findings supported previous claims that the tap/trill contrast has neutralized in Equatoguinean Spanish (Granda, 1984; Lipski, 1985).

Although the intervocalic rhotic contrast has not been acoustically analyzed in NBS,

Hagerty’s (1979) phonological analysis of Belizean Spanish provides valuable insight into rhotic production in Northern Belize7. In his work, Hagerty (1979, p.81) noted that there was “an ongoing process of phonemic merger…resulting in a complete lack of opposition,” particularly among younger Belizeans; suggesting, therefore, a neutralization of the intervocalic rhotic contrast in NBS. Crucially, he observed that in NBS the merger was always in the direction of the retroflex approximant and not in the direction of the tap, as is the case for Afro-Bolivian

Spanish (Lipski, 2008) and Equatoguinean Spanish (Lipski, 1985, p.44; O’Brien, 2013, p.58).

More than thirty five years have passed since Hagerty made these observations. Whether the intervocalic rhotic contrast has neutralized in NBS or not was the main question this study sought to answer. Given that language contact has increasingly become more intense in Northern

Belize, it may be that the use of a converged rhotic category has become more widespread.

Alternatively, it may be that what Hagerty observed in the 1970s was simply the presence of an alternative rhotic contrast (i.e., one based on segment duration) among NBS speakers.

The current study contributes to our understanding of the distribution and realization of rhotics in the Northern Belize context, while also providing an insight into contact outcomes in a context that has undergone a drastic increase in the frequency of CS during the last four to five decades (Balam, 2015, 2016b).

7 Hagerty collected his data (between 1975 and 1978) through a picture identification task and semi-structured interviews, where participants spoke about Belizean legends and folk tales of supernatural beings. The final 38 interviews that were analyzed were each approximately 30 minutes in length.

112 3.4 Methodology

3.4.1 Participants

In order to acoustically determine whether there was phonological convergence in the intervocalic rhotic contrast in NBS, ten adolescents from Orange Walk, Belize were recruited to participate in the present study. Consultants were divided into two groups: the freshman high school group (4 males, 1 female) and the senior high school group (1 male, 4 females)8. High school students were chosen as they represent the social group that receives the most Spanish instruction in the Belizean school system. In addition, in previous sociolinguistic work, adolescents have been described as the ‘movers and shakers’ in linguistic change (Eckert, 1997a,

2000), particularly as it relates to sound innovations (Eckert, 1997b)9. Thus, any phonological patterns that are indicative of language change or innovation in progress may most likely be manifested in adolescent speech. Academic level was considered to determine whether there was a significant difference in rhotic realization between the freshman and senior group. Given that the senior group had taken a standard Spanish class for a longer period of time, it was expected that they would more successfully maintain a normative intervocalic rhotic contrast. On the other hand, freshman students who have received less classroom instruction in Spanish were expected to evince greater difficulty maintaining the intervocalic rhotic contrast.

Table 3-3 presents information on speakers’ proficiency in their languages, as assessed via the widely-used DELE Spanish proficiency test (cf. Slabakova & Montrul, 2003)10. It also

8 In the current study, the effect of gender (male vs. female) was not examined. However, a larger sample could allow in-group and across-groups examination of gender effects in relation to rhotic production, especially since gender has been attested as a crucial factor in phonological production (e.g., Bradley & Willis, 2012; Díaz-Campos, 2008; Henriksen & Willis, 2010). A more balanced group of participants could have allowed a clearer picture of NBS rhotic production vis-à-vis gender and sociophonetic attitudes.

9 In line with Eckert’s (1997a) contention, Díaz-Campos (2008, p.54) found that younger speakers of favored innovative trill variants such as .

113 presents speakers’ self-rated proficiency in their languages (where ‘1-3’ indicates ‘poor’ and ‘6-

7’ indicates ‘excellent’), and their patterns of language use.

Table 3-3. Participant information on self-reported proficiencies and language use Speaker/ DELE Self-rated Language Use Gender Proficiencies SP BK EN % per day % per day w/closest friends w/family members F1/m Low 3.5 5.9 5.7 90BK/10S 80BK/20S F2/f Low 5.6 3.6 5.7 80S/20K 90S/10BK F3/m Low 6.2 - 5.9 80S/20E 100S F4/m Low 6.2 4.6 5.9 70S/30BK 80S/20BK F5/m Inter 6.3 - 6.1 90S/10E 100S S1/m Inter 5.2 5.0 4.8 60S/40BK 70S/30BK S2/f Inter 5.8 3.5 5.6 70S/30K 90S/10BK S3/f Low 3.6 6.5 6.3 80K/20S 70BK/30S S4/f Inter 5.9 4.1 5.8 50S/40BK/10E 80S/20BK S5/f Inter 5.7 3.6 5.5 70S/30BK 90S/10BK *SP = Spanish, BK = Belizean Kriol, EN = English, inter = intermediate

The self-reported data showed that most speakers rated both their Spanish and

English proficiencies similarly. Speakers F2, F4, S2, S4 and S5 reported being less proficient in

BK than in Spanish or English, whereas speakers F1 and S3 rated their Spanish proficiency as markedly lower than both their BK and English proficiencies. The latter two consultants were

Kriol-dominant Spanish speakers who primarily used BK when speaking with closest friends and family members. The DELE results were consistent with speakers’ self-rated proficiencies only in some cases (e.g., speakers F1 and S3, but see F4). These discrepancies, however, are not surprising given the fact that although the DELE is a widely used measure, it is not the most adequate test to assess the proficiency of NBS speakers. The DELE test largely reflects

Peninsular Spanish rather than Latin American varieties of Spanish, and as such, its results must be interpreted with caution, especially when examining bilingual populations.

10 The DELE proficiency test consists of 50 questions. Generally, speakers who score between 41-50 are placed into the advanced proficiency group. Those who score between 30-40 are placed in the intermediate proficiency group. Lastly, speakers who score below 30 are placed in the low proficiency category.

114 In terms of language use, with the exception of speakers F1 and S3, all speakers reported using Spanish more frequently than English or BK both with their closest friends and family members. Given that F1 and S3 went to an urban primary school, it is not surprising that they were more dominant in Kriol, as the shift to Kriol in Northern Belize is more evident in urban rather than rural areas (Balam, 2013; Balam & Prada Pérez, in press; Brockmann, 1979). In contrast, speakers F3 and F5, who went to a primary school in a rural area, reported not being proficient in BK. In line with Brockmann’s (1979) observation, the self-reported data suggest that although the use of BK is increasing, Spanish is still the language that is used more frequently in Orange Walk villages.

In the present study, adolescent consultants are described as ‘emergent’ bidialectal speakers of Spanish because -- in contrast to other Latin American countries like where Spanish is the language of instruction -- Spanish co-exists with English as an official language in Belize. Hence, speakers start developing proficiency in standard Spanish primarily in the high school classroom context. Notably, although Spanish was officially incorporated into the primary school curriculum in 2001, not all schools include Spanish as an academic subject11 , and emphasis on standard Spanish primarily takes place at the high school level.

3.4.2 Tasks

A total of three tasks, eliciting semi-spontaneous and read-aloud data, were used. Elicited oral production data were obtained via a picture description task, where participants used two

11 In Belizean primary schools, Spanish is included as one of the content subjects. However, Spanish instruction is largely limited to translation and to the use of vocabulary words and/or lists, with less emphasis on communicative methods of teaching. Although Belize does have a Spanish Curriculum for lower, middle and upper division levels (http://www.moe.gov.bz/index.php/education-services/quality-assurance-development-services-qads/curriculum- exams), their implementation in primary schools across Belize varies (A. Pérez, personal communication, March 16, 2015). In Belizean high schools, Spanish is a compulsory core subject that is taken for a minimum of three or four contact hours per week. It is only in the classroom context that the use of standard Spanish is emphasized. Outside of the classroom, students normally switch among NBS, Spanish/English CS, Spanish/English/Kriol CS, and BK.

115 six-quadrant picture stories from Timm and Eccott’s (1972) picture story textbook to narrate a story (Figure 3-3). In order to encourage the use of vernacular speech, participants were asked by the author, a trilingual native of the community, to narrate the story in Spanish as it is spoken in

Belize.

Figure 3-3. Picture Story ‘The Flood’ (Timm & Eccott, 1972, p.64)

Oral production data were also obtained via a translation task of “The Pink Crocodile,” a story that I wrote to elicit rhotic production in different phonetic contexts (see Appendix A).

Since participants focused more on the translation aspect of the task as opposed to their own language use, speakers focused less on their pronunciation than in the subsequent reading task. In oral reading, speakers often engage in careful speech and are more aware of their pronunciation

(for relevant discussion, see Ramos-Pellicia, 2007). In the reading task, data were obtained via a

Spanish reading task (see Appendix B), which was the Spanish translation of the English story

“The Pink Crocodile”. Since the Spanish reading task was in monolingual Spanish, it was expected that this would trigger careful speech; hence, consultants would give greater attention to normative Spanish pronunciation. The English story and its Spanish translation, written by the researcher, were deemed as appropriate to elicit the production of rhotics, as they were in line

116 with lexical and morphosyntactic patterns of variation attested in the varieties spoken in this bi/multilingual context.

3.4.3 Procedure

With their parents’ consent, speakers were audio-recorded in their homes, where they felt at ease and comfortable. This facilitated the use of vernacular speech. Care was taken to ensure that there was silence in the room as the interview was being conducted. Each speaker was engaged in a preliminary, short informal interview in NBS, where they spoke about their school life and briefly narrated the story of Los Tres Cerditos ‘The Three Little Pigs’. Rhotics produced in this section of the interviews were not analyzed. Participants subsequently completed the picture description task, followed by an interval between this task and the translation task.

During the interval, consultants read the English story on their own. Participants were given an average of fifteen minutes before they began translating the story “The Pink Crocodile” to

Spanish. Finally, participants were simply asked to read the Spanish version of the story aloud.

After the recordings, the researcher helped some consultants to complete the Spanish linguistic background questionnaire. Since students typically fill out forms and applications in

English, the researcher clarified any doubts or queries students had regarding items in the questionnaire. Lastly, participants completed the DELE Spanish proficiency test.

Recordings were done in Orange Walk, Belize, using a Marantz PDM 620 and a Shure

SM 10-A head-mounted microphone with a sampling rate of 44.1kHz. Recordings were orthographically transcribed by the researcher, and tokens were identified and extracted for further acoustic analysis.

117 The reading text contained a total of sixteen intervocalic taps and nine intervocalic trills12.

Eight taps occurred in the onset of a stressed syllable, as in amarillo ‘yellow’ and orilla ‘edge’; two taps occurred in the onset of an unstressed syllable that was followed by a stressed syllable

(e.g., corazón ‘heart’ and caracol ‘seashell’), and six occurred in the onset of an unstressed syllable preceded by a stressed one, as in pero ‘but’. Intervocalic taps in monosyllabic words and the word para ‘to’ were excluded from the analysis as they showed a strong propensity to lenite

(i.e., para lenited to pa).

While all Spanish and English r-words were extracted from the elicitation tasks and the read-aloud task, results reported here are based primarily on the acoustic analysis of intervocalic rhotics13. A total of 193 taps and 137 retroflex approximants from elicited oral production were acoustically analyzed to examine the maintenance of the NBS divergent intervocalic rhotic contrast. Furthermore, 160 taps and 90 rhotics in medial trill position from the reading task were acoustically analyzed to determine whether there was neutralization of the normative rhotic contrast (i.e., tap/trill contrast). Analysis of the extracted tokens was carried out using the acoustic software program Praat (Boersma & Weenink, 2008).

A rhotic was considered a tap when there was full or partial occlusion indicated in the spectrogram (i.e., break between vocalic sounds). A tap was considered ‘overgeneralized’ if it occurred in a non-target phonetic context (i.e., generalized in the context of an intervocalic trill or retroflex approximant). In line with Blecua (1999), Bradley and Willis (2012), Face (2006),

12 The preceding and following vowels were not controlled for, as previous work has shown that this variable does not have an effect on the realization of rhotics (for further discussion, see Blecua, 2001). Solé (2002) found, however, that the vowel which follows a rhotic does have an effect on trill production.

13 No interrater procedure was employed to cross-check acoustic measurements. In line with previous studies on rhotics (e.g., Willis & Bradley, 2008; Bradley & Willis, 2012; O’Brien, 2013), the present study assumes that the acoustic measurement conducted by the researcher is scientifically solid enough to draw conclusions from the data analyses.

118 Hualde (2005), Solé (2002), and Willis (2007), rhotics were only considered trills when two or more closures were visible in the spectrogram. In line with previous acoustic research (Díaz-

Campos, 2008, p.49; also see Bradley, 2006, p.12), two lingual contacts were chosen as the minimum number of contacts required for the segment to be considered a normative trill. It is well established that three-contact trills are generally infrequent in Spanish varieties (Bradley,

2006).

Lastly, following Zhou et al. (2008) and Olsen (2012), approximants with a bunched or retroflex manner of articulation were distinguished by measuring the average F4 and F5 distance.

This was obtained by averaging the F4 and F5 frequencies at three of four even intervals, depending on the segment length, and then subtracting the F4 average from the F5 average. An average F5 – F4 distance of 700 Hz or less indicated a bunched approximant, whereas a distance of 1400 Hz or more indicated a retroflex approximant (Zhou et al., 2008).

3.5 Results

3.5.1 Rhotics in Northern Belizean Spanish

The data revealed that in NBS, the intervocalic rhotic contrast is generally maintained through the systematic use of the tap and a retroflex approximant /ɻ/ variant. In NBS, the tap appears in intervocalic position. In contrast to normative Spanish, however, the retroflex approximant appears in canonical trill contexts. Figure 3-4 shows a NBS canonical intervocalic tap in the word corazón ‘heart’, characterized by a single closure lasting 25 ms.

119

Figure 3-4. Canonical tap in corazón ‘heart’ produced by speaker S2

The most salient phonological feature of NBS was the retroflex variant /ɻ/, which may be the phoneme that contributes to this Spanish variety’s “apparent English accent” (Hagerty, 1996, p.137). Importantly, this variant was not always produced as retroflex, an issue I discuss in more detail in the following section.

It is notable that since NBS speakers routinely engage in Spanish/English CS (Balam,

2013; Brockmann, 1979, p.175), the retroflex approximant is employed in both English and NBS lexemes in mixed discourse. In this regard, we can see the parsimonious use of rhotics, as the intervocalic rhotic contrast is maintained in NBS, but the retroflex approximant variant is also employed in Spanish/English CS. Although there is a popular belief that this variant is primarily used by L2 learners of Spanish, the present data add to the findings from previous work

(Hagerty, 1979), which show that native speakers of Spanish also employ the retroflex approximant /ɻ/.

120 Figure 3-5 exemplifies the NBS intervocalic retroflex approximant, which clearly appears as a vowel-like segment. English-like rhoticization is acoustically evidenced in the lowering of the third formant. In the acoustic literature, this drop in the third formant is what distinguishes bunched and retroflex approximants produced by speakers of American English (Bullock &

Gerfen, 2004; Hagiwara, 1995; Ladefoged & Maddieson, 1996; Stevens & Blumstein, 1975;

Olsen, 2012; Zhou et al., 2008). Thus, in contrast to other varieties of Spanish where other variants such as the assibilated rhotic (e.g., Bradley, 1999; Vásquez Carranza, 2006) or the fricative trill (Colantoni, 2006) appear in the intervocalic trill position, NBS speakers employ an approximant rhotic variant. In the case of agarró ‘caught’ below, the average F5 – F4 distance was 1,601 Hz; thus indicating a strong retroflex manner of articulation.

F5

F4

Figure 3-5. Retroflex approximant in agarró ‘caught’ produced by speaker F1

121 A distinguishing feature between the retroflex approximant and the bunched variant is the

F5 – F4 distance. As Figure 3-6 below illustrates, in bunched approximants, there is visibly less

F5 – F4 distance. In the case of cigarro ‘cigarette’, the average F5 – F4 distance was 667 Hz.

F5 F4

Figure 3-6. Bunched approximant in cigarro ‘cigarette’ produced by speaker S4

Overall, trill production was very low. In contrast to Costa Rican Spanish, where an incidence of 2-3% of trill production has been reported (Umaña Aguilar, 1981), none of the ten speakers in the current study produced a canonical trill in elicited oral production; thus confirming that in NBS, the trill is not used, especially in more spontaneous speech. In the read aloud data, only three trills were produced, in each case by a different speaker in the senior high school group. In intervocalic tap position, a non-canonical, two-closure trill was produced in the word morada ‘purple’. The other two normative trills were produced in word-initial position, for

122 the word rana ‘frog’ by S1, the speaker who code-switched the most, and by speaker S2 (see

Figure 3-7). The highly infrequent use of the trill is similar to O’Brien’s (2013) findings.

Figure 3-7. Two-closure trill in rana ‘frog’ produced by speaker S2

Acoustic analysis of the read-aloud data confirmed that the phonetic context where the

NBS system seems most vulnerable is the intervocalic position where speakers sometimes produced a retroflex approximant or tap to the same degree, a pattern which is further discussed in the following section.

Notably, in word-final position, the tap was in free variation with a null rhotic. of the rhotic was pervasive, particularly in uninflected light or main verbs (see Figure 3-8).

Impressionistic analysis of 251 infinitive verb tokens from the elicited oral production data showed that deletion of the word-final rhotic occurred 59% of the time (e.g., [a.ˈsɛ] hacer ‘to do,

[in.bi.ˈta] invitar ‘to invite’, [sa.ˈli] salir ‘to go out’). The deletion of the tap in infinitive verbs is not common in Spanish varieties, but it has been attested in Cantabrian Spanish in Northern

123 Spain (Penny, 1969, cited in Bradley, 2005), Ecuadorian Spanish (Bradley, 1999) and Central

Valley Costa Rican Spanish (Vásquez Carranza, 2006). It is also a characteristic feature of Afro-

Hispanic varieties of Spanish (Hualde et al., 2010).

Figure 3-8. Null rhotic in uninflected light verb hacer ‘do’ in bilingual light verb construction asé come down ‘to come down’

Lenition of the word-final tap in NBS can be analyzed both as a means to reduce articulatory effort and as a way to maintain prosodic continuity in bi/multilingual speech. If the word-final tap is elided in infinitive verbs, this implies that there is no occlusion to interrupt the prosodic flow of phonetic segments. For instance, prosodic continuity is achieved in bilingual light verb constructions, as in ‘asé come down’ in (12). In these hybrid constructions (see

Chapter 4), the Spanish light verb hacer ‘do’ is typically pronounced as [a.ˈsɛ] and not [a.ˈsɛɾ].

124 This facilitates the flow of sounds and words between Spanish and English in these hybrid grammatical structures, particularly in cases where the initial phoneme of the lexical verb is an

English approximant rhotic (e.g., asé register, asé reach, asé reinforce, etc.). Thus, the lenited tap facilitates the more rapid transition between segments in switched discourse.

There were only few instances where the retroflex approximant was overgeneralized. A freshman, high school participant overgeneralized the retroflex approximant in the word caracol

‘shell’ in the read-aloud task. Another speaker employed the retroflex variant three times with verbs in elicited production. However, these were isolated cases. Also worth mentioning is that in NBS speakers’ oral production, there was no instance of the voiceless English affricate /tʃ/ in

Spanish words such as otro ‘another’ or tres ‘three,’ a realization that has been attested in both highland Ecuadorian Spanish and Central Valley Costa Rican Spanish.

Overall, results confirmed that the retroflex approximant rhotic continues to be a salient sound of NBS. This raises the question whether its use is also overgeneralized in intervocalic contexts; thus, leading to a phonological merger in semi-spontaneous speech or reading. We examine this issue in the ensuing section.

3.5.2 Neutralization of the NBS Intervocalic Rhotic Contrast

To examine neutralization of the intervocalic rhotic contrast in NBS, I first present the results from elicited oral production. Subsequently, I focus on the data from the read-aloud task.

3.5.2.1 Elicited oral production

Acoustic analysis of 193 taps and 137 approximants showed that in elicited oral production, speakers successfully maintained their native variety’s divergent tap/approximant contrast. In the intervocalic tap context, speakers consistently produced the tap with a complete or partial closure. In fast speech, taps were sometimes produced as lenited taps. As Figure 3-9 illustrates, a lenited tap is a continuant rhotic realization (Rose, 2010), also described in the

125 literature as a perceptual tap (Willis & Bradley, 2008; Bradley & Willis, 2012), which lacks a visual closure. In these cases, the reduction of intensity was useful in isolating the tap in order to measure its duration. This reduction of intensity reveals that even though the closure element is is not produced, a continuant rhotic segment is nonetheless present. The acoustic analysis revealed that the mean duration for the intervocalic NBS tap across speakers was 26.7 ms (see

Table 3-4).

Figure 3-9. Lenited tap in miró ‘saw’ produced by speaker F1

An English-like approximant was also produced in elicited production, although variation was attested. In contrast to Calvo Shadid and Portilla Chaves (1998), who reported that retroflex allophones accounted for 54% of their total occurrences in intervocalic trill position, English-like approximants in the present dataset accounted for 94% of the intervocalic trill tokens, which suggests that the use of an English-like approximant is more stabilized in the Northern Belize context than in the Costa Rican context. The ‘overgeneralized’ tap was attested in 9 out of 146

126 target retroflex approximants (see Table 3-4). In general, the production of overgeneralized taps was minimal. A McNemar’s chi-squared test confirmed that there was a significant difference in the proportions of NBS taps produced and the number of overgeneralized taps produced in elicited production [χ2 (1) = 88.3, p = .000]. Thus, target taps were produced as NBS taps and not as taps in non-target phonetic environments.

Table 3-4. Rhotic production in elicited oral production Speaker Tokens Tap Tokens with Tap Tokens with Mean with NBS duration overgeneralized duration retroflex F5-F4 tap in ms tap in ms approximant distance in Hz F1 25 25.6 0 0 18 1048 F2 14 28.7 2 24.0 9 1497 F3 15 23.5 0 0 14 1529 F4 13 24.9 0 0 5 1262 F5 30 27.6 0 0 27 1583 S1 15 25.4 1 32.0 13 974 S2 28 27.0 2 24.0 11 1072 S3 16 28.4 4 28.7 9 1258 S4 13 27.3 0 0 14 992 S5 24 27.9 0 0 17 1270 Total 193 26.7 9 27.1 137 1248

In terms of mean duration across speakers, little variation was found. Speakers generally produced mean durations longer than 20 ms (i.e., the mean reported by Quilis, 1993). In the case of speakers F2, S1, S2 and S3, although there was more variation in speakers’ individual average durations, the mean duration for the NBS tap and the overgeneralized tap across speakers were similar (i.e., 26.7 and 27.1 respectively). Note that overgeneralized taps were produced both by speakers who reported a higher proficiency in BK than SP (i.e., F2 and S3) and by speakers who reported being more proficient in SP (i.e., S1 and S2). It may be that this overgeneralization occurred as a result of the semi-experimental nature of the tasks. In future work, sociolinguistic interviews should be carried out to further overcome the “Observer’s Paradox” (Labov, 1966),

127 which remains a challenge for researchers investigating vernacular speech, especially in communities where varieties have been or are stigmatized14 (Wardhaugh, 2006, p.19).

In relation to the intervocalic approximant variant, it is noteworthy that there does not seem to be a correlation between BK and degree of retroflection. Although speakers F3 and F5 were not proficient in BK, they nonetheless produced the approximants with the highest degree of retroflection. Thus, we cannot attribute the degree of retroflection to proficiency in BK. In this regard, results from this study are contrary to Zimmer’s findings (2011, p.203), where a higher degree of retroflection was reported for Afro-Costa Rican, /Spanish bilinguals than for Hispanic Costa Rican Spanish speakers15. The mean F5 – F4 distance for the current group of speakers was 1248 Hz, which is considerably greater than the distance of 1057 Hz reported by Olsen (2012), for his participants who were native English speakers learning L2

Spanish.

Crucially, none of the consultants’ overall production of approximant rhotics, in terms of mean F5 – F4 distance, showed a bunched manner of articulation. At the same time, not all speakers produced approximants with a retroflex manner of articulation. In fact, only three speakers produced intervocalic approximant rhotics with an overall mean F5 – F4 distance of

1400 Hz or more (see Table 3-4). Thus, contra the idea that NBS speakers produce retroflex approximants (Hagerty 1996), the acoustic analysis revealed that speakers generally produced rhotics that are neither clearly retroflex nor bunched. For example, speakers F1 and S2 produced approximants that were intermediate in manner of articulation, with a mean F5 – F4 distance of

14 The retroflex approximant is a stigmatized variant, often associated in the literature with speakers that have low levels of formal education (Adams, 2002; Vásquez Carranza, 2006).

15 Zimmer’s contention, however, is mostly based on impressionistic analyses. It is not clear how she determined degree of retroflection in her data.

128 1048 Hz and 1072 Hz respectively. These results may be indicative of an intermediate approximant rhotic that is unique to the NBS variety. Alternatively, in line with Olsen’s (2012) findings, it may simply be that retroflex approximants employed by native or native-like speakers of English have a degree of retroflection that is actually lower than 1400 Hz.

In summary, the elicited oral production data showed that contra Hagerty’s (1979) observation, adolescent consultants maintained their native variety’s intervocalic tap/approximant contrast, and overgeneralization of the retroflex approximant in target tap contexts was minimal. For instance, speaker S4 produced the retroflex approximant in the word empezaron ‘started’ twice and in the word llamaron ‘called.’ Tap overgeneralization was also minimal, accounting for 4% of the target intervocalic approximant rhotics. Based on the mean durations of the NBS tap and the overgeneralized tap produced by speakers F2, S1, S2 and S3

(see Table 3-4), however, the data revealed that speakers did not acoustically differentiate these two variants. Thus, neutralization was attested but in a small percentage of the elicited oral production data.

3.5.2.2 Read-aloud

In the read-aloud data, it was also found that speakers produced NBS taps in their target contexts. Results from an item analysis that was carried out showed that the NBS intervocalic tap was produced 95% of the time in the read aloud task (see Table 3-5). The other variants included and overgeneralized retroflex approximants in the words caracol ‘seashell’16 and orilla

‘edge’.

Mean durations ranged from 24 to 32 ms. For caracol ‘seashell’ and amarillo ‘yellow,’ the mean durations of the NBS tap were substantially longer than for the words corazón ‘heart’ and pero ‘but’. Only the latter tokens had mean durations similar to Blecua’s (2001) reported

16 In NBS, caracol refers to a ‘shell’ or ‘seashell.’

129 average duration of 23 ms for the normative tap in Peninsular Spanish. Overall, however, the mean tap duration across tokens was 27.7 ms. Similar to the pattern observed in the elicited oral production data, there was minimal tap overgeneralization.

Table 3-5. Phonetic distribution of the intervocalic tap in reading Target Total NBS Other Tap mean words N Tap /ɾ/ Variants duration N N in ms pero (4)* 40 37 3 25 amarillo (2) 20 20 0 32 caracol (1) 10 8 2 31 morada (1) 10 8 2 28 miro (2) 20 20 0 29 quiero (2) 20 20 0 25 pajarito (2) 20 20 0 27 corazon (1) 10 10 0 24 orilla (1) 10 9 1 28 Total N 160 152 8 27.7 * Number in parentheses indicates the number of times the target word appeared in the read-aloud task.

On the other hand, phonetic instability was clearly evident in the intervocalic trill/ approximant context, where overgeneralization of the tap was attested (see Table 3-6). Although we would generally anticipate more careful, normative speech (Ramos-Pellicia, 2007) and the maintenance of phonological contrasts in reading, results revealed that particularly in reading consultants had difficulty maintaining either of their rhotic contrasts (i.e., native and standard

Spanish). Acoustic analysis of the read-aloud data confirmed that the NBS intervocalic tap was used to a similar degree as the NBS intervocalic approximant; thus, the NBS approximant rhotic was not predominantly employed in reading as in semi-spontaneous speech.

130

Table 3-6. Phonetic distribution of the intervocalic approximant rhotic in reading Target Total Approximant Mean Overgeneralized Mean words N rhotic F5-F4 tap Duration N Distance N in ms in Hz perro (5) 50 24 1272 26 26 corrió (1) 10 3 1182 7 29 arriba (1) 10 3 1405 7 32 arroz (1) 10 5 1630 5 32 cerrado (1) 10 2 1600 8 26 Total 90 37 1418 53 29.0 * Number in parentheses indicates the number of times the target word appeared in the read-aloud task.

In 41% of the tokens, the intervocalic rhotic was produced as a NBS approximant, whereas in remaining 59%, the intervocalic approximant rhotic was realized as an overgeneralized tap, as in cerrado ‘closed’ (Figure 3-10), with a visual closure, followed by no frication, post-approximantization and/or r-coloring. A McNemar’s chi-squared test considering phonetic realization (retroflex approximant and overgeneralized tap) and speaker group

(freshman and senior) showed that there was a significant difference in the proportions of overgeneralized taps versus approximant rhotics produced [χ2 (1) = 44.9, p = .000]; thus, revealing that the overgeneralized tap was produced significantly more than the target approximant rhotic in the read-aloud task.

In previous work (Kouznetsov & Pamies, 2008, p.674), it has been shown that Spanish trills are sometimes produced with a single closure followed by a vocalic component before the next vowel (Henriksen & Willis, 2010). However, as illustrated by Figure 3-10, there is no such vocalic component before the open, /a/; thus, the NBS tap and the overgeneralized tap are strikingly similar not only auditorily but acoustically as well.

131

Figure 3-10. Overgeneralized tap in cerrado ‘closed’ produced by speaker S2

Overgeneralized taps have been documented in the trill production of intermediate and advanced L1 speakers of American English learning L2 Spanish (Face, 2006; Rose, 2010). They have also been reported for some speakers of Afro-Costa Rican Spanish (Zimmer, 2011). In the case of the current study’s consultants, despite the differences in years of Spanish instruction or their language dominance, consultants did not differ in their ability to maintain the normative tap/trill phonemic contrast. Contrary to what was hypothesized, both groups showed similar patterns of phonetic realization. There was a tendency, however, for the senior group to produce overgeneralized taps more frequently. The senior group accounted for 60% of the overgeneralized taps.

In relation to the NBS intervocalic approximant, data from the read-aloud task showed a two-fold effect. First, there was a substantial decrease in the frequency of the approximant rhotic, from 94% to 41%. There was also a corresponding increase of more than 100 Hz in the overall mean degree of retroflection, more than likely as a result of more careful speech. A paired-

132 samples t-test revealed that there was a significant difference in the mean degree of retroflection for approximant rhotics produced in elicited production versus oral reading, t(8) = -2.608, p =

0.03 [Elicited production mean degree of retroflection = 1248 Hz; Oral reading mean degree of retroflection = 1352 Hz].

Most importantly, an increase in tap overgeneralization was evidenced in the read-aloud data, which contributed to the attested partial neutralization of the intervocalic rhotic contrast.

Analysis of individual speaker data further confirmed that there was no difference between the

NBS tap and the overgeneralized tap in terms of segmental duration. Table 3-7 summarizes speakers’ tap durations across the two tasks. The elicited oral production data show that six out of ten participants maintained the NBS intervocalic tap/approximant contrast, whereas only two did so in the read-aloud task.

Table 3-7. Mean durations of taps across tasks Speaker Elicited oral production Read-aloud

NBS /ɾ/ overgeneralized NBS /ɾ/ overgeneralized /ɾ/ /ɾ/ Duration in SD Duration SD Duratio SD Duration SD ms in ms n in ms in ms

F1 25.6 4.5 0 0 33.5 12.4 0 0 F2 28.7 5.0 24.0 1.4 29.2 6.8 32.8 5.8 F3 23.5 4.4 0 0 24.9 5.3 28.6 9.4 F4 24.9 4.0 0 0 27.3 4.6 25.3 4.6 F5 27.6 6.0 0 0 27.0 7.8 27.0 8.4 S1 25.4 3.3 32.0 5.9 25.1 3.9 26.9 5.1 S2 27.0 6.0 24.0 0 30.2 8.7 32.7 7.8 S3 28.4 7.3 28.7 0 31.4 4.3 30.2 2.3 S4 27.3 6.1 0 0 23.1 5.3 23.3 3.8 S5 27.9 5.7 0 0 23.3 2.9 0 0

A 4 (condition: elicited production tap, elicited production overgeneralized tap, read- aloud tap, read-aloud overgeneralized tap) by 2 (speaker type: freshmen vs seniors) repeated

133 measures ANOVA revealed a main effect for condition, F(1.9, 15.4)= 8.474, p = .004, no main effect for speaker type, F(1,8)= .507, p =.497, and no condition by speaker type interaction,

F(1.9, 15.4)= 1.436, p =.267. When considering the entire data set, a Bonferroni post-hoc analysis revealed that the NBS tap and the overgeneralized tap were significantly different from each other in elicited production (p = .033) but not in the read-aloud task (p > .05), thus confirming that it is particularly in the reading task that tap overgeneralization led to neutralization of the normative tap/trill contrast and partial loss of the speakers’ native variety’s intervocalic rhotic contrast.

Since only four speakers in elicited oral production and eight in the read-aloud task produced overgeneralized taps, paired-samples t-tests were subsequently carried out to determine if there were significant differences between the means of the NBS and overgeneralized taps that were produced by the these speakers. The paired-samples t-tests revealed that the mean durations of the NBS taps and the overgeneralized taps produced by speakers F2, S1, S2 and S3 in elicited oral production were not significant, t(3) = .080, p = .941 [NBS tap mean = 27.4 ms; overgeneralized tap = 27.1 ms]; and no statistical difference was found between the means for the taps produced by speakers F2, F3, F4, F5, S1, S2, S3 and S4 in the read-aloud task, t(7) = -

1.414, p = .200 [NBS tap = 27.3 ms; overgeneralized tap = 28.4 ms]. Furthermore, no statistical difference was found between freshmen and senior speakers in their production of the NBS taps and the overgeneralized taps, p > .05.

Thus, in the read-aloud data, insofar as the intervocalic tap/trill contrast is concerned, eight out of ten speakers did not consistently maintain either their native Spanish variety’s contrast or the normative intervocalic rhotic contrast. In terms of segmental duration, there was evidence of neutralization in the production of the NBS tap and the overgeneralized tap. Figure

134 3-11 illustrates the tap durations in the minimal pair pero/perro ‘but/dog,’ which are 22 ms and

18 ms respectively. In Figure 3-12, the non-continuant taps in cerrado ‘closed’ and caracol

‘seashell’ lasted 23 ms and 25 ms respectively. Note that tap duration differences are minimal, and there is no intervening vocalic component between the rhotic and the vowels that follow.

Figure 3-11. Speaker S2’s neutralized production of pero ‘but’ and perro ‘dog’

135

Figure 3-12. Speaker S3’s production of cerrado ‘closed’ and caracol ‘seashell’

In the read-aloud data, the use of the overgeneralized tap was also attested in word-initial position, where either a trill or a retroflex approximant is expected. Figure 3-13 shows an overgeneralized tap, lasting 27 ms, in word-initial position.

136

Figure 3-13. Overgeneralized tap in la rata ‘the rat’ produced by S1

In sum, the data showed that there were marked differences between intervocalic rhotic realizations in elicited oral production versus reading. While in elicited oral production, the NBS tap/approximant phonemic contrast was more consistently maintained, in the read-aloud task, there was a decrease in the frequency of the NBS approximant rhotic and an increase in the use of the overgeneralized tap, which resulted in partial loss of the intervocalic phonemic contrast.

Participant groups did not significantly differ in their phonetic realizations. Freshman participants did not exhibit more phonetic variation and/or greater difficulty in maintaining the intervocalic rhotic contrast as it had been hypothesized. Speakers from both groups overgeneralized the NBS tap in the intervocalic approximant context. In the read-aloud data, the mean segmental durations for the NBS tap and the overgeneralized tap (i.e., 27.5 ms vs. 28.4 ms) showed that there was a striking similarity between the two taps. Thus, neutralization of the

137 normative tap/trill contrast led to the partial loss of the NBS intervocalic tap/approximant contrast. As to why this may have occurred is further elaborated in the following section.

3.6 Discussion

I set out to determine whether there was evidence of phonological convergence vis-à-vis the intervocalic rhotic contrast among adolescent speakers of NBS. Acoustic analysis of data collected from ten consultants showed that speakers maintained their native variety’s divergent intervocalic tap/approximant contrast in elicited oral production but not in oral reading. Contra

Hagerty’s (1979) claim that in NBS, the phonemic merger of intervocalic rhotics is in the direction of the retroflex approximant, results revealed that the NBS intervocalic rhotic contrast is maintained in semi-spontaneous speech. Overgeneralization of the approximant phoneme accounted for less than 1% of the entire data set. In the read-aloud data, however, speakers overgeneralized the NBS tap, and this contributed to variable neutralization (i.e., neutralization of the normative tap/trill contrast and partial loss of speakers’ divergent intervocalic rhotic contrast). These results could simply be dismissed as another case of “near-neutralization” (for further discussion, see Silverman, 2012, p.69), where variability in the maintenance of a phonemic contrast sometimes occurs as a way to salvage the listener from semantic ambiguity.

However, given the marked differences between the elicited oral production data versus the read- aloud data, there is no doubt that NBS speakers’ rhotic production warrants more careful consideration of relevant social and cognitive factors. In the ensuing sections, I discuss factors that may explain the variable neutralization attested in the present study. I also discuss the implications that findings have on our current understanding of convergence.

3.6.1 A Closer Look at Task Type and Phonetic Outcomes

One possible explanation that can account for the variable neutralization of the intervocalic rhotic contrast in the read-aloud data is that speakers’ phonetic production may have

138 been affected by their heightened awareness of pronunciation. From the works of Labov (1966), and researchers such as Tarone (1983) and Elliott (2003), we know that when a bilingual speaker is aware that he/she is being observed, he/she employs a more conservative speech style. In the

Northern Belize context, this awareness of pronunciation would become even more pronounced, given that oral reading in monolingual Spanish is specifically associated with classroom discourse practices.

In previous work, it has been shown that the nature of a task does have an effect on speakers’ phonetic production. For example, in her study on rhotic production in coda position,

Ramos-Pellicia (2007, p.55) found that Lorain Puerto Rican Spanish speakers used the normative rhotic variant in “their most monitored form of speech.” In more informal conversation, however, speakers employed the less prestigious variant, the lateral rhotic [l], a distinguishing yet stigmatized feature of Puerto Rican Spanish (e.g., hablar ‘to speak’ is pronounced as

[a.ˈβlal]).

In the present study, a similar pattern was attested in the use of the native rhotic variant.

In elicited oral production, which is less monitored than reading-aloud, speakers used the NBS approximant to maintain their variety’s intervocalic rhotic contrast. However, the more formal nature of the read-aloud task triggered more conservative speech, which was evidenced in (i) speakers’ attempt to switch to more normative patterns of rhotic distribution and (ii) an increase in the degree of retroflection.

The fact that some speakers produced the overgeneralized tap even in elicited speech suggests that even elicited oral production may have been perceived as semi-formal (i.e., picture stories can be associated with classroom teaching practices); consequently, speakers did not categorically employ their native variety’s intervocalic rhotic contrast. We can surmise that if the

139 acoustic analysis had been based on naturalistic speech alone, obtained via sociolinguistic interviews, the use of the NBS approximant rhotic would have been more consistent.

Changes in speakers’ phonetic production according to task type bring to the fore the importance of analyzing acoustic data from multiple sources, as different tasks may lead to different phonetic outcomes. Analyzing both (semi-)spontaneous and reading data is crucial to our understanding of speakers’ multidimensional phonological patterns, as it provides a more holistic picture of how bi/multilinguals’ phonologies change depending on the kind of discourse

(i.e., more naturalistic versus less naturalistic). Note that had only semi-spontaneous data been analyzed, the acoustic analysis would have revealed little to no evidence of phonological convergence, since it is the reading data that provided evidence of variable neutralization.

A pertinent question that remains unanswered is whether adolescent NBS speakers are able to produce the apico-alveolar trill or not. It may be that the production of the overgeneralized tap in the intervocalic trill context occurred as a result of articulatory undershoot, which occurs when a speaker intends to produce a phoneme but fails to reach the intended articulatory target (Jaworski & Gillian, 2011; Liljencrants & Lindblom, 1972). This would be supported by the fact that it is particularly when speakers engaged in more monitored speech that their phonetic performance changed, indicating that they did in fact try to switch to a more standard variety of Spanish. However, given consultants’ emergent Spanish bidialectalism, speakers were unable to successfully maintain the normative tap/trill intervocalic contrast. In other words, speakers may have tried to produce the normative trill in the reading task, but they produced an overgeneralized tap more than 50% of the time instead, which in terms of segmental duration, did not differ from the NBS tap.

140 In contact acquisition studies, it has been shown that children successfully acquire phonological contrasts, and that they have impressive control over their ability to switch not only between languages but also between language varieties. Khattab (2009), for example, showed how three /English bilingual children living in England accommodated their phonetic patterns to their mother’s Arabic-accented English by making detailed phonetic adjustments to their utterances. There is compelling evidence, however, which shows that as children mature, it becomes increasingly more challenging for them to acquire phonological features not present in their first language (Chambers, 1992; Trudgill, 1981). The older speakers are when they acquire a second language, the less native-like their accent is in that second language (for relevant work, see Lenneberg, 1967; Neufeld, 1980; Oyama, 1976; Patkowski, 1990; Scovel, 1969, 1988;

Thompson, 1991, and references therein).

In view of phonetic accommodation, the results from this study suggest that adolescent speakers of NBS, insofar as the intervocalic rhotic contrast is concerned, seem to be unable to switch from non-standard phonology to standard , which requires the tap/trill phonemic contrast instead of the tap/approximant contrast. This may be related to limited formal instruction in standard Spanish. Thus, the fact that adolescent NBS speakers are emergent bidialectal speakers of Spanish may be the primary reason why they are seemingly unable to maintain their rhotic intervocalic contrast, specifically when they try to accommodate to a more standard monolingual Spanish mode. In view of the fact that only two normative trills were produced in the read-aloud task, it may be that only at higher proficiency levels do NBS speakers produce the distinctive tap/trill phonological contrast found in most Spanish dialects.

At the same time, however, we must consider the unmarked status that standard Spanish has in Northern Belize and speakers’ predispositions to their languages (Balam, 2013; Balam &

141 Prada Pérez, in press), as the patterns observed in the read-aloud data may also be indicative of the role that attitudinal factors play in speakers’ choice and ability to produce the tap/trill contrast. Recall that even though speakers primarily learn standard Spanish in the classroom, they receive ample exposure to normative forms via Mexican Spanish media and interpersonal communication with speakers of other varieties of Latin American Spanish that do employ the

Spanish trill.

In the following section, I elaborate on the potential role that social-indexical factors may have played in the phonetic patterns attested.

3.6.2 Social-Indexical Factors in Northern Belize

The social nature of Spanish bidialectalism in Northern Belize must be considered as an important factor that contributes to speakers’ speech patterns, and thus, their phonetic capabilities. Whereas Spanish/English CS is unmarked in Northern Belize, the use of standard

Spanish is marked (Balam, 2013; Koenig, 1975) and adolescents’ attitudes toward Spanish are increasingly becoming negative (Balam & Prada Pérez, in press). We must question, therefore, how the marked status of standard Spanish may have affected speakers’ speech patterns.

A key factor that could have had an effect on phonetic production is the interlocutor’s and speakers’ shared linguistic and ethnic background. We must bear in mind that the accepted linguistic norms of a community constitute a fundamental aspect of the social knowledge speakers have of how language varieties are and should be used (Myers-Scotton, 1993). Thus, speakers’ non-usage of the trill and subsequent variable neutralization could have been triggered by speakers’ attempt to conform to expected linguistic norms (i.e., avoidance of standard

Spanish).

In Communication Accommodation Theory (CAT), Giles and colleagues (1973, 1991) posit that speakers often unconsciously use strategies to negotiate the ideal social distance they

142 want to have with interlocutors, and thus, their changes in accent and speech patterns are driven by motivations such as the desire for social acceptance, maintenance of linguistic identity, etc.

Trudgill (1986) underscores that dialect accommodation especially takes place when a sound is a salient marker of identity, and speakers are acutely aware of these linguistic markers.

In Spanish/English bilingual contexts, the ability to articulate the normative trill is typically associated with ‘standard’ Spanish and the ‘correct’ way of speaking Spanish (for this belief among teachers of Spanish in Belize, see Balam & Prada Pérez, in press). At the same time, variants such as the retroflex approximant and the assibilated rhotic are stigmatized, but they nonetheless persist due to their status as markers of local or regional identity (Adams, 2002;

Vásquez Carranza, 2006).

The phonetic patterns of rhotic production observed in the reading task may have occurred as a direct result of two accommodation strategies in conflict. In CAT, the divergence strategy is used to reinforce linguistic differences; in this case, the nature of the reading task would have inevitably caused students to modify their phonetic patterns so that they would seem more ‘standard’ and ‘formal’. However, since the study’s participants and the interlocutor shared the same ethnic and linguistic identity, divergence would have been in direct conflict with convergence and maintenance strategies. This analysis would be supported by previous observations that among NBS speakers, language choice is affected by the perceived ethnicity of the speaker (Brockmann, 1979; Koenig, 1975, 1980). Recall that in the Northern Belize context,

CS is unmarked, while the use of monolingual varieties is marked. Thus, the use of standard varieties among Northern Belize Maya/Mestizos is generally disparaged, especially among adolescents (Balam & Prada Pérez, in press; Koenig, 1975).

143 It can be argued that in an effort to conform to linguistic norms and maintain a shared linguistic and ethnic identity with the interlocutor (i.e., solidarity), adolescent NBS speakers may have tried to reduce the social distance between themselves and the interlocutor by keeping some of the rhotic distributions of their home language. In the process, speakers unconsciously tried to avoid normative Spanish patterns, while also trying to sound as ‘standard’ as possible. It has been shown in previous sociophonetic work that sometimes bilinguals can choose to realize a category or not, and this occurs as a means of indicating ethnic affiliation (Hirson & Sohail,

2007) or as a way of distancing oneself from a certain type of stigmatized variant or accent

(Simonet, 2010). For example, Ramos-Pellicia (2007) points out that since the rhotic variant [l] is negatively perceived in Island Puerto Rican Spanish (IPRS), first generation Lorain Puerto

Rican Spanish speakers in her study used the retroflex approximant instead of the lateral rhotic variant in order to dissociate themselves from the stigmatized lateral variant.

In the current study, it may be that the speakers wished to distance themselves from standard Spanish because of prevalent negative attitudes to standard Spanish among adolescents.

If the trill is associated with standard Spanish, to which speakers are negatively predisposed, then more than likely, they will be more hesitant to produce it, especially if the speaker and the interlocutor are from the same speech community, and are thus, equally aware of the community’s established linguistic norms. Recent research conducted with Belizean teachers of

Spanish suggests that the trilled /r/ is negatively perceived among Northern Belize Maya/Mestizo adolescents (Balam & Prada Pérez, in press). Thus, the use of standard Spanish phonology is one that clearly many Maya/Mestizo adolescents do not embrace even in the classroom context.

It is this negative attitude toward the use of standard Spanish among native Northern

Belizean Maya/Mestizo students that may partly explain why students choose not to use

144 canonical apico-alveolar trills. If the rejection of standard Spanish phonological forms occurs even in the Spanish classroom, then we can surmise this rejection may become even much more pronounced in informal contexts. Negative attitudes towards standard varieties have been attested in other Creole contact situations. For instance, Craig (1983), who conducted extensive research on Jamaican Creole/English speakers, emphasized that part of the problem in teaching the standard variety to non-standard speakers of English is that they have negative attitudes toward Standard English, particularly its phonology.

Overall, the present study has revealed that despite the drastic increase in the frequency of Spanish/English CS in Northern Belize in the last four to five decades (Balam, 2015, 2016b), more intense language contact and greater proficiency in English and BK among younger generations have not led to phonological convergence in NBS rhotics. Had this been the case, adolescent speakers would not have maintained their native variety’s intervocalic tap/approximant contrast, but they did. In semi-spontaneous speech, consultants generally maintained their native variety’s intervocalic rhotic contrast. It was only in oral reading that there was variable neutralization (i.e., neither the divergent nor the normative intervocalic rhotic contrast was maintained).

These findings suggest that convergence is a complex, multidimensional phenomenon that must be analyzed in light of speakers’ bidialectal and bi/multilingual linguistic repertoire, different types of discourse and relevant cognitive (e.g., age, emergent versus dynamic bi/multilingualism, etc.) and social factors (e.g., status of languages, speakers’ attitudes, etc.).

Although no evidence was found for the presence of a converged retroflex approximant, as suggested by Hagerty (1979), a certain level of synchronic convergence was attested in reading,

145 which I attribute both to adolescents’ emergent Spanish bidialectalism and other social factors such as speakers’ attitudes toward standard Spanish.

In this regard, therefore, convergence is not analyzed as an ‘optimization strategy’ (in the sense of Muysken (2013) or Toribio (2004)), but as a natural outcome of speakers’ emergent

Spanish bidialectalism. There is the possibility that older Northern Belize Maya/Mestizos that have more fully developed their multilingual repertoire are able to phonetically switch from divergent to normative patterns of rhotic distribution in naturalistic speech. Based on this study’s findings, however, it remains inconclusive whether older consultants from Northern Belize would indeed successfully switch from a divergent to a normative intervocalic rhotic contrast in more formal discourse. It could be the case that some speakers would also produce patterns of variable neutralization. Further work needs to investigate the maintenance of the intervocalic rhotic contrast among older bi/multilinguals from Northern Belize.

One potential explanation that can explain why there was no strong evidence of diachronic convergence between Spanish and English intervocalic rhotics in NBS is the status of languages. Despite the favorable change in the status of BK, and higher proficiencies in English and BK among younger generations, NBS has undoubtedly remained as the dominant language in Northern Belize (see section 1.1 in Chapter 1). Although there has been a radical increase in the frequency and density of Spanish/English CS in the last four to five decades, the status of languages in Northern Belize has not created the sociolinguistic conditions that are most conducive to diachronic convergence. Myers-Scotton (2002, p.172) highlights that convergence especially takes place when a dominant language that has more socioeconomic prestige impinges on the grammar and lexicon of a subordinate language. In the last few decades, the process of

146 English “impinging” on NBS has simply not been favored by the marked status that English has maintained in Northern Belize informal discourse.

3.7 Concluding Remarks

In sum, the data from NBS partially support Hammond’s (1999) suggestion that in

Spanish dialects, there is neutralization of the intervocalic rhotic contrast. The findings revealed that variable neutralization is a possibility, especially in contexts where speakers are still in the process of becoming bidialectal speakers of Spanish and are unable to phonetically switch from one Spanish variety to the other under certain social circumstances. In the case of Northern

Belize, it may be that only after years of formal instruction in a Spanish-speaking country and/or continuous naturalistic exposure to varieties such as Mexican Spanish that NBS speakers develop the necessary phonological competence in standard Spanish, which would allow them to successfully switch from NBS to standard Spanish. Future investigation should examine whether in cases such as these, NBS speakers are able to successfully maintain the tap/trill contrast in reading and more naturalistic discourse. Such a study could contribute to our current understanding of phonetic outcomes among native Spanish speakers who acquire/learn normative patterns of rhotic distribution and realization in post-adolescence. Lastly, future acoustic studies on NBS need to examine other age groups, and further investigate the role of proficiency in BK in relation to the maintenance of the intervocalic rhotic contrast.

3.8 Limitations

Despite the present study’s acoustic analysis of segmental duration, it remains unclear whether in NBS other features are used to maintain this phonological contrast (e.g., of vowel preceding or following rhotic). Future work could explore these features. Future studies will also need to take into consideration certain variables that were not controlled for in the present study. For instance, even though commonly used nouns in NBS were used in the reading

147 task, the frequency of these words containing the target phonetic contexts was not systematically controlled for. This could have potentially led to frequency effects, as some words may not have been frequently or commonly used by certain speakers (e.g., Kriol-dominant speakers). The use of a familiarity scale, in which speakers rate how familiar they are with words, or self-ratings on frequency of use of lexical items could have mitigated frequency effects on speakers’ phonetic speech patterns (R. Wayland, personal communication, April 28, 2016). Last but not least, although social factors (e.g., speakers’ attitudes) are argued to affect speakers’ phonetic production, a quantitative analysis of both linguistic and social factors is necessary to either substantiate or refute the contentions made in the present analysis.

148 CHAPTER 4 BILINGUAL LIGHT VERB CONSTRUCTIONS

4.1 Introduction

In this chapter, I examine the use of bilingual light verb constructions (henceforth

BLVCs) among adolescent, post-adolescent, and adult Northern Belize bi/multilinguals. The chapter consists of two inter-related studies, which together provide an investigation of these bilingual constructions vis-à-vis three restrictions that have been proposed to constrain their use in mixed speech; namely, (i) verb frequency (Fuller Medina, 2005; Jenkins, 2003; Reyes, 1982),

(ii) stativity (Fuller Medina, 2005) and (iii) passivization (González-Vilbazo & López, 2011). In

Study 1, I analyze Northern Belize code-switchers’ acceptability judgments of 553 BLVCs, extracted from semi-structured interviews with 38 consultants, in order to determine whether verb frequency and stativity constrain the incorporation of ‘hacer + V’ in Spanish/English CS1.

In Study 2, I quantitatively examine speakers’ judgments from another sample of 46 consultants in order to assess the compatibility that the light verb hacer ‘do’ has with stative and eventive passives.

I argue that in the Northern Belize context, relaxed language norms, positive attitudes towards bilingual language practices, and the development of CS as an identity marker are factors that contributed to the highly productive and innovative use of these constructions. The skillful use of ‘hacer + V’ among younger generations in present-day Northern Belize suggests that the creative use of these innovations is a more recent phenomenon, one which was catalyzed by the aforementioned sociolinguistic factors.

1 In the present investigation I use the terms BLVCs and ‘hacer + V’ interchangeably, as in Spanish/English contact, hacer ‘do’ is the only light verb used to incorporate English verbs in CS. In other contact situations, other light verbs such as be, go, come or stay play a similar role in mixed speech (e.g., Sarnami/Dutch BLVCs, see Muysken, 2000, p.198; for Panjabi/English BLVCs, see Romaine, 1995, p.133). These light verbs often co-exist with ‘do’ and may develop a specialized function to encode stativity or transitivity.

149 This chapter is organized as follows. In section 4.2, I first provide a general overview of the morphosyntactic structure under examination, and I situate the two inter-related studies within the larger aims of this dissertation. In section 4.3, I summarize previous descriptive and quantitative studies on ‘hacer + V’, especially those that relate to the restrictions being tested. In sections 4.4 and 4.5, I describe the methodology employed in the first study, and I present the results. In sections 4.6 and 4.7, I describe the methodology employed in the second study, and I present the results. Lastly, in sections 4.8 and 4.9, I provide a discussion of main findings from both studies and offer concluding remarks.

4.2 Bilingual Light Verb Constructions

Bilingual light verb constructions, otherwise known as ‘do-constructions’ (Myers-Scotton,

2002; Myers-Scotton & Jake, 2013; Versteegh, 2009), ‘bilingual compound verbs’ (Edwards &

Gardner-Chloros, 2007; Vergara Wilson & Dumont, 2015) or ‘mixed compound verbs’ (Chan,

2008) have been attested in typologically diverse language contact situations across the world

(for BLVCs involving different language pairs, see Muysken, 2000; Edwards & Gardner-

Chloros, 2007; Wichmann & Wolgemuth, 2008).

In Spanish/English CS, as (1) exemplifies, canonical BLVCs are hybrid morphosyntactic structures where the fully inflected Spanish light verb hacer ‘do’ co-occurs with an English infinitive verb, which contributes the semantic content (Balam, 2014, 2015, 2016a; Vergara

Wilson, 2013; Vergara Wilson & Dumont, 2015; Jenkins, 2003). In (1), hacer is the carrier of number, tense, aspect and mood features, whereas the ‘appreciate’ and ‘spend’ only contribute meaning to the BLVC. In Myers-Scotton’s and Jake’s (2000) terms, therefore, it is the

Spanish light verb that bears all of the outsider late system morphemes from the ML, whereas the lexical verb that is provided by the EL bears the semantic component of the hybrid content morpheme.

150

(1) Yo hago appreciate el time que hago spend con ella2

I do.1SG.PRS appreciate.INF the time that do.1SG.PRS spend.INF with her3

‘I appreciate the time that I spend with her.’

(PA15, female, 18)4

In monolingual Spanish, the closest equivalent to ‘hacer + V’ is the causative hacer construction (Balam, 2015, 2016a; Jenkins, 2003; Vergara Wilson, 2013). As the following examples from Mexican American Spanish illustrate, causation (i.e., somebody makes someone do something) is involved in (2), as there is a causer (i.e., He) and a causee (i.e., me) who performs the act of studying. This is not the case in (3), however, where the verbal structure contains only one thematic role (i.e., agent = daughter), and no causation is involved.

(2) Me hizo estudiar

CL.INDOBJ do.3SG.PST study

‘He made me study.’

(Pfaff, 1979, p.300)

(3) Su hija hace teach allá en San José

His daughter do-2SG.PRS teach there in San Jose

‘His daughter teaches there in San Jose.’

(Pfaff, 1979, p.301)

2 In code-switched utterances in this dissertation, normal font is used for English words. Spanish words are italicized, whereas Belizean Kriol words are emboldened.

3 Key to glosses: 1SG, 2SG, 3SG = 1st, 2nd and 3rd person singular; 1PL, 2PL, 3PL = 1st, 2nd and 3rd person plural; CL = clitic; F = feminine gender; FUT = future, IMP = imperfect tense; INF = infinitive; M = masculine gender; PASTPART = past ; PRET = tense; PROG = progressive tense; REFL = .

4 In linguistic examples from the Northern Belize corpus, within brackets appear the speaker’s number, speaker’s gender and speaker’s age.

151 Noteworthy is that in some CS contexts BLVCs do have a monolingual template in at least one of the component languages, as is the case of Bengali in Bengali/English CS

(Chatterjee, 2014, 2016) and Persian in Persian/English CS (Moinzadeh, 1999). In these contexts, the light verb construction is already a productive form in the monolingual variety, and the light verb may co-occur either with nominal or verbal elements. In (6), for example, the

Persian light verb kard ‘do’ co-occurs with a Persian infinitive bāzdid ‘visit’, but the sentence does not have a causative interpretation. Crucially, in the case of both Spanish and English, there is no structural equivalent that exactly parallels the periphrastic ‘hacer + V’ in Spanish/English

CS (Balam, 2015, 2016a; Jenkins, 2003; Wilson Vergara, 2013).

(4) Az šāhr - e - ma bāzdid kard

from city us visit do.PST

‘He visited our city.’

(Moinzadeh, 1999, p.49)

Although BLVCs have been attested in many CS varieties (e.g., Tamil/English:

Annamalai, 1989; Turkish/Dutch: Backus, 1996; Shona/English: Bernsten, 2000;

Bengali/English: Chatterjee, 2016; Hindi/English: Bhatia & Ritchie, 1996, 2016;

Panjabi/English: Romaine, 1995; Spanish/Maya: Suárez-Molina, 1996), their generally low frequency in bilingual corpora has precluded scholars from thoroughly investigating and/or understanding them (e.g., Pfaff, 1979: 3 tokens; Toribio et al. 2012: 12 tokens). Backus (1996, p.216) rightly observes that “[o]riginally the ‘do-construction’ did not receive a great deal of attention in the literature on CS, since the Spanish-English data on which much of it was based, did not include it to a great extent”.

152 The extensive research on Spanish/English CS conducted in the last four decades, suggests that the preferred strategies for the incorporation of English verbs in Spanish contact situations is either through the morphophonological integration of verbs, as in (5), or simply through a non-specialized auxiliary strategy, as in (6), where the English verb is the complement of the Spanish auxiliary (Backus, 1996).

(5) Los hombres me trustearon

The men CL.DIROBJ trust.3PL.PST

‘The men trusted me.’

(Pfaff, 1979, p.275)

(6) Va a reenlist

Ir.3SG.FUT to reenlist.INF

‘He is going to reenlist.’

(Reyes, 1982, p.154)

Considered a universal property of CS (Edwards & Gardner-Chloros, 2007; González-

Vilbazo & López, 2012), the origin and status of these structures are still contested. Some researchers analyze them as borrowings (Fuller Medina, 2005; Moinzadeh, 1999; Reyes, 1982;

Wichmann & Wohlgemuth, 2008), whereas others view them as illustrative of CS (Crutchley,

2015; González-Vilbazo & López, 2011; Myers-Scotton & Jake, 2013; Sebba, 1998), with features of creolization (Balam, 2015; Edwards & Gardner-Chloros, 2007; Gardner-Chloros,

2010; Pfaff, 1979). For some scholars, BLVCs are neither borrowings nor switches but emergent hybrid structures that may belong to a third (Romaine, 1995, p.160) or new grammar that develops as a result of bilingual discourse practices (Wilson Vergara, 2013, p.132). Thus, linguists still do not concur on how to classify BLVCs.

153 Furthermore, the questions of when and why these bilingual structures are incorporated in bi/multilingual discourse still remain largely unanswered. Some researchers have attributed the incorporation of these structures to lexical deficits (Fuller Medina, 2005; Jenkins, 2003; Reyes,

1982), whereas others have contended that BLVCs are particularly used to incorporate phrasal verbs (Myers-Scotton, 2002, p.136; Balam & Prada Perez, 2016) or concepts that are culturally novel (Edwards & Gardner-Chloros, 2007). BLVCs have also been found to be employed in specific discourse contexts. For example, in a pilot study where they examined BLVCs in three different data sets involving CS with the Greek light verb kano ‘do’, Gardner-Chloros and Finnis

(2006) found that among young Greek-Cypriots in , BLVCs were only used in humorous or mocking contexts. In contrast, the London Greek-Cypriot bilinguals from the other data set and young Greek-French bilinguals in Brussels also used BLVCs in serious contexts.

Other scholars have highlighted changes in the evolution of BLVCs. For Pfaff (1979), for example, the ‘hacer + V’ structure simply represents an extension of the causative infinitival construction to non-causative contexts. For Backus (1996) and Muysken (2000), BLVCs evince a developmental progression across time. Backus (1996) examined the cross-generational differences in the use of BLVCs, particularly vis-à-vis the occurrence of the Turkish light verb yap- ‘do’ with different kinds of complements. Noteworthy was that whereas the first generation used the light verb yap- with Turkish nominal elements, the second generation used it particularly with Dutch verbs. In Backus’ view, across generations, the ‘NP + yap-’ construction further evolved and grammaticalized into ‘infinitive verb + yap-’ constructions. Thus, this pattern “reflect[s] different stages of development” (Muysken, 2000, p.208).

More recently, from a theoretical point of view, Myers-Scotton and Jake (2013) suggest that BLVCs may be cross-linguistically attested in CS because the incorporation of non-finite

154 verbs incurs less psycholinguistic cost than finite verbs in CS production. They base their analysis on the Abstract Level model (Myers-Scotton, 2002), a model of the abstract lexical structure of entries in the mental lexicon. According to this model, the levels of abstract structure include lexical-conceptual structure, predicate–argument structure, and the level of morphological realization patterns. Initially, Myers-Scotton and Jake (1995) proposed that in CS, any EL form had to be checked twice: once at the conceptualizer and once at the mental lexicon.

At the conceptual level, the activation of the EL form had to satisfy the semantic/pragmatic features of the speaker’s intentions. At the mental lexicon level, the EL form had to be checked for congruence at the three levels of abstract structure.

In more recent work, however, Myers-Scotton and Jake (2013) posit that EL verbs only need to be checked for congruence at the lexical-conceptual structure level (i.e., they have to meet the requirements of the speaker’s intentions). Given that these non-finite verbs (e.g., infinitives and in BLVCs) are used only to express speaker’s intentions (i.e., meaning), they do not have to be checked at the predicate–argument structure and the morphological realization levels. Precisely because congruence checking between the EL infinitive verb and the ML grammatical frame is not necessary, “production cost associated with language switching remains minimal” (p.11). Thus, Myers-Scotton and Jake posit that the use of

BLVCs promotes savings in psycholinguistic cost, and this may be why bi/multilingual speakers from different CS varieties are predisposed to their use5.

In terms of the sociolinguistic contexts where BLVCs emerge, scholars often associate these structures with intense language contact (Backus, 1996; Boumans, 1998; Muysken, 2000).

Boumans (1998) particularly contends that these constructions arise in cases of migrant

5 Note that in the case of finite verbs, if they were switched, the EL would have to be checked with the ML at the three levels to ensure there is sufficient congruence. This would require greater psycholinguistic cost.

155 bilingualism and intense language contact in industrialized societies, but he also points out that they may appear in cases of language loss. Verteegh (2009) points out the case of Rapa Nui

Spanish (Makihara, 2005) in Easter Island, where younger speakers who are deficient in Rapa

Nui employ the light verb hacer, as in (7), to incorporate Rapa Nui elements without its complex pre- and post-verbal morphology (for a case involving language death and the incorporation of

Spanish verbs in Chontal Maya, see Knowles-Berry, 1988).

(7) Hizo more el ŋao

Do.PST rip apart the neck

‘The neck (tube) was ripped.’

(Makihara, 2005, p.748)

Undoubtedly, BLVCs have been studied from a variety of empirical and theoretical perspectives, as they raise important questions about the existence of universal CS constraints, the nature of intra-sentential CS, and the relationship between CS and other contact phenomena such as convergence and language change. Although scholars now concur that CS is systematic and rule-governed (Ritchie & Bhatia, 2004, p.339; Toribio, 2001, 2004), there is still disagreement as to whether CS is inclusive of other concomitant phenomena such as convergence, creolization and borrowing, as was reviewed in Chapter 2. In addition, while some researchers envision CS as the combination of two lexicons (MacSwan, 2005) or grammars

(Myers-Scotton, 2002), for others, CS triggers convergence (e.g., Bullock & Toribio, 2004) and/or co-occurs with linguistic creativity (Gardner-Chloros & Edwards, 2004; Lakshmanan et al., 2016; Sebba, 1998, 2009).

BLVCs continue to generate scholarly interest among linguists, as they are puzzling CS structures that often exhibit features that cannot be easily attributed to either of the component

156 languages (for relevant discussion, see Gardner-Chloros & Edwards, 2004). Northern Belize offers fertile ground to examine these constructions in relation to current debates on the nature of

CS and the existence of universal CS constraints. Firstly, as sociohistorical data show (see section 1.2 in Chapter 1), bi/multilingualism in Northern Belize has remained rather stable but has become increasingly more intense in recent decades. Given that Spanish/English CS has become much more commonplace than in the 1970s, due to higher levels of bilingual and

English proficiency, a pertinent question is how verb switching is instantiated among present-day younger generations who perceive CS positively and associate it with their Northern Belizean

Maya/Mestizo identity.

Although Hagerty (1996) highlighted the frequent use of ‘hacer + V’ in Belizean

Spanish, no previous study has specifically examined BLVCs in the variety of Spanish spoken in

Northern Belize. The present investigation fills this gap and contributes to our knowledge of this understudied CS phenomenon in an unexamined Spanish contact situation, while also contributing to the more general understanding of contact outcomes and syntactic microvariation in CS contexts.

My central focus here is the examination of three linguistic factors that have been proposed in the variationist (i.e., verb frequency and stativity) and syntactic-theoretical literature

(i.e., passivization). The examination of these restrictions provides insight into the existence of universal CS constraints, and the nature of intra-sentential CS in a contact situation which, unlike the U.S. Hispanophone context, has several of the key ingredients (i.e., relaxed language norms, positive attitudes toward CS, etc.) that favor the use of dense CS practices. Furthermore, the present investigation offers preliminary yet informative comparisons between general contact

157 outcomes in Northern Belize versus those in Spanish/English bilingual communities in the U.S. that have been more extensively examined during the last four decades.

Consonant with recent work on CS, which advocates for a more integrative approach to the study of bilingual speech phenomena (MacSwan & McAlister, 2010; Rodriguez-González &

Parafita-Couto, 2012), I examine both speakers’ acceptability judgments and use of BLVCs in spontaneous oral production in order to test the three proposed restrictions and provide a more holistic picture of ‘hacer + V’ production in Orange Walk, Northern Belize.

4.3 Bilingual Light Verb Constructions in Spanish Contact Situations

I now turn to the restrictions that have been proposed to constrain the incorporation of

‘hacer + V’ in bi/multilingual speech. First, I provide an overview of constraints based on verb frequency and stativity, followed by an elaboration of the syntactic restriction on the formation of eventive passive BLVCs.

4.3.1 Verb Frequency and Stativity

In the U.S. Spanish/English context, Pfaff (1979, 1982) and Reyes (1982) were among the first to report the occurrence of BLVCs in the speech of Mexican American Spanish speakers6. Reyes, however, was the first to contend that ‘hacer + V’ was constrained by verb frequency. Basing his descriptive analysis on informal conversations with Mexican American

Spanish speakers from California, Reyes described ‘hacer + V’ as an unassimilated borrowing frame, which was used to borrow only certain types of verbs.

Reyes argued that bilingual constructions like *hacer eat and *hacer walk never occur because in this case, the highly frequent Spanish equivalents (i.e., comer ‘to eat’ and caminar ‘to walk’, respectively) would be used instead. Thus, according to Reyes, the ‘hacer + V’ syntactic

6 In contexts of Spanish/English CS, BLVCs have only been attested in Southwest U.S. (Jenkins, 2003; Reyes, 1982; Toribio et al., 2012; Wilson Vergara, 2013; Vergara Wilson & Dumont, 2015) and Belize (Balam, 2014, 2015, 2016; Fuller Medina, 2005; Hagerty, 1996).

158 frame was only employed with English verbs that had low frequency Spanish verb equivalents

(e.g., mejorar ‘to improve’, dibujar ‘to draw’, dirigir una petición ‘to petition’, etc.).

Jenkins (2003) subsequently concurred with Reyes’ (1982) contention that BLVCs are constrained by verb frequency. Jenkins’ descriptive analysis was based on interview data from

15 Spanish/English bilinguals from and Southern (between ages 25-51).

Jenkins observed that BLVCs were largely triggered by lexical need. For instance, he highlighted that this construction occurred after repair phenomena, when speakers attempted to utter a

Spanish verb, only to doubt the formation, pause, and then employ ‘hacer + V’. In addition, he noted that hacer was attested with verbs related to English-language domains (i.e., verbs related to school or occupation), and with low-frequency, or “new” verbs (e.g., Tengo que hacer vacuum

‘I have to vacuum’). Following Reyes (1982), Jenkins concluded that “there is no functional need for a bilingual to shift to English for a frequently used verb” (2003, p.198).

It is important to note that Reyes’ and Jenkins’ proposed constraint on verb frequency follows a line of reasoning that has been well-known in the contact literature vis-à-vis lexical borrowings. In previous work, it has been argued that ‘core’ (Teschner, 1974) or ‘basic’ (Hock &

Joseph, 1996, p.257; Thomason, 2001, p.71-72) vocabulary items are most resistant to borrowing. Unlike cultural borrowings, which are incorporated in bilingual speech to designate new or novel concepts (i.e., technology-related words), core borrowings have existing lexical items that are readily available in the recipient language lexicon (Myers-Scotton, 2002, p.41) and are frequently used. The use of core borrowings, thus, often duplicates existing native words and may appear superfluous. Given the general assumption that frequently used words referring to

“essential human activities [and] needs…such as eat [and] sleep...” (Hock & Joseph, 1996, p.257) are most resistant to being borrowed or incorporated into the recipient language, it is not

159 surprising Reyes and Jenkins posited that ‘hacer + V’ occurs only with English lexical verbs that have low frequency rather than high frequency Spanish equivalents.

In more recent work, Vergara Wilson (2013) examined other aspects of this construction in his analysis of 62 BLVCs (with different lexical verbs), extracted from the New Mexico

Spanish/English Bilingual (NMSEB) corpus (Torres Cacoullos & Travis, in preparation).

Vergara Wilson argues that there is strong evidence of grammaticalization in the fact that ‘hacer

+ V’ behaves independently of its closest monolingual Spanish equivalent (i.e., causative construction such as hacer llorar, ‘to make cry’), and that it behaves remarkably productively.

Vergara Wilson notes that the V slot can be expanded to different types of verbs.

In New Mexican Spanish/English CS, ‘hacer + V’ was attested with transitive, intransitive and reflexive forms. Notably as well, hacer was attested with English verbs that had high frequency Spanish equivalents (e.g., decidir ‘to decide’, caminar ‘walk’, bailar ‘to dance’, etc.). Overall, 16% of the overall tokens (i.e., 10/62) occurred with high frequency verbs, suggesting that ‘hacer + V’ does not only occur with lexical verbs that have low frequency

Spanish equivalents (Vergara Wilson, 2013, p.128). In other words, it is not necessarily triggered by a lexical deficit or need.

As it relates to Belize, Hagerty (1996, p.136) was the first to observe that “[BLVCs are] an Anglicized feature of particular interest because of [their] frequency in Belizean Spanish”. In his brief descriptive analysis, he observed that these constructions were often used to fill lexical gaps in the Spanish language, as in hacer jogging ‘to jog’, where there is no exact equivalent in

Spanish to refer to the concept ‘to jog’. Given that Hagerty did not provide any quantitative data such as the number of ‘hacer + V’ tokens in his corpus, however, the frequency with which

‘hacer + V’ occurs in Belizean Spanish in comparison to other CS contexts still remains unclear.

160 In a more recent study on Belizean Spanish, Fuller Medina (2005) collected data from native speakers of Belizean Spanish in order to determine whether the use of ‘hacer + V’ was governed by constraints based on verb frequency (Reyes, 1982; Jenkins, 2003) and stativity. She also examined whether BLVCs occurred with unfamiliar or new verbs (e.g., technology-related verbs). Fuller Medina collected oral production data via a picture description task from 21

Spanish/English bilinguals, whose ages ranged between 18 – 89 years old, from different parts of

Belize. The 18 pictures that were used in the picture description task included actions that corresponded to dynamic verbs (e.g., paint), high frequency verbs (e.g., eat), low frequency verbs (e.g., testify), images depicting technology use, and images depicting activities that are associated with rural life. To analyze verb frequency, Fuller Medina conducted a search in the

Corpus del Español (Davies, 2002), and classified all verbs that had fewer than 2,000 occurrences in the 20th century as low frequency, whereas those that had more were classified as high frequency. For the stative/dynamic distinction, she classified realizations, accomplishments and activities as dynamic verbs, and verbs that referred to a state were considered stative.

Overall, speakers produced a total of 35 BLVCs, which contained 22 different verb types.

Results showed that 90% of tokens belonged to the class of dynamic verbs (e.g., Está haciendo graduate ‘She’s graduating’), while only 10% were stative verbs. Fuller Medina concluded that

‘hacer + V’ is not likely to occur with stative lexical verbs. Stronger evidence was found for the restriction based on verb frequency, as 95.5% of BLVCs occurred with lexical verbs whose

Spanish translation equivalents were low frequency verbs (e.g., hacer plant ‘to plant’).

Furthermore, none of the participants produced BLVCs with high frequency verbs such as eat, drink, get up, etc. Instead, they used the Spanish verb equivalent to describe such activities; thus,

161 supporting Reyes’ (1982) and Jenkins’ (2003) claim that verbs borrowed through the ‘hacer + V’ template are restricted by verb frequency.

Her data also showed that slightly over 50% of BLVCs (n = 18) were associated with technology (e.g., hacer chat ‘to chat’, hacer print ‘to print’, hacer download ‘to download’, etc.).

This finding supported Jenkin’s (2003) claim that ‘hacer + V’ is used to borrow English verbs that are associated with unfamiliar or new domains. Based on her findings, Fuller Medina concurred with Jenkins’ (2003, p.8) contention that “if the lexical item in Spanish is available to the speaker then there will be no need for ‘hacer +V’ but if it is unavailable then ‘hacer +V’ is used”. In line with Reyes, Fuller Medina also suggested that lexical need seems to be a driving force in the incorporation of ‘hacer + V’.

Although Fuller Medina’s (2005) seminal work provided valuable insight into the use of

BLVCs in Belize, her analysis had some limitations. For one, she did not report on judgment data, so we do not know how speakers react to the use and acceptability of BLVCs. Additionally, she did not control for potential dialectal differences among speakers. Recall that in his work on

Belizean Spanish, Hagerty (1979) identified a dialect boundary that separates Northern from

Western Belize. It is anticipated, therefore, that Belizean Maya/Mestizos who are not native speakers of NBS, or who have less exposure to NBS (i.e., speakers who live in the Cayo district) will incorporate less BLCVs in their naturalistic discourse, given that their native Spanish variety more closely approximates Guatemalan or . Another issue is that Fuller

Medina did not analyze naturalistic oral production data from interviews, and this may partially explain why her corpus of nine hours of recorded speech yielded only 35 tokens, which does not reflect the frequency of BLVCs Hagerty (1996) had previously observed for Belizean Spanish.

162 While the picture description task may have motivated the use of ‘hacer + V’, it also limited the description to a specific action; thus, restricting the overall use of verbs.

4.3.2 Passivization

The restriction on passivization was proposed by González-Vilbazo and Lopéz (2011), who examined BLVCs in Spanish/German CS. Their data, collected via questionnaires, interviews and informal conversations, were obtained from 82 Spanish/German bilinguals, between ages 16-18, at the German School of Barcelona.

In their Minimalist analysis of the light verb hacer, they examined four puzzling syntactic characteristics of ‘hacer + V’; namely, (i) the asymmetric nature of this structure (i.e., the light verb can only be realized in one language while the lexical verb has to be taken from the other language); (ii) the light verb’s incompatibility with the passive voice; (iii) feature spreading (i.e., the selected VP has features such as word order and of hacer; and (iv) the absence of

‘hacer + V’ in the input grammars.

My primary focus here is on the second characteristic, as this is a restriction that

González-Vilbazo and Lopéz (2011, p.843) describe as “imposed on the computational system rather than a language-specific - or code-switching specific - type of restriction.” Thus, this is a universal restriction that applies to both monolingual and bilingual speech. As (8) illustrates, in

Spanish/German CS, the light verb hacer co-occurs with the intransitive verb schlafen ‘sleep’. It can also occur with transitive (e.g., hace lesen ein Buch ‘reads a book’) and unaccusative (e.g., se hizo zerbrechen ‘it broke’) predicates. Importantly, this is only possible in the active voice.

(8) Juan hace schlafen

Juan do.1SG.PRS sleep

‘Juan sleeps.’

(González-Vilbazo & Lopéz 2011, p.843)

163 In (9) and (10), we see that the light verb hacer cannot be used in a bilingual passive construction in Spanish/German CS. By passive, González-Vilbazo and Lopéz (2011, p.836) specifically refer to the eventive Spanish passive form, as in (9), where the verb ser ‘to be’ is used alongside a past participle. This is similar to the English analytical passive form (e.g., The house was built).

(9) *El libro fue hecho lesen

The book was done sold

‘The book was read.’

(10) *Das Buch ha sido hecho verkaufen

The book has been done sold

‘The book has been sold.’

(González-Vilbazo & Lopéz, 2011, p.843)

In line with Bhatia and Ritchie’s (1996) analysis of BLVCs in Hindi/English CS,

González-Vilbazo and Lopéz argue that the ‘hacer + V’ construction is only motivated as a last resort, similar to ‘do-insertion’ in English.

In the production of a BLVC, when a German verbal root, such as lauf- ‘run’ is brought into the computational system (see Figure 4-1), it cannot value or satisfy Spanish little v’s conjugation class feature because it lacks this feature. Thus, the light verb hacer must be inserted to save the derivation. In light verb constructions, the configuration is grammatical as hacer does bear a conjugation feature, so it is able to satisfy the [uConj] feature of vsp (see Figure 4-2).

164

Figure 4-1. German root lauf- cannot value Spanish little v’s conjugation feature

Figure 4-2. The light verb hacer values the [uConj] feature of vsp

In the case of , the light verb hacer spells out as little v, and it does not incorporate to the lexical verb that it selects. Note that with monolingual Spanish verbs (e.g., cant-ar ‘to sing’, camin-ar ‘to walk’, etc.), incorporation occurs because the verbal roots have matching features. Thus, the roots value the [uConj] feature in little v and then move up, as

Figure 4-3 shows, to incorporate to little v. The vsp then acquires a conjugation class feature and becomes a morphological base to which TAM (Tense/Aspect/Mood) features are attached.

Figure 4-3. Root cant- values the [uConj] feature in little v and then move up

In the case of passive BLVCs, the lexical verb remains unincorporated, as little v is realized as hacer. However, this is where the problem lies: in a grammatical construction, the

165 internal argument would be able to move up to Spec, v or subject position. Recall that in a monolingual Spanish verbal construction, the lexical verb would satisfy the conjugation class feature and move up to adjoin to little v; hence, creating a space for a constituent to move up in the syntactic tree.

In the case of passives with hacer, this space is not available given that little v is spelled out as hacer. In González-Vilbazo and Lopéz’s view, the raising of the internal argument to subject position would give rise to a violation of the Relativized Minimality/Minimal Link

Condition. Thus, as a result of this restriction, the light verb hacer is incompatible with eventive passives7 and examples such as *El event fue hecho organized por ellos…‘The event was organized by them’ are not grammatical in Spanish/English CS.

4.4 Study 1 Methodology: Verb Frequency and Stativity

I first provide the details on the methodology employed in Study 1, which sought to examine the restrictions on verb frequency and stativity. Specifically, I investigated whether verb frequency (Reyes, 1982; Jenkins, 2003) and stativity (Fuller Medina, 2005) are constraining factors in the use of BLVCs in Northern Belize. To this end, both acceptability judgment data and oral production data were analyzed.

If linguistic/structural factors are deterministic, we could surmise that results would support previous findings for Spanish/English CS data from the Southwest U.S. (Jenkins, 2003;

Reyes, 1982) and Belize (Fuller Medina, 2005). At the same time, it must be taken into consideration that the task could have an effect on patterns observed. For instance, it is possible that in oral production the ‘need’ to switch with infrequent verbs (i.e., in Zentella’s (1997, p.97) terms, ‘crutch-like code-mixing’) is a deterministic factor. In a judgment task that requires

7 Syntactic raising of the theme does not apply to stative passives (Wasow, 1977, Levin & Rappaport, 1986, among others).

166 reading, however, verb frequency or stativity might not be of importance as there is nothing in the syntax (i.e., its derivation in Minimalist terms) to ban the switch into a frequent verb.

On the other hand, if social factors are more influential, then we would anticipate different patterns than those attested in the Southwest U.S. Recall that in Muysken’s view (2000,

2013), it is more likely that dense CS practices and congruent lexicalization will be found in contexts with a low degree of normativity or linguistic purism. Following Muysken (2000, 2013) and Bullock and Toribio (2004), this would mean that because of code-switchers’ greater capitalization of convergence or congruent lexicalization in these contexts, then CS patterns may evince greater density and productivity across a variety of lexical and syntactic environments.

4.4.1 Consultants

A total of 38 consultants completed the acceptability judgment task (AJT). For the AJT,

25 adolescent participants (ages 14–17), of which 15 were males and 10 were female, were recruited from Muffles High School and Orange Walk Technical High School in Orange Walk,

Belize. Additionally, 13 post-adolescent participants (ages 19–26), of which 8 were males and 5 were females, were recruited via purposeful sampling, from the Orange Walk community at large. All participants were Orange Walk-born natives and had either completed or were currently obtaining their secondary education in Orange Walk.

Data on consultants’ linguistic background were collected via a language background questionnaire (Appendix C). In terms of self-rated speaking proficiency, consultants gave an overall higher rating to Belizean Spanish (Mean: 5.4), followed by English (Mean: 5.2), CS

(Mean: 5.1) and BK (Mean: 4.9). Only adolescents reported having higher proficiency in Kriol than in Spanish (see Table 4-1), a rating that may reflect their attitudinal predisposition to Kriol,

Belize’s lingua franca. Given that high school students spend a significant part of their daily lives within an academic domain where the adolescent culture demands linguistic conformity to the

167 prevalent use of Kriol, it is not surprising that they perceive themselves more dominant in BK than in Spanish.

Table 4-1. Linguistic background of consultants for Study 1 Self-rated speaking proficiency Frequency of language use Belizean English Belizean Mixture Belizean Belizean English Mixture Spanish Kriol Spanish Kriol Mean Mean Mean Mean Mean Mean Mean Mean (SD) (SD) (SD) (SD) (SD) (SD) (SD) (SD) AD 4.3 4.8 4.5 4.2 7.0 7.1 6.3 5.7 females (2.6) (1.3) (1.8) (1.5) (1.0) (0.9) (1.3) (1.9) (n = 10) AD 5.1 5.1 5.5 5.1 7.1 7.1 5.7 5.7 males (1.8) (1.2) (2.2) (1.9) (1.6) (1.8) (2.2) (1.9) (n = 15) PA 5.8 5.0 4.8 5.6 7.4 7.0 5.4 6.6 females (2.2) (1.0) (2.5) (1.9) (1.3) (1.0) (0.5) (1.5) (n = 5) PA 6.3 5.8 4.8 5.6 7.4 7.1 7.1 7.0 males (1.2) (1.5) (1.8) (2.0) (0.9) (0.6) (0.6) (0.9) (n = 8) Total 5.4 5.2 4.9 5.1 7.2 7.1 6.1 6.3 (n = 38) (2.1) (1.3) (2.0) (1.8) (1.3) (1.3) (1.7) (1.7) *AD = adolescent, PA = post-adolescent

In terms of language use, the overall means revealed that consultants reported using NBS and (Mean: 7.2) and BK (Mean: 7.1) more than CS (Mean: 6.3) or English (Mean: 6.1), on a scale where 1 indicated ‘Never’ and 8 indicated ‘Everyday, most of the time during the day.’

Across the four participant groups, consultants reported using primarily Belizean Spanish and

Belizean Kriol. Only adolescent females reported using CS more frequently than English. Note that some speakers’ conceptualization of Belizean Spanish may actually refer to CS, as in the interviews, consultants often associated the variety spoken in Northern Belize to the practice of switching between languages (see Balam, 2013, p. 257).

168 4.4.2 Data

Data were collected via two tasks: an acceptability judgment task and a semi-structured interview. The Acceptability Judgment Task (AJT) elicited speakers’ well-formedness judgments regarding code-switched utterances containing BLVCs (see Appendix D). The task included a total of 54 items, which consisted of 29 test items and 25 fillers. All test items consisted of sentences where participants rated randomized verbal phrases on a 1-4 Likert scale (1= totally unacceptable, 2 = unacceptable, 3 = acceptable, 4= totally acceptable).

As Figure 4-4 exemplifies, there were two types of BLVCs in the AJT: canonical (i.e., the lexical verb is an infinitive) and non-canonical (i.e., control structures.)

Figure 4-4. Sample items from the AJT

For canonical BLVCs, as in (11a), the dependent variable was the form of the verb: a

BLVC, which was represented at least twice (i & ii), with different switching patterns in the object, an English verb (iii), or a Spanish verb (where the switch is post-verbal as in (iv)).

Although the effect of the switching patterns in the object was included in the design, it is not discussed here. The focus is on the comparison between the BLVC sentence that receives the higher rating, switching into an English verb, or switching after the Spanish verb. Therefore, for

169 (11a), sentences (i), (iii), and (iv) were compared, since (i), with a mixed DP, was rated higher than (ii). In the case of non-canonical BLVCs, as in (11b), the levels of the dependent variable were: no BLVC (Spanish control verb with an English main verb), BLVC in the control verb,

BLVC in the main verb or ‘double’ BLVC (control and main verbs).

Sentences were manipulated for predicate type (transitive, unaccusative, control), stativity, and verb frequency. The design, thus, included the following conditions: BLVCs with

(i) transitive verbs, manipulated for frequency, (ii) stative accusative verbs, manipulated for frequency, (iii) stative unaccusative verbs, and (iv) control structures. BLVCs with (i) through

(iii) were all canonical constructions which have been mentioned in the antecedent literature

(González-Vilbazo & López, 2011; Vergara Wilson, 2013), whereas condition (iv) was the only non-canonical BLVC explored in the current study. Note that in the case of the double BLVC,

‘hacer + V’ appears consecutively, in the control verb and lexical verb contexts. Speakers’ intuitional data were analyzed using SPSS, version 21.

To determine verb frequency of the lexical verbs, Fuller Medina’s (2005) methodology was followed. Frequency was operationalized according to word searches (for all conjugated forms of the translation equivalent) in the Corpus del Español (Davies, 2002). Verbs were classified as low frequency verbs if they had fewer than 2000 occurrences in the 20th century data, whereas verbs with more than 2000 occurrences in the on-line database were considered high frequency. All sentences were controlled for length and plausibility.

Oral production data were obtained via 10-20 minute, semi-structured code-switched interviews with a total of 43 native bi/multinguals (25 adolescents and 18 post-adolescents) from

Northern Belize. Interviews were audio-recorded using a Marantz PDM 620. In order to collect speech samples that were reflective of the community linguistic norms, interviews were

170 administered in pairs or small groups of peers, which was intended to ensure the production of authentic speech behaviors (Blom & Gumperz, 1972; Wassink, 1999). All sentential contexts where ‘hacer + V’ occurred were extracted from the interviews and orthographically transcribed for coding (for categories, see Table 4-2) and quantitative analysis.

Table 4-2. Categories used for Verb Type Verb Type Example Activity Lo usas pa’ hace garnish un plate ‘You use it to garnish a plate.’

Change of state Solo creo que hizo broaden el vocabulary ‘I just think the vocabulary broadened.’

Psychological Hay unos word que hago medio understand ‘There are some words that I kind of understand.’

Communication Yo no me hago express así cuando voy abroad ‘When I go abroad, I don’t express myself like that.’

Other Si no hiciste pass, like, tienes que agarra el course over ‘If you didn’t pass, you have to take the course again.’

Stative Puedes hace live el life del past en el present ‘You can live the life of the past in the present.’

Intransitive motion Yo garre y hice cras ovah dah Banquita’s ‘I crossed over to Banquita’s (House of Culture).

Exchange Sus amigos hicieron rent un golf cart ‘His friends rented a golf cart.’

Reverse Psychological Me hace annoy como hablan el Spanish ‘It annoys me how they speak Spanish.’

Transitive motion Quisiera conseguir un scholarship para hacer further mis studies ‘I would like to get a scholarship to further my studies.’

Perception Ellos no pueden hace como distinguish si de veras es ella ‘They cannot determine whether it’s really her.’

171 To examine stativity, I first coded lexical verbs following Fuller Medina’s (2005) classification, in which dynamic verbs referred to realizations, accomplishments and activities whereas stative verbs referred to verbs of state. In the interest of providing a more detailed breakdown of the types of lexical verbs in BLVCs, verbs were also coded according to categories/verb classes used in the variationist literature (e.g., Aaron, 2010). Table 4-2 provides the categories used to code the data.

Thus, stativity was subsumed under verb type. In contrast to previous studies, which have been limited by a relatively small number of tokens (e.g., Fuller Medina, 2005: 35 tokens;

Romaine, 1989: 77 tokens; Toribio et al., 2012: 12 tokens; Vergara Wilson, 2013: 62 tokens), the present quantitative analysis is based on 553 ‘canonical’ BLVCs extracted from the interviews.

For the analysis of stativity, all 553 occurrences were examined independently, as every BLVC essentially occurs in possible variation with an English, BK and/or Spanish verb; thus, every

BLVC represents an instance where the speaker uses the bilingual structure over the English,

Spanish or Kriol verb or near-equivalent verbal form.

However, for the analysis of frequency in the oral production data, following Fuller

Medina (2005) and Vergara Wilson (2013), only the 299 different verb types (not the overall 553 occurrences) were considered as the number of times that a verb was repeated has no bearing on the established frequency of the verbs (i.e., following Fuller Medina’s criteria). The number of times a lexical verb was used only becomes relevant in the examination of level of diffusion, which I also explore.

Verbs with multi-word Spanish equivalents (e.g., start over, volunteer, babysit, campaign, notice, agree, etc.) and verbs with Spanish reflexive equivalents (e.g., behave, fit in, get used to, base on, focus, fall, get along, lay down, etc.) were excluded from the frequency analysis as they

172 constitute a class of verbs whose Spanish equivalents and/or near-equivalents differ both morphologically and semantically. Also excluded were verbs with no Spanish equivalents (e.g., google, double-ride, type in, run out (of air), etc.). In total, 75 verb types were excluded, leaving

224 verb types for the analysis of frequency. Results from the AJT and the oral production data are presented in the following section.

4.5 Results

I first present the results from the statistical analysis of the AJT data, followed by a distributional analysis of the oral production data.

4.5.1 Speakers’ Intuitions: Verb Frequency and Stativity in BLVCs

Participants’ ratings of BLVCs in stative vs. non-stative transitives (both frequent and non-frequent) are summarized in Table 4-3.

Table 4-3. Descriptive statistics for stativity and frequency in BLVCs with transitive predicates Descriptive Stativity Speaker group Mean Std. N statistics: Deviation Stativity in BCVs Non-Frequent Non-Stative Adolescent females 3.55 0.33 5 transitives Adolescent males 3.41 0.53 8 Post-adolescent females 3.00 0.79 10 Post-adolescent males 3.13 0.72 15 Total 3.21 0.67 38 Stative Adolescent females 3.40 0.29 5 Adolescent males 3.44 0.42 8 Post-adolescent females 2.88 0.66 10 Post-adolescent males 3.02 0.45 15 Total 3.12 0.53 38 Frequent Non-Stative Adolescent females 3.15 0.65 5 transitives Adolescent males 3.47 0.56 8 Post-adolescent females 2.88 0.69 10 Post-adolescent males 2.78 0.44 15 Total 3.00 0.61 38 Stative Adolescent females 3.50 0.47 5 Adolescent males 2.56 0.35 8 Post-adolescent females 2.93 0.68 10 Post-adolescent males 2.82 0.53 15 Total 3.09 0.61 38

173

It is important to note that all participant groups’ ratings fall within the acceptance rate, suggesting the widespread acceptance and use of these hybrid structures in Northern Belize. All groups accepted BLVCs with both stative (e.g., hizo need un laptop ‘s/he needed a laptop’) and non-stative transitive (e.g., hizo cook el food ‘s/he cooked the food’) lexical verbs. At the same time, BLVCs were also accepted with frequent (e.g., hizo eat el pork meat’ ‘s/he ate the pork meat’) and infrequent (e.g., hizo raffle un gold chain) transitive lexical verbs. Interestingly, whereas previous work reported a stativity and a frequency effect, no effects were anticipated as there should be no syntactic restriction on the stativity or frequency of the verb in order to switch using hacer ‘do’.

In order to examine stativity, transitive predicates that were stative were compared to transitive predicates that were non-stative. Both frequent and infrequent transitive predicates were included. For frequent transitive predicates, a 2 (stativity: stative vs. non-stative) by 4

(speaker group: adolescent females, adolescent males, post-adolescent females, post-adolescent males) repeated measures ANOVA revealed a main effect for speaker group, but no main effect for stativity, and no interaction.

Table 4-4. Stativity in BLVCs ANOVA Main effect for Main effect for Stativity by results: stativity speaker group speaker group Stativity interaction Frequent F(1,34)= 2.471, F(3, 34)= 4.421, F(3,34)= .586, transitives p= .13, p=.01, p=.63, partial η2 = .068 partial η2 =.281 partial η2 =.049 Non- F(1,34)= 1.175, F(3, 34)= 1.882, F(3,34)= .235, frequent p=. 286, p=.151, p= .87, transitives partial η2 =.033 partial η2 = 0.142 partial η2 =.020

174 Male post-adolescent participants rated sentences significantly lower (M: 2.8, SD: .128) than high-school males (M: 3.5, SD: .176). For non-frequent transitive predicates, a similar analysis was performed but found no main effect for stativity, speaker group or stativity by speaker group interaction.

Therefore, the prediction that stativity would not be significant in the AJT bears out in the data. While in oral production speakers may be more sensitive to stativity in their CS, it is not due to the syntax of BLVCs as hacer selects for a V of any type. It could be a general trait of CS, and thus, we would expect to find this stativity effect in CS dialects where instead of BLVCs code-switchers use a finite English verb, an issue we leave for further research.

As was the case with stativity, even though an effect has been reported in spontaneous oral production data, we predicted the syntax would not restrict BLVCs due to frequency.

Therefore, we anticipated all groups would exhibit no frequency effect. To test this hypothesis, frequent and infrequent transitive predicates (both stative and non-stative) were compared. The descriptive statistics indicated that all speaker groups accepted BLVCs with both frequent and non-frequent transitives.

Two separate 2 (frequency: high vs low frequency verbs) by 4 (speaker group) repeated- measures ANOVAs were conducted; one for stative and one for non-stative transitives. For stative transitives, the analysis revealed no main effect for frequency (although the effect size was small), a main effect for speaker group, and no frequency by speaker group interaction (see

Table 4-5). Adolescent males rated BLVCs with stative transitives significantly higher than post- adolescent males. For non-stative transitives, no main effect for frequency, speaker group or frequency by speaker group interaction was found.

175 Table 4-5. Frequency in BLVCs ANOVA Main effect for Main effect for Frequency by results: frequency speaker group speaker group Frequency interaction Stative F(1,34)= .059, F(3, 34)= 4.295, F(3,34)= 1.334, transitives p= .81, p=.01, p=.28, partial η2 = .002 partial η2 =.275 partial η2 =.105 Non-stative F(1,34)= 3.712, F(3, 34)= 2.012, F(3,34)= 1.051, transitives p=. 06, p=.13, p= .38, partial η2 =.098 partial η2 = 0.151 partial η2 =.085

In summary, verb frequency and stativity were not significant variables in the acceptance of BLVCs among Northern Belizean adolescent and post-adolescent males and females. All

BLVCs were accepted at the same rate with transitive predicates. One salient trend was that male adolescent speakers’ mean ratings were significantly higher that their post-adolescent counterparts’ ratings, specifically in the stative versus non-stative contrast with frequent transitives and in the frequent versus infrequent contrast with stative predicates.

4.5.2 BLVCs in Control Structures

For control sentences, participants were asked to rate four related sentences: one with no

BLVC (Spanish control verb with an English lexical verb), one with a BLVC in the control verb

(with an English lexical verb), a BLVC in the main verb (with a Spanish control verb) or a double BLVC (‘hacer + V’ in both control and main verbs). The descriptive statistics show that conditions with BLVCs are within the acceptance rate while the condition without a BLVC is not

(see Table 4-6), a pattern reflective of the marked status of monolingual Spanish in Northern

Belize (Balam, 2013, 2015).

The data also revealed that post-adolescent females rate BLVCs lower than the other groups. Post-adolescent females consistently gave the lowest mean ratings to BLVCs in control structures. On the other hand, sentences without BLVCs were generally rated as the least

176 acceptable by adolescent females, consistent with research that shows that the use of standard varieties are marked among Northern Belize adolescent Maya/Mestizos (Balam, 2013; Balam &

Prada Pérez, in press; Koenig, 1975).

Table 4-6. Descriptive statistics for control structures Descriptive statistics: Speaker group Mean Std. N BLVCs in control Deviation structures No BLVCs Adolescent females 1.45 0.60 5 Adolescent males 1.75 0.45 8 Post-adolescent females 1.80 0.55 10 Post-adolescent males 1.80 0.51 15 Total 1.74 0.51 38 BLVC in the control Adolescent females 2.45 0.69 5 verb Adolescent males 2.46 0.39 8 Post-adolescent females 1.90 0.50 10 Post-adolescent males 2.25 0.47 15 Total 2.23 0.52 38 BLVC in the lexical Adolescent females 2.95 0.33 5 verb Adolescent males 3.03 0.41 8 Post-adolescent females 2.75 0.58 10 Post-adolescent males 3.02 0.49 15 Total 2.94 0.48 38 Double BLVC Adolescent females 2.90 0.38 5 Adolescent males 3.03 0.39 8 Post-adolescent females 2.45 0.80 10 Post-adolescent males 2.58 0.51 15 Total 2.68 0.59 38

In order to examine these comparisons, a 4 (sentence type: no BLVC, BLVC in the control verb, BLVC in the main verb, and double BLVC) by 4 (speaker group) repeated- measures ANOVA was performed, which revealed a main effect for sentence type, but no main effect for speaker group and no sentence type by speaker group interaction (Table 4-7). The

Bonferroni post-hoc test revealed that sentences with a BLVC in the main verb and sentences with a double BLVC were rated significantly higher than sentences with a BLVC in the control verb and these, in turn, were rated significantly higher than those without a BLVC.

177 Table 4-7. ANOVA results for sentence type in control structures ANOVA Main effect for Main effect for Sentence type by results: sentence type speaker group speaker group Sentence interaction type Control F(2.8,102)= 49.400, F(3, 34)= 1.530, F(8.3,102)= 1.402, structures p= .00, p=.23, p=.20, partial η2 = .592 partial η2 =.119 partial η2 =.110

Overall, speakers rated BLVCs significantly higher than sentences with a switch into an

English verb. Predicate type, stativity, and frequency did not restrict the use of BLVCs.

Importantly, the intuitional data revealed that speakers may not always produce the BLVCs that they rate as acceptable. For instance, control structures were rated higher when the Spanish control verb is followed by a BLVC or when ‘hacer + V’ is used in both the control and main verbs. The use of ‘hacer + V’ in control structures, however, was highly infrequent in the interviews. In the following section, we take a closer look at BLVCs in oral production.

4.5.3 BLVCs in Oral Production

A total of 564 BLVCs were extracted from naturalistic speech, of which 98% were canonical ‘hacer + V’ constructions, as in (12), where hacer co-occurred with an English infinitive verb rather than with a participle (i.e., stative passive BLVCs).

(12) Es cuando el country hizo gain su freedom

Be.3SG when the country do.PST gain.INF its freedom

‘It’s when the country gained its freedom.’

(PA15, female, 18)

As Table 4-8 illustrates, the vast majority of these tokens were produced by post- adolescent rather than adolescent speakers. The most prolific users of ‘hacer + V’ were two post- adolescent female speakers, one who produced 133 occurrences and the other 114 occurrences, comprising 23.6% and 20.2% respectively of the total number of ‘hacer + V’ tokens (n = 564).

178 On the other hand, there were two adolescent speakers who only produced one token of ‘hacer +

V’.

Table 4-8. BLVC production across two speaker groups Speaker group N % Adolescents 71 12.8 Post-adolescents 482 87.2 Total 553 100

Among non-canonical ‘do-constructions’ that were attested in the production data,

BLVCs in stative passives (as in (13)) and in control structures (as in (14) and (15)) were found, primarily produced by post-adolescent speakers. In the case of (14), produced by an adolescent female speaker, we see a trilingual verbal construction, where the Spanish light verb co-occurs with an English control verb, followed by a BK lexical verb.

(13) En Sixth Form,‘tas hecho considered un adult

In Sixth Form, be.2SG done considered.PASTPART an adult

‘In junior college, you are considered an adult.’

(PA14, female, 19)

(14) En veces, like, hago try taak Spanish

Sometimes, like, do.1SG.PRS try.INF to speak.INF Spanish

‘Sometimes, like, I try to speak Spanish.’

(AD09, female, 16)

In (15), we see an example of a ‘double’ BLVC, where two BLVCs occur consecutively.

In this case, hacer is used along with the control verb ‘allow’ and with the lexical infinitive

‘chat’, which is a complement to the control verb. These are innovative constructions that have not been previously reported elsewhere in the Spanish/English CS literature on ‘hacer + V’.

179 (15) Hay un program que hace allow hacer chat

There is.3SG a program that do.3SG allow.INF do.INF chat.INF

‘There is a program that allows chatting.’

(PA12, male, 19)

Further analysis of the 553 canonical BLVCs revealed that the ‘hacer + V’ occurred with

299 different lexical verbs. These constructions were attested in a variety of argument structures including intransitive (e.g., develop, get along), transitive (e.g., ban, enjoy), ditransitive (e.g., explain, ask), copulative (e.g., act mature, sound similar), Spanish reflexive (e.g., vestirse ‘dress up’, reunirse ‘gather’), passive (e.g., hecho involved ‘to be involved’, hecho banned ‘to be banned’) and control verbs (e.g., try, allow).

The total number of phrasal verbs in the corpus accounted for 10% of the lexical verbs, in contrast with 7.5% in Fotiou’s (2012) Greek Cypriot/English corpus and less than 1% in Vergara

Wilson’s (2013) New Mexican Spanish data. Even though they do not seem to comprise a significant portion of the type of lexical verbs that co-occur with hacer, BLVCs with phrasal verbs were used in innovative ways.

In terms of verb frequency and stativity, although the adolescent group produced markedly fewer BLVCs, they nonetheless employed the light verb hacer with lexical verbs that were frequent (e.g., lo hicimos lose ‘we lost him’, hace change ‘it changes’) and stative (e.g., hago realize… ‘I realize…’, hago understand ‘I understand…’), suggesting that even at a developmental stage of bilingual proficiency in this bi/multilingual context, these two linguistic factors do not seem to ban the incorporation of BLVCs. In the ensuing sections, we take a closer look at verb frequency and stativity by looking at the patterns in the overall group data.

180 4.5.4 Restriction of Frequency in Oral Production

In line with speakers’ judgments, the analysis of the oral production data further revealed that frequency does not constrain the incorporation of BLVCs. Figure 4-5 illustrates the distribution of frequency measures of lexical verbs (following Fuller Medina’s criteria) attested in the interviews. It shows that although ‘hacer + V’ was attested primarily with low frequency verbs (i.e., 64.1%), high frequency verbs accounted for 35.9% of the lexical verbs, which is considerably higher than the percentage reported by Fuller Medina (i.e., 4.5%).

10000 9000 8000

7000 6000 5000

frequency lexical verb 4000 3000 2000 1000 0 50 100 150 200 250 bilingual light verb constructions

Figure 4-5. Frequency of lexical verbs in BLVCs

A small percentage of low frequency verbs were highly infrequent (i.e., 9%), with fewer than 100 occurrences. On the other hand, 15% of the lexical verbs (not included in the scatter chart due to space limitation) were highly frequent, exceeding 10,000 occurrences in the Davies

Corpus. Thus, BLVCs occurred with both highly infrequent verbs (e.g., hacé garnish ‘to garnish’, hacen discriminate people ‘they discriminate people’, hacerles revoke sus licence

‘revoke their license’, hacemos socialize ‘we socialize’) and highly frequent lexical verbs (e.g.,

181 estan haciendo have fun ‘they are having fun’, empiezas a hacé follow ‘you start to follow’, hacé live ‘to live’, puedes hacer wear… ‘you can wear…’).

While the preceding analysis takes into consideration monolingual data, another way of examining BLVCs is by focusing on level of diffusion. Level of diffusion can be of use to the present study, as it reveals whether lexical verbs in BLVCs only consist of idiosynctratic lexical verbs, whose use may be more reflective of ‘crutch-like code-mixing’ (Zentella, 1997), in which bi/multilinguals switch as a result of difficulty with lexical access, a genuine lexical gap, etc.

Following Reyes’ (1982) and Fuller Medina’s (2005) analyses, we would not anticipate that the

English translations of highly frequent Spanish verbs such as ‘drink’ and ‘eat’ would become in any way diffuse or conventionalized in the bilingual community. Recall that “if the lexical item in Spanish is available to the speaker then there will be no need for ‘hacer +V’” (Jenkins, 2003, p.8).

Notably, the examination of the level of diffusion of lexical verb types revealed that

BLVCs occurred with what other researchers would consider ‘diffuse’ (Aaron, 2014) or

‘recurrent’ (Torres Cacoullos & Aaron, 2003) items. Type frequency (i.e., the number of speakers using a lexical type) and token frequency (i.e., the number of times a lexical item occurs) are two frequency measures that can give us some insight into the distribution of lexical items in a bi/multilingual community (Torres Cacoullos & Aaron, 2003, p.162). Following the work of Poplack, Sankoff, and Miller (1988), Torres Cacoullos and Aaron (2003) classify lexical types used by five or more speakers as ‘widespread’, ‘recurrent’ those that are used by two to four speakers, ‘idiosyncratic’ those lexical types that appear more than once but used by one speaker, and ‘nonce’ those items that are attested only once. Aaron (2014) distinguishes between

182 ‘singletons’ (i.e., those that appear only once) versus ‘diffuse’ (i.e., those that are used at least three times by three different consultants) lexical items.

Following these criteria, I distinguished between diffuse (i.e., lexical verb types used by at least three different speakers), non-diffuse (i.e., used three times or more, but by fewer than three speakers) and idiosyncratic lexical verbs (i.e., used two times or less by same or different speakers). As Table 4-9 illustrates, this analysis revealed that the vast majority of lexical verbs were idiosyncratic (81%). The remaining 19% of verbs, however, comprised of both lexical verb types that were used three times but by fewer than three speakers (e.g., appreciate, celebrate, criticize, develop, function, influence, manage, participate, pass, reinforce, etc.) and verbs which were used three times by at least three different speakers (e.g., describe, follow, socialize, transfer, enjoy, focus, graduate, meet, gain, etc.).

Table 4-9. Lexical verb type and level of diffusion

Lexical Verb Types Tokens N % N % Idiosyncratic 242 81 293 53 Lexical verb appears 3 times or more 57 19 260 47 Diffuse (> 3 speakers) 32 165 Non-diffuse (<3 speakers) 25 95 57 260 299 100 553 100

Overall, the lexical verbs that were most diffuse included change of state verbs (i.e., increase: 8 times; decrease: 12 times; mix: 9 times), psychological verbs, which can be considered stative (i.e., understand: 9 times; learn: 11 times), and an activity verb (i.e., teach: 12 times). The group of ‘diffuse’ lexical verbs comprised 10.7% of the total 299 different verb types analyzed. Although the percentage of lexical verbs that are ‘diffuse’ or more frequent is

183 considerably lower than idiosyncratic verbs, it is nonetheless noteworthy that ‘diffuse’ verbs did include frequently used stative verbs such as entender ‘to understand’ and aprender ‘to learn’.

Furthermore, when the overall number of tokens is taken into consideration (n = 553), there is a roughly equal distribution in the naturalistic production of lexical verbs that were idiosyncratic versus those that were diffuse or non-diffuse (see Table 4-9).

It is important to note that in order to ensure comparability, frequency was operationalized following Fuller Medina (2005). It must be pointed out, however, that a primary challenge for researchers who investigate bi/multilingualism is that frequency counts based on online corpora (e.g., Davies Corpus) reflect monolingual Spanish varieties rather than bilingual

Spanish/English varieties (for further discussion on challenges on the quantification of frequency, see Erker & Guy (2012)). Thus, in the present frequency investigation, not all lexical verbs that were ‘diffuse’ in the bilingual data were necessarily ‘frequent’ according to the Davies

Corpus del Español. From a total of 31 diffuse lexical verbs, only 15 were considered high frequency verbs (i.e., > 2,000 occurrences). There were 11 diffuse verbs that were low frequency

(i.e., < 2,000 occurrences), whereas 5 lexical verbs were excluded because their translation equivalents were reflexive verbs or multiword equivalents.

Notably, the diffuse verbs that were low frequency according to the Davies Corpus included verbs that were school-related such as fail, transfer (from one school to another), teach, socialize (often used in relation to school), and describe. It is unsurprising that these particular verbs would be diffuse in the Northern Belize context. Although Spanish is spoken by the majority as a first language, English is the official language of instruction in schools. Thus, lexical items that specifically relate to this domain permeate the daily lives of Northern Belize

Maya/Mestizos, particularly students. In this regard, we see that the frequency of verbs is

184 intricately tied to the social and educational realities of a bi/multilinguals (for relevant discussion, see Jenkins, 2003, p.197).

Importantly, the case of ‘hacer + V’ is more difficult to analyze, as it is not the prototypical case of morphosyntactic variation. One cannot assume a priori that a monolingual

Spanish variant is present in the speaker’s lexicon or in the community lexical repertoire, as it may simply be the case that, for example, ‘hacer transfer’ is the conventionalized construction employed to describe ‘transferring from one to another’. In this case, one cannot speak of a lexical gap. If there is no functional need for a Spanish verbal equivalent in a community, particularly one where CS is clearly the norm and socially unmarked, then such equivalent will simply not be used. These patterns are not reflective of lexical gaps, but rather of the unmarkedness of bilingual language practices in a certain community. We must also highlight that an analysis where BLVCs are merely viewed as variants of monolingual translation equivalents fails to capture the hybrid and innovative nature of these structures and the multiple factors involved in their usage.

4.5.5 Restriction of Stativity in Oral Production

The data analysis revealed that stativity was also not a constraining factor in the incorporation of BLVCs. The distributional analysis, following Fuller Medina’s classification, showed that both dynamic (48.6%, n = 269) and stative (23.0%, n = 127) lexical verbs were employed with hacer (see Figure 4-6). Importantly, there was a sub-group of lexical verbs

(28.4%, n = 157) that could not be easily accounted for using Fuller Medina’s broad classification system.

185 Lexical Verbs

Dynamic Stative Other

Figure 4-6. Distribution of lexical verbs following Fuller Medina’s classification

These ‘Other’ verbs included those that were used to express ideas (e.g., Te hacen allow bastantes cosas ‘They allow you to have many things’; Está haciendo increase ‘It is increasing’;

Hace cause un big issue ‘It causes a big issue’) that did not easily fit within the dynamic/stative binary distinction due to the absence of observable action. The more fine-grained categories employed in the present analysis give us a more clear insight into the different types and distributions of lexical verbs attested in speakers’ naturalistic discourse.

The analysis following the categories outlined in Table 4-2 revealed that stative verbs

(e.g., hacer agree ‘to agree’, hacen keep ‘they keep’, hacemos need ‘we need’) accounted for less than 5% of the English lexical verbs. Notably, BLVCs with psychological verbs (e.g., hago feel ‘I feel’, hago realize ‘I realize’, haces understand ‘you understand’) accounted for a considerably higher percentage of lexical verbs. Given the overlap in argument structure between these two types of verbs, if we considered stative and psychological verbs along with perception verbs as a single class, they accounted for 20% of the total number of lexical verbs. Clearly, we see that stative verbs are not banned from being incorporated in BLVCs.

186 The data did show a tendency for BLVCs to primarily occur with activity verbs. When combined as the class of dynamic verbs, activity verbs (e.g., hacerlo carve ‘to carve it’, hacen clean ‘they clean’), intransitive motion verbs (e.g., haces gather ‘you gather’, hago lay down ‘I lay down’) and transitive motion verbs (e.g., hacer knock on sus heels ‘to put on heels’, hacé park sus car ‘to park their car’) account for 34.8% of lexical verbs. There was also a high rate of

BLVC occurrence with change of state verbs (e.g., hizo decrease ‘it decreased’, hacer reinforce

‘to reinforce’, hacen shorten ‘they shorten’, se hizo burst ‘it burst’), whereas ‘hacer + V’ was least attested with perception verbs (e.g., Si haces spot a alguien… ‘It you see/recognize somebody…’). Note, however, that if we combine change of state and ‘other’ verbs with reverse psychological predicates and the class of stative, psychological and perception verbs, which all lacked clearly observable agentive action, this accounts for more than 50% of lexical verbs.

Table 4-10. Lexical verb distribution in canonical BLVCs Verb Type N % Activity 156 28.2 Change of state 107 19.3 Psychological 82 14.8 Communication 55 9.9 Other 50 9.0 Stative 26 4.7 Intransitive motion 23 4.2 Exchange 22 4.0 Rev. psychological 16 2.9 Transitive motion 13 2.4 Perception 3 0.5 Total 553 100.0 *Rev = Reverse

Overall, the oral production data revealed that that there is no inherent feature in English lexical verbs that constrain their incorporation in BLVCs. The light verb hacer occurred with a wide variety of lexical verb types; thus, revealing BLVCs are used when speaking about a variety of actions, which may be overt or abstract in nature. The productivity of this CS strategy

187 was also evinced in the argument structures with which ‘hacer + V’ occurred, which included intransitive, transitive, ditransitive, reflexive, passive and control structures. In the following section, I elaborate on the innovative uses of BLVCs.

4.5.6 Beyond Verb Frequency and Stativity: Hybridity and Innovation in BLVCs

A salient characteristic of BLVCs that warrants further attention is the skillful use of these hybrid structures to build semantic and syntactic convergence (in Bullock & Toribio’s

(2004) terms). The naturalistic production of BLVCs revealed that in switched discourse speakers capitalized on their lexical resources; in other words, speakers combined different features and elements of their languages to encode their bilingual utterances with semantic nuances.

In (16) and (17), for instance, we can see the multiple ways in which a highly frequent verb such as ‘speak’ is employed in Northern Belize switched discourse. Note that in addition to the literal meaning of speak (e.g., haciendo speak ‘engaging in conversation’), this lexical verb is also employed to underscore a personality trait. In (16), the BLCV incorporates the English expression ‘speak one’s mind’, which specifically refers to the speaker’s perception of herself as having the quality of directly and candidly expressing how she feels about someone of something. In Spanish, a potential equivalent would be ser franco.

(16) Yo naturally hago speak mi mind

I naturally do.1SG.PRS speak.INF my mind

‘I naturally speak my mind.’

(PA18, female, 23)

In (17) below, we see another innovative use of the verb ‘speak’. In this case, ‘speak’ is used to underscore the notion that something is reflective of the people. Hence, in this case,

188 ‘speak’ has nothing to do with the communicative act of holding a conversation. The concept is clearly more abstract in nature.

(17) Eso hace speak overall de toda la gente

That do.3SG.PRS speak.INF overall of all the people

‘That speaks overall, of all the people.’

(PA07, male, 19)

The polysemic use of English lexical verbs is also attested in (18) and (19). In (18), the speaker employs ‘struggle’ to refer to a physical type of struggle involving two persons. It entails physical contact. In Spanish, a possible equivalent would be forcejear ‘to struggle’, which implies some sort of physical struggle between individuals.

(18) No estaba haciendo struggle con él

Not be.IMP do.PROG struggle.INF with him

‘I wasn’t struggling with him.’

(PA15, female, 18)

In (19), the same verb is used to refer to a different kind of struggle, one that is more abstract. In this case, the struggle relates to the male referent’s emotional or socioeconomic strife. We can infer this from the utterance where the speaker expresses her admiration for her friend’s efforts to overcome the dire conditions or obstacles that are the source of his ‘struggle’.

In this case, the Spanish verb luchar ‘to struggle’ would more closely align with the way the verb is used in (19).

(19) El hace struggle y I admire di fact he di push ihself

He do.1SG.PRS struggle.INF and I admire the fact he be push himself

‘He struggles, and I admire the fact he is striving to better himself.’

189 (PA18, female, 23)

In the following two examples, the verb ‘get’ is used in different ways as well. Whereas in (20), it refers to the act of acquiring or obtaining jobs, in (21), the verb ‘get’ is synonymous to the stative verb ‘understand’. Thus, the verb ‘get’ is used to refer to conseguir un trabajo ‘get a job’ and entender ‘understand’.

(20) Aquí los que tienen mostly stripes hacen get jobs

Here those that have.3PL mostly stripes do.3PL get.INF jobs

‘Here, those who have connections usually get jobs.’

(PA05, female, 18)

(21) Hacemos get porque

Do.3PL.PRS get.INF why

‘We understand why.’

(PA18, female, 23)

Lastly, in (22) and (23), we can see how English phrasal verbs are skillfully employed in

BLVCs. Note that in (22), the use of ‘run out’ implies specifically running out of somewhere, which is precisely the way in which the speaker used it. In an anecdote that the speaker shared, he was elaborating on a school fight that took place inside a classroom. When he saw that the teacher was coming, he ran out so as not to be associated with the altercation.

(22) Yo solo hice run out… desde que miré el teacher

I just do.1SG.PST run out.INF as soon as that see.PST the teacher

‘I just ran out…as soon as I saw the teacher.’

(AD11, male, 15)

190 In (23), on the other hand, ‘run out (of air)’ is an English idiomatic expression that refers to the physical condition in which a person has difficulty breathing either due to exhaustion or some other cause. Notably, whereas ‘run out’ in (22) could potentially have a Spanish near- equivalent (i.e., salir corriendo ‘run out’), the use of ‘run out’ in (23) does not have a literal translation equivalent in Spanish (i.e., *salir afuera de aire ‘run out of air’). In the latter case, the non-usage of the pronominal se suggests that the construction has a more English-like quality rather than a Spanish-like one (i.e., se estaba haciendo run out de air ‘was running out of air’, where se derives from se le estaba acabando el aire).

(23) Dis pickney ya estaba haciendo run out de air

This boy.3SG already was.IMPV do.PROG run out.INF of air

‘This boy was already running out of air.’

(PA18, female, 23)

Other idiomatic expressions used with hacer included ‘hit the books’ (i.e., Usually hago hit mi books ‘I usually study hard’) and ‘watch one’s or someone’s back’ (i.e.,…estás haciendo watch over tu back ‘…you are guarding yourself from danger’). Thus, the idiomatic quality of certain BLVCs can render these constructions unintelligible to monolingual Spanish speakers or even English L2 learners with low levels of speaking proficiency.

Taken together, these examples show that code-switchers from Northern Belize capitalize on the semantic features that are available to them in their English lexical inventory, demonstrating that they make full use of the lexical resources available to them. The prevalent use of polysemy in switched discourse demonstrates parsimonious and efficient use of their lexical resources, as concepts that would often require multiple Spanish equivalents are expressed through polysemic English verbs. The use of idiomatic expressions in BLVCs also

191 adds complexity to speakers’ bi/multilingual speech. Although a complete syntactic analysis falls outside the purview of the current chapter, the skillful merging of syntactic features from English and Spanish is also noteworthy, as this evinces not only the rule-governed nature of CS, but the linguistic creativity that characterizes it as well. Consider examples (24) and (25):

(24) Una se hizo give up

One CL.3SG do.PST give up.INF

‘One of them gave up.’

(PA16, female, 22)

(25) Se te hace pile up tu work

CL.2SG do.PRS pile up.INF your work

‘Your work accumulates.’

(PA14, female, 19)

In both cases, phrasal verbs are used. Crucially, the phrasal verbs are embedded alongside clitics, which reveals that certain pronominal features of Spanish morphosyntax were incorporated in the derivation of these utterances. In the case of (24), the se more than likely is from darse por vencido ‘to give up’, whereas the se in (25) is derived from acumularse ‘to accumalate’. In both cases, Spanish pronominal features are employed to encode the utterances with middle voice. In other cases, clitics are incorporated to encode reflexivity or reciprocity.

Examples such as these illustrate bi/multilingual speakers’ skill at establishing semantic equivalence across language varieties and their ability to seamlessly merge different elements from their rich linguistic repertoire in order to devise novel, hybrid structures8. In the discussion

8 For analyses on the syntactic or clausal incorporation of BLVCs in Northern Belize switched discourse, see Balam, 2015, 2016; Balam & Prada Pérez, 2016).

192 section, I return to this issue and further explore how this aspect of BLVCs relates to our current understanding of their use in code-switched speech.

4.6 Study 2 Methodology: Passivization

I provide here the details of the methodology employed in the second study, which specifically examined the compatibility of hacer with passive BLVCs9. The goal of the study was threefold. Firstly, I examined Northern Belize bi/multilinguals’ judgments of bilingual stative and eventive passive structures with the light verb hacer. Although González-Vilbazo and

López’s (2011) prediction applies specifically to eventive passive BLVCs (i.e., passives with ser

‘to be’), I tested whether stative passive BLVCs (i.e., passives with estar ‘to be’) would be more favorably accepted, given that they are not subject to any syntactic restriction. Secondly, I examined gender agreement between the past participle of hacer and the feminine-marked subject. Although previous studies have pointed out that hacer is a carrier of TAM features

(Balam, 2015; González-Vilbazo & Lopéz, 2011; Jenkins, 2003; Muysken, 2000; Reyes, 1982), hacer in passives is potentially a carrier of grammatical gender in addition to TAM (i.e., hecho

‘domasc’; hecha ‘dofem’), as past participles are gender marked in Spanish eventive passives.

Lastly, I tested whether distance had any effect on the acceptability of passive BLVCs.

4.6.1 Participants

A total of 46 native speakers of Northern Belizean Spanish (29 females, 17 males), whose ages ranged from 18 to 36, participated in the second study. Only 6 participants (4 females and 2 males) from Study 1 participated in Study 2 as well. Given that the use of ‘hacer + V’ is clearly

9 Further analysis was also conducted, comparing the judgments of three groups of bilinguals; namely, Northern Belize (n = 46), Florida (n = 46) and New Mexico (n = 11) bilinguals. This yielded an insight into the acceptability of these innovative structures across speakers from different Spanish/English contexts. In the present chapter, however, only the data from Northern Belize consultants is reported (for further details on the findings from the comparative analysis across CS varieties, see Balam & Prada Perez (2013)).

193 conventionalized in Northern Belize, I did not anticipate significant differences between the participants from Study 1 to differ from the other consultants that were recruited for Study 2.

Data obtained from the language background questionnaire (see Appendix E) showed that consultants were all proficient in Spanish, Belizean Kriol and English. Based on reported language use, most participants were Spanish dominant speakers, who had learned NBS as a first language and started learning English and/or Belizean Kriol in elementary school. As Table 4-11 shows, for self-rated language proficiency (where ‘1 – 3’ indicates low proficiency; ‘4 – 5’ intermediate proficiency and ‘6 – 7’ advanced proficiency), males rated their proficiency higher in Belizean Spanish than in CS, whereas the opposite trend was attested among females. Both males and females, however, reported having the lowest proficiency in Belizean Kriol.

Table 4-11. Linguistic background of consultants for Study 2 Self-rated language proficiency Language use (frequency) Belizean English Belizean Mixture Belizean English Belizean Mixture Spanish Kriol Spanish Kriol Mean Mean Mean Mean Mean Mean Mean Mean (SD) (SD) (SD) (SD) (SD) (SD) (SD) (SD) males 6.1 6.0 5.4 5.9 6.2 4.7 4.7 5.5 (n = 17) (0.9) (0.9) (1.4) (1.0) (0.8) (1.6) (1.8) (1.4) females 5.5 5.8 5.2 5.9 5.6 5.0 5.0 6.2 (n = 29) (1.2) (0.8) (1.2) (0.8) (1.5) (1.3) (1.2) (1.0)

Patterns of language use (where ‘1 – 3’ indicates infrequent use of the language, ‘4 – 5’ sporadic use, and ‘6 – 7’ frequent use of the language) were in line with reports of language proficiency, as males reported more frequent use of Belizean Spanish, whereas females reported more frequent use of CS. Noteworthy is that in Study 1, post-adolescent females also reported more frequent use of CS, suggesting that overall, in the Northern Belize community, females are more positively predisposed to CS than males, a trend that is consonant with recent work on CS that demonstrates this trend (e.g., Dewaele & Wei, 2014).

194 4.6.2 Data

Data for the analysis were collected via an acceptability judgment task (AJT), which was administered to gain an insight into consultants’ well-formedness judgments of passive BLVCs

(see Appendix F). Recall that Study 1 revealed that in Northern Belize CS, infrequent examples of stative passive BLVCs, as in (26) and (27), were attested in the oral production data. To the best of my knowledge, ‘hacer + V’ in stative passives has not been reported in any

Spanish/English community in the U.S. The examples provided illustrate that in contrast to canonical Spanish/English LVCs, lexical verbs in these structures always appear in the past participle rather than the infinitive form. Also, in both (26) and (27), the light verb is marked with the canonical masculine morpheme –o.

(26) Estabamos hecho involved en sports

Be.3PL.IMP do.PSTPTCP.M involved.PTCP in sports

‘We were involved in sports.’

(Speaker AD15, male, 15)

(27) Usually, ‘ta hecho banned en school

Usually, be.3SG.PRS do.PSTPTCP.M banned in school

‘Usually, it is banned at school.’

(Speaker PA14, female, 19)

González-Vilbazo and Lopéz (2011) highlight that eventive passives are not compatible with hacer because of a computational restriction from Universal Grammar on the raising of the passive subject through a light verb. Importantly, in Spanish, the use of the copulas estar ‘to be’ and ser ‘to be’ distinguishes between stative (adjectival) and eventive (verbal or analytical) passives, a verbal difference encoded in Spanish but not in English (Bruhn de Garavito &

Valenzuela, 2008; Luján, 1981; Varela, 1992). Given that stative passives were attested in the

195 oral production data in Study 1, they were included to determine whether they would be accepted more than eventive passive BLVCs. Since they are not subject to the restriction from Universal

Grammar, it was anticipated that they would get higher ratings of acceptability.

Note that as a result of the general low frequency of passives in both monolingual (Bruhn de Garavito & Valenzuela, 2008, p.326) and bilingual Spanish, examining speakers’ judgments was essential to obtain a more fine-grained insight into the use of ‘hacer + V’ in both stative and eventive bilingual passive structures. Since few examples were attested in Study 1, it remains a question whether eventive passives with BLVCs are not present in the corpus because they are infrequent constructions, because they are ungrammatical, or because they may be an idiolectal phenomenon, only found in the speech of certain code-switchers10.

In the generativist literature, it is well established that among Spanish/English bilinguals, contact phenomena such as lexical borrowing and CS are systematic, and speakers make consistent well-formedness judgments based on the grammatical principles that underlie their competence (Anderson & Toribio, 2007; González-Vilbazo & López, 2012; MacSwan &

McAlister, 2010; Toribio, 2001). Given that these passive BLVCs are rare structures in naturalistic data, the present method was deemed appropriate to gain an insight into speakers’ competence.

Given that disparaging attitudes toward bilingual language practices are not prevalent in

Northern Belize, we did not expect speakers’ attitudes to be influenced by negative evaluations of CS. At the same time, Study 1 revealed that speakers’ generally positive attitudes toward CS

10 In Study 1, participants did not spontaneously produce any eventive passive BLVCs. In subsequent fieldwork, however, there was one example that was naturalistically produced by one of the consultants’ friends, as they informally spoke about a documentary they had watched (El army fue hecho defeated ‘The army was defeated’). Eventive passive BLVCs were not produced even in the cross-sectional interview data (see Chapter 5, for a cross- sectional analysis of determiner phrases). Thus, while ‘hacer + V’ with true passives is attested in Northern Belize, they are more infrequent than stative passive BLVCs or doble hacer.

196 did not translate to acceptance of sentences across the board, suggestive of an overall effect from their more positive predisposition to CS. In the previous study, clear differences were found as to acceptability of structures in the condition with control structures versus other contexts. Thus, there were different degrees of acceptance, which aligned with consultants’ production of forms in naturalistic speech (i.e., control structures were exceptional and infrequent forms).

The AJT in Study 2 sought to mainly examine speakers’ intuitions regarding passive

BLVCs. As in Study 1, items (54 test items and 54 fillers) consisted of a sentence on a 1-4 Likert scale (1 = lowest degree of acceptability; 4 = highest degree of acceptability). Unlike the AJT in

Study 1 where each item was followed by the four conditions examined, items in the AJT in

Study 2 were simply presented in a randomized order. Given that overt grammatical gender in

Spanish – one of the variables examined – is salient orthographically, it was essential to not present conditions in such a way that consultants would immediately be aware of the masculine- marked versus feminine-marked conditions. This could have affected results. In addition, the design of the AJT in Study 2 was more complex in terms of the variables examined; thus, it was best to present conditions singly rather than in groups.

Test sentences in the AJT in Study 2 were controlled for tense, animacy of the lexical subject, and frequency of the nouns. Since eventive passives are canonically expressed in the preterite of ser ‘be’ when referring to a past action, and stative passives are canonically expressed in the imperfect form of estar ‘be’ (Bruhn de Garavito & Valenzuela, 2008), all eventive passives were presented in the preterite and all stative passives in the imperfect form to avoid participants’ judging tense. Only commonly used feminine inanimate subjects that were morphologically marked with -a were used in the test items (e.g., casa ‘house’, carta ‘letter’, mesa ‘table’, etc.).

197 The variables examined included passive type (stative versus eventive), as in (28), grammatical gender on the light verb (feminine -a versus masculine -o), as in (29), and distance

(determiner phrase + BLVC versus determiner phrase + relative clause + BLVC), as in (30).

(28) Passive type

a. Stative passive: Antonio se molestó porque la portada no estaba hecha

locked bien.

‘Antonio got angry because the gate was not locked properly.’

b. Eventive passive: Cristina se molestó porque la casa no fue hecha decorated bien.

‘Cristina got angry because the house wasn’t well decorated.’

(29) Grammatial gender

a. Feminine: Cristina se molestó porque la casa no fue hecha decorated bien.

‘Cristina got angry because the house wasn’t well decorated.’

b. Masculine: Cristina se molestó porque la casa no fue hecho decorated bien.

‘Cristina got angry because the house wasn’t well decorated.’

(30) Distance levels

a. Level 1 (D1): Paulina se molestó porque la carne no estaba hecha seasoned

bien.

‘Paulina got angry because the meat wasn’t well seasoned.’

b. Level 2 (D2): Raúl se molestó porque la carta que escribió la semana pasada no fue

hecha mailed.

‘Raúl got angry because the letter that he wrote last week wasn’t

mailed.’

198 The examples in (28) show BLVCs in stative versus eventive passive structures, respectively. Although the sentences are similar in terms of number of words, (28a) employs estar ‘be’ whereas (28b) employs ser ‘be’, yielding two different structures: stative or adjectival vs eventive or true passives. The exemplars in (29) differ in the light verb gender marking. In the case of (29a), the light verb hecha ‘done’ is morphologically marked to agree with the embedded, corresponding feminine NP la casa ‘the house’ whereas in (29b), the masculine light verb hecho ‘done’ does not agree with the feminine-marked lexical subject, thus yielding a gender mismatch. Note that the test sentences only included feminine nouns that were frequent and overtly marked (e.g., casa ‘house’, cuenta ‘bill’, mesa ‘table’, falda ‘skirt’, etc.). The exemplars in (30) differ in the presence of a relative clause intervening between the light verb and the noun it agrees with in D2, and its absence in D1.

In total, 24 different past participial English verbs were used, which included frequent verbs used in the home domain such as cleaned, ironed, cooked, closed, washed, unplugged, etc.

Each of the 24 test sentences had a masculine and a corresponding feminine condition that differed only in the light verb’s gender (e.g., hecho decorated/hecha decorated). In the AJT, test items and fillers were presented in a randomized order.

After participants completed the task, the data were systematically coded and submitted to statistical analysis using SPSS, version 21. The results are presented in the next section.

4.7 Results

The data revealed that consultants accepted all conditions with BLVCs (see Table 4-12).

Recall that the data were manipulated for PASSIVE TYPE, GRAMMATICAL GENDER, and DISTANCE, which combined resulted in the eight conditions examined. In addition, the language-external variable GENDER OF THE PARTICIPANT was included. A repeated-measures ANOVA was conducted to examine participants’ behavior across the 8 conditions with the between-subjects

199 variable GENDER. Since sphericity was not assumed, a Greenhouse-Geisser correction was used.

The data analysis returned a main effect for condition, F(4.4, 192.5) = 6.746, p = .000, but no main effect for gender (of the participant), F(1, 44) = .018, p = .894, and a condition by gender interaction, F(4.4, 192.5) = 2.423, p = .044.

Table 4-12. Mean group ratings of BLVCs across two distance levels stative Eventive feminine masculine feminine masculine BLVC BLVC BLVC BLVC Mean (SD) Mean (SD) Mean (SD) Mean (SD) Distance 3.14 (0.6) 3.33 (0.5) 3.07(0.5) 3.17(0.5) level 1 Distance 3.25(0.5) 3.26 (0.5) 3.01 (0.6) 3.17(0.5) level 2

Regarding the variable passive type, the posthoc test revealed that there were no significant differences across conditions except for feminine BLVCs in D2, where they were rated significantly higher with stative passives. Recall that we predicted a passive type effect, as stative passives are syntactically compatible with a light verb while eventive passives are not

(González-Vilbazo & López, 2011). These data seem to indicate, however, that BLVCs are in fact compatible with eventive passives (contra González-Vilbazo & López, 2011). In these data, an effect was also found only with feminine BLVCs, which are, in general, rated lower than masculine BLVCs (even though this trend did not reach significance in our data). Therefore, we can explain this result by assuming that BLVCs are less marked with the masculine-marked light verb in stative passives, although they are acceptable with eventive passives but to a lesser degree.

Regarding the variable grammatical gender on the light verb, there was no effect in all conditions except for stative D1, where the masculine-marked BLVC was rated significantly higher than the feminine-marked BLVC. Finally, for the variable DISTANCE, the Bonferroni post-

200 hoc test revealed no distance effects. Notably, as Table 4-12 shows, the lowest rated condition was feminine BLVCs with eventive passives in D2, which was rated significantly lower than masculine BLVCs with stative passives in D1 (highest ranked) and feminine and masculine

BLVCs with stative passives in D2.

Even though not many significant results between counterbalanced conditions were attested, there were some significant differences across conditions that indicate these variables do indeed have an effect on speakers’ intuitions. Specifically, gender had an effect where the masculine-marked BLVC was rated significantly higher than its feminine counterpart.

Masculine-marked BLVCs with stative passives in D1 were rated significantly higher than the feminine eventive D1 and the feminine eventive D2 conditions. Regarding passive type, the feminine-marked BLVC with eventive passives in D1 was not significantly different from its stative counterpart, but it was rated significantly lower than the masculine stative D1 condition and the feminine stative D2 condition. Thus, it was rated significantly lower than the stative conditions. This offers some statistical support to the claim that stative passive BLVCs are rated higher than eventive passive BLVCs, even though this difference did not reach significance across comparisons, and both passive BLVCs were within the acceptance range. Distance, on the other hand, did not seem to have an effect on speakers’ intuitions.

The condition by gender interaction was further explored by examining male and female data separately. Given that there were notable differences between males and females in terms of language use (see section 4.6.2), we sought to investigate whether there would be any inter- group trends. Within the male data, condition was returned as significant, F(2.5, 42.2)= 4.533, p

= .010. Figure 4-7 illustrates male consultants’ ratings of passive BLVCs across the two distance levels.

201 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 feminine LV masculine LV feminine LV masculine LV stative eventive males D1 males D2

Figure 4-7. Male consultants’ ratings of passive BLCVs for distance levels 1 and 2

In relation to gender, only two conditions were returned as significant: masculine- versus feminine-marked BLVCs with stative passives in D1, where the masculine condition was rated higher, and masculine- versus feminine-marked BLVCs with eventive passives in D2, where the masculine was also rated higher than the feminine. With respect to passive type, an effect was only found with feminine BLVCs in D2, where stative passives were rated higher than eventive passives. Lastly, among the counterbalanced tokens, the only significant difference was between

D1 and D2 with feminine-marked BLVCs in stative passives, where D2 was rated significantly higher, an unexpected result.

The female data also returned condition as a main effect, F(5, 140.5) = 2.993, p = .013.

Gender, however, was not returned as significant in any of the counterbalanced tokens. The bars in Figure 4-8 illustrate that female consultants generally gave similar ratings to both masculine- and feminine-marked BLVCs. Regarding passive type, masculine-marked BLVCs with stative passives in D1 were rated significantly higher than eventive passives. Masculine-marked BLVCs

202 with stative passives in D2 were rated significantly higher than eventive passives. Lastly, there was no distance effect.

4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 feminine LV masculine LV feminine LV masculine LV stative eventive females D1 females D2

Figure 4-8. Female consultants’ ratings of passive BLCVs for distance levels 1 and 2

Therefore, comparing males and females, although in the data from males more variables were returned as significant, both groups behave in similar ways in many respects. Specifically, both groups rate statives higher than eventives, masculine higher than feminine BLVCs (even though gender does not reach significance in the female group), and have no clear effect for distance. For both groups, the highest rated condition was masculine-marked BLVCs with stative passives in D1 and the lowest one is feminine-marked BLVCs with eventive passives in D2.

Therefore, even though consultants accepted all conditions, they did not accept all to the same degree. Although not across the board, the masculine BLVC was preferred to the feminine one, the stative to the eventive passives, and D1 to D2.

The preference for masculine-marked passive BLVCs was particularly evident in the stative passive condition in distance level 1. As Figure 4-9 illustrates, male consultants clearly showed a preference for the masculine-marked BLVCs, revealing that for these participants, not

203 only is lack of gender agreement between the light verb and the head noun in the utterance acceptable, but oftentimes is considered more acceptable than utterances where the light verb hacer agrees with the gender of the head noun.

4.0

3.5

3.0

2.5

2.0

1.5

1.0 fem masc fem masc fem masc fem masc fem masc fem masc BLVC BLVC BLVC BLVC BLVC BLVC males females

Figure 4-9. Ratings for stative passive BLVCs at distance level 1

Male consultants’ general preference for masculine-marked BLVCs was also attested in the individual mean scores. As Table 4-13 illustrates, 50% or more of the consultants rated the masculine BLVCs as more acceptable in the stative passive condition in distance level 1 and the eventive condition in distance level 2.

204 Table 4-13. Consultants’ ratings of feminine- versus masculine-marked passive BLVCs

Distance level 1 Distance level 2 Stative Eventive Stative Eventive N % N % N % N % Feminine BLVC 9 19.6 14 30.4 17 37.0 6 13.0 Masculine BLVC 25 54.3 14 30.4 16 34.8 23 50.0 Equal rating 12 26.1 18 39.2 13 28.2 17 37.0 Total 46 100 46 100 46 100 46 100

Noteworthy was that for these conditions, there was a tendency for more males than females (i.e., for D1 stative passive: 65% males versus 48% females; for D2 eventive passive:

53% males versus 48% females) to rate the masculine conditions more favorably. Only in the stative condition in distance level 2 did the majority of consultants give preferential ratings to the feminine condition. Overall, mean scores across conditions showed that consultants rated the masculine conditions as more acceptable than their corresponding feminine counterparts for 71%

(i.e., 17/24, 6 pairs of sentences x 4 conditions) of the passive BLVCs. Only 12% (i.e., 3/24) of the sentences that were feminine-marked were rated higher than their masculine counterparts, whereas 17% (i.e., 4/24) of sentences were rated as equally acceptable.

In summary, the data analysis showed that consultants accepted BLVCs with stative passives. Crucially, and contra González-Vilbazo and López’s (2011) claim, they also accepted

‘hacer + V’ with eventive passives. They rated stative passives higher than eventive passives with feminine gender agreement and in D2, a result that could be due to the frequency of stative versus eventive passives in general. Furthermore, consultants did rate the masculine conditions higher than the feminine ones but not to a statistically significant degree across the board. Lastly, distance did not have an effect in these data.

205 4.8 Discussion

The primary goal of the two inter-related studies on BLVCs was to determine whether verb frequency, stativity, and passivization either disfavor or ban the incorporation of BLVCs, as has been contended in previous work. In Study 1, contra Reyes (1982), Jenkins (2003) and Fuller

Medina (2005), results revealed that both adolescent and post-adolescent speakers accepted and produced BLVCs with stative and high frequency verbs. Noteworthy was that the non-canonical

‘hacer + V + hacer + V’ construction was rated as acceptable but to a lesser degree in comparison to canonical constructions, suggestive of the fact that this innovation is still in early stages of diffusion or that it is only employed by code-switchers with the highest levels of bi/multilingual proficiency11. In Study 2, contra González-Vilbazo’s and López’s (2011) claim, consultants’ intuitions revealed that eventive passive BLVCs were rated as acceptable. A salient trend was that masculine-marked BLVCs were generally rated as more acceptable than corresponding feminine-marked BLVCs, a pattern which aligns with the non-canonical use of grammatical gender in NBS (see Chapter 5).

Taken together, these findings indicate that in the Northern Belize context, the use of

‘hacer + V’ is not constrained by the linguistic factors previously proposed in the literature, raising important questions about the reasons why the Spanish/English CS variety in this context markedly differs from CS varieties in other contact situations, in particular the U.S.

Hispanophone context.

Under a Mimimalist approach to CS (MacSwan, 1999, 2005), there should be no differences between Spanish/English code-switchers in Belize and other contact situations (i.e.,

11 In a recent cross-generational analysis of BLVCs, Balam (2015) found that adolescent bi/multilinguals patterned more with the eldest group in their more simplistic use of ‘hacer + V’. In the study, more complex forms of BLVCs (i.e., ‘hacer + V’ in passive and control structures) were only attested among post-adolescent and adult code- switchers.

206 all code-switchers should reject eventive passive BLVCs due to a universal restriction), as CS patterns, according to this perspective, depend solely on formal features of the individual lexical items and not on any social factors. Recall that in MacSwan’s (2000, p.45) view, CS is “the simple consequence of using lexical items from two languages in the course of a derivation”.

Whereas the computational system is invariant, the properties of the lexical items are satisfied as they would in monolingual syntax.

Assuming that the core grammars of Spanish and English are largely similar in Northern

Belize and other Spanish/English contexts (e.g., New Mexico), a Minimalist approach would not predict marked differences in CS outcomes. In light of the present findings and previous work on

BLVCs in New Mexico (Jenkins, 2003; Reyes, 1982; Wilson Vergara, 2013; Vergara Wilson &

Dumont, 2015), however, we can see that there are in fact differences in the use of ‘hacer + V’ in both contexts, bringing into question the validity of a purely lexicalist or Minimalist approach to the analysis of CS phenomena.

Based on the study’s findings, we could question whether the Northern Belize data simply demonstrate the fact that BLVCs in this context may alternatively be analyzed as

‘complex predicates’, as previously proposed for Persian (see Folli, Harley & Karimi, 2004, p.1395). A perennial debate in work on Persian passives is whether Persian actually has a structural passive equivalent to the passive construction attested in English. Researchers such as

Moyne (1974) posit that Modern Persian lacks passives, and cases that have been analyzed as passives are simply constructions where the predicate employs the use of the inchoative, light verb shodan ‘become’. It must be pointed out, however, that whereas Folli et al. (2004) can argue in their complex predicate analysis of Persian passives that the past participle serves as the

207 non-verbal (i.e., adjectival) element that attaches to the light verb, the same cannot be proposed for eventive passive BLVCs in Spanish/English CS.

Clearly, as shown previously in section 3.3.2, eventive passive BLVCs in

Spanish/English CS are structurally similar to analytical passives in both Spanish and English, where the copula ‘be’ is used alongside a past participle that is clearly a verbal element.

Additionally, no inchoative verb (e.g., become) functions like the light verb hacer in Northern

Belize Spanish/English CS. Unlike other language pairs where across time, different light verbs may emerge to indicate an active/passive voice distinction in bi/multilingual speech (for examples, see Muysken, 2000), in Spanish/English CS, only hacer is used in both active and passive constructions. In the bilingual passive, the light verb hecho may be motivated not only by prominent patterns in bi/multilingual speech (i.e., extensive use of BLVCs) but by the need to encode grammatical gender, a feature that the English lexical verb cannot bear (e.g., Fue hecho organized/Fue organizadomasc ‘It was organized’). Further syntactic-theoretical work needs to further examine this issue.

Findings from the present study support Gardner-Chloros’ (1995, 2010) and Sebba’s

(1998, 2009) assertion that CS outcomes are shaped by the historical and social nature of bi/multilingualism in a given context. Sebba (1998, p.7-8) underscores that “the locus of congruence is the mind of the speaker, but community norms determine, by and large, the behavior of individual speakers.” Thus, sociolinguistic aspects of the community largely determine how speakers will, in Bullock, Hinrich and Toribio’s (2014) terms, perceive and establish congruence and convergence between their languages. Although there are syntactic constraints in CS, bi/multilingual speakers are able to not only “avoid the straitjacket of

208 grammatical rules,” (Gardner-Chloros & Edwards, 2004, p.108), but they can also build structural congruence in an effort to facilitate CS (Sebba, 1998).

In the Northern Belize context, this active construction of congruence is clearly evinced in the wide variety of lexical (i.e., dynamic, stative, psychological, etc.) and syntactic contexts

(i.e., transitive, intransitive, passive, etc.) where ‘hacer + V’ occurs. In particular, what is striking are the novel contexts where ‘hacer + V’ is used (i.e., control and passive structures).

While it may seem that the use of BLVCs among Northern Belize bi/multilinguals is unconstrained, I argue here that both linguistic and social factors have played important roles in

CS outcomes. Crucially, however, Northern Belize represents a contact situation where sociolinguistic factors seem to override ‘grammatical’ or structural factors (for relevant discussion, see Gardner-Chloros & Edwards, 2004). Thus, it is not purely linguistic factors and/or formal features that have determined contact outcomes in Belize, but rather, the singularity of historical and sociolinguistic circumstances that have contributed to the distinctive productivity and creativity attested in the use of BLVCs.

The influential role of structural or linguistic factors is also clearly evidenced in certain aspects of BLVCs in Northern Belize. For instance, like in the U.S. Hispanophone context

(Vergara Wilson & Dumont, 2015), the light verb bearing morphosyntactic information comes from Spanish, the language that is morphologically richer. It is not the case that an English light verb is productively employed alongside Spanish lexical verbs. In terms of the lexical verbs, it is notable that in both the New Mexican (Vergara Wilson, 2013) and Northern Belize contexts, lexical verbs include a variety of argument structures including reflexive verbs, hence revealing that overall, there are similarities in the use of ‘hacer + V’ irrespective of the sociolinguistic milieu. At the same time, it is the differences that warrant more careful attention. Why is it that

209 in Northern Belize, the use of BLVCs seems markedly more productive, creative and unconstrained by linguistic factors such as stativity and passivization?

I argue that this outcome can only be attributed to historical and sociolinguistic aspects of the contact situation. In the next section, I elaborate on crucial aspects of the Northern Belize context, which may help to explain the highly productive and innovative use of ‘hacer + V’; namely, speakers’ attitudes toward their language varieties and the development of

Spanish/English CS as a marker of Northern Belize Maya/Mestizo identity. I subsequently explore how these factors may have contributed to the co-occurrence of other contact phenomena such as convergence and creolization.

4.8.1 Speakers’ Attitudes to their Language Varieties and Bi/Multilingualism

An important sociolinguistic aspect that distinguishes Northern Belize from many other

Spanish contact situations in the U.S. Hispanophone context and Latin America is precisely speakers’ attitudes toward their language varieties. Recall that the younger generation generally exhibits pejorative attitudes toward standard varieties rather than CS or Belizean Kriol (see section 1.2 in Chapter 1). In contrast, the use of CS and Kriol have become increasingly unmarked in the last four to five decades12. It is important to bear in mind, however, that these overall positive attitudes toward bi/multilingualism may have actually been passed down from the previous generation, many of whom were Maya/Spanish bilinguals. As a result of relaxed language norms among the previous generation (for relevant discussion, see Balam, 2015), it is not unsurprising that the younger generation has a natural positive predisposition to bi/multilingual language practices. Thus, in Northern Belize, bi/multilingual practices have

12 Status-planning efforts from the National Kriol Council of Belize and the Belize Kriol Project have solidified Belizean Kriol’s status as a language of prestige. This starkly differs from other Spanish contact situations, where standard varieties are often associated with linguistic hegemony (Mrak, 2011), whereas Creoles and CS remain largely stigmatized (for relevant discussion on the status on the status of CS and/or U.S. Spanish, see Ramirez & Milk (1986); Valdés, González, López García, & Márquez (2003)).

210 comprised an important communicative resource for generations (for relevant discussion, see

Balam, 2015).

If previous generations perceived bilingualism as a normal social phenomenon and developed positive attitudes to the use of non-standard varieties, then more than likely, positive attitudes to the use of syntactic innovations would have thrived as well. We know that in situations where there are “flexible identity boundaries” and lax parental and community attitudes to the use of non-standard innovative forms, there is a higher probability that innovative forms will spread across a bi/multilingual community (Matras, 2010, p.72). Thus, speakers’ positive language attitudes to bilingual language behaviors may have licensed the ubiquitous use of non-standard, innovative forms such as ‘hacer + V’. Based on oral histories of older speakers

(ages 50+) in the present sample, it certainly seems to be the case that Maya/Spanish/English bi/multilingual language practices were rather common and pervasive in rural primary schools in colonial Belize.

In relation to the contemporary younger generation’s attitudes to their languages, consider the following young Belizean multilingual speaker’s description of her language practices:

(31) hablamos English, Kriol y Spanish todo hecho mixed,

speak-3PL English, Kriol and Spanish all done.M mixed.PSTPTCP

like something natural, like the three languages, like if it’s our first language

like something natural, like the three languages, like if it’s our first language

like you know, you have that privilege

like you know, you have that privilege

‘We speak English, Kriol and Spanish all mixed together, like

211 something natural, like the three languages, like if it’s our

first language, like you know, you have that privilege!’

(Speaker AD10, female, 15)

It is clear that the widespread perception of their rich linguistic repertoire as a “privilege” has certainly allowed many speakers from Northern Belize to fully exploit their linguistic resources and to further create congruence between their languages. In the words of Matras

(2010, p.83), “bilingual speakers pursue a limited range of strategies in order to facilitate management of the bilingual repertoire, among them the creation of new constructions.” Thus, the innovative use of BLVCs in novel syntactic contexts such as passive and control structures can be viewed as a facilitative strategy employed to make component languages more congruent in bilingual discourse.

To Minimalist researchers, the use of BLVCs in eventive passive constructions may appear to violate universal restrictions. This is a clear example, however, “[of] rule-breaking behavior, which should be seen not in relation to static norms but in terms of language change and convergence” (Gardner-Chloros, 2010, p.202). My view is that the causative relationship between CS and convergence is especially instantiated in cases where speakers are able to freely and actively create congruence between their languages, without the societal pressure imposed by a dogmatic tradition of linguistic purism and/or language separation (i.e., languages relegated to different contexts). Thus, it is in a context like Northern Belize, with favorable social conditions for CS, that we are best able to capture and investigate the nature and evolution of CS practices.

4.8.2 Speakers’ Identification with Spanish/English Code-Switching

Notably, among the younger population in Northern Belize, the highly productive and creative use of BLVCs may have also been catalyzed by speakers’ identification with

212 Spanish/English CS (Balam, 2013). Recent attitudinal research suggests that Spanish/English CS has assumed an important identity function in this multilingual community (Balam, 2013, 2015).

For Belizeans in general, asserting one’s national identity has taken particular precedence given that for decades, Guatemala has claimed Belize (for relevant discussion, see Le Page &

Tabouret-Keller (1985)). Le Page (1992) rightly highlights that the threat to Belize’s territorial sovereignty was one of the crucial factors that “shaped the direction of the search for new identities among the younger generation, and the linguistic attributes with which to project those new identities” (p. 87). Among younger generations of Corozaleños and Orangewalkeños, this search translated to a dissociation from Standard Spanish. On the other hand, speakers embraced

Spanish/English CS as a means to to ‘project’ their ethnic and national identities. In the concluding chapter, I further explore this issue.

The combination of positive attitudes toward CS and the emergence of Spanish/English

CS as an identity marker may have both contributed to the conventionalized, productive and creative use of BLVCs. Among younger generations, the close identification with bi/multilingual language practices legitimizes the prevalent use of CS, and thus, BLVCs are employed not only in different social but grammatical contexts as well13.

4.8.3 BLVCs, Creolization, Convergence and Language Change

The case of ‘hacer + V’ in Northern Belize supports the view that rather than representing a ‘static’ or ‘opportunistic nonce strategy’ (Edwards & Gardner-Chloros, 2007, p.80), BLVCs evince features of convergence, creolization14, and language change (Pfaff, 1979;

13 BLVCs are also open to modifiers as they occur with intervening material such as fillers, quantifiers and adverbs (e.g. se hacen easily fall ‘they easily fall’, hago medio understand ‘kind of understand’, hacen más try hacé get along ‘they try to get along more’).

14 For the purposes of this dissertation, creolization refers to a process of language evolution (Mufwene, 2008) and linguistic creativity (Le Page & Tabouret-Keller, 1985; Siegel, 2005, p.149; Baker, 2000) where there is the creation and re-creation of novel productive morphosyntactic structures (DeGraff, 2009). Although the creative exploitation

213 Edwards & Gardner-Chloros, 2007). Gardner-Chloros (2010) contends that “[BLVCs] show features of creolization, as they involve grammatical convergence and an analytic approach to vocabulary” (p.198). In the oral production data from this chapter, this ‘analytic approach’ was evident in speakers’ skillful use and combination of lexical resources from both Spanish and

English to encode semantic nuances such as reflexivity or reciprocity in BLVCs (also see Balam

(2015) and Balam & Prada Pérez (2016)).

Moreover, we must also consider the presence of BK. As a CS context where there is extensive CS alongside the pervasive use of BK, Northern Belize fits Gardner-Chloros’ (1995) description of contexts where there is “linguistically intense and constraint-defying forms of code-switching” (p.79). The seemingly constraint-free kinds of innovation present in Northern

Belize beg us to question whether BK has somehow indirectly affected CS outcomes in Northern

Belize. The concurrent development of Spanish/English CS as an identity marker alongside BK as a nascent prestige variety in the last four to five decades may have contributed to the ubiquitous use of BLVCs.

Le Page and Tabouret Keller (1985, p.11) remind us that particularly in cases of creolization, language becomes a game wherein speakers are not only the players who invent the rules of the game (Brown, 1958) but also act as the umpires as well. There is the possibility, therefore, that the widespread acceptance and use of Belizean Kriol may have further licensed the creative use of linguistic structures among the younger generation, especially if speakers associate or equate Spanish/English CS with Kriol (for an insight on this phenomenon, see

Balam, 2013, p.257).

of linguistic resources is also present in second language acquisition (Winford, 2003), the notion of linguistic creativity is particularly underscored in cases of creolization and the emergence of mixed languages (Thomason, 1995).

214 Creolization in the use of BLVCs is most evident in the case of hacer in control structures (e.g., no he hecho learn hacer cook nada), as there is no exact structural counterpart in

Spanish (i.e., *no he hecho aprender hacer cocinar) or English (i.e., *I have not did learn do cook anything). This shows that hacer has evolved across generations, as speakers now use

‘hacer + V’ in novel lexical and syntactic environments, which were unattested in the speech of older generations (Balam 2015, 2016a).

In the case of passive BLVCs, findings suggest that language change takes places in a systematic pattern. The fact that Northern Belize bilinguals accept and sometimes give higher ratings to passive BLVCs marked with the prototypical masculine morpheme –o is a telling piece of evidence, as speakers’ judgments align with the non-canonical use of the grammatical gender in NBS (see Chapter 5). At the same time, however, speakers’ acceptance of BLVCs marked with the feminine morpheme –a shows their awareness that feminine-marked BLVCs are grammatical within a more normative verbal paradigm where agreement is maintained between syntactically related constituents.

Montes-Alcalá and Lapidus Shin (2011, p.137) point out that in judgment tasks, “the greater level of consciousness and metalinguistic awareness” may result in a disparity between speakers’ judgments and their spontaneous speech. However, in the current study, despite the metalinguistic nature of the task, we see that the forms that lacked agreement were generally rated as more acceptable than those that showed overt (canonical) gender agreement. Thus, speakers’ judgments aligned with gender agreement patterns in NBS.

In this regard, the data further confirm that even the most atypical of innovations occur in a principled and systematic way rather than haphazardly. They develop in tandem with broader trends in the speakers’ language(s). In the case of Belize, the productive use of BLVCs and the

215 pervasive use of the masculine gender in monolingual and bilingual discourse align with speakers’ overall preference for masculine-marked passive BLVCs. The case of passive BLVCs also constitute evidence that, despite their infrequent use in modern Spanish, passives are also vulnerable to syntactic innovation in bilingual speech. Thus, light verbs are in fact compatible with both stative and eventive passives. What ultimately determines whether this development takes place, however, are sociolinguistic conditions rather than structural or syntactic constraints.

It must be underscored that although González-Vilbazo and López’s (2011) restriction holds for their Spanish/German data, passives have in fact been reported to be compatible with light verbs in previous work (e.g., Annamalai, 1989, for Tamil/English; Backus, 1996, for

Turkish/Dutch; Tamis, 1986, for Greek/English). In light of these data and the present findings, the Relativized Minimality/Minimal Link Condition explanation that González-Vilbazo and

López argue for cannot be universal. The current data, therefore, support the view that CS is not derived merely from formal aspects of grammar (e.g., phi features) alone (Hebblethwaite, 2007) but is instead a language practice that is profoundly connected to speakers’ historical and sociolinguistic realities. It may be that in the Spanish/German context in Barcelona, speakers have not yet perceived and created this CS juncture. Alternatively, there may be certain typological factors involved in certain language pairings that disfavor, but not ban, the occurrence of light verbs with passives.

In sum, there is no doubt that linguistic factors are delimiting factors in CS outcomes, but socio-linguistic and historical factors seem to have had a greater effect in the Northern Belize context. Given the favorable conditions for CS in Northern Belize, BLVCs have served as a CS strategy that allows bi/multilingual speakers to exploit their linguistic resources and capitalize on grammatical parsimony (Matras, 2010, p.83; Toribio, 2004). Although BLVCs seem superfluous

216 and more morphologically complex, they actually evince a regularization of morphological forms in bilingual speech. It can be analyzed as a form of ‘adaptive simplification’, in the sense of Otheguy and Lapidus (2003), since speakers no longer have to deal with phonologically- driven exceptions to the Spanish inflectional system, but rather, their focus lies only on inflecting hacer. Thus, the Spanish verbal paradigm from this perspective is used more parsimoniously.

Furthermore, the creation of new switch sites and skillful incorporation of different features from their languages reveal that speakers establish “common ground” (Edwards & Gardner-Chloros,

2007; Sebba, 1998, p.8) despite the semantic and syntactic incongruences that may exist in the verbal systems of the language varieties.

4.9 Concluding Remarks

The studies presented in this chapter showed that in Northern Belize, BLVCs are not constrained by verb frequency, stativity and/or passivization. Following the works of Le Page and Tabouret-Keller (1985), Sebba (1998, 2009) and Gardner-Chloros (1995, 2010), the current study postulates that code-switchers’ use of BLVCs in Northern Belize was catalyzed by speakers’ positive predisposition toward and identification with bilingual language practices. The current study opens an avenue for further research on interrelated social and linguistic phenomena related to the use of ‘hacer + V’ in Northern Belize, particularly its potential use in classroom contexts. Most importantly, more in-depth, cross-regional investigation of social factors must be carried out (e.g., Northern Belize versus New Mexico), as this kind of comparative research will provide a more lucid, global understanding of how bi/multilinguals employ bi/multilingual language practices vis-à-vis the sociolinguistic atmosphere in which they are found. This would yield a clearer picture of the effect of sociolinguistic factors on the use of these innovative contact phenomena.

217 CHAPTER 5 SEMANTIC CATEGORIES AND GENDER ASSIGNMENT

5.1 Introduction

In this chapter, I investigate two aspects of determiner phrases (henceforth DPs) that have been previously investigated in Spanish/English CS; namely, the incorporation of semantic domains and gender assignment in monolingual versus code-switched speech. To this end, the quantitative analysis of naturalistic, oral production data from 62 native speakers of NBS was conducted. Results revealed both similarities and differences vis-à-vis previous findings for varieties of Spanish/English code-switching in the U.S.

In view of the incorporation of semantic domains, domains which favored non-native nouns in Spanish/English DPs included abstract concepts, work/money-related terms, linguistics/language terms, and everyday items; hence, revealing that bi/multilingual discourse practices are employed in conversations about a variety of topics. For gender assignation, assignment patterns in monolingual DPs were canonical whereas a near categorical preference for the masculine gender was attested in mixed DPs. In contrast to previous findings for CS varieties in the U.S. Hispanophone context, biological gender was not found to be a deterministic factor in switched DPs. The analysis highlights the important role that type of code-switching has on contact outcomes in bi/multilingual communities, as speech patterns are reflective of the status and resourcefulness that code-switching is afforded at societal and idiolectal levels.

This chapter is organized as follows. In section 5.2, I introduce the structures under examination, and I situate the present study within the larger aims of the dissertation. Section 5.3 summarizes previous research on semantic domains and gender assignment. Section 5.4 describes the methodology employed in the present study. In section 5.5, I present the results. In sections 5.6 and 5.7, I provide a discussion of main findings, and I offer some conclusions.

218 5.2 Semantic Category Incorporation and Gender Assignment in Determiner Phrases

In this chapter, I analyze two aspects of DPs that have been previously examined in relation to Spanish/English CS; namely, (i) the incorporation of semantic categories and (ii) gender assignment in monolingual versus code-switched speech. While the former is a semantic feature and the latter a lexical one, together their examination contributes to our understanding of linguistic variation in Spanish and Spanish/English CS varieties. Furthermore, the study of these concomitant phenomena provides insight into the deterministic role of linguistic versus social factors.

Previous work has found that across CS contexts, nouns often constitute the majority of other-language items in naturalistic bilingual discourse (Clegg, 2010, p.223; Gardner-Chloros,

2009, p.31; Haugen, 1950; Herring, Deuchar, Parafita-Cuoto, and Moro, 2010; Jake, Myers-

Scotton, and Gross, 2002, p.72; Moyer, 1992; Muysken, 2000; Myers-Scotton & Jake, 2013;

Otheguy & Lapidus, 2003; Pfaff, 1979; Poplack, 1980; Poplack, Sankoff and Miller, 1988, p.62;

Weston, 2013, but see Hebblethwaite, 2010).

As the example in (1) from Spanish/English CS illustrates, these nouns or noun phrases are often used to express novel ideas and/or concepts (in Myers-Scotton’s (2002) terms, ‘cultural borrowings’).

(1) Ya tenemos un shipping register nuevo

Already have.3PL a.M.S shipping register new

“We already have a new shipping register.”

(Moyer, 1992, p.202)

Similar to the case of BLVCs, mixed DPs have also raised questions as to potential restrictions on what can be borrowed or switched (i.e., semantic categories that are most likely to

219 be borrowed or switched in bilingual speech). In previous work, it has been suggested that particularly education or food terms (Teschner, 1974) or words that are technology-related are incorporated in switched DPs (De la Cruz Cabanillas, Tejedor Martínez, Díez Prados, & Cerdá

Redondo, 2007; Lipski, 2008; Teschner, 1974). In contrast, the general assumption is that semantic categories that belong to frequent, ‘core’ or ‘basic’ vocabulary items, such as kinship terms as in (2), are rarely borrowed or incorporated in bilingual speech (Hock & Joseph, 1996, p.257; Smead, 2000, but see Aaron, 2014).

(2) Mi husband sigue tomando el cuécaro in the mornings.

My huband still eat.3SG.PROG the quaker oats in the mornings

“My husband still eats quaker oats in the mornings.”

(Moyer, 1992, p.206)

When English-origin items such as shipping register in (1) are incorporated in bi/multilingual speech, an inter-related phenomenon that arises is gender assignment, which has also been of interest to scholars. Gender assignment is a lexical process in which other-language nouns are assigned to a gender class (Corbett, 1991)1. In the case of Spanish/English CS,

English-origin nouns are assigned either to the masculine gender, as in (1), or the feminine gender, as in (3).

(3) Yo empecé la high school aquí en Nueva York.

I start.1SG.PRET the.F high school here in New York

“I started high school here in New York.”

(Montes-Alcalá & Lapidus Shin, 2011, p.129)

1 I maintain the distinction between gender assignment and agreement phenomena. At the syntactic level, it has been argued that agreement entails feature checking between an interpretable gender feature on the noun and uninterpretable gender features on other constituents such as determiners and adjectives (Carstens, 2000).

220 It is important to note that whereas Spanish nouns are encoded with grammatical gender,

English nouns are not; thus, presenting a case of structural incompatibility between the two languages in bilingual discourse. Understanding how bilinguals address this structural difference in mixed speech has been of particular interest, given that it reveals how code-switchers bring their languages together at the intra-sentential level in cases of grammatical incongruence.

My main concern here is with the lexical process whereby English-origin nouns and/or noun phrases are assigned to one of the two main Spanish noun classes. Also of relevance to the present analysis is the choice of the determiner in mixed DPs. Recall that whereas Moro (2001) contends that mixed DPs where an ungendered article occurs with a Spanish noun are banned due to syntactic restrictions, Jake et al. (2002) predict that these are possible switches in cases where the ML is English (see Chapter 2). I examine this issue in section 5.4.

In Spanish, animate nouns that refer to people or some animals are assigned to a noun class based on biological sex (e.g., el niño ‘the boy’, la niña ‘the girl’). The vowels –o and –a are the prototypical masculine and feminine morphemes respectively, but there are many exceptions

(Montrul & Potowski, 2007). Notably, some Spanish words ending with –a can be masculine

(e.g., el problema ‘problem’), whereas some words ending in –o can be feminine (e.g., la mano

‘hand’). Similarly, some nouns that end in –e can be either masculine (e.g., el equipaje

‘luggage’) or feminine (e.g., la gente ‘people’), and some nouns that end in –a can be either masculine or feminine, as in la belga/el belga ‘the Belgian woman/the Belgian man’ (Parafita

Couto, Munarriz, Epelde, Deuchar, & Oyharçabal, 2014, p.3). In addition, there are some consonants or consonant combinations that favor the masculine gender, i.e., /n/, /r/, /s/, /l/, whereas others favor the feminine gender, i.e., /ad/, /ión/, /is/ (Clegg & Waltermire, 2009). Thus,

221 there are both semantic and phonological criteria that are operant in the gender assignment process in Spanish.

In Spanish/English CS, several linguistic factors have been proposed to determine gender assignment. Whereas some researchers have found biological sex to be an overriding factor when assigning gender to animate English-origin nouns (Clegg, 2006; Clegg & Waltermire, 2009;

DuBord, 2004; Smead, 2000), the analogical gender criterion (i.e., gender is assigned according to the gender of the translation equivalent) has also been found to be the most deterministic factor (Jake et al., 2002; Liceras, Fernández Fuentes, Perales, Pérez-Tattam, & Spradlin, 2008;

Morin, 2006; Poplack et al., 1982; Smead, 2000).

Importantly, several researchers have also found the overwhelming use of a default gender strategy, where English-origin nouns are assigned to the masculine gender irrespective of the gender of their translation equivalents. This pattern has been attested among Spanish/English bilinguals (Aaron, 2014; Clegg & Waltermire, 2009; Clegg, 2010; Montes-Alcalá & Lapidus

Shin, 2011; Otheguy & Lapidus, 2003, 2005; Smead, 2000; Valdés Kroff, 2016) and L2 learners of Spanish (Franceschina, 2001; Liceras et al., 2008).

In example (4), from Spanish/English CS in New York, the speaker assigns the masculine gender to the compound noun social worker even though the antecedent is transparently feminine.

(4) ¿Y tu mamá? Ella es un social worker, una trabajadora social...

And your mother she be.1SG a social worker, a worker social

“And your mother? She’s a social worker, a social worker…”

(Otheguy & Lapidus, 2003, p.216)

222 Given that the prevalent use of the masculine gender in mixed DPs has been attested in the oral production of bilinguals from different Spanish/English CS communities in the U.S.

Hispanophone context, this phenomenon raises important questions about the ways in which naturalistic code-switchers employ their linguistic resources in bilingual discourse. In addition, it offers fertile testing ground to compare Spanish/English contact outcomes in contexts in which crucial differences exist vis-à-vis the historical and sociolinguistic milieu. Most importantly, given the ubiquitous presence of mixed DPs in naturalistic speech, they provide insight into the nature of CS and convergence. Recall that Bullock and Toribio (2004) and Gardner-Chloros and

Edwards (2004) argue that CS triggers convergence. In light of the sociolinguistic conditions in

Northern Belize (i.e., positive attitudes toward CS, etc.), Muysken’s grammatical typology predicts that there should be evidence of ‘congruent lexicalization’ in the Northern Belize context, particularly among speakers with high bilingual proficiency.

To date, studies on both semantic domains and gender assignment have primarily been carried out in bilingual communities in the U.S. Hispanophone context. In contrast, no previous study has examined these phenomena in Northern Belize. The fact that CS is unmarked and that

BK is ascribed overt prestige in Northern Belize (Balam, 2014) poses important empirical and theoretical questions regarding CS patterns and the status of grammatical gender in Northern

Belize, as grammatical gender is a phi-feature known to be vulnerable to diachronic simplification (Camacho & Sánchez, 2002) and loss in cases of contact (Cornips, 2008, and references therein).

The aim of the present study was two-fold. I was interested in investigating whether the incorporation of semantic domains evinces productivity. If there is no restriction that constrains the borrowing or switching of English-origin nouns, then mixed DPs should show full

223 productivity in terms of the types of semantic categories that are attested in mixed DPs. Thus, mixed DPs should occur freely with different types of semantic domains. Secondly, I was interested in examining the general patterns of gender assignment in monolingual versus switched discourse, especially in relation to previous findings on gender assignment in the U.S.

Hispanophone context. To this end, I examined novel oral production data from 62 native consultants from Orange Walk, Northern Belize.

5.3 Previous Studies

In the ensuing sections, I summarize previous work on semantic category incorporation and gender assignment in Spanish/English CS.

5.3.1 Semantic Domains in Spanish/English Code-Switching

Whereas several studies have examined gender assignment in bilingual discourse, limited research has been carried out to understand the openness of semantic domains to English-origin nouns in bilingual speech. To my knowledge, there are only two studies that have examined semantic categories in Spanish/English CS, and both have analyzed this phenomenon in the

Spanish/English CS variety from New Mexico.

Clegg (2010) analyzed a total of 1875 borrowings2 extracted from 30 interviews with native speakers of New Mexican Spanish. These interviews were from the New Mexico

Colorado Spanish Survey (NMCOSS). Following Poplack et al. (1988), Clegg classified these nouns into 23 different categories: automobiles and transportation (e.g., troca ‘truck’); drugs, cigarettes and alcohol (e.g., juisque ‘whiskey’); food and food preparation (e.g., bacon); music and dance (e.g., club); games and toys (e.g., marbles); sports and hobbies (e.g., cheerleader);

2 Clegg included borrowings that were morphophonologically adapted (e.g., cuara ‘quarter’) and those that were not adapted (e.g., marbles). He did not distinguish between established and spontaneous borrowings. It remains unclear whether a separate analysis of adapted versus unadapted borrowings in his corpus would reveal similar or different patterns.

224 computers, television and technology (e.g., internet); politics, law and crime (e.g., gang); religion and superstition (e.g., temple); medical and health (e.g., pink eye); housing and domestic items

(e.g., garach ‘garage’); farming and nature terms (e.g., bil ‘bail of hay’); relationships, friends and family (e.g., daddy); work-related terms (e.g., business); education (e.g., prom); war (e.g., army); events and holidays (e.g., Thanksgiving); financial and monetary (e.g., cuara ‘quarter’; numbers and dates (e.g., July); units of measure (e.g., yarda ‘yard’); ethnic and people terms

(e.g., Hispanic); and other (e.g., beach).

Overall, Clegg found that no specific domain markedly favored the incorporation of

English-origin nouns. In Clegg’s analysis, the categories with the largest number of loanwords included farming and nature terms (e.g., elque ‘elk’), computers, television and technology (e.g., internet ‘Internet’), education (e.g., rula ‘ruler’), housing and domestic items (e.g., blender), and relationships, friends and family (e.g., grampa ‘grandpa’), which comprised 11.4%, 11.4%,

10.2%, 9.2% and 8.8% respectively, of the total corpus. In contrast, semantic categories that were less open to English-origin nouns included religion and superstition (1%), numbers and dates (0.9%) and drugs, cigarettes and alcohol (0.9%).

Based on his findings, Clegg concluded that lexical need does play a role in the borrowing process, but there are also other factors involved. He noted, for instance, that there is no lexical need to borrow kinship nouns such as daddy or grampa as there are Spanish equivalents. This is also the case of numbers, such as ‘forty five’ (referring to 1945), as there are readily available Spanish terms that can be used instead. According to Clegg, insufficient differentiation and subtle cultural differences between word meanings also play a crucial role, and this is what can account for English-origin forms such as daddy.

225 Bilinguals, Clegg argues, are able to encode fine semantic and cultural distinctions in their choice of English-origin words. Thus, the semantic concept daddy may convey a different meaning than papá. In the traditional New Mexican context, the father would typically work outside of the home and consequently use more English. Thus, daddy may highlight the cultural difference between fathers who speak English to their children versus those who speak Spanish to their children.

More recently, Aaron (2014) also analysed English-origin nouns in the CS variety in New

Mexico. Her analysis was based on 15 interviews from the New Mexico Spanish-English

Bilingual (NMSEB) corpus (Torres Cacoullos & Travis, in preparation). In her study, Aaron classified English-origin nouns into 17 categories. In contrast to Clegg, however, Aaron analysed Spanish (n = 856) and English (n = 608) monolingual data sets and a Spanish/English data set (n = 477) to further examine the openness of semantic categories across different language modes.

Aaron’s analysis revealed that English-origin nouns occurred relatively more often in four domains, namely, kinship terms (e.g., dad), years and number, technology (e.g., treadmill), and vehicles (e.g., troca). In contrast to Clegg’s findings, the frequency with which these concepts were produced as English-origin nouns was clearly more pronounced. When taking into consideration all nouns in monolingual and mixed DPs, kinship terms were produced in mixed

DPs 57% of the time, whereas years and numbers, vehicles and technology where produced

56%, 52% and 40% of the time, respectively. On the other hand, domains that least favored mixed DPs included work/money terms (e.g., spending money), abstract concepts (e.g., weekend), and linguistic terms (e.g., Spanish), which accounted for 7%, 7% and 3% respectively, of the overall nouns in the corpus examined.

226 Thus, although Clegg did find technology-related terms and kinship terms to be frequent

(in line with Aaron), items that relate to years and numbers were found to be infrequent, which suggests that even within the New Mexican context, different communities may differ in their bilingual behaviours. Also recall that Clegg had 21 categories; thus, his more fine-grained distinctions may have affected the frequency counts of certain categories. For example, Clegg distinguished between food and drinks/smoking whereas Aaron did not. Although Aaron used fewer semantic domains, it is noteworthy nonetheless that for three domains, mixed DPs accounted for more than 50% of the total DPs. Based on her findings, Aaron highlighted that

English-origin nouns are used to perform locally determined discourse functions such as naming kin (e.g., daddy).

5.3.2 Gender Assignment in Spanish/English Code-Switching

Several researchers have examined how bilinguals reconcile the structural difference between Spanish and English when assigning gender to English-origin nouns (for Texas Spanish:

Chaston, 1996; García, 1998; for New Mexican Spanish: Clegg, 2006, 2010; Clegg &

Waltermire, 2009; Aaron, 2014; for Southern Arizona Spanish: Dubord, 2004; for New York

Spanish: Montes-Alcalá & Lapidus Shin, 2011; Otheguy & Lapidus, 2003; Poplack, Pousada, &

Sankoff, 1982; for Miami Spanish: Valdés Kroff, 2016). Of relevance to the present analysis are the following studies that have examined gender assignment in naturalistic Spanish/English CS among adult bilinguals.

Jake, Myers-Scotton and Gross (2002) examined oral production data from 10

Spanish/English bilinguals (of South American background) working in the U.S. or studying in

U.S. graduate schools. In their examination of 151 NPs consisting of English nouns and Spanish determiners overtly marked with gender (i.e., articles, quantifiers and demonstratives), 89.4% of

English-origin nouns (135/151) were assigned the masculine gender. In terms of the factors that

227 determine the gender assignment process, their examination revealed that analogical gender was the strongest predictor of gender assignment. Whereas analogical gender accounted for 36% of

English-origin nouns (55/151), the phonological criterion (i.e., gender is determined by the terminal phoneme in loanword) only accounted for 23% of mixed NPs (35/151).

Jake et al. noted that 15% of the data (23/151) could be accounted by both analogical and phonological gender (e.g., este man ‘this man’), whereas the remaining 25% (38/151) could be accounted by neither analogical gender nor phonological gender (e.g., el research/lafem investigaciónfem ‘the research’). For this last group of tokens, they noted that 37 out of 38

English-origin nouns were assigned the masculine gender by default. Thus, when there is no factor that can indicate gender in mixed NPs, the masculine default gender determines gender assignment in bilingual discourse. Jake et al. further argued that biological gender is not a deterministic factor in the gender assignment process. All 13 mixed NPs that were encoded with biological gender were masculine-marked; this occurred even with English-origin nouns that had feminine translation equivalents (e.g., un people person ‘una persona sociable’, unos people ‘una gente/unas personas’). It remains unclear, however, why Jake et al. did not analyze this pattern as an instantiation of the masculine default gender.

Otheguy and Lapidus (2003) also found evidence for the application of the masculine default gender in their examination of 477 English-origin NPs from a corpus of interviews with

33 Spanish/English bilinguals from New York. Otheguy and Lapidus (2003) found that out of

234 English-origin nouns that occurred with a gender-marked article or modifier, 87% of loanwords were treated as masculine, whereas only 13% were treated as feminine, revealing that the analogical criterion (i.e., when gender is determined by the gender of the translation equivalent) was not the most influential factor for gender assignment in their data (e.g. el

228 swimming for ‘lafem natacionfem’ ‘swimming’). They conducted further analysis of three of the interviews they examined and found that for Spanish NPs, masculine- and feminine-marked NPs were almost equally distributed (i.e., 53% and 47% respectively). Thus, the overextension of the masculine gender was only attested in mixed DPs.

In their analysis, they argue that phonology and biological sex are two factors that can account for feminine gender assignment in mixed DPs. The role of phonology is apparent in cases where Spanish speakers perceive -er English words (i.e., boiler) as ending in the prototypical feminine phoneme /a/. These words, when borrowed, are assigned the feminine gender. They suggest a similar process occurs with words that end with /-ay/. Secondly, the role of biological gender can be observed in cases where the biological sex of the referent determines the gender of the loanword (e.g., una teenager ‘the female teenager’). Importantly, they pointed out that biological gender was losing its deterministic quality among some Spanish/English bilinguals from New York, as some speakers assigned the masculine gender even to female referents (e.g., los midwives ‘the midwives’, un social worker ‘a female social worker’).

In light of their findings, Otheguy and Lapidus (2003) highlighted that their consultants have “essentially suspended the application of the common Spanish pattern of arbitrary, memory-dependent gender distinctions” in bilingual speech (p. 216). Bilinguals applied the

“routines of arbitrary, memory-based gender assignment” only to Spanish NPs, but not to mixed

NPs (p. 217). In relation to feminine gender assignment, only the rule-based assignment based on phonology and biological sex (to a lesser degree) remained operant in the assignment of the feminine gender in mixed NPs. Otheguy and Lapidus noted that this phenomenon evinces

‘adaptive simplification,’ as bilinguals simplify only certain aspects of the Spanish gender system in an effort to promote savings in cognitive load when switching between languages.

229 Montes-Alcalá and Lapidus Shin (2011) made similar observations in their analysis of gender assignment in both written and oral production data. For the written data, Montes-Alcalá and Lapidus Shin examined bilingual literary texts by contemporary bilingual Cuban American,

Nuyorican, and Spanish authors. For oral production, they examined interview data from 44 fluent Spanish/English bilinguals from New York. In total, they examined 315 English

NPs from the written corpus and 608 English NPs from the oral corpus.

In their quantitative analysis, they found that the most common strategy in both corpora was to assign the masculine default gender regardless of the gender of the Spanish equivalent noun (e.g., el vegetable soup/lafem sopafem ‘the soup’, un appointment/unafem citafem ‘an appointment’), followed by not assigning any gender at all. A statistical difference was attested between the written and oral production data. Feminine NPs were more common in the written than in the oral data. For NPs that were assigned gender, the masculine default gender was also used more consistently in the oral production data (i.e., 91.2%) versus the written data (i.e.,

79.4%).

Further analysis of 90 feminine-marked mixed DPs showed that in the oral production data, multiple strategies were operative, including the phonological criterion (e.g., una pact), biological gender (e.g., una teenager ‘a teenager’) and the analogical criterion (e.g. la diversity

‘the diversity’) and some that could be attributed to both phonological and analogical criteria

(e.g., la van ‘the van’). In contrast, in the written data, analogical gender largely determined assignment of the feminine gender (i.e., la magic city ‘the magic city’, la big house ‘the big house’). Thus, the analogical criterion was a stronger predictor of feminine gender assignment in written bilingual discourse. Based on their findings, Montes-Alcalá and Lapidus Shin (2011, p.135) concluded that “bilinguals not only avoid assigning gender to loanwords, but also avoid

230 the contexts that make use of the function of gender.” Along the lines of the argument put forth by Otheguy and Lapidus (2003), Montes-Alcalá and Lapidus Shin analyzed this avoidance strategy as a means to lighten cognitive load in bilingual discourse and increase communicative efficiency.

In a recent study on English-origin DPs in Miami Spanish, further evidence was found for the use of a masculine default gender. Valdés Kroff (2016) analyzed a total of 316 mixed DPs extracted from the Bangor Miami Corpus. Overall, his analysis revealed that feminine-marked mixed DPs were infrequent, comprising only 2.5%, lower even than English determiner +

Spanish nouns (e.g., the muebles ‘the furniture’), which accounted for 3.8% of the data. Thus, an overwhelming preference for the masculine default gender was attested (i.e., 93% or 97% considering only gender-marked DPs). Notably, Valdés Kroff also examined the potential role of biological gender in feminine-marked mixed NPs (n = 11) and found that whereas male referents

(n = 8) were categorically assigned the masculine gender, female referents were not categorically assigned the feminine gender (e.g., un Renaissance woman ‘a Renaissance woman’). Only 27% of female referents (n = 3) were assigned the feminine gender. Based on the limited number of feminine-marked mixed DPs, Valdés Kroff posits that the adoption of the default gender strategy occurs in planned CS, whereas feminine-marked mixed DPs evince exceptional switches that are unplanned. He further hypothesizes that the latter DPs are highly restricted to single switches rather than cases of alternational switching, and switches leading to these DPs should be preceded by more disfluencies, repetitions and pauses.

The role of biological gender in the gender assignment process, however, cannot be undermined, as previous work has shown its prominent role in gender assignment. DuBord

(2004)’s examination of 174 DPs extracted from semi-spontaneous interviews with 18 bilinguals

231 from Southern Arizona, for instance, revealed that biological gender is the most consistently applied criterion. In her study, DuBord analyzed the roles of biological gender (e.g., Batman), phonological gender (e.g., dishwasher, terminal phoneme /r/ is a masculine phoneme) and analogical gender (e.g., el mall ‘the mall’, translation equivalent: centro comercialmasc).

DuBord’s results revealed that biological gender was the most deterministic factor in the gender assignment process. In the 35 tokens that were marked with biological gender, male referents were assigned the masculine gender 96.2% of the time, where female referents were assigned the feminine gender 75% of the time. DuBord points out, however, that the two female referents that were not assigned the feminine gender were debatable, as they were both ambiguous (i.e., el baby ‘the baby’; el Bachelorette ‘the Bachelorette’, potentially referring to the program).

Thus, overall, biological gender of the English-origin noun consistently determined gender assignment in cases where nouns were marked with biological sex. For DPs without biological sex, the phonological and analogical criteria were found to have the same influence on gender assignment when analyzed separately. When phonological and analogical genders were analyzed together, however, phonological gender was found to have greater influence on gender assignment than analogical gender (contra Jake et al.’s results). In line with Otheguy and

Lapidus’ findings, analogical gender was not found to play a crucial role in the gender assigned to English-origin words. Importantly, DuBord also found that 7.9% of the corpus data showed variation in gender assignment. This was particularly evident with unestablished loans such as high school and club in the speech of English dominant bilinguals or bilinguals who use Spanish less frequently, suggesting that frequency of use of Spanish also has an effect on the gender assignment process.

232 Clegg and Waltermire (2009) made similar observations about the dominant role of biological gender. In their variationist analysis, they examined 204 NPs extracted from spontaneous conversations with 15 Spanish/English bilinguals from Northern New Mexico.

They tested three criteria: namely, biological gender (e.g., una norse ‘a nurse’), synonymic gender (i.e., the ‘analogical criterion, e.g. un flavour ‘a flavour’ from unmasc sabor), and the terminal phoneme (i.e., the last phoneme in loanword) or phonological criterion. In line with

DuBord’s findings, their VARBRUL analysis revealed that the most accurate predictor of gender assignment was biological gender. The second most deterministic factor was the gender associated with the terminal phoneme, whereas the least deterministic factor was the analogical criterion. In particular, they found that the masculine gender was assigned to 89% of the single,

English-origin nouns. Crucially, overgeneralization of the masculine gender was only operant in bilingual noun phrases but not with native Spanish nouns (n = 123), where there was equal distribution (i.e., masculine-marked DPs: 50%; feminine-marked DPs: 50%).

In contrast to previous analyses, however, Clegg and Waltermire argued that gender is not assigned by any sort of default strategy. Instead, they attributed their participants’ use of the masculine gender to Spanish phonotactic patterns. They point out that observed patterns can be attributed to the fact that typically masculine terminal phonemes for Spanish (i.e., loanwords ending with n, o, r, s, e, l) are more productive than feminine terminal phonemes, which only have one productive terminal phoneme (i.e., a). English-only nouns with new or different terminal phonemes are more than likely associated with the masculine gender, given that this is the group that accepts non-native lexical elements. Furthermore, given that English infrequently uses the terminal phoneme /a/, then fewer English nouns are assigned the feminine gender in bilingual speech. Thus, given that these bilinguals treat English-origin nouns like Spanish nouns,

233 Clegg and Waltermire argue these mixed NPs are illustrative of nonce borrowings rather than

CS.

More recently, Aaron (2014) analyzed 820 lone English-origin nouns from 18 interviews from the New Mexico Spanish English bilingual corpus (NMSEB). In her analysis, Aaron comparatively examined three data sets, namely, a Spanish/English data set (n = 477), a Spanish- only dataset (n = 856) and an English-only data set (n = 608), extracted from the same interviews. Aaron further subdivided the bilingual dataset into ‘singletons’ (i.e., used only once by one speaker) versus ‘diffuse’ items (i.e., used by three or more speakers). Her results revealed that when considering only the NPs that were gender-marked (n = 117), the masculine gender was used 86% of the time whereas the feminine gender was only employed in 14% of singleton items. Though less overwhelmingly, the masculine gender was also employed for diffuse items the majority of the time (i.e., 67%). It is important to highlight that Aaron does not attribute the preference for the masculine gender, however, to CS per se or to a default gender strategy.

Aaron further examined her data sets in terms of non-referential functions that nouns perform (i.e., orienting, classifying, predicating). Her analysis showed that English nouns and singletons are more likely to perform a classifying function, especially vis-à-vis occupation.

Thus, Aaron argues that this partly accounts for bilinguals’ predisposition to use the masculine gender. For Aaron, the use of the English-origin noun social worker in the mixed DP un social worker in (4) does not refer to the person herself, but the category to which she belongs. Given that the category is not marked for gender (i.e., both men and women can be social workers), then it follows that its gender assignment is not necessarily linked to a female. Thus, Aaron concludes that English-origin nouns perform specific, locally determined discourse functions; in

234 the case of the New Mexican context, these functions include naming kin and classifying a person as belonging to a certain occupation.

Overall, the previous studies reveal a clear asymmetry in the gender assignment process in Spanish/English CS. By and large, the masculine gender is overwhelmingly chosen as the default gender, but this may vary depending on the animacy of the loanword. For animate nouns, the biological gender generally remains operant to some extent. It is important to highlight that researchers do not concur on what motivates the pervasive use of the masculine gender or the limited use of the feminine gender in switched DPs. Otheguy and Lapidus (2003) claim that in contact situations, “a process that normally incurs a heavy memory cost is automatized or eliminated in a portion of its range of applicability, thus producing considerable savings in cognitive load” (p.216). Hence, in order for gender assignment to be less taxing on memory in bilingual speech, gender assignment is wholly established on the basis of phonological ending

(i.e., –a for feminine, -o for masculine) and referent sex. This explains why among

Spanish/English bilinguals in New York, the distinction between masculine and feminine nouns is fully maintained for native, Spanish nouns but not for English-origin nouns.

In contrast, Clegg and Waltermire (2009) and Aaron (2014) do not interpret the pervasive use of the masculine gender as indicative of a default gender strategy. Clegg and Waltermire attribute gender assignment patterns in their data to Spanish phonotactic patterns. Aaron, on the other hand, asserts that the preference for the masculine gender is not related to the practice of

CS itself, but to discourse-functional tendencies internal to Spanish. In her analysis, English- origin nouns were employed to perform certain discourse functions such as expressing certain kinship terms (e.g., dad) and classifying individuals as belonging to a certain occupation (e.g.,

Era también farmer ‘He was a farmer too’). These functions, Aaron argues, are determined by

235 community linguistic norms. These views are contra Otheguy and Lapidus (2003), and Montes-

Alcalá and Lapidus Shin (2011), who analyze the adaptive simplication of the gender system as a phenomenon that particularly occurs in bilingual speech.

Although studies on gender assignment in Spanish/English CS have examined production data from bilinguals that are from different communities across the U.S. Hispanophone context, they have all converged on rather similar findings. In the present chapter, we examine how the understudied multilingual Northern Belize context fits into this puzzle of contact Spanish and mixed DPs. To this end, I analyzed non-switched and switched DPs in an effort to elucidate (i) the openness of semantic domains to native and non-native nouns and (ii) gender assignment in monolingual versus code-switched stretches of discourse.

In the next section, I describe the methodology employed in the present analysis.

5.4 Methodology

In this section, I first provide details on the consultants that participated in the sociolinguistic interviews, followed by the criteria that were used to extract and code the data for the quantitative analysis.

5.4.1 Participants

The oral production data analyzed were obtained via sociolinguistic interviews, lasting between 25 and 60 minutes, with 62 consultants (21 males, 41 females) of different ages from

Orange Walk, Northern Belize. Following Tagliamonte (2006, p.46), participants were asked to elaborate on a variety of topics including memorable childhood experiences, past or present school or job experiences, customs and traditions in Northern Belize, folktales, hobbies, food recipes, etc. The author, a multilingual code-switcher from Northern Belize, conducted the interviews, thus diminishing potential experimenter effects on speakers’ linguistic behaviors

(Grosjean, 1998).

236 Information on consultants was collected via a Language Background Questionnaire (see

Appendix F)3. In the case of the older participants (> 50 years old), this information was collected orally and written down on the questionnaires. Given that older consultants did not obtain a high school education (n = 9), this was deemed the most appropriate method to elicit information on their linguistic background. All consultants in the present sample reported being exposed to Spanish from birth. In general, consultants started to learn English between the ages of one and six. The majority reported that they started to learn English at age 5 (n = 36), followed by age 4 (n = 17) and age 6 (n = 5). Only four participants reported that they started learning English between ages 1 and 3. Thus, the majority started learning English at school (n =

51). Only seven consultants reported that they started to learn English at home, whereas four reported that they started learning both at school and home.

In relation to BK, only the younger consultants were fluent or proficient in BK (n = 52, <

40 years old). The majority reported that they started to learn BK between the ages of 10 and 14

(n = 30), whereas the remaining consultants started learning it between the ages of 3 and 9 (n =

22). Importantly, most consultants started learning BK at the primary school (n = 20) or high school level (n = 26). Only four consultants reported that they started to learn BK at home, whereas two reported school and home. Overall, consultants were all Spanish-dominant bi/multilinguals who learnt Spanish as a first language. In the majority of cases, English was learned as a second language, whereas BK was learned as a third language.

In terms of language use, on a Likert scale of 1 – 7 (where 1 indicated ‘rarely’ and 7

‘very often’), the first generation4 unanimously reported using NBS the most (mean = 7). The

3 The LBQs used in the studies were similar but not identical. With every study, minor modifications were made to improve the questionnaire. 4 I do not use the term ‘generation’ as it is traditionally used in the antecedent literature on immigrant bilingualism (e.g., Valdés, 2001, p.6-7). Given that the migration of Yucatec Maya/Mestizos to Belize took place during the

237 second generation reported using NBS (mean = 6.5) and Spanish/English CS (mean = 5.8) the most. In contrast, the post-adolescent group in the third generation reported using BK (mean =

6.4) and Spanish/BK CS (mean = 6.0) the most. Lastly, the adolescent group in the third generation reported more frequent use of Spanish/English CS (mean = 6.1) and NBS (mean =

6.0). For self-rated language proficiencies, on a scale of 1 – 7 (where 1-3 indicated ‘poor’ and

6-7 indicated ‘excellent’), all age groups rated their proficiency in their native language the highest, with group means ranging from 5.9 for the second generation to 7.0 for the first generation.

Information from the LBQ was used to divide participants according to different age groups. Given that the present study’s participants were largely homogenous in terms of ethnic and linguistic background, participants were divided into three main groups, according to the domains of family, work, and education. The first generation (n = 10) comprised the older group

(ages 50 – 99), which included only Spanish-dominant bilinguals whose daily conversations were relegated to the home or familial domain. The second generation (n = 27) included dynamic

Spanish/English/Kriol bi/multilinguals (ages 21 – 40) who held a full-time job and whose daily interactions primarily took place in the sphere of work. Lastly, the third generation (n = 25) or the younger group (ages 14 – 20) included Spanish/English/Kriol bi/multilinguals whose interactions particularly occurred in the school domain, either at the secondary or (associate degree) college level. Although the associate degree group was small, the distinction between adolescents and post-adolescents was maintained as Balam (2013) found in previous attitudinal

Caste War of Yucatan (1847-1901), the present study’s first generation actually comprise second, third or even fourth generation descendants. Maintaining this generational distinction was important, however, given that that there were salient differences between older and younger age groups in terms of their English proficiency and their use of bi/multilingual language practices.

238 work that there were some significant differences in Northern Belizean adolescents’ and post- adolescents’ perception of and predisposition toward the use of CS.

5.4.2 Data

In this section, I describe the criteria that were used to select tokens and the categories that were used to code tokens for quantitative analysis.

5.4.2.1 Semantic categories

Given that gender marking did not have bearing on the analysis of semantic domains, I considered all non-switched and switched DPs containing overt determiners (i.e., both gender- marked determiners such as definite articles and those not marked with gender such as possessive pronouns). A total of 14,865 tokens extracted from the interviews were considered for the distributional analysis which included a Spanish data set (n = 8447), a Spanish/English/Kriol data set (n = 4739), and an English/Kriol data set (n = 1679). A subsequent logistic regression analysis was conducted on VARBRUL using only the Spanish and Spanish/English/Kriol data sets to further examine trends observed in the distributional analysis. Tokens that were excluded from the gender assignment analysis were also excluded from the semantic analysis.

For the analysis of semantic domains, tokens were coded using Aaron’s (2014) classification system with minor modifications (see section 5.4.2.2). Table 5-1 below provides the semantic categories and examples (non-switched and switched) from the present corpus.

239 Table 5-1. Semantic categories and examples from Northern Belize bi/multilingual speech Semantic Domain Examples Kinship terms mis primos ‘my cousins’; mi dad ‘my dad’ Everyday items la bola ‘the ball’; el mirror ‘the mirror’ Time, events and places el parque ‘the park’; un house bash ‘a house party’ Person (not kin) el vecino ‘the neighbour’; el counselor ‘the counselor’ Year, number or quantity un poquito ‘a little bit’; el otro half ‘the other half’ Work- or money-related nuestro salario ‘our salary’; un loan ‘a loan’ Related to the land/earth el cerro ‘a hill’; el river ‘the river’ Technology mi teléfono ‘my phone’; sus atm ‘their atm’ Academia (or school) una clase ‘ a class’; el word bank ‘the word bank’ Food, drink, and smoke la arina ‘the flour’; su beer ‘her beer’ Vehicle/transport-related un barco ‘a ship’; los four tires ‘the four tires’ Institution la ley ‘the law’; el government ‘the government’ Animal la serpiente ‘the snake’; un horse ‘a horse’ Abstract concepts una vida feliz ‘a happy life’; tus goals ‘your goals’ Health and body su espalda ‘her back’; el heart ‘the heart’ Domestic life la cocina ‘the kitchen’; el stove ‘the stove’ Linguistics or language el inglés ‘English’; su Spanglish ‘her Spanglish’ Other la cosa ‘the thing’; el same thing ‘the same thing’

Although previous work on both semantic domains (Clegg, 2010) and gender assignment

(Clegg & Waltermire, 2009; Jake et al., 2002; Valdés Kroff, 2016) have examined English-only nouns in relation to potential corresponding Spanish equivalents, I do not focus on this aspect.

Clegg (2010, p.224) aptly highlights in his study that one of the challenging issues that arises when identifying whether a loanword represents a lexical gap or not is that “there is no reliable way of knowing what words do and do not form part of the lexicon of [a] speech community”.

To have an insight into this phenomenon (i.e., determine whether there was an existing counterpart in the language), Clegg gave the list of English-origin nouns extracted from his interviews to a native speaker of New Mexican Spanish so that the speaker could provide corresponding Spanish equivalents when possible. The speaker provided equivalents only for

48% of the noun types. In a recent study on bilingual light verb constructions, Balam & Prada

Pérez (2016) conducted a similar follow-up task. In this task, two prolific users of BLVCs were

240 asked to provide potential Spanish equivalents for the bilingual verbs they had used during interviews. Speaker PA14 only provided verbal equivalents for 52% of 90 different BLVCs, whereas PA18 provided equivalents for 85% of 85 different BLVCs.

These findings reveal lexical gaps are a normal aspect of the bilingual lexicon, but these

‘gaps’ should not be interpreted as a linguistic liability. In many cases, it simply reflects the unmarkedness and dynamic nature of CS in bi/multilingual communities. The assumption that all balanced Spanish/English bilinguals have equivalents for every single English-origin item in their switched discourse is not only unrealistic, but it essentially propagates the notion of the mythical bilingual, which as Valdés (2001) rightly highlights, does not exist (for further discussion, see Chapter 6).

Following Clegg (2010) and Balam & Prada Pérez (2016), in an effort to better understand analogical gender patterns in the present corpus, a follow-up task was conducted with

WO51, a second generation consultant5, where she was asked to provide potential equivalents for the 3,587 English-origin nouns that were gender-marked. Note that in an ideal investigation of lexical availability, all consultants would complete such a task, providing a more fine-grained picture not only of within-sample but interspeaker variation as well6. Here, I provide only an insight into this phenomenon. Results are presented in section 5.5.6.

5 A second generation consultant was randomly chosen given that the second generation constitutes the group of speakers who have more fully developed their bi/multilingual proficiencies in this context. In contrast to adolescent and post-adolescent participants, who are still developing their skills in a classroom setting, second generation participants are dynamic bilinguals who make everyday use of their linguistic repertoire in a variety of social situations.

6 In order to determine the status of English-origin items, one would have to thoroughly investigate not only whether an equivalent or near-equivalent actually exists, but one would also have to determine whether it is used in a community, how frequently it is used, whether the equivalent is known to individual speakers and subgroups of speakers, and whether individual speakers regard the word as belonging to one language versus another (Pfaff, 1979).

241 5.4.2.2 Gender assignment

For the analysis of gender assignment, I examined only non-switched and switched DPs that were overtly gender-marked, as in (5) and (6) respectively. I excluded English/Belizean

Kriol DPs as English and BK nouns are not marked for grammatical gender.

(5) ‘Taba haciendo snow esa noche

‘It was snowing that night.’

[AD17, female, 17]

(6) Es un special bond que se hace develop entre students.

‘It’s a special bond that develops among students.’

[WO44, female, 31]

In total, 14,865 DPs containing overt determiners in the interviews were extracted and coded for presence/absence of gender marking on the determiner, gender on the determiner, gender on the noun, and semantic category of the noun. Not considering exclusions in the non- switched (n = 299) and switched (n = 148) data sets, a total of 13,186 tokens (8,447 non- switched DPs, 4,739 switched Spanish/English DPs) were extracted from the 62 interviews.

Further quantitative analysis of overtly gender-marked DPs in the non-switched (e.g., esa noche) and switched data sets (e.g., un special bond), comprising 68% and 75% of total tokens respectively, was conducted to examine gender assignment in Spanish monolingual and code- switched DPs. In contrast to previous studies that have examined relatively small data sets of

Spanish/English DPs that are overtly gender-marked (e.g., Clegg & Waltermire, 2009: 204 tokens; DuBord, 2004: 174 tokens; Jake et al., 2002: 151 tokens; Otheguy & Lapidus, 2003: 234 tokens; Valdés Kroff, 2016: 315 tokens), the present gender assignment analysis is based on

3,587 overtly gender-marked mixed DPs.

242 The first criterion for the extracted tokens was that they minimally comprise an overt determiner and a noun or canonical null nominal. All Spanish monolingual and switched phrases where a noun was not preceded by a determiner, such as nouns in prepositional phrases as ‘entre students’ in (6) were not included in the analysis. For both non-switched and switched DPs, all cases with pronominal forms were excluded. For Spanish DPs, I excluded cases such as el otro uno ‘the other one’ and similar ‘uno constructions’, which constitute a potential syntactic calque from English (Balam, 2014). To ensure consistency, I also excluded these cases in the switched data (i.e., el worst one, el first one, los same one, los main ones, el busiest one, etc.).

For non-switched tokens, I also excluded DPs with names of people and places (e.g. una diferente Megan ‘a different Megan’, el Cayo ‘Cayo district’) and non-canonical DPs where the nominal element was a quantifier referring to time (e.g., las dos ‘2 p.m.’). I did include, however, canonical null nominals (e.g., los dos (chiquitos) ‘the two boys’, las dos (muchachas) ‘the two girls), as the gender assignment process was operative in these cases. Following Aaron (2014), I also included nominalized adjectives in both non-switched (e.g., la última ‘the youngest one’, los buenos ‘the good students’) and switched DPs (e.g., los middle division ‘the middle division students’, un permanent ‘a permanent job’).

For non-switched DPs, I did keep tokens with Mayan-derived nouns (e.g., ese wech ‘that armadillo’, la xtabai ‘the forest spirit’, el xix ‘leftovers’), as these nouns were already lexicalized in the variety of Yucatan Spanish initially brought by the Yucatec Maya and Mestizos who originally migrated to Northern Belize in the 1840s. I excluded all cases, however, where the

Spanish head noun was modified by an English adjective (e.g., su own comida ‘her own food’, un big hecho ‘a big fact’), as the focus of this examination was on examining DPs that were entirely Spanish monolingual versus bilingual DPs containing English-origin nouns. While the

243 presence and/or placement of modifiers in switched DPs in an interesting area of study, I do not explore this issue here.

Following Otheguy and Lapidus (2003) and Montes-Alcalá and Lapidus Shin (2011), in the switched data, I also excluded all English-origin nouns that are potentially utilized in monolingual Spanish-speaking countries such as Internet, tablet, iPad, gas, taxi, etc. All food- related nouns such as relleno and recado were also excluded from mixed DPs, as these nouns are often phonetically indistinguishable in monolingual versus switched discourse in Northern

Belize. They are also used by Belizeans of other ethnic backgrounds (e.g., Creoles, Garinagu, etc.), who may have limited or no proficiency in NBS. Lastly, the small number (n = 29) of attested Kriol nouns (e.g., un Spanish gyal ‘a Hispanic girl’, mi grani ‘my grandmother’) were included in the analysis, but not distinguished from English nouns, as most BK nouns are derived from English (Holm, 1977, p.1). Thus, many English lexical items in Belize could potentially be considered both English and Kriol due to some of the phonological overlaps between the two systems (Balam, 2014, p.88).

I subscribe to Haust’s (1995), Gardner-Chloros’ (2009, p.31), and Myers-Scotton’s

(2002) view that loans and switches occur on a continuum, as discussed in Chapter 2. Whereas the borrowing/CS distinction was crucial in previous studies (e.g. Aaron, 2014), it does not play a central role in the present investigation. Consonant with similar work on the study of gender assignment (Otheguy & Lapidus, 2003; Parafita et al., 2014), my main concern was investigating how other-language nouns are assigned Spanish gender, irrespective of their status as borrowings or switches.

5.5 Results

Before examining semantic category incorporation and gender assignment in monolingual versus code-switched speech, I first provide a background on the status and use of

244 grammatical gender in NBS. Although a full analysis of gender agreement is beyond the scope of this study, it is advantageous to have a general insight into the use of grammatical gender agreement in NBS.

5.5.1 Grammatical Gender in Northern Belizean Spanish

To my knowledge, no previous study has investigated the use of grammatical gender in

NBS. We do get insights into this phenomenon, however, in Hagerty’s (1996, p.137) brief descriptive analysis of Belizean Spanish, where he observes that a common characteristic of

Belizean Spanish are cases of “gender confusion” (e.g., el primer vez ‘the first time’, el gente

‘the people’, la problema ‘the problem’). Hagerty’s observation suggests not only that that non- standard patterns of gender assignment are pervasive in this variety of contact Spanish, but that the gender system may have been undergoing linguistic convergence in the direction of English.

Hagerty asserts that “since the English noun phrase has no for gender, this gender confusion could be seen as a step in the direction of the English structure” (p.137). This begs the question of whether there is some degree of convergence toward English in NBS, specifically as it relates to gender assignment in monolingual NBS stretches of discourse.

In a description of NBS morphosyntactic characteristics, Balam (2014) further points out that the lack of gender agreement is indeed a salient feature of the Spanish variety spoken in

Northern Belize. In oral production data from adolescent speakers, he noted examples such as

(7), where there is lack of gender agreement both at the DP and verbal phrase levels.

(7) Encontró una rana morado que estaba sentado en un shell

Find.3SG.PRET a frog.F purple.M that was sitting.M on a shell

“He found a purple frog sitting on a shell.”

(Balam, 2014, p.90)

245 In this case, the adjective morado and the progressive verb form sentado do not agree with the head noun rana, which is overtly marked with the feminine gender.

In the naturalistic oral production data from this study, it was rather notable that the overextension of the masculine gender was also attested in other grammatical contexts.

Table 5-2. Gender dis(agreement) in Northern Belizean Spanish

Type of phrase and examples Adjectival (within-phrase) La escuelita antes era unafem.sg escuelitafem.sg chicomasc.sg [EL56, 51] “At that time, the school was a small school.” Me gusta ver películasfem.pl románticosmasc.pl [PA20, 21] “I like to watch romantic movies.” Con lafem.sg mismafem.sg hojafem.sg secomasc.sg haces el fuego. [AD11,16] “With the same dry leaf, you make the fire.” …porque habían muchasfem.pl cosasfem.pl buenosmasc.sg allá [EL57, 52] “…because there were many good things there.”

Predicative (across-phrase) Hay familiasfem.pl que estan divididosmasc.pl… [WO45, 27] “There are families that are divided…” …pero lafem.sg comidafem.sg allá es más sabrosomasc.sg [PA20, 21] “…but the food there is nicer.” El problema….es que lafem.sg comunidadfem.sg está muy separadomasc.sg [WO48, 30] “The problem…is that the community is too separate.” Había bastante ruínasmasc.pl que nisiquiera ‘taban tocadomasc.sg [EL58, 71] “There were many ruins that were not even excavated.”

Clitic Mi papá consigue lafem.sg hojafem.sg…y mi hermano lomasc.sg limpia [PA50, 19] “My father gets the leaf…and my brother cleans it.” Mi niñezfem.sg lomasc.sg pasé en una aldea. [WO45, 27] “My childhood, I spend it in a village.” ‘Ta bonito lafem.sg mayafem.sg pero ahora bastantes no lomasc.sg saben. [EL53, 99] “The Mayan language is beautiful but now many people can’t speak it.” …pero ella, ese tiempo, ella lomasc.sg tostaba…lafem.sg canelafem.sg… [EL54, 51] “…but she, at that time, she would toast it…the cinnamon…” *Within brackets appears consultant number followed by speaker’s age.

As Table 5-2 illustrates, lack of agreement in within-phrase (i.e., adjectival) and across- phrase (i.e., predicative and clitic) contexts was attested in the speech of the present sample of

246 speakers. Notably, this overuse of the masculine gender was found across generations. Thus, variability in the use of gender agreement does not seem to be restricted to any particular age group.

It is important to underscore that speakers variably employ canonical patterns of gender agreement. Here, I provide only examples with lack of gender agreement in an effort to illustrate how the use of variable gender agreement has become the norm in this community. This is a striking phenomenon given that the lack of gender agreement is typically associated with speakers of Spanish Creoles (Baptista & Guéron, 2007; Holm, 2000, 2006; Lipski, 2005; Lipski

& Schwegler, 1993; Porras, 2013, among others) and Afro-Hispanic varieties (Lipski, 2006;

Sessarego, 2013), but not with balanced bilinguals, especially Spanish-dominant bilinguals.

In fact, for Spanish/English bilinguals in the U.S., Lipski (1993, p.161) contends that in contrast to vestigial Spanish speakers and second language learners, “native Spanish speakers, including true bilinguals whose Spanish exhibits massive English structural and lexical interference, virtually never commit errors of adjectival inflection.” In the case of native Spanish speakers from Northern Belize, however, we see that incongruences in agreement constitute simply another distinctive aspect of their Spanish variety.

It is not surprising, therefore, that in the analysis of passive BLVCs (Chapter 4), there was a tendency for speakers to give higher ratings to masculine-marked passive BLVCs (where hacer did not match with the feminine head noun). While masculine-marked passive BLVCs and utterances such as those in Table 5-2 illustrate the non-canonical use of gender agreement, they bring to the fore the possibility that the overgeneralization of the masculine gender has become more pervasive in the gender assignment system of NBS; hence, evincing convergence. This is

247 one of the concomitant issues I explore in the present cross-generational analysis of monolingual and switched DPs produced by Northern Belize bi/multilinguals.

In the next section, I present an overview of overall DP production across the four age groups. I subsequently examine mixed DPs in relation to predictions made by Moro (2001) and

Jake et al. (2002). Lastly, I present the quantitative analyses for semantic domains and gender assignment in monolingual versus code-switched speech.

5.5.2 Non-Switched versus Switched DPs across Age-groups

In line with the reported patterns of language use obtained in the LBQ, the extracted DPs revealed that younger groups frequently employed Spanish/English CS, whereas the ten oldest participants generally stayed in a Spanish monolingual mode. Table 5-3 shows that that there is considerable interspeaker variation in terms of the number of DPs produced per interview across the different age groups. The standard error7 and range for Spanish DPs is particularly pronounced among the post-adolescent group (ages between 18-20) and the eldest group (ages between 50-99). In the case of the latter group, this pattern can be attributed to the fact that their interviews were generally longer.

Table 5-3. Production of determiner phrases across age groups Age group Spanish Spanish/English English/Kriol Mean SE (Range) Mean SE (Range) Mean SE (Range)

14 – 17 (n = 18) 90 9.4 (20-166) 76 9.3 (18-180) 12 3.4 (0-59) 18 – 20 (n = 7) 105 28.6 (31-226) 81 12.1 (39-139) 41 15.3 (1-116) 21 – 40 (n = 27) 114 9.1 (35-197) 99 8.0 (42-215) 43 8.2 (1-165) 50 – 99 (n = 10) 303 29.7 (94-502) 15 6.4(0-70) 1 0.7 (0-7)

7 Due to the very marked differences across speakers in the production of DPs, the standard error measure was employed to better illustrate the variation of DP production across the three conditions (i.e., Spanish, Spanish/English, English/Kriol).

248 350

300

250

200

150

100 Mean DPs of Mean number 50

0 14 - 17 18 - 20 21 - 40 50 - 99 Spanish DPs Spanish/English DPs English/Kriol DPs

Figure 5-1. Mean production of determiner phrases by age group and type of determiner phrase

Overall, as Figure 5-1 shows, the post-adolescent group is the group that had the most variation in DP production across the three conditions (Spanish DPs, Spanish/English DPs,

English/Kriol DPs). A tendency for less variation was attested in the production of English/Kriol

DPs, particularly in the youngest and eldest group, a trend that is in line with the fact that in

Belize, the use of English is marked. This is also consistent with the speakers’ reported patterns of language use.

Figure 5-1 also reveals that the eldest speakers, who had limited speaking proficiency in

English, produced the fewest Spanish/English DPs, whereas younger speakers produced a significantly greater number of Spanish/English DPs, in line with the fact that younger generations in Northern Belize have a higher proficiency in English than older consultants (i.e., ages 50+). Across groups, English/Kriol DPs were produced the least, consistent with the marked status of monolingual English use in Northern Belize informal contexts (Balam, 2013a,

2015).

249 Notably, in their use of English/Kriol DPs, the youngest group seemed to pattern more with the eldest group than with the intermediate younger groups. In both the adolescent and eldest speaker groups, there were consultants who did not produce any English/Kriol DPs. On the other hand, in their use of Spanish/English DPs, the youngest group patterned more with the intermediate age groups.

A simple linear regression analysis was conducted to examine the extent to which the mean proportions of mixed DPs produced by the different consultants could be predicted from age. The model including age as a predictor accounted for 42% of the variance in the production of mixed DPs in the present corpus. Thus, the older the participant, the smaller the proportion of mixed DPs in comparison to Spanish and English/Kriol DPs s/he produced (ß = -.74, t-value = -

6.65, p < .001). Figure 5-2 below illustrates the bi-modal distribution of mixed DPs across speakers, revealing that there were two main linguistic groups, i.e., the younger groups, which behaved more or less homogenously and the eldest group, which produced few or infrequent

Spanish/English DPs.

Figure 5-2. Mean proportions of switched DPs across different age groups

250 In sum, the cross-generational use of DPs was markedly different between younger and older speakers in the overall production of mixed DPs, a finding which suggests that there is indeed a correlation between bilingual proficiency and the prevalence of CS practices. It is primarily the young speakers who are proficient in English and Kriol, i.e., those between ages14

– 40, who frequently produced Spanish/English DPs.

In the next section, I examine the production of these mixed DPs more closely by investigating speakers’ choice of language when switching from determiners to noun phrases. As we saw in Chapter 2, this topic has received considerable attention in the CS literature and is thus of interest to the present study.

5.5.3 Choice of Determiner in Mixed DPs

The data revealed that the vast majority of mixed DPs occurred with Spanish rather than

English determiners, in line with both the Bilingual NP Hypothesis (Jake et al., 2002) and the

Minimalist approach to CS (Moro, 2001; MacSwan, 2005). The occurrence of DPs consisting of an English or BK determiner and Spanish nouns (e.g., di hoja ‘the leaf’, my suegra ‘my mother- in-law’, fu dem novelas ‘their novelas’) were highly infrequent, but they were nonetheless attested, contra Moro’s (2001) claim that they should not occur.

There were only 17 English/Spanish DPs, accounting for 0.4% of the overall switched data set (n = 4746) involving Spanish and English. This was a smaller percentage than that reported by Herring et al. (i.e. 4%, 15/373) and Valdés Kroff (3.8%, 12/316). In the present corpus, the majority of English/Spanish DPs (n = 14) were produced by one second generation, female speaker who frequently switched between Kriol, Spanish and English. Thus, there is the possibility that this pattern is more frequent among Kriol-dominant bi/multilinguals in Northern

Belize or that it is an idiosyncrasy of some speakers. Further research is needed to investigate this issue.

251 Importantly, a Minimalist approach to CS is unable to account for these switches. In contrast, Jake et al. predict that switches with English or Kriol articles are possible provided that the ML is English or Kriol rather than Spanish. The 14 switches produced by the second generation speaker offer support for Jake et al.’s claim, as they generally occurred in switched discourse where the ML was either English or Kriol. In line with the MLF, one way of determining the ML is by identifying the language of the finite verb in the mixed constituent

(Herring et al., 2010, p.562). In example (8), produced by the Kriol-dominant trilingual speaker, the switch occurs from Kriol to Spanish. In the bilingual constituent containing the switch olla, the Kriol verbal construction wah baada ‘will bother’ is employed, revealing that Kriol is the

ML.

(8) Cuando entro, ai noh smel noting yu noo, pero vi que el stove ‘ta on an ai seh, “Ai noh

even wah baada lif up di olla, que se quede allí. Ya sta charcoal.”

‘When I go in, I don’t smell anything you know, but I saw that the stove was on, and I

said, ‘I will not even bother to lift up the pot. Let it stay there. [The beans] are already

burnt’’.

(WO27, female, 29)

In (9), similarly, the speaker smoothly switches between English, Spanish and Kriol. In this case, however, the mixed phrase di masa ‘the corn dough’ is embedded in a bilingual constituent where the main finite (control) verb is in Spanish (i.e., usas ‘you use’), whereas the lexical verb is in Kriol (i.e., fu wet ‘to dampen’). In the case of di hoja ‘the leaf’, the finite verb is in English (i.e., spread).

(9) Some people use manteca de cochino, but if yuh noh have dat, usas aceite, normal

aceite fu wet di masa, y you spread it pan di hoja…

252 ‘Some people use lard, but if you don’t have that, you use oil, the regular cooking oil to

make the corn dough wet, and you spread it over the leaf…’

(WO27, female, 29)

The example in (10) also shows that the finite verb (i.e., do) used in the mixed utterance matches with the language of the determiner in the mixed DP (i.e., mai ‘my’), in line with the

MLF’s prediction. The case of (11) is a bit unusual, in that within the bilingual constituent where the mixed DP occurs, the speaker verbalizes the Spanish noun masa ‘corn dough’, yet the structure is Kriol-like, as these uninflected verbal forms are distinctive of Kriol but not Spanish or English (e.g., Kriol sentence: Shee laik cook ‘She likes to cook’). Thus, the verb amasar ‘to make corn dough’ is morphologically Spanish, but syntactically, it behaves like a Kriol verb.

Thus, in this case, it is not clear whether Spanish or Kriol would be the ML.

(10) But den, wen ai do tamales wit mai suegra, ella quick thing! Laik...wee cuda staat

nain ah'claak an wee don di eet laik eleven tirti, twelv.

‘But then, when I do tamales with my mother-in-law, she does it quickly! Like…we

could start making them at 9:00 a.m. and we are already eating tamales at 11:30 a.m. or

midday.’

(WO27, female, 29)

(11) Wen ai mi di stay wit mai ma, umm no puedo está solo sentada. Ai mean, she seh,

"Haz algo, haz algo!" ¿Pero quéee? Like you know, I can't just relax? Tengo que ‘ta

cocinando, tengo que está esto y esto…like my suegra, she amasar everyday, Ai goh

laik, “Mhm-hm. Dat’s wai dehn gaat Robi's bred” [laughter].

253 ‘When I was at home with my mother, um, I cannot just be sitting down. I mean, she

says, “Do something! Do something! But what? Like you know, I can’t just relax? I have

to be cooking, I have to be doing this and that….like my mother-in-law, she kneads flour

everyday. I say, “Mhm-hm. That’s why there is bread at Robi’s bakery.’

(WO27, female, 29)

There were three examples of English/Spanish DPs (each produced by a different speaker) in the corpus that could not be accounted for by Jake et al.’s analysis. These mixed DPs occurred with determiners that were quantifiers (i.e., one or two, eighty five, certain), as in (12).

Unlike the examples from the Kriol-dominant trilingual, these were cases where the finite verbs were in Spanish rather than English, contra Jake et al.’s prediction.

(12) Sí, so cada maistro tendrá como one or two niños que really um necesitan ese, ese full

attention

‘Yes, so every teacher will have, like, one or two children who really, um, need that, that

full attention.’

(WO41, male, 30)

Overall, although the MLF does not correctly predict every single switch where an

English determiner occurs alongside a Spanish noun, it must be highlighted that the Northern

Belize data support the view that these kinds of switches are grammatical (contra Moro’s

Minimalist (2001) analysis) and perhaps commonly used among certain sub-groups of Northern

Belize bi/multilinguals. Based on examples such as (8) – (11), it may be that Kriol-dominant bi/multilinguals who switch between Kriol and Spanish employ this type of mixed DP more frequently. This is a hypothesis that merits further investigation.

254 5.5.4 Semantic Category Incorporation

To analyze the openness of semantic categories to English-origin nouns, tokens in the present corpus were coded using Aaron’s (2014) classification system. I examined Spanish monolingual, Spanish/English DPs, and English/Kriol DPs in order to obtain a more holistic view of the openness of semantic domains in the data.

Note that in the case of English/Kriol DPs, however, I did not distinguish between

English (e.g., nine years, a big problem, etc.), Kriol (e.g., wah gyalfren ‘a girlfriend’, fu she prablem ‘her problem’, etc.) or English/Kriol DPs (e.g., my bredda ‘my brother’), as the study’s main focus was on Spanish/English CS. For the purpose of comparability, exclusions in the English/Kriol DPs were done following the same criteria outlined in section 5.4.2.2.

During the coding process, some modifications were done to a few categories. For example, quantity (e.g., un poco ‘a little bit’, la mayoría ‘the majority’) was added to year and number. Abstract concepts included only canonical abstract nouns (e.g., energy, life, awareness) whereas abstract time/space concepts were classified as time.

Although temporal nouns such as ‘hour’, ‘week’, ‘weekend’, and ‘year’ are abstract, they constitute a sub-set of fixed concepts that are typically used in reference to a specific moment in time. They also do not have the semantic element of individual relativity that applies to canonical abstract nouns such as ‘life’ and ‘happiness’. Therefore, these nouns are subsumed under time, events, and places in the present study.

Table 5-4 presents the distribution of DPs across the 17 categories used in the present investigation.

255 Table 5-4. Openness of semantic domains to nouns in different types of DPs

Semantic Domain Spanish Spanish/ English/ Overall % DPs English DPs Kriol DPs Spanish/ English DPs N % N % N % Total % Kinship terms 1410 16.7 146 3.1 96 5.7 1652 9 Everyday items 257 3.0 116 2.4 24 1.4 397 29 Time, events and places 1734 20.5 693 14.6 401 23.9 2828 25 Person (not kin) 1482 17.5 642 13.5 204 12.2 2328 28 Year, number or quantity 249 2.9 87 1.8 47 2.8 383 23 Work- or money-related 227 2.7 336 7.1 116 6.9 679 49 Related to the land/earth 522 6.2 127 2.7 19 1.1 668 19 Technology 100 1.2 165 3.5 47 2.8 312 53 Academia (or school) 430 5.1 894 18.9 142 8.5 1466 61 Food, drink, and smoke 406 4.8 193 4.1 59 3.5 658 29 Vehicle/transport-related 45 0.5 84 1.8 8 0.5 137 61 Institution 45 0.5 25 0.5 23 1.4 93 27 Animal 152 1.8 31 0.7 9 0.5 192 16 Abstract concepts 584 6.9 891 18.8 323 19.2 1798 50 Health and body 233 2.8 82 1.7 38 2.3 353 23 Domestic life 40 0.5 8 0.2 4 0.2 52 15 Linguistics or language 190 2.2 169 3.6 52 3.1 411 41 Other 341 4.0 50 1.1 67 4.0 458 11 Total 8447 100 4739 100 1679 100 14865 32

Overall, across conditions, nouns occurred in a variety of semantic contexts. In contrast to Clegg’s (2010) findings for Spanish/English CS in New Mexico, there were some marked tendencies in the favorability of certain semantic contexts to native and non-native nouns in the

Northern Belize data. For example, nouns related to academia occurred more often in switched

Spanish/English DPs (18.9%) rather than in Spanish (5.1%) or English/Kriol DPs (8.5%). Nouns related to time, events and places occurred more often in Spanish (20.5%, e.g., la hora ‘the hour’, al otro día ‘the next day’, la cárcel ‘prison’, una fiesta ‘a party’) and English/Kriol DPs

(23.9%, e.g., eleven years, the next month, fifteen minutes, an appointment, etc.).

256 Notably, abstract concepts seemed to favor English nouns in both mixed Spanish/English

DPs (18.8%) and English/Kriol DPs (19.2%). Unlike Aaron’s (2014) findings, however, kinship terms favored monolingual Spanish nouns but not nouns in switched Spanish/English DPs

(16.7% vs. 3.1% respectively). This indicates that the propensity for kinship terms to favor

English-origin nouns (e.g., mi dad ‘my dad’) is specific to New Mexican communities.

The last column in Table 5-4 shows the overall proportions of Spanish/English DPs in relation to the total number of DPs produced across semantic domains. It reveals that the most favorable contexts for non-native nouns in Spanish/English DPs include work- or money-related terms (49%), technology (53%), academia (61%), vehicles/transport (61%) and abstract concepts

(50%). Whereas vehicles and technology were found to favor English-origin nouns in Aaron’s

(2014) study (52% and 40% respectively), work/money and abstract concepts were not found to be favorable contexts (7% in both cases).

In the present data, abstract concepts was a favorable domain for English-origin nouns in both Spanish/English (e.g., superstitions, lifestyle, childhood, dreams, trouble, style, culture, mistake, miracle, freedom, etc.) and English/Kriol DPs (e.g., fun, mentality, opportunity, reality, challenge, ambition, mischief, faith, identity, soul, etc.), whereas in Aaron’s study, abstract time/space concepts favored, though not markedly, Spanish nouns more (12%).

In the present analysis, it was noteworthy that favorable domains for nouns in

Spanish/English DPs included academia and work/money. This suggests that in the Northern

Belize context, speakers routinely switch when having conversations that pertain to the spheres of school or professional work. In the interviews, consultants spoke about both school-related

(e.g., mi class ‘my class’, ese subject ‘that subject’, un low grade ‘a low score’, mi law studies

‘my law course’ etc.) and work-related topics in code-switched discourse (e.g., un profit, el

257 currency, un financial report, al office, el market price, el whole department, un promotion, etc.).

In (13) below, for example, WO36 who is a bank teller comments on errors bank tellers sometimes make in their jobs, such as placing a cash deposit in a wrong account.

(13) En veces hay simple mistakes que tú haces y eso puede hace cause que tengas un

difference at the end of the day…en veces si no te haces concentrate bien haces tontera,

like…si tú no ‘tas haciendo pay attention, puedes hace un lee mistake, lo pones en un

wrong account…

‘Sometimes there are simple mistakes that you do, and that can make you have a

difference at the end of the day…sometimes if you don’t concentrate well, you do

mistakes, like...if you are not paying attention, you can do a simple mistake, you place it

[cash deposit] in a wrong account…’

[WO36, female]

In (14), a microfinance field officer elaborates on a community outreach program, where a local credit union motivates lower-income families in rural areas to save money by matching the savings in their accounts once they save a certain amount of money.

(14) El program es como un incentive donde hacemos encourage que hagan save y como

hacemos encourage que hagan save? Pues nosotros hacemos match el amount de

savings de ellos. Tiene un maximum de eighty dollars, pero cuando ya llegues a eighty

dollars, entonces ya vas a tené el double right?

‘The program is like an incentive where we encourage people to save, and how do we

encourage them to save? We match the amount of savings they have. There is a

maximum of eighty dollars, but when you have eighty dollars, you will have double the

amount right?’

258 [W034, male]

In (15), a primary school teacher candidly explains his thoughts on the benefits of employing drama as a teaching strategy in the classroom.

(15) Yo veo que más hacen capture lo que toy tratando de hacer bring across porque lo puedo

ver en sus grades, y um, they can quickly hace recall lo que yo estuve hablando talvez

last week, because solo les doy un lee hint…remember hicimos esto. ‘Oh yes!’ y se

recuerdan de lo que uno estaba hablando because they were involved.

‘I see that they can better learn what I am trying to bring across, because I can see it in

their grades, and um, they can quickly recall what I was talking about last week, because

you just give them a little hint…remember we did this. ‘Oh yes!’ and they remember

what we were talking about because they were involved.’

[W041, male]

These examples vividly illustrate how second generation speakers comfortably employed bi/multilingual language practices when speaking about job-related topics.

To confirm the trends attested in Table 5-4 vis-à-vis the favorability of certain semantic domains to English-origin nouns in Spanish/English DPs, a logistic regression analysis was run in VARBRUL on the Spanish and Spanish/English tokens, which together constituted 89% of the overall DPs. I examined DPs in relation to the two linguistic factors that are the focus of the present examination (i.e., gender on the determiner and semantic domain) and one non-linguistic factor (i.e., age group).

To avoid interactions between the different types of semantic domains, I collapsed the 17 semantic categories into 11 categories. Vehicle/transport-related nouns were included under technology, as vehicles are a certain type of modern technology. Year, number and quantity

259 nouns and institution nouns were re-classified as abstract concepts. Food, drink and smoke, and domestic life items refer to items that are seen and used on a quotidian basis; thus, they were regrouped with everyday items. Health and body terms, which were generally infrequent, were subsumed under ‘Other’ nouns. Lastly, given that animals are related to the earth, they were subsumed under land/earth nouns.

Given that the simple logistic regression analysis revealed that the youngest and eldest groups behaved differently, I also collapsed the four age groups, dividing them simply into the younger and older generation. I was interested in examining the differences in CS practices within the younger generation since previous work on semantic categories and gender assignment has not explored the potential differences among sub-groups of bi/multilinguals.

Endeavouring to better understand linguistic patterns among bi/multilinguals in the younger generations (ages 40 and younger), I recoded the three younger age groups based on their overall proportions of Spanish/English DPs, as there were consultants whose proportions of switched

DPs comprised more than 50% of their total DP production. This group I classified as the translanguagers (n = 18), a term I adopt from García and colleagues (2009, 2014) to refer to speakers who engage in dense bilingual language practices. Translanguagers were not restricted to any specific age group, as they were attested in the third generation adolescent (n = 6) and post-adolescent groups (n =2), and the second generation (n = 10). On the other hand, I labelled speakers whose overall proportions of mixed DPs were 50% or less as code-switchers (n = 34)8.

8 In future work, a more detailed breakdown of speaker type can be conducted in order to have a more clear insight into the distribution of the relative frequency of mixed DPs across speakers. With the current method, speakers whose relative production of mixed DPs were 13%, 15% or 49% would be classified as code-switchers. Intead of a binary system (i.e., less or more than 50%), it would be informative to divide the mixed DP data according to a four- way classification system (i.e., 0%-25%; 26%-50%; 51%-75%; 76%-100%). Such classification could yield useful information on the relative production of mixed DPs. For instance, in the data from the present analysis, most consultants fell within the second (i.e., 25/52) and third range (i.e., 17/52), whereas only one adolescent speaker fell within the uppermost range; hence, suggesting that the relative frequency of mixed DPs in the naturalistic discourse of a Northern Belize bi/multingual may mostly likely lie between 26% and 75%.

260 In Table 5-5, I provide the results from the VARBRUL analysis.

Table 5-5. Logistic regression analysis of factors that favor non-switched or switched Spanish/English DPs FACTOR Weight % Total N % Data Gender on determiner Masculine 0.79 55 6485 49 Feminine 0.66 30 3890 30

Ungendered 0.02 .6 2811 21 Range 77

Speaker type Translanguagers 0.82 65 3221 24 Code-switchers 0.58 37 6786 52 Older generation 0.10 5 3179 24 Range 72

Semantic domain Academia 0.84 68 1323 10 Technology 0.83 63 394 3 Work/money-related 0.74 60 563 4 Abstract concepts 0.65 53 878 14 Linguistics/language 0.58 47 359 3 Everyday items 0.53 31 1020 8 Person (not kin) 0.43 30 2124 16 Time, places or events 0.41 29 2427 18 Land/earth 0.40 19 832 6 Other 0.40 19 706 5 Kinship terms 0.14 9 1556 12 Range 70

Total N = 13186, significance = 0.000, input = .165 Log likelihood = -5048.761, total chi-square = 552.2642 , chi-square/cell = 5.5784

The analysis returned all three factors as significant, with gender on the determiner as the strongest predictor of switched DPs, followed by speaker type and semantic domain. It is not surprising that gender on the DP is the strongest predictor. We will see in the following section that the masculine gender is overwhelmingly used in the Northern Belize context. The factor speaker type was also significant, with both translanguagers (factor weight 0.82) and code-

261 switchers (factor weight 0.58) favoring mixed DPs and the older generation disfavoring it (factor weight 0.10), a trend consistent with the simple linear regression analysis.

For semantic domain, the analysis revealed that there were several categories that favored

English-origin nouns. Overall, academia (factor weight 0.84), technology (factor weight 0.83), work/money-related terms (factor weight 0.74), abstract concepts (factor weight 0.65), linguistics/language terms (factor weight 0.58), and everyday items (factor weight 0.53) favored switched DPs, whereas kinship terms was the category that least favored switched DPs (factor weight 0.14). Thus, in contrast to Clegg’s (2010) and Aaron’s (2014) findings, a wider variety of semantic contexts favored English-origin nouns, suggesting that in the naturalistic discourse of bi/multilinguals from Northern Belize, mixing at the DP level is generally more pervasive.

The openness of semantic domains to English-origin nouns was particularly attested in the speech of translanguagers. The cross-tabulation below shows the distributions of mixed and monolingual Spanish DPs across domains for translanguagers and code-switchers. As Table 5-6 illustrates, for code-switchers, there were only three semantic domains where mixed DPs clearly comprised more than 50% of total DP production; namely, academia (71%), technology (64%), and work/money-related terms (72%). Translanguagers, on the other hand, clearly demonstrated the unmarkedness of CS in their language practices, as their mixed DPs largely comprised the majority (> 50%) across most domains. For this sub-group of bi/multilinguals, therefore, CS is very productive vis-à-vis the semantic contexts where English-origin nouns are incorporated.

262 Table 5-6. Cross-tabulation of DP production by speaker type and semantic domains Semantic Domain DP Type Code-switchers Translanguagers Total N % N % N % Academia mixed 533 71 315 91 893 67 monolingual 216 29 31 9 430 33 Total 749 346 1323 Technology mixed 135 64 103 98 249 63 monolingual 76 36 2 2 145 37 Total 211 105 394 Work/money-related mixed 154 72 168 81 336 60 monolingual 61 28 39 19 227 40 Total 215 207 563 Abstract concepts mixed 476 49 520 82 1004 53 monolingual 494 51 115 18 878 47 Total 961 1882 Linguistics/language mixed 81 43 87 86 169 47 monolingual 107 57 14 14 190 53 Total 188 101 359 Everyday items mixed 188 37 119 63 317 31 monolingual 321 63 70 37 703 69 Total 509 189 1020 Person (not kin) mixed 339 29 289 53 642 30 monolingual 835 71 254 47 1482 70 Total 1174 543 2124 Time, events or places mixed 392 32 282 52 693 29 monolingual 840 68 260 48 1734 71 Total 1232 542 2427 Land/earth mixed 79 26 70 59 158 19 monolingual 230 74 49 41 674 81 Total 309 119 832 Other mixed 64 18 65 41 132 19 monolingual 282 82 93 59 574 81 Total 346 158 706 Kinship terms mixed 82 9 60 22 146 9 monolingual 810 91 216 78 1410 91 Total 892 276 1556 Total mixed 2514 37 2078 65 4739 36 monolingual 4272 63 1143 35 8447 64 Total 6786 3221 13186

263 Crucially, nouns in the ‘Other’ category (41%), which included health/body terms, did not follow this pattern. Kinship terms also deviated from this pattern (22%), as they were infrequently incorporated even among translanguagers (23%), indicating that across different types of bi/multilingual speakers in Northern Belize, the semantic domain least likely to accept non-native nouns is kinship terms. Thus, we can see that unlike the New Mexican data where one of the main functions of English-origin nouns is to name kin (Clegg, 2010; Aaron, 2014), this pattern did not bear out in the Northern Belize data.

Among the eldest consultants (data not depicted in Table 5-6), work/money-related terms

(10%), technology (14%) and academia (20%) were the semantic contexts where English-origin nouns most frequently occurred, consonant with the pattern observed in the code-switcher group.

This offers support to Teschner’s (1974) assertion that semantic domains that are most open to loanwords include both education- and technology-related terms. This pattern bears out across types of bilinguals from different age groups. Importantly, among translanguagers, there is a notable expansion in the semantic domains that are open to English-origin nouns. Kinship terms, however, notably remain as an unfavorable semantic context for English-origin nouns across speaker groups.

In sum, the previous results show that although speakers from the second and third generations seem to generally produce mixed DPs to the same degree, there are notable differences between code-switchers and translanguagers. In the naturalistic discourse of the latter type of bi/multilinguals, more semantic domains are open to English-origin nouns, with the clear exception of kinship terms, which seem to be disfavored not only at the level of the community, but at the level of individual speakers as well.

264 In the following section, we take a closer look at gender assignment in order to better understand how English-origin nouns are incorporated in Northern Belize bi/multilingual speech.

I first present results regarding the overall distribution of monolingual versus switched DPs.

5.5.5 Gender Assignment in Non-Switched and Switched DPs

For the analysis of gender assignment, I examined a total of 5,738 monolingual Spanish

DPs and 3,587 switched DPs where the determiner was overtly gender marked. Results for the monolingual Spanish DPs revealed that masculine- and feminine-marked DPs were almost equally distributed (see Table 5-7), in line with previous findings for New York Spanish

(Otheguy & Lapidus, 2003) and New Mexican Spanish (Clegg & Waltermire, 2009). Thus, contra Hagerty (1996), data from NBS show that canonical patterns of gender assignment in

Spanish monolingual DPs are still maintained in this variety of contact Spanish.

Table 5-7. Distribution of masculine- and feminine-marked DPs Determiner Phrases Non-switched Switched Total N % N % N % Masculine-marked DPs 2963 52 3573 99. 6522 70 6 Feminine-marked DPs 2777 48 14 0.4 2805 30 Total 5740 100 3587 100 9327 100

In contrast to findings for New York Spanish and New Mexican Spanish, however, the data showed near-categorical use of the masculine gender in switched DPs. Only 14 nouns in switched DPs were assigned the feminine gender. In the ensuing sections, I examine the roles of analogical gender, biological gender and animacy in the gender assignment process.

5.5.6 Analogical Gender

The data revealed that in Northern Belize Spanish/English CS, the criterion of analogical gender was not deterministic in the gender assignment process. Recall that in the follow-up task, a second generation consultant from the study was asked to provide potential equivalents for

265 nouns in switched DPs. As Table 5-8 illustrates, the speaker provided equivalents for 76% of switched DPs, whereas no gender equivalents were provided for 24% of the data.

Table 5-8. Translation equivalents for English-origin nouns N % Masculine gender equivalents 1633 45.5 Feminine gender equivalents 1094 30.5 No gender equivalents 860 24.0 Total 3587 100

For 24% of the cases, no equivalents were provided simply because there was no exact equivalent in Spanish. These included English expressions such as upper hand, no-no, hard time, etc., and school-related terms such as first years and 5s, which refer to groups of students in a certain academic level (i.e., first year college level students and standard 5 primary school students respectively). These examples reveal, thus, that cases that appear as lexical gaps sometimes evince the incorporation of English-specific colloquial expressions, and/or new categories such as first years, where an NP that initially functions as a modifier is subsequently nominalized to refer to a certain group of students.

In light of these data, I concur with Clegg (2010) in that bi/multilingual discourse practices are not driven only by lexical need. In the present corpus, switched DPs contained

English-origin nouns whose Spanish translation equivalents are commonly used words. These include examples such as food ‘comida’, school ‘escuela’, job ‘trabajo’, village ‘aldea’, person

‘persona’, girl ‘muchacha’, etc. Although feminine gender equivalents accounted for 30.5% of the data, it is noteworthy that English-origin nouns were overwhelmingly assigned the masculine gender, regardless of the feminine gender of the translation equivalent, contra Jake et al.’s (2002) argument that analogical gender is the strongest predictor of gender assignment in code-switched discourse. These results indicate that for bi/multilinguals in Northern Belize, the analogical

266 gender criterion is generally not applied even in cases where Spanish counterparts are available in speakers’ lexical repertoire. Thus, what accounts for gender assignment in bilingual discourse is simply the employment of a masculine default gender strategy.

5.5.7 Biological Gender and Animacy

I also analyzed overtly gender-marked DPs in terms of biological gender and animacy

(Clegg, 2006; DuBord, 2004; Otheguy & Lapidus, 2003; Montes-Alcalá & Lapidus Shin, 2011;

Poplack et al., 1982; Sánchez, 1995; Valdés Kroff, 2016), as these are two factors that have been shown to be operative in the gender assignment process. Researchers like Clegg (2006) postulate that biological sex of the animate referent overrides any other potential factor. Both Otheguy and

Lapidus (2003, p.225) and DuBord (2004, p.38) also found that biological gender was an important factor in the gender assignment process in bilingual speech.

As Table 5-9 illustrates, in the case of the Northern Belize data, there are two salient patterns that warrant further elaboration.

Table 5-9. Gender assignment according to biological gender and animacy Non-switched DPs Switched DPs Masculine- Feminine- Masculine- Feminine- marked marked marked marked N % N % N % N % male noun 566 19.1 0 0.0 60 1.7 0 0.0 female noun 3 0.1 427 15.4 50 1.4 1 7.1 animate generic noun 295 10.0 251 9.0 589 16.5 10 71.4 animate non-human noun 141 4.8 73 2.6 38 1.1 0 0.0 inanimate noun 1958 66.0 2026 73.0 2836 79.4 3 21.4 Total 2963 100 2777 100 3573 100 14 100

First and foremost, the use of the masculine default gender in switched DPs is more categorical than in previous studies, which have reported between 87% (Otheguy & Lapidus,

2003) and 97% (Valdés Kroff, 2016). In the present data, masculine-marked DPs accounted for

267 99.6% of tokens. Secondly, contra previous findings, biological gender was not found to be a deterministic factor in switched DPs.

In monolingual Spanish DPs, gender assignment based on biological gender was fully operative with male nouns (e.g., el maestro ‘the male teacher’, el vecino ‘the male neighbor’; el muchachito ‘the young boy’) and female nouns (e.g., la maestra ‘the female teacher’, una niña terca ‘a stubborn girl’; una viejita ‘an old lady’). Only in three cases was the masculine gender overgeneralized with female nouns (3 of the 23 assignment errors, i.e., *un señora ‘a lady’; *los mujeres ‘the women’, and *otro maestra ‘another female teacher’).

In contrast, in Spanish/English DPs, the masculine default gender was used with all types of nouns. All animate generic human nouns such as social worker, chef, leader, teacher, vice- principal, administrator, superintendent and supervisor were assigned the masculine gender.

More strikingly, both male (n = 60) and female (n = 50) referents were assigned the masculine gender. The masculine gender was assigned to female nouns, which included singular, female

(e.g., un mom ‘a mother’, el bigger sister ‘the older sister’, un housewife ‘a housewife’) and plural, female referents (e.g., los nuns ‘the nuns’, los women ‘the women’).

In addition, mythical beings (e.g., los mermaids ‘the mermaids’) and religious entities

(e.g., el Virgin Mary ‘the Virgin Mary’) that are transparently feminine were also assigned the masculine gender. Notably, the use of the masculine gender was even attested in preposed masculine adjectives, as in (16) and (17), where the masculine-marked adjective único ‘only’ modifies the female referent.

(16) Yo crecí con puro boys, yo era el unico girl….yo era la más chica

‘I grew up with boys only, I was the only girl…I was the youngest.’

[PA21, female]

268 (17) Eramos tres…yo era el único female, y em, bajábamos en Yo Creek

‘There were three of us…I was the only female, and um, we used to stop at Yo Creek.’

[WO46, female]

Thus, unlike data from New Mexican Spanish (Aaron, 2014), the preference for masculine gender assignment does not appear to be related to nonreferentiality and/or general patterns of gender assignment attested in monolingual Spanish discourse. Both (16) and (17) show that speakers PA21 and WO46 use the masculine gender to refer to themselves. In (16), it is clear that the gender distinction is fully maintained for monolingual Spanish DPs (i.e., la más chica) but suspended only for switched DPs (i.e., el único girl), strongly suggesting that the choice of the masculine default is in fact directly related to an aspect of CS itself.

Lastly, within the small feminine-marked DP data set, the feminine gender was mostly based on analogical gender and animacy rather than biological gender. There was only one case of a female referent that was assigned the feminine gender (i.e., una black woman ‘a black woman’). Thus, even with feminine-marked DPs, biological gender had a lesser deterministic role than animacy. In the following section, we take a closer look at feminine-marked mixed

DPs.

5.5.8 Feminine Gender Assignment

Table 5-10 below provides the list of feminine-marked switched DPs in the present corpus. To further examine the potential effect of different assignment criteria, I coded for analogical, phonological and biological gender. I also examined each utterance to determine whether the switch was preceded by pauses and/or repair phenomena, as Valdés Kroff (2016) hypothesizes that feminine gender assignment typically occurs in unplanned switching. The analysis revealed that the phonological criterion clearly had little deterministic effect. In fact,

269 most English-origin words in the feminine-marked data set (n = 10) had terminal phonemes that are associated with the masculine gender (i.e., /n/, /l/, /s/), yet they were nonetheless assigned the feminine gender. In the masculine-marked data set, the masculine gender default was applied across the board, irrespective of terminal phonemes in English-origin nouns.

Table 5-10. Feminine gender assignment Mixed DP L1 equivalent terminal animacy biological Disfluency (analogical gender) phoneme sex prior to switch la person persona MP √ X X la bell campana MP X X √ la principal (2) directora MP √ X X la nurse (2) enfermera MP √ X X la find trouble busca pleito MP √ X √ una black woman mujer morena MP √ √ X una highty tighty mujer rica VP √ X X la parent pariente AP √ X √ la majority mayoría VP X X X una nurse (2) enfermera MP √ X X la otra half pound media libra FP X X X *MP = masculine phoneme, VP = vowel phoneme, AP = ambiguous phoneme, FP = feminine phoneme

Overall, 86% of tokens (n = 12) matched the gender of their translation equivalents, whereas 79% of feminine-marked mixed DPs (n = 11) were animate. In terms of biological gender, there was only one token that was inherently marked with biological sex (i.e., woman), as nouns such as person and principal could potentially refer to a male. Thus, the analogical gender criterion was the strongest predictor of feminine gender assignment in this dataset, followed by animacy.

Unplanned or ‘unintended’ switching (as evidenced by disfluencies, pauses or repetitions leading up to the switch) also seemed to play a role in some cases, in line with Valdés Kroff’s

(2016) claim. There were three examples that show evidence of ‘unplanned’ switching. In (18), for instance, speaker AD08 produced a one-second pause before switching into the English noun

270 bell. The switch may have been caused by CS ‘slip’. By slip, I mean the switched element may have been unintended or triggered by lexical need in spontaneous speech. Although campana

‘bell’ is a rather commonly used word in Northern Belize, the speaker may have failed to quickly retrieve it while speaking, and this triggered the switch.

(18) Luego tocaba la…bell…veníamos acá. Comíamos. Íbamos otra vez.

‘Then the…bell would ring…we would come here. We would eat. We would go again.’

[ADO8, male]

At the same time, however, feminine-marked mixed DPs were also present in cases of alternational or dense CS, where there were no hesitations or repair phenomena leading to the feminine-marked, switched DP. Such examples do not reveal any kind of unplanned switching.

In (19), for instance, the mixed DP ‘la person’ is embedded in a compound sentence where the first independent clause is densely switched.

(19) Hacen try de imitate el way que lo hacen, y, like, puedes ve que la person se dio cuenta

y que se siente mal.

‘They try to imitate the way they speak, and, like, you can see that the person notices it

and that he/she is offended.’

[AD07, female]

In (20), dense switching is even more apparent. In this case, the mixed DP ‘una black woman’ is embedded in a stretch of speech where speaker WO27 fluidly switches between

Spanish, English and Kriol.

(20) De los single mothers’ one? Ai mi laik it porque, um, son de diferentes mamás de

totally different lives. Of course deh have to include una black woman [laughs]…

271 ‘The single mothers’ movie? I liked it because, um, it is about different mothers who live

different lives. Of course, they have to include an African-American female character

[laughs]…’

[WO27, female]

Overall, results confirmed that for feminine-marked mixed DPs, the analogical criterion and animacy were the factors that had the strongest predictive effects in the gender assignment process. In contrast, the least predictive factors were phonological gender and biological sex.

5.5.9 Overgeneralization of the Masculine Gender?

A relevant question which arises as a result of the pervasive use of the masculine gender in the present data is whether non-canonical patterns of gender assignment in monolingual speech are also relatively frequent among speakers. This becomes especially relevant given that in the Northern Belize context, as shown in section 5.5.1, the masculine gender is generalized in gender agreement contexts. Thus, in light of the fact that gender disagreement is commonly attested in NBS, there is the possibility that gender assignment incongruences are relatively common.

Note that across time, it is not uncommon for subsequent generations of Spanish speakers to show gender assignment incongruences. For example, in her examination of L2 learners of

Spanish and heritage speakers of Spanish, Martínez-Gibson (2011, p.187) found gender assignment error rates of 7.5% and 16.75% for first and second generation heritage speakers, respectively. Importantly, these speakers were from households where Spanish was the dominant language. In other Spanish contact situations undergoing language shift, notable changes have been found in the gender assignment system as well. Romero (2012) reports that among Judeo-

Spanish speakers in Istanbul, gender shift (i.e., masculine nouns assigned feminine gender and

272 feminine nouns assigned masculine gender) has dramatically increased in the younger generation, oscillating between 5% and 25% in speakers younger than 50.

To further explore potential cross-generational differences in gender assignment in

Spanish DPs, I examined cases of gender incongruencies, in other words, non-switched DPs where the marked form of ambiguous nouns, as in la mar ‘the sea’ and la calor ‘the heat’, are used instead of their more canonical counterparts (i.e., el mar, el calor), and also, cases of disfluencies and assignment errors.

Table 5-11 below provides the patterns of distribution for gender incongruences across the different age groups.

Table 5-11. Gender assignment incongruences across three generations 1st 2nd 3rd AD 3rd PA (50-99) (21-40) (14-17) (18-20) N % N % N % N % Overgeneralized feminine 5 25 10 56 6 29 3 33 gender Overgeneralized masculine 15 75 8 44 15 71 6 67 gender Total 20 100 18 100 21 100 9 100 Phonologically-conditioned or 12 60 13 72 10 48 7 70 ambiguous nouns Disfluency 0 0 2 11 2 10 0 0 Errors 8 40 3 17 9 43 3 30 Total 20 100 18 100 21 100 10 100 *PA = post-adolescent, AD = adolescent

273 With the exception of the second generation, the other age groups showed a stronger tendency to generalize the masculine gender (e.g., los gentes ‘the people’, mucho dedicación ‘a lot of dedication’, el foto ‘the photograph’) rather than the feminine gender (e.g. la área ‘the area’, la problema ‘the problema’, la agua ‘the water’) across different types of nouns.

Only 4 cases of disfluencies were identified, where the speaker produced a brief pause before switching into the English noun. There were few gender assignment errors (e.g. *el

*comida ‘the food’, *el copa ‘the wine glass’, el mentalidad ‘the mentality’, etc.), accounting for

32% of gender assignment incongruences (n = 22). As it relates to the corpus of overtly marked non-switched DPs gender (n = 5740), assignment errors only accounted for 0.3%, distributed across the three generations.

The analysis also revealed that, across generations, speakers had particular difficulty with certain types of nouns. These included transvestite nouns that end in –a but are masculine, such as problema ‘problem’; hermaphroditic nouns, such as agua ‘water’, which are phonologically- conditioned in the gender assignment process; and ambiguous nouns, such as azúcar ‘sugar’ and mar ‘sea’ (Eddington & Hualde, 2008). Given that these nouns, particularly hermaphroditic nouns, are problematic for native speakers from other varieties of monolingual Spanish

(Eddington & Hualde, 2008, p.15), it is not surprising that this pattern also bears out in the current data.

These results offer evidence that among Spanish-dominant bi/multilinguals from

Northern Belize, instances of non-standard gender assignation are not as “common” as Hagerty

(1996, p.137) observed. Hagerty’s examples of non-standard gender assignment include el primer vez ‘the first time’ and el gente ‘the people’. In the present data, there was only one case of un vez ‘one time’ and five instances where the noun gente was assigned the masculine gender.

274 Hence, while there is a marked cross-generational increase in the number of mixed

Spanish/English DPs (see Figure 5-1), this development is not accompanied by a corresponding increase in gender assignment errors in the younger generations.

In sum, the gender assignment analysis revealed that contra previous work that has found support for the analogical criterion or ‘synonymic gender’ (Fuller & Lehnart, 2000; Jake et al.,

2002; Liceras et al., 2008; Sánchez, 1995; Smead, 2000), the present data align with studies that have not done so (Otheguy & Lapidus, 2003; Valdés Kroff, 2016). It also showed, however, that biological gender was not operant in the masculine-marked mixed DP data set. In the feminine- marked mixed DP data set, it was operant, but it was not the strongest predictor of feminine gender assignment. Across generations, patterns of gender assignment were canonical. Only few cases of gender incongruences were attested.

5.6 Discussion

The naturalistic, oral production data from 62 native speakers of NBS revealed several similarities and differences with respect to findings in previous studies. Consonant with Herring et al.’s (2010) findings, mixed DPs where an English determiner co-occurred with a Spanish noun were infrequent. Nonetheless, contra Moro’s (2001) contention, these switches are shown to be possible in Northern Belize Spanish/English CS. In the semantic analysis, results revealed that several semantic domains favored English-origin nouns. In contrast to the New Mexican data, however, the least favorable context for English-origin nouns is kinship terms. In relation to gender assignment, two salient patterns emerged. Consonant with previous work on varieties of

U.S. Spanish (Otheguy & Lapidus, 2003; Clegg & Waltermire, 2009), in the monolingual

Spanish DPs, canonical patterns of gender assignment were attested, dispelling the notion that

Belizean Spanish (as it relates to gender assignment) is moving in the direction of English morphosyntax (Hagerty, 1996). In switched DPs, an overwhelming preference for the masculine

275 gender in overtly gender-marked DPs was attested. In contrast to data from U.S. Spanish/English communities (e.g., Clegg, 2006; DuBord, 2004), however, biological gender was largely inoperative.

In the following sections, I discuss the implications of these findings vis-à-vis our current understanding of CS practices (at the DP level) across Spanish/English contexts. I first discuss what motivates the attested patterns in the incorporation of semantic domains and in the gender assignment process in monolingual versus switched discourse. Subsequently, I elaborate how type of CS may account for the results that have been attested thus far in studies of semantic domains and gender assignment in Spanish/English CS.

5.6.1 Semantic Domains

Clearly, the openness of semantic domains to English-origin nouns is determined by both idiolectal and social factors. In terms of proficiency, we see that the eldest speakers (ages 50+) who were least proficient in English produced the least mixed DPs. On the other hand, among younger consultants (ages 14 – 40) with higher levels of bi/multilingual proficiency, the production of mixed DPs was significantly higher. Notably, whereas the three younger age groups behaved largely homogenously, there were two sub-groups in terms of openness to

English-origin nouns in their naturalistic discourse. In contrast to code-switchers, translanguagers’ speech evinced favorability to English-origin nouns across most semantic domains. At the same time, however, the powerful effect of community linguistic norms was evinced in the fact that kinship terms did not favor English-origin nouns across even among translanguagers. Thus, we see that bi/multilinguals, regardless of age and switching practices, abide by certain community linguistic norms. Social factors also played an important role. In section 5.6.4, I return to this issue and further elaborate how social factors may account for the

276 favorability of semantic domains to English-origin nouns in the speech of bi/multilinguals. In the next section, we discuss what determines gender assignment in Spanish/English CS.

5.6.2 What Determines Gender Assignment in Code-Switched Discourse?

Results revealed that the analogical criterion was clearly not a deterministic factor in the gender assignment of the vast majority of English-origin nouns in the present study. The effect of biological gender in mixed DPs also proved to be almost non-existent. Thus, the employment of a masculine default gender is what accounts for the assignment patterns in the present corpus.

While the use of this default gender has been attested in multiple Spanish/English CS contexts, there is also variation (e.g., 87% versus 99.6%).

Findings suggest that across Spanish/English bilingual communities, the deterministic role of the masculine default gender in U.S. Hispanophone communities is not as strong as in

Northern Belize, where the default gender strategy is consistently applied even to English-origin nouns that are transparently feminine. These are differences that can be attributed to both typological/linguistic (e.g., word order similarities between Spanish and English, frequency of

CS, instantiation of convergence, etc.) and social factors (e.g., the status of CS, degree of linguistic normativity, etc.). In section 5.6.4, I discuss this in more detail.

In line with Otheguy and Lapidus’ (2003) study, the Northern Belize CS data showed that it is only in switched discourse that the Spanish patterns of arbitrary, memory-dependent gender distinctions are not applied. Thus, the use of the masculine gender default in Spanish/English CS is indeed a CS-specific phenomenon, contra Aaron’s (2014) view. Results revealed that specifically as it relates to gender assignment in Northern Belize bi/multilingual speech, gender is not being lost, as the gender distinction was only neutralized in bi/multilingual speech but not monolingual discourse.

277 This gender neutralization in CS, which was complete in the case of 86% of consultants, can be analyzed in different ways. For Pfaff (1979, p.305), when the two-way distinction is neutralized, this is “some indication that grammatical gender is beginning to be lost for English nouns borrowed or switched into Spanish.” For Montés-Alcalá and Lapidus Shin (2011), on the other hand, it is viewed as a means to increase communicative efficiency. My view, consonant with Otheguy and Lapidus’ analysis, is that speakers use the masculine default gender to facilitate switching between two structures that would otherwise be less compatible due to structural differences between the languages (see Sebba, 1998).

Findings from this study support Toribio’s (2004, p.167) claim that “when bilinguals’ languages are simultaneously deployed, as in code-switching, their interaction may be more pronounced and parsimony more urgent.” The non-application of analogical gender in masculine-marked, switched DPs showcases speakers’ parsimonious use of their lexical resources. Even if Northern Belize bi/multilinguals had a translation equivalent in their mental lexicon, they nonetheless used the masculine default gender, as this was the choice that was most efficient and economical. Toribio (2004) aptly asserts that particularly in CS, bilinguals enhance the structural similarities between their linguistic systems to optimize bilingual discourse and reduce the cognitive costs required when simultaneously using two linguistic systems. One way of enhancing Spanish/English CS is precisely by neutralizing the gender distinctions in switched discourse. By not following Spanish morphosyntax, switched discourse progresses more smoothly and seamlessly, as it is not restricted by gender nuances that are difficult even for native, monolingual speakers (see section 5.5.9).

Some researchers assert that in CS, bilinguals simultaneously use two linguistic systems, resulting in an increase in cognitive load (but see Wilson Vergara and Dumont (2015)). In order

278 to reduce these cognitive demands, bilinguals develop strategies to simplify this process (Silva-

Corvalán, 1994), such as ‘adaptively’ simplifying the Spanish gender system (Otheguy &

Lapidus, 2003). If indeed bi/multilinguals strive to reduce the cognitive load of simultaneously processing two or more linguistic systems, then we can surmise that the more unmarked and pervasive CS is in a given speech community, then the more evidence we should find that bilinguals optimize the linguistic resources available to them by employing convergence, even if the resultant structures they use fail to strictly align with their dominant language’s morphosyntax (i.e., the Spanish gender distinction, hence, the analogical criterion).

The data from Northern Belize suggest that in contexts with intense language contact, low levels of linguistic prescriptivism, and positive attitudes toward CS, the Spanish gender distinction can be completely neutralized in bilingual discourse. This raises key questions about the role that convergence plays in CS. I analyze the partial or complete neutralization of the

Spanish gender distinction in Spanish/English CS as a clear case of ‘composite’ CS (Myers-

Scotton, 2002). In Toribio’s (2004) terms, this constitutes an example of convergence (akin to

Muysken’s 2013 conceptualization of congruent lexicalization). Crucially, if we analyze the variable or complete neutralization of the gender distinction in Spanish/English CS as reflective of composite CS or convergence, then it follows that even in the Spanish/English CS varieties of

New York, New Mexico and Miami, there are traces of the same mechanism at play.

Among younger consultants in the present sample, CS has incontrovertibly become more pervasive. The preference for the masculine gender has regularized in an effort to promulgate the optimal use of linguistic resources. This is especially the case among speakers who produced the most switched DPs. Overall, 89% of speakers (i.e., 54/61) who produced Spanish/English DPs exclusively employed the masculine default gender in switched DPs. Out of the seven speakers

279 who produced some feminine-marked mixed DPs, six of them were code-switchers. In contrast, only one translanguager produced a feminine-marked switched DP (i.e., la person ‘the person’).

Thus, it is noteworthy that translanguagers not only employed CS in more semantic contexts, but that they also employed the masculine default gender more consistently.

Although Jake et al. (2002, p.83) argue that the analogical criterion is the strongest predictor of gender assignment, we know from previous work and the present study that this is not the case in naturalistic Spanish/English oral production across different sociolinguistic contexts. Myers-Scotton’s Matrix Language Frame model (1993, 2002) predicts that in cases of classic CS, the matrix language (ML) would primarily determine the gender of English-origin nouns. Following Myers-Scotton’s (1993, 2002) conceptualization of classic CS and Jake et al.’s

(2002) analysis of mixed DPs, when a switched English noun such as ‘girl’ enters the language production process, we would assume that the lemma supporting the translation equivalent (i.e., niña, chamaquita, chiquita) would also be activated.

In a classic CS context where Spanish is the ML, the gender assignment process would proceed as it would in Spanish and not English, in line with Jake et al.’s prediction. Notably, however, even with frequent nouns such as ‘girl’ or ‘housewife’ whose counterparts can easily be accessed, the grammatical procedures from Spanish were not applied to English-origin nouns in the present data. Instead, nouns were incorporated as ungendered content morphemes; thus, revealing a composite structure, exhibiting features from both Spanish and English.

In sum, findings from the present study strongly support the idea that CS induces convergence (Backus, 2004; Bullock & Toribio, 2004; Gardner-Chloros & Edwards, 2004;

Myers-Scotton, 2002; Thomason, 2001; Toribio, 2004, among others). My view is that while CS

280 and convergence indeed go hand and hand, social conditions ultimately determine to what degree this relationship is accentuated and manifested in switched discourse.

5.6.3 What Accounts for Feminine-Marked Switched DPs?

Herring et al. (2010, p.570) aptly note that there are different pragmatic, sociolinguistic and phonological factors that can account for infrequent CS patterns. In the case of feminine gender assignment in the present corpus, several factors were at play. Analogical gender and animacy (when referring to human agents) were strong predictors. There were other notable phenomena at play as well. There were a few mixed DPs (e.g., la bell, la parent) that may have resulted from unplanned switching (Valdés Kroff, 2016). These feminine-marked switched DPs could be unintended CS slips, similar to infrequent slips found in monolingual speech. Evidence for this comes from the fact that the noun ‘parent’, for example, appears in singular or plural forms 49 other times in the corpus, and in all cases, it is masculine-marked. Thus, it is not the case that this noun shows intra- or interspeaker variation in terms of gender assignment in bilingual speech. The application of the feminine gender in the mixed DP ‘la parent’, therefore, was more than likely unintended.

Other cases revealed a combination of factors. For instance, the noun ‘nurse’ appeared five times in the corpus, and in four cases, it was feminine-marked. This shows that this English- origin noun has a higher propensity to be assigned feminine gender. In some cases, a combination of social and biological factors can account for the conventionalization of some feminine-marked nouns. In Belize, until recently, nurses were usually women. This could explain why the noun ‘nurse’ is associated at a social and linguistic level with the feminine gender. In a pilot study that was previously conducted with another group of ten adolescent speakers (Balam,

2014, p.89), the only two cases where English-origin nouns were feminine-marked were precisely with the noun ‘nurse’.

281 Another notable case is the use of ‘highty-tighty’ in (21). In this case, the pause does not necessarily indicate unplanned switching, but rather, the speaker’s attempt to retrieve the appropriate English word. In this case, speaker WO31 emphasizes the mother-in-law’s , by specifically referring to her ability to cook expensive meals. Note that the idiomatic expression ‘highty-tighty’ is not encoded with biological sex. In Belize, however, it is used to specifically refer to an arrogant rich girl or woman.

(21) Su suegra de ella era una…highty-tighty que sabía cociná…todo esos rich foods.

‘Her mother-in-law was a ‘highty-tighty’ who knew how to cook…all those expensive

meals.’

[WO31, female]

Thus, while the use of the feminine gender in switched DPs may be related to unintended switching and/or other related linguistic factors, it is clear that the feminine gender also affords bi/multilinguals a way to encode certain semantic and pragmatic nuances that directly relate to the way some English-origin nouns have historically been lexicalized in the speech community

(for relevant discussion, see Clegg, 2010, p.234).

5.6.4 The Importance of Type of Code-Switching

Findings from the present study bring to the fore that the type of CS attested in a bi/multilingual community is an important factor we must consider when examining both the openness of semantic domains to other-language nouns and gender assignment in switched discourse. By type, I mean the nature of CS, both as a linguistic and sociocultural phenomenon.

We must not forget that CS outcomes are shaped by intricately interconnected linguistic, social and cognitive factors (Muysken, 2013). In certain cases of intense societal bi/multilingualism, as

Winford (2013, p.365) reminds us, “social factors play a significant, and in some cases a more important role than linguistic factors, in shaping the consequences of language contact”.

282 In particular, the status that CS is ascribed in a community and the degree of linguistic prescriptivism are extremely important concomitant social factors that indelibly affect how CS is perceived and employed as a sociolinguistic resource within a community. Type of CS can account for the attested differences between this study and Aaron (2014). Whereas Aaron (2014) found that English-origin nouns were particularly used for certain discourse functions such as expressing kinship terms, the opposite was found in the Northern Belize data. Kinship terms did not favor English-origin nouns. Crucially, more semantic domains favored English-origin nouns in the Northern Belizean context. Why do we find this difference? Note that in both the New

Mexican and the Northern Belizean data, technology and vehicles/transport are favorable contexts for English-origin nouns in switched DPs. Thus, there are commonalities between the two sociolinguistic contexts, which highlight that typologically, Spanish/English contact do lead to some common outcomes in different sociolinguistic contexts.

Crucially, there are marked differences as well. I argue here that the attested differences relate to the fact that CS in Northern Belize is more pervasive and unmarked, and hence utilized in more social and grammatical contexts. Evidence for this claim is found not only in the semantic contexts where English-origin nouns occur, but also in the variety of syntactic and clausal contexts where bilingual light verb constructions (e.g., No hago recall ‘I don’t recall’) are incorporated in Northern Belize code-switched discourse (Balam & Prada Pérez, 2016).

It is worth noting that there is no linguistic constraint per se that bans the openness of certain semantic domains to other-language nouns. The greater openness of semantic domains to

English-origin nouns in the Northern Belize context is simply a direct reflection of the type of

CS (dense, unmarked, more composite) attested in the community. It is also reflective of the

283 levels of linguistic prescriptivism present in the society and the level of acceptability that is afforded to bi/multilingual language practices.

In the U.S. context, linguistic hegemony is rigidly associated with standard English and standard Spanish varieties (García, 2014; Mrak, 2011; Palmer & Martínez, 2013; Zentella, 1997, among many others), while bi/multilingualism is often perceived negatively. On the other hand, in Belize, the appreciation for linguistic diversity is evidenced in a national language policy that recognizes the status of English as the official language of instruction, but which also promotes the use of first languages as “important vehicles” that create a vital bridge between home and school domains (Government of Belize Language Policy, 2000, p.183). Thus, Belize does not have a subtractive language policy, but rather an additive one that promotes the development of speakers’ rich, linguistic repertoire (Balam & Prada Pérez, in press).

A fundamental difference between New Mexican communities and Northern Belize is that in the latter context, CS has been able to historically develop and thrive in a context with low levels of linguistic purism (for relevant discussion, see Balam, 2013a, 2015; Balam & Prada

Pérez, in press). As a result of its unmarkedness and societal acceptance in different social domains, language alternation is used more productively in the Northern Belize context. In the present oral production data, speakers (many of whom were professionally trained elementary school teachers) comfortably spoke in code-switched speech about their work duties, as CS in

Belize is not banned from the workplace or even academic contexts (also see Balam & Prada

Pérez, in press).

In contrast, Clegg’s (2010) and Aaron’s (2014) data suggest that the use of

Spanish/English CS in New Mexico is more marked and restricted to less productive semantic contexts (i.e., kinship terms which constitute a specified set of nouns). In the U.S., the use of

284 bilingual language practices remains largely marked. This may explain why certain semantic domains do not favor English-origin nouns in New Mexican bilingual discourse, as conversations that take place in a professional sphere may more than likely be relegated primarily to monolingual modes of discourse (either English or Spanish). This could also account for the fact that whereas bilingual light verb constructions are attested in both contexts, they are more conventionalized and ubiquitous in Northern Belize (Balam, 2015, 2016a; Balam and Prada

Pérez, 2016).

In sum, the deterministic effect of linguistic factors was clearly evident in the preference for a masculine default gender, which has been found across several Spanish/English contexts. In contrast, community differences in the favorability of semantic contexts to English-origin nouns can be attributed to social factors such as the status of CS and the degree of linguistic prescriptivism (Muysken, 2013). In the Northern Belize context, sociolinguistic conditions have facilitated dense CS practices, which have in turn resulted in the more frequent use of intra- sentential CS, evidenced in the richer variety of semantic contexts that favor English-origin nouns in switched speech.

5.7 Concluding Remarks

Naturalistic data from Northern Belize revealed that several semantic categories (most domains in the case of translanguagers) favored non-native nouns in switched Spanish/English

DPs, consistent with the idea that CS is unmarked in Northern Belize. For gender assignment, the data revealed that the Spanish gender distinction was only neutralized in mixed DPs. I posit that the type of CS in bi/multilingual communities is essential to our understanding of outcomes attested across CS contexts with a similar language pair but different sociolinguistic conditions.

We must not forget that in certain cases, sociolinguistic conditions can indeed be more deterministic than linguistic factors (Gardner-Chloros & Edwards, 2004; Thomason, 2008).

285 Although the present study makes a valuable contribution to the study of contact Spanish outcomes in Central America, a limitation is that cross-dialectal analyses on semantic domains are limited only to work that has been conducted on CS data from New Mexico and Northern

Belize. More empirical work has to be conducted on the openness of semantic categories to

English-origin nouns in other Spanish/English communities. Future research can determine whether in contexts where CS is marked and linguistic purism prevails, the maintenance of separate grammars takes precedence and other criteria (e.g., analogical and biological gender) remain more fully operative, consonant with a classic CS scenario.

286 CHAPTER 6 IMPLICATIONS OF FINDINGS

6.1 Conclusion

Through the examination of judgements and/or oral production data from over 100

Northern Belize bi/multilinguals, the present dissertation provided insight into linguistic aspects of NBS and Spanish/English CS in Northern Belize. Endeavoring to elucidate the wider implications that the current dissertation has on the antecedent literature on contact linguistics and CS, I first outline the main findings of the present dissertation in section 6.2. Subsequently, in sections 6.3 to 6.5, I discuss the implications of main findings in relation to three issues that were raised in the introductory chapter; namely, (i) the nature of language practices in Northern

Belize; (ii) the prevailing view of CS as the combination of two discrete languages and the relationship between CS and convergence; and lastly, (iii) the roles of linguistic versus social factors in contact outcomes. In sections 6.6 and 6.7, I offer concluding remarks, identify limitations in the present quantitative study, and provide avenues for further research.

6.2 Summary of Main Findings

By examining both phonetic/phonological and morphosyntactic phenomena, this dissertation has provided a multidimensional view of the patterns of language use, language change, and innovation in Northern Belize contact Spanish.

In Chapter 3, the analysis of data from elicited production revealed that adolescent consultants maintained their home variety’s tap/approximant intervocalic contrast. In the read- aloud task, however, this distinction was partially lost, as consultants’ attempt to switch to standard Spanish resulted in intervocalic tap/trill neutralization and inconsistent use of their native variety’s intervocalic rhotic contrast. Thus, rhotic neutralization was found to be a

287 variable rather than a global phenomenon. No evidence was found of a converged retroflex approximant category in intervocalic position, as suggested by Hagerty (1979).

In Chapter 4, the quantitative analysis of BLVCs revealed that verb frequency, stativity, and passivization did not disfavor or ban the use of ‘hacer + V’ in switched discourse. The creative use of ‘hacer + V’ was noted for its features of convergence and creolization, which both attested to speakers’ optimal use of linguistic resources when switching at the verb phrase level. I argue that historical and social conditions have facilitated the productive and creative use of these hybrid structures in Northern Belize.

Lastly, in Chapter 5, the cross-sectional analysis of naturalistic speech data from 62 consultants revealed that several semantic domains favored the incorporation of English-origin nouns in Spanish/English DPs. Semantic domains were most open to English-origin nouns among translanguagers. Notably, for gender assignment, it was found that the Spanish gender distinction was only neutralized in switched DPs but not in monolingual Spanish DPs; hence, revealing that this is a ‘composite’ (Myers-Scotton, 2002) mechanism specifically induced by

CS.

Based on the aforementioned findings, one of the main contributions this dissertation makes is that it provides quantitative insight into patterns of language use among Northern

Belize Maya/Mestizos who are native speakers of NBS. In the ensuing section, I discuss how findings elucidate our understanding of language practices in a context where sociolinguistic conditions have favored the frequent use of intra-sentential CS.

6.3 Language Practices among Maya/Mestizos in Northern Belize

Findings revealed the complex phonetic/phonological and morphosyntactic nature of language practices in Northern Belize, while also contributing to our understanding of contact outcomes in linguistic situations where bi/multilingual language practices have been

288 commonplace for generations and where monolingualism is not privileged over bi/multilingualism.

In the first study, results revealed that the use of intervocalic rhotics is affected by task type, as different tasks had different effects on speakers’ pronunciation, and hence, the maintenance of the intervocalic rhotic contrast. In previous work (e.g., Willis & Bradley, 2008;

Bradley & Willis, 2012), the tap/trill contrast in Spanish varieties has generally been examined from a unidimensional perspective. In other words, the maintenance of the intervocalic rhotic contrast has been investigated as a static phenomenon that is either present or absent and not as a phenomenon that is flexible (i.e., changes according to discourse mode) and multidimensional

(i.e., involves the co-existence of two or more phonological contrasts/patterns). The present findings suggest, however, that rhotic neutralization is a phonemenon that may be variably instantiated under certain contexts. We can expect to find similar results among speakers in other bilingual communities, where rhotics are also marked sounds due to their perceptual salience

(see, for example, the work of Ramos-Pellicia, 2007).

In previous acoustic studies on the tap/trill contrast, one of the main limitations is that analyses relied only on one source of data, thus presenting an incomplete picture of rhotic production in different types of speech (i.e., formal versus more naturalistic). For instance, scholars typically only examine semi-spontaneous speech from oral narratives (e.g., Bradley &

Willis, 2012; Henriksen & Willis, 2010; Willis & Bradley, 2008), leaving into question whether in more careful speech (e.g., oral reading), speakers might approximate normative Spanish patterns, especially since classroom settings often use oral reading activities to draw learners’ attention to the accuracy of their pronunciation.

289 In research conducted by Bradley and Willis (2012), for example, there is the possibility that in more careful speech (i.e., read-aloud task) their consultants who were university and professional degree students would more frequently employ two- or three-closure trills. Data from an oral reading task could have substantiated whether indeed consultants do or do not employ canonical trills, as it may be the case that in formal social situations (e.g., conversations in the workplace, formal presentations, etc.), these speakers’ phonological patterns accommodate to a more normative ‘phonological’ style.

In the present analysis, it was shown that semi-spontaneous speech and oral reading had different effects on rhotic production among adolescent speakers of NBS. Variable neutralization of the intervocalic rhotic contrast was only attested in oral reading but not in semi-spontaneous speech. This showed that speakers’ phonological patterns changed depending on task type or discourse mode (i.e., semi-spontaneous speech versus reading). We can surmise that particularly in Spanish bidialectal contexts where the home variety and the classroom standard variety markedly differ, speakers would engage in phonetic/phonological accommodation under situations where a more normative variety of Spanish is expected to be used. In the case of

Northern Belize, as previously highlighted, a more systematic analysis of the different factors that influence phonetic/phonological outcomes (e.g., proficiency in spoken standard Spanish, patterns of language use, speakers’ attitudes toward their language varieties, etc.) needs to be carried out.

In relation to bilingual language practices, given the paucity of work on Northern Belize

Spanish/English CS in the contact Spanish literature, little was known about this innovative CS variety. Although a few researchers had noted the frequent practice of Spanish/English CS in

Belize, quantitative evidence to support this claim was lacking (Brockmann, 1979; Hagerty,

290 1996). Moreover, in previous descriptive work, it was even suggested that Spanish/English CS may not be pervasive in Belize. In particular, Lipski (2005, p.2)1 observed that “the effortless interweaving of languages within a single clause occurs more frequently and with a higher density among Spanish-English bilinguals in the than in communities involving similar Romance languages in contact with English…even Spanish-English in Gibraltar and creole English-Spanish in Belize”.

However, the findings from the the quantitative analyses of ‘hacer + V’ and mixed DPs in the present dissertation suggest otherwise. Among Spanish/English bilinguals in Northern

Belize, CS practices are so dense and unmarked that the bi/multilingual mode constitutes the primary mode through which many speakers express themselves on a daily basis. Contra Lipski’s

(2005, p.2) assertion, CS at the intra-sentential level may in fact be more ubiquitous in Northern

Belize than in most Latino communities in the U.S. Hispanophone context. I argue here that a confluence of historical and sociolinguistic conditions have collectively rendered Northern

Belize a more favorable context for Spanish/English intra-sentential CS to thrive. This is an important finding given that Northern Belize allows us to have comparative insight into CS outcomes in this context versus the U.S. Hispanophone context, where extensive work has been conducted on CS since the 1980s.

Noteworthy is that contexts like Northern Belize, where CS is unmarked, problematize traditional notions of how bi/multilinguals employ their linguistic resources. In Grosjean’s

(2001) influential work on the language mode continuum, for instance, the inherent assumption is that bilinguals move across a continuum that ranges from monolingual to bilingual discourse.

1 Lipski (2005) based this conclusion on previous descriptive work (Lipski, 1986). It may be that Lipski’s observations are based on his observations of speakers from other parts of Belize and not Northern Belize. Among speakers in Central Belize, for example, CS is less frequent given the more frequent use of monolingual varieties of . Alternatively, it could be that Lipski was basing this on observations of older speakers who engaged in limited CS due to lower levels of bi/multilingual proficiency.

291 The starting point in this continuum, therefore, is monolingual discourse. In contexts like

Northern Belize, however, the monolingual mode is actually marked given the prevalence of widespread multilingualism. In these situations, thus, what we find is that the bilingual or bi/multilingual mode itself may constitute the starting point in the continuum (i.e., the ‘unmarked choice’ in Myers-Scotton’s (1993) terms).

The fact that the bi/multilingual mode is the starting point for many speakers in the

Northern Belize context has important implications on how speakers deploy their linguistic resources and what cognitive costs this incurs. Generally, linguists debate whether CS is cognitively costly or not (for relevant discussion, see Myers-Scotton & Jake, 2013; Vergara

Wilson & Dumont, 2015, and references therein), based on the inherent assumption that speakers move from a monolingual to a bilingual mode. In a multilingual context like Belize, however, where monolingualism is not privileged over bi/multilingualism and where the norm is to engage in language mixing, speakers’ conscious or unconscious attempt to separate their languages is what may be more cognitively costly.

Particularly in the case of rhotics, it becomes relevant to question whether speakers were able to suppress co-activation of their phonological systems and focus only on employing the normative intervocalic contrast in oral reading (more reflective of a monolingual Spanish mode).

As the results showed, speakers were unable to switch to the normative tap/trill intervocalic contrast, which may be reflective of a more global cognitive process that becomes especially taxing for these speakers when switching from a bi/multilingual to a monolingual mode. In other words, for speakers that are naturalistic code-switchers/translanguagers, engaging in dense CS practices comes more naturally. Thus, suppressing the bi/multilingual mode and switching to a monolingual mode may be more cognitively costly than seamlessly mixing their linguistic

292 resources. This costly switch to a monolingual mode is what may have affected speakers’ phonology in the study on rhotics.

At the same time, it must be undercored that social factors are also relevant, as patterns attested cannot be attributed only to cognitive factors. Based on previous attitudinal research

(Balam, 2013; Balam & Prada Pérez, in press), it is likely the case that strong negative attitudes toward monolingual varieties bar or discourage speakers’ use of normative Spanish forms in general. It may be that even if Northern Belize adolescents eventually develop the ability to produce the normative tap/trill contrast, they would still not employ it in informal contexts. This would partially result not only from speakers’ negative attitudes toward standard Spanish, but from the fact that standard Spanish is simply ‘non-functional’ within the Northern Belize context. By this, I mean that standard Spanish is the variety that is used the least, as it is the variety that is most marginalized, not only in social but academic contexts as well. Hence, if adolescents or speakers feel that there is no need to use standard Spanish, they will simply not use it. Thus, the non-use of normative patterns may be due not only to cognitive factors and speakers’ emergent Spanish bidialectalism, but to the fact that standard Spanish does not play a prominent role in Northern Belize.

In relation to the differential use of intra-sentential CS across age groups, it must be underscored that language proficiency is a crucial factor. In particular, the markedly differential use of mixed DPs between the older and younger generations (Chapter 5) primarily results from differences in bilingual proficiency. In previous work, similar cross-generational patterns have been attested, where bilingual proficiency has been found to be a key factor in CS frequency and interspeaker variation (e.g., Backus, 1996; Jenkins, 2003; Weston, 2013).

293 The findings from this dissertation support the idea that bilingual proficiency is one of the most important factors in the differential cross-generational use of CS practices, as it is particularly among bi/multilinguals that were proficient in NBS, English (and/or Kriol) that CS was found to be the most unmarked. Importantly, bilingual proficiency alone does not translate into the conventionalization of CS structures or into the development of CS as an unmarked medium. This is where the deterministic role of historical and sociolinguistic factors takes precedence. If bi/multilingual language practices are outrightly or subtely regarded as deviant in a community, speakers will simply avoid or limit their use of such practices. It is vital, therefore, for scholars who study language contact and CS to take into consideration not only linguistic and cognitive factors, but social ones as well.

One sociohistorical event in Northern Belize within the last four to five decades that certainly merits further elaboration is the development of Spanish/English CS as an identity marker, as I argue that this is one of the crucial factors that allowed bi/multilingual language practices to thrive in this context. The fact that Spanish/English CS has assumed an identity function in Northern Belize partly explains why CS has become prevalent among Northern

Belize Maya/Mestizos, particularly among younger generations.

This is a phenomenon that I specifically address here, as it has not been explored in any previous linguistic, anthropological or sociological analysis of Belize. In previous research on

Belize (e.g., Le Page & Tabouret-Keller, 1985; Ravindranath, 2009; Seitz, 2005), it has been underscored that BK is the linguistic marker of a ‘Belizean’ identity. Based on the present dissertation’s findings regarding speakers’ identities and their discourse practices, however, I posit that for the majority of Northern Belize Maya/Mestizos, Spanish/English CS plays a more prominent role in identity construction and negotiation.

294 In his work on speaker attitudes and identities among Belizeans, Seitz (2005) argued that

Mestizos in Belize assert their national identity by employing BK. He further claimed that the use of BK is driven by Belizean Mestizos’ desire to distinguish themselves from Central

American Mestizo immigrants. In his work, Seitz is puzzled by one of his consultants’

“…‘negative’ reference to the Spanish language…despite the fact that she also says that Spanish is the language mainly spoken in her own home. With this brief, but loaded, statement she culturally disconnects herself and Belize from the ‘language of Central America (p. 101).’”

Notably, although Seitz makes keen observations about this Northern Belizean female’s desire to dissociate herself from both Spanish and the “Hispanic cultural realm” (Seitz, 2005, p.102), he does not mention how bilingual language practices fit into the lives of these multilingual speakers. The erasure of this vital sociolinguistic aspect of the cultural heritage of present-day Northern Belize Maya/Mestizos reveals how little attention has been given to

Spanish/English CS by researchers who have investigated different sociohistorical, sociological or linguistic aspects of Belize. Seitz, like Brockmann (1979), makes reference to the practice of

CS among Northern Belize Mestizos, but he does so only in passing. This is rather surprising given that he discusses the roles of language (i.e., Kriol) in the projection of identity.

In relation to Seitz’s observation about the Maya/Mestizo female from Corozal who dissociates herself from Spanish, an alternative analysis could be that she was highlighting the fact that she does not speak ‘monolingual Spanish’ but rather a distinctive contact Spanish variety that markedly differs from monolingual varieties of Central American Spanish, not only in terms of phonology, but also in terms of morphosyntax. Thus, the speaker distances herself from monolingual Spanish in particular, which is linguistically associated with officially

Spanish-speaking Central American countries. This may be indicative of her acute awareness of

295 the fact that monolingual Spanish and Spanglish or Spanglishiol (combination of Spanish,

English and Kriol) are markedly different.

Pertinent to the foregoing discussion is that the available literature on Belize’s linguistic landscape has always been told from a Creole perspective. By this, I mean that anthropologists and linguists who have examined aspects of have particularly examined Belize as a Creole contact situation in the Anglophone Caribbean (e.g., Le Page & Tabouret-Keller,

1985; Le Page, 1992), and not necessarily as a Spanish contact situation in the heart of a

Spanish-speaking region. This may have partly led to the continuous oversight of the CS practices employed by Northern Belize Maya/Mestizos.

Previously, several researchers have suggested that BK is the symbol of “Belizeanness”

(Batty, García, & Cucul, 2011; Caritas Lawrence, 2001; Decker, 2005; Escure, 1983; Hellinger,

1974; Le Page, 1992; Le Page & Tabouret-Keller, 1985; Ravindranath, 2009; Salmon & Gómez

Menjívar, 2014; Seitz, 2005). Previous data support the idea that Belizeans generally give greater importance to their national identity. Seitz, for example, found that 74.3 % of his consultants reported that they most strongly think of themselves as ‘Belizean’. Survey data from the present dissertation also revealed that the majority of consultants rated ‘Belizean’ as the ‘most important’ label that defines who they are.

My view, however, is that the notion of “Belizeanness” vis-à-vis Northern Belize

Maya/Mestizos cannot be associated primarily with the use of BK. Recall that although the consultants from the second sample concurred that BK is a language (see Chapter 1), they also agreed that BK does not represent who they are. Therefore, they do not claim a Kriol linguistic identity. On the other hand, consultants from the first (n = 38) and second (n = 52) samples agreed that Spanish/English CS is reflective of who they are. Therefore, Seitz’s (2005, p.142)

296 contention that “[l]ittle doubt can remain that for many native-born Belizeans of all ethnic backgrounds, fluency in Creole is the first sign of ‘Belizeanness’” is a misleading statement.

The present dissertation strongly suggests that for the majority of Northern Belize

Maya/Mestizos, Spanish/English CS has developed into the marker that speakers primarily use to assert their “Belizeanness”, contra Seitz’s (2005) contention that the use of BK is the means through which Mestizos in Belize assert their national identity. The cultural distance that Seitz

(2005, p.106) argued Northern Belize Mestizos feel in relation to , , and

Hondurans is actually more linguistic in nature. It is, after all, the linguistic aspect of Northern

Belize Maya/Mestizos – Spanish/English bilingual language practices – that individualizes them in Central America.

Notably, in remote areas of Northern Belize, there is still limited competence in BK.

Thus, Maya/Mestizos in rural areas can only assert their ‘Belizean’ identity through the use of

Spanish/English CS, which is what distinguishes them from Central Americans. Given that the majority of Maya/Mestizos in Northern Belize are Spanish-dominant bilinguals, it would be counterintuitive to argue that the younger Mestizo generation’s primary way of showing their

‘Belizean’ identity is through the use of BK. This would essentially suggest a radical shift from

Spanish to BK, which clearly has not taken place in Northern Belize (see census data on language use in section 1.2.1 in Chapter 1). Although there are many Maya/Mestizos who speak

BK and Spanish as first languages in Northern Belize urban areas, Spanish still has a stronghold in rural communities. For Spanish-dominant bilinguals in rural areas, the primary means of asserting a Maya/Mestizo Belizean identity is through the use of Spanish/English CS and not

BK. For Maya/Mestizos proficient in BK, both Spanish/English CS and/or BK may be used as linguistic markers to assert a Belizean identity.

297 In this regard, we can analyze the Northern Belizean Maya/Mestizo identity as a complex, fluid, multidimensional phenomenon, in line with Bailey’s (2002, p.99) view that

“identities are not reified dichotomies but rather involve multiple alignments and oppositions…vis-à-vis other individuals or [ethnic] groups”. Linguistically, CS enables Northern

Belize Maya/Mestizos to assert both their national and mixed ethnic identities. Given the strong anti-colonial sentiments that were prevalent during the 1960s and thereafter, CS may have also thrived not only because it was an “act of identity” (Le Page & Tabouret-Keller, 1985), but because it was an “act of resistance” (Meakins 2008, p.70) against colonization and linguistic purism as well. In post-colonial societies, the search for a unique identity is a crucial endeavor for people, and it is through language that speakers either reject or embrace different identities.

6.4 Toward a Reconceptualization of Code-Switching

I now return to the debate on what intra-sentential CS entails. Recall that whereas some scholars view CS as being categorically distinct from other phenomena such as borrowing (e.g.,

Poplack and colleagues), others argue that CS is a “chameleon-like” phenomenon (Gardner-

Chloros, 1995, p.86) that is inclusive of other contact phenomena. Findings from the present dissertation support a dynamic, multidimensional view of CS, which, as proposed by Gardner-

Chloros (1995, 2009), is inclusive of other interlingual phenomena such as borrowing, creolization and convergence. We could envision these interlingual phenomena as ‘strategies’ that become instantiated in CS.

In the ensuing sections, I discuss how findings from this dissertation support the view that CS co-occurs with other contact phenomena (Gardner-Chloros, 1995, 2009). Drawing on findings from the present dissertation, I explain why it is best to conceptualize CS as a complex phenomenon that comprises different strategies that are not mutually exclusive (Muysken, 2000,

2013).

298 6.4.1 Convergence and Creolization

The bi/multilingual data from Northern Belize revealed that CS can co-occur with convergence and creolization. Thus, CS is best analyzed as a complex phenomenon that comprises different optimization strategies (i.e., insertion, alternation, congruent lexicalization, backflagging) that are not mutually exclusive (Muysken, 2000, 2013). Rather than simply representing the mere combination of grammatical systems, CS is a dynamic phenomenon that is often characterized by features of convergence and linguistic creativity (for relevant discussion, see Gardner-Chloros & Edwards, 2004; Lakshmanan, et al., 2016). While the latter phenomenon has often been associated with the emergence of mixed languages and Creoles, it is imperative to not undermine this feature of intra-sentential CS in bi/multilingual communities (for relevant discussion, see Lakshmanan et al. (2016)), as CS can lead to cross-generational linguistic innovation and language change (Backus, 1996, 2004, 2005; Balam, 2015, 2016a; Lakshmanan et al., 2016).

In relation to convergence, Northern Belize bi/multilinguals created parallels between lexical items and structures from their languages, and they merged them into hyrid, syntactic and semantic forms, exhibiting elements from Spanish and English. Doǧruöz and Backus (2009, p.58) underscore that in the process of copying (in the sense of Johanson (2002), referring to the importation of elements from another language), speakers’ perception of semantic equivalence plays a key role, “[particularly] where speakers perceive a transparent link between translation equivalents.”

From BLVCs where speakers merged English phrasal verbs and/or idiomatic expressions and Spanish pronominal forms (e.g., se hizo give up ‘she gave up’, me hacen advise ‘they advise me’, se hacen behave mal ‘they misbehave’, etc.), it is evident that speakers consciously or unconsciously engaged in a search for semantic and syntactic equivalence and subsequently

299 produced innovative structures. BLVCs, thus, revealed an element of convergence, which we cannot simply conceptualize as a simple combination but rather as skillful “enhancement”

(Bullock & Toribio, 2004) of linguistic systems, which enables bi/multilinguals to fully exploit the rich repertoire of features available to them.

This enhancement was particularly evinced in mixed DPs, where the Spanish gender distinction was neutralized in bi/multilingual speech. Although Myers-Scotton (2002, p.105) argues that composite CS results when “speakers – because of psycholinguistic or sociopolitical factors – do not have access to the morphosyntactic frame of the participating language that is the desired source of the Matrix Language”, I adopt the view that linguistic convergence toward

English structure in bi/multilingual speech can be a strategy that bi/multilinguals employ to

“ease” (Clyne, 1987) or “facilitate” (Sebba, 1998, 2009) bi/multilingual switching. Based on previous research, this strategy is instantiated in Spanish/English CS across sociolinguistic contexts (i.e., New York, New Mexico, Miami, Belize), suggesting that the parsimonious use of linguistic structures in bilingual speech is a natural aspect of the bi/multilingual mind (Toribio,

2004).

Social factors, I argue, ultimately determine to what extent convergence is manifested in intra-sentential CS. In the case of Miami and particularly Northern Belize, the neutralization of the Spanish gender distinction seems more pervasive, as English-origin nouns that are transparently feminine are generally masculine-marked. In the U.S. Hispanophone context,

Miami stands out, as it has long been considered “a Spanish-speaking city” (Le Page &

Tabouret-Keller, 1985, p.74)2, where Cubans have “prov[en] that English fluency is not a sine

2 Stepick and Stepick (2002, p.79) report that in Miami “[i]t is easier to find a job, to shop, and just to get things done if one knows Spanish”. As it relates to language use, they highlight that the 1990 Census revealed that Spanish had replaced English in Miami-Dade, as the language most often spoken at home. According to Stepick and Stepick,

300 qua non for economic advancement”. Thus, the pressure to assimilate to an English-speaking society is not as strong in Miami as in other parts of the U.S. This balances the status of Spanish and English more, allowing speakers to more freely engage in switched discourse. Based on

Miami’s unique historical and sociolinguistic profile, therefore, it is not surprising that Miami would provide favorable conditions for Spanish/English CS patterns to regularize more than in other U.S. cities where English is the language of unparalleled prestige and dominance.

Importantly, the fact that even in New Mexico the overwhelming use of the masculine gender is attested suggests that in cases of classic Spanish/English CS (e.g., Clegg & Waltermire,

2009), traces of composite CS are also present. This supports Gardner-Chloros’ (1995, p.68) view that the rigid maintenance of the separate identities of linguistic systems in CS may in fact be the exception rather than the norm. It also supports Toribio’s contention that irrespective of the sociolinguistic context, “bilinguals select the most parsimonious grammar that serves both languages” (p.167). Thus, convergence is an integral element of classic CS contexts and not restricted only to cases of composite CS, as generally thought. The reason why the use of the masculine default gender is found across sociolinguistic contexts may simply relate to the bi/multilingual mind’s reflex to prioritize a principle of economy when there is co-activation of different languages.

One piece of evidence that supports the notion that bi/multilinguals do not necessarily strive to strictly apply the grammatical procedures of Spanish in CS is that a feminine default gender may be employed, as in the case of Basque/Spanish mixed DPs (Parafita-Couto et al.,

2014, p.5). In a total of 92 mixed DPs, Deuchar, Epelde, Oyharçabal and Parafita Couto (2010) found that the feminine gender occurred with Basque nouns 78.3% of the time. In this case, the

what makes Miami the de facto capital of Latin America is not only the number of Latinos and the dominant use of Spanish, but the “political and economic clout exercised by Miami Cubans” (p.79).

301 feminine gender was attested with both feminine translation equivalents (48.6%) and masculine translation equivalents (51.4%). This reveals that even in naturalistic CS involving Spanish and other languages, faithfully maintaining the grammatical procedures of the gendered language

(Spanish) does not take precedence in the gender assignment process in CS. Instead, speakers adopt a default gender.

The Northern Belize data, therefore, reveal that convergence constitutes an important strategy in CS, which may lead to language innovation and language change. Recall, however, that I am using this term as envisioned by Toribio (2004), where convergence is not necessarily indicative of a matrix language ‘turnover’ (Myers-Scotton, 1998). Instead, convergence is a CS- induced reflex or strategy that is specifically triggered in bi/multilingual speech. In the case of mixed DPs, for instance, there was no evidence that gender assignment in monolingual Spanish

DPs revealed instability or approximation toward English (i.e., no gender distinction or overuse of the masculine gender). The neutralization of the Spanish gender distinction was a phenomenon that was only operative in switched DPs.

I take convergence to be a CS strategy in its own right (for an alternative view, see

Muysken, 2014), but it is more easily observable in certain contexts, where historical and sociolinguistic conditions are conducive to the employment of dense CS practices, and hence, convergence as an optimization resource in bi/multilingual speech. I do not anticipate, for instance, for composite CS structures to be prevalent in bilingual contexts with a monoglossic tradition where CS is marked or where speakers engage mostly in intersentential CS. In the U.S.

Hispanophone context, for instance, language contact change has mainly manifested itself in the preferential use of pre-existing structures and in the loss of semantic and discourse-pragmatic constraints in the use of these forms (Muysken, 2013; Silva-Corvalán, 1993, 1994, 2008;

302 Toribio, 2004). In contrast, in a post-colonial context like Northern Belize where bi/multilingualism has been the norm for generations and where speakers do not have ultra- normative language attitudes, there has been the cross-generational emergence of innovative, convergent structures (for relevant discussion, see Balam, 2015, 2016a).

We can also find hybrid CS structures in diverse, multilingual contexts like the Indian sub-continent (for relevant work, see Bhatia & Ritchie, 2008, 2016; Chatterjee, 2016; Gumperz

& Wilson, 1971; Romaine, 1995, among many others), where language mixing is, as Bhatia and

Ritchie (2016) aptly highlight, “a grass-roots phenomenon…that is stable, inevitable and time- tested” (p.1). In , like in Belize, although English has official status, it is CS that is unmarked, particularly among younger generations.

It is in contexts like these where CS is ‘a way of life,’ as Lakshmanan describes it (U.

Lakshmanan, personal communication, June 10, 2016), that we are most likely to find the more dynamic, “chameleon-like” (Gardner-Chloros, 1995, p.86) nature of intra-sentential CS. Note that I do not take the lack or limited use of convergence as a ‘reflex’ of bi/multilingual speech in certain cases of classic CS as indicative of the non-existence of convergence as a strategy of CS, but rather as a reflection of larger historical and sociolinguistic conditions (e.g., the prestige of language varieties, the marked presence and use of bi/multilingual language practices, speakers’ language attitudes, etc.).

As it relates to creolization, speakers produced syntactic innovations that have no exact structural equivalents in either Spanish or English (e.g., ‘hacer + V + hacer + V’) and that are unattested in previous generations (Balam, 2015). It is worth noting that the cross-generational increase in the use of Spanish/English CS in the last four to five decades in Northern Belize has resulted in speakers’ creation of novel, hybrid structures. To my knowledge, BLVCs in passive

303 and control structures have not been reported for any other varieties of Spanish/English CS. The fact that novel forms of BLVCs do not have structural equivalents in either Spanish or English shows that across generations CS can indeed serve as a mechanism of morphosyntactic innovation and language change (Balam, 2015; Backus, 2005; Gardner-Chloros & Edwards,

2004; Myers-Scotton, 2002; Thomason, 2001).

Whether we view the emergence and development of novel BLVCs as reflective of a semantic extension (Pfaff, 1979), reanalysis (Backus, 2005) or grammaticalization (Backus,

1996; Vergara Wilson, 2013), what remains clear is that we cannot describe or conceptualize CS as the mere combination of two monolingual grammars or lexicons (Gardner-Chloros &

Edwards, 2004), as CS often involves the creation of new, hybrid syntactic (i.e., bilingual light verb constructions) and semantic structures (i.e., mixed DPs with a neutralized gender distinction) that are exclusive to a mixed or switched mode (Edwards & Gardner-Chloros, 2007;

Gardner-Chloros, 1995; Sebba, 1998, 2009). This necessarily implies speakers’ access to and creative use of both universal principles and the language-specific resources available to them in their bi/multilingual linguistic repertoire (for further discussion, see Bhatia & Ritchie, 2016;

Lakshmanan et al., 2016).

In relation to ‘hacer + V’ in Northern Belize, it is pertinent to note that BLVCs have been attested in Spanish/Maya CS (Suárez-Molina, 1996). Crucially, however, while we can attribute canonical forms of ‘hacer + V’ to a pre-existing template in Spanish/Maya CS (Suárez-Molina,

1996), we cannot do so for BLVCs in passive and control structures. The syntactic structure in bilingual passive BLVCs, for instance, is based on the analytical passive (i.e., ‘be’ verb + past participle), characteristic of both English and contact Spanish (Escobar, 2010, p.478; Lipski

2008, p.67), but not Yucatec Mayan, which lacks verbal copulas (Armstrong, 2009).

304 In Yucatec Mayan, passivization is morphologically marked post-verbally rather than pre-verbally (Bohnemeyer, 2009; Lehmann, 2015); suggesting that it is unlikely that passive

BLVCs were attested in the pre-contact context (i.e., Yucatan peninsula, Mexico), where hacer primarily occurred with nominal elements rather than verbs (Suárez-Molina, 1996). Given the marked structural differences between Spanish and Yucatec Mayan passives, we can surmise that

Spanish/Maya contact may not have facilitated the construction of congruence and hence a switch site. In contrast, in the Northern Belize contact situation, convergence or congruent lexicalization (Muysken, 2000, 2013) was certainly catalyzed by the similarities between

Spanish and English analytical passives.

Passive BLVCs are, thus, indicative of CS-driven convergence (Backus, 2009; Clyne,

2003; Sebba, 1998). The creation of this novel switch site has also taken place in this context given the favorable social conditions for bi/multilingual behaviors. Toribio (2004, p.172) rightly asserts that “congruent lexicalization is predicted to be more favored in communities in which code-switching is practiced and valued (e.g., as a marker of in-group membership)”. Thus, in the case of Northern Belize, historical and sociolinguistic conditions have allowed hybrid structures that evince convergence not only to spontaneously arise but to gain ‘currency’ (in the sense of

De Graff, 2009, p.897) and thrive, both at the societal and idiolectal levels (Balam, 2015, 2016a;

Balam, Prada Pérez & Mayans, 2014). While convergence is attested as a reflex of bilingual speech (e.g., neutralization of Spanish gender distinction in CS), we may also see elements of convergence from a more diachronic dimension (i.e., novel forms of BLVCs, which exhibit features from multiple languages, may develop across time).

In the theoretical literature, both BLVCs and mixed DPs are often argued to be illustrative of classic CS (e.g., Myers-Scotton & Jake, 2013), a problematic idea given that both

305 of these CS structures evince elements of convergence (Gardner-Chloros, 2009). A positive development in Myers-Scotton and Jake’s work is that now they highlight the malleability or flexibility of the ML in relation to the nature of EL elements it accepts, a position that differs from earlier postulations of classic CS (e.g., Bilingual NP Hypothesis), where the ML was conceptualized as carrying out a more inflexible role in terms of the grammatical procedures underlying the derivation of mixed constituents.

This flexibility is now able to account for BLVCs that appear more Spanish-like (e.g., Se tienen que hace behave en un certain way ‘They have to behave in a certain way’) or English- like (e.g., Tienen que hace behave en un certain way ‘They have to behave in a certain way’). It also captures the dynamism, complexity and multidimensionality that characterize CS (Gardner-

Chloros, 2009, 2010; Gardner-Chloros & Edwards, 2004; Muysken, 2000, 2013; Sebba, 1998,

2009). Most importantly, it underscores the dual/multiple activation of grammars in bi/multilingual speech, where linguistic resources remain available so that speakers employ different features in their rich repertoire to devise novel structures and meanings (Bhatia &

Ritchie, 2008, 2016).

Although Myers-Scotton and Jake (2013) do not classify BLVCs as representative of composite CS, their recent analysis is more in line with the contention that in CS, bi/multilinguals focus on the parsimonious use of their linguistic resources (Bullock & Toribio,

2004; Toribio, 2004), even if this means suppressing aspects of monolingual language production. This is a positive step away from the notion that in CS, the same processes that apply in monolingual production are exactly replicated. Myers-Scotton and Jake underscore that bilinguals can fine-tune their cognitive control of both languages. In the ensuing section, I

306 further explore why embracing a dynamic view of CS is necessary to further understand bi/multilingualism.

6.4.2 Re-examining the Nature of Code-Switching

Relevant to the foregoing discussion is Poplack and Dion’s (2012) criticism of previous researchers (e.g., Field, 2002; Heath, 1989; Myers-Scotton, 1993, 2002, 2006; Thomason, 2003;

Treffers-Daller, 2005; Winford, 2009) who contend that the distinction between borrowing and

CS is inherently fuzzy and not categorical. Poplack and Dion (2012, p.280) observe that “[i]t would seem that this question should be decidable on the basis of systematic empirical enquiry, but the cumulative advances of what little quantitative research there is in this field have apparently had little if any impact on the debate.”

The most pressing issue at hand, however, which the present findings highlight, is not whether or how CS and borrowing are different, but that “the dynamic character of CS, which is a major vehicle of language change and convergence, is not accounted for” (Gardner-Chloros &

Edwards, 2004, p.104) in mainstream models of CS. Recall that models such as the MLF

(Myers-Scotton and colleagues) and the Minimalist approach to CS (MacSwan and colleagues) champion the notion that CS is the combination of two grammars or lexicons that strictly maintain their identities in mixed discourse. As Otheguy et al. (2015) aptly highlight, “the notion of code switching still constitutes a theoretical endorsement of the idea that what the bilingual manipulates, however masterfully, are two separate linguistic systems” (p. 282).

On the other hand, convergence is not conceptualized as a key element or strategy of CS.

While influential CS models have undoubtedly enriched our understanding of intra-sentential

CS, they have not embraced a dynamic view of CS in which other interlingual phenomena such as convergence comprise an integral component of ‘classic’ CS. Hybridity and convergence have been consistently excluded from mainstream definitions of CS, which in itself is a larger

307 theoretical issue that must be addressed if we are to develop linguistic theories that move beyond a narrow view of bilingualism and which can account for linguistic outcomes in different CS contexts.

Given the widespread conception that a bilingual is a person who possesses two discrete grammars or lexicons, it is unsurprising that many linguists simply assume in their work that the starting point for bilinguals is the monolingual mode. It is also unsurprising that educational practices, specifically as they relate to language minority students, continue to align with a view of bilingualism that is ‘parallel’ (Heller, 1999) or ‘dual’ (Fitts, 2006). Since the 1980s, it has been argued that a bilingual is not the sum of two monolinguals (Grosjean, 1989); ironically, language programs in the U.S. continue aiming at “producing students who function similarly to monolingual speakers of two distinct languages” (Palmer & Martínez, 2013, p.274).

The dogmatic ideology of language separation that García and colleagues argue dominates current classroom practice aligns with an elitist view of bilingualism (i.e., fluent speaker of language A and language B), which is essentially in tandem with the mainstream view of CS (i.e., the alternation of two systems that remain discrete in bilingual discourse). Recall that the mainstream view of CS completely excludes convergence; hence, suggesting that hybridity is not necessarily a natural outcome of CS. In second and heritage language classrooms, this linguistic ideology fails to promote dynamic bilingualism, and frames hybridity (i.e., convergent structures) in students’ home varieties (i.e., CS, Spanglish, Spanglishiol, etc.) as an anomaly that needs remediation (for relevant discussion, see García, 2009, 2014).

The reconceptualization of bilingualism and CS, hence, is necessary because it has important ramifications not only for theoretical linguistics, but also for language education programs and pedagogical practices. Palmer and Martínez (2013) highlight that current

308 approaches to teach Latino students in the U.S. are inadequate, given that they are guided by monolingual perspectives of language and bilingualism. For many educators, only monolingualism is normalized, whereas Spanish/English CS or Spanglish continues to be vilified as an unwanted language practice that has no place, value or function within the classroom.

Thus, the translanguaging practices of bi/multilinguals are not only suppressed in schools

(Otheguy et al., 2015), but they are still treated as deviant.

García (2014, p.69) highlights that dual-language programs do not “give any space to the language practices of the bilingual Latino community, stigmatizing their practices and compartmentalizing English and Spanish strictly.” This separation of languages in these programs follows a purist, monolingual tradition, aimed at ensuring that one language does not

‘contaminate’ the other (Otheguy et al., 2015, p. 302). In García’s view, these practices promulgate the notion that native bilinguals’ hybrid language practices need remediation, and that mythical bilingualism (i.e., bilingual who is equally fluent in two languages) should be the end goal of language education. García and colleagues, thus, propose that “[r]ather than view bilingualism as the combination of two separate, bounded languages…it is more profitably understood as a repertoire of related language practices” (Palmer & Martínez, 2013, p.277).

I concur with García and colleagues in that there must be a shift away from monolingual frameworks of bilingualism that alienate hybridity. This shift will need to be one that

“normalizes hybridity and that acknowledges the creativity, flexibility, and skill embedded in bilingual students’ everyday language practices” (Palmer & Martínez, 2013, p.288). The findings from the present dissertation have precisely revealed this aspect of Northern Belize bi/multilinguals – the hybridity, creativity, and flexibility evinced in speakers’ skillful use of their rich linguistic resources. In line with sociolinguistic and sociocultural approaches to

309 language learning (e.g., Canagarajah, 2004, 2007; Norton & McKinney, 2011; Norton &

Toohey, 2001, 2004; Toohey, 2000), it is vital that speakers’ hybrid language practices be recognized as resources rather than deficits in the language learning process.

Importantly, what scholars like Gardner-Chloros and García have in common is that they advocate for a dynamic view of bilingualism, in which complex, hybrid aspects of bilingual language practices constitute a natural and essential component of bilinguals’ linguistic resources. Thus, bilinguals have a unified repertoire of bilingual language practices that they develop to different degrees (García, 2009, 2014; García & Wei, 2014).

In summation, findings from this dissertation support the idea that CS leads to convergence and language change (Backus, 2004, 2005; Clyne, 1987; Fuller & Lehnert, 2000;

Gardner-Chloros & Edwards, 2004; Myers-Scotton, 2002; Sebba, 1998, 2009; Thomason, 2001;

Toribio, 2004). More importantly, this dissertation is in line with previous sociolinguistic work that advocates for a more dynamic view of CS (Clyne, 2003; Gardner-Chloros & Edwards, 2004;

Gardner-Chloros, 1995, 2009, 2010; McCormick, 2002; Sebba, 1998, 2009), language contact

(Clyne, 2003; Muysken, 2013), and bilingualism (García, 2009, 2014; García & Kleifgen, 2010;

García & Wei, 2014; Otheguy et al., 2015; Pennycook, 2010). A primary yet challenging step in this process of reframing what CS and bilingualism are will be accepting and recognizing hybridity as a normal aspect of bilingualism and CS.

In essence, if we take bilingualism and CS to be dynamic processes, and if we agree that languages are not lifeless entities that remain static across time, there is no principled reason to a priori assume that CS will not co-occur with (a certain degree) of convergence and contact- induced change and innovation. Hence, classic and composite CS should not be mutually exclusive, as they are intricately connected contact phenomena. While previous work on the

310 borrowing/CS dichotomy has contributed to our understanding of intra-sentential CS, it has nonetheless consistently excluded and/or undermined the relationship between CS, convergence and language change. As more language educators start recognizing the value of translanguaging as an instructional resource in second and heritage language classrooms, it is imperative that linguists also re-evaluate their conceptualization of intra-sentential CS, which for decades has been viewed and examined as the mere combination of two discrete systems that should not exhibit any convergence and/or other interlingual phenomena.

6.5 Weighing in on Social and Linguistic Factors

The last issue at hand relates to the deterministic roles of linguistic versus social factors.

The present findings suggest that speech patterns are not just the outcome of linguistic factors and/or syntactic features. Contact outcomes are intricately connected to the historical and social realities of people; hence, rendering a scenario approach, or a socially contextualized interpretation of language contact outcomes (Muysken, 2013, p.710), as the most viable way to examine and understand linguistic outcomes.

Findings from the present dissertation have shown that linguistic factors (e.g., verb stativity) and/or formalist features (e.g., phi features) alone cannot account for language contact outcomes in Northern Belize. As it relates to the intervocalic rhotic contrast, the presence of variable neutralization in reading but not in semi-spontaneous speech cannot be attributed only to speakers’ proficiencies and emergent Spanish bidialectalism. We must take into consideration speakers’ predispositions toward standard Spanish in this post-colonial context.

The effect of attitudinal factors on speakers’ discourse practices was substantiated by the

Spanish/English CS data examined, where it was found that CS was productive and unmarked among consultants, in line with a positive predisposition toward and identification with bilingual language practices. In the analysis of the CS data, it was found that both González-Vilbazo’s and

311 López’s (2011) Minimalist analysis of BLVCs and Moro’s (2001) Minimalist analysis of mixed

DPs underpredicted switches. Eventive passives BLVCs were rated as acceptable, contra

González-Vilbazo and López’s (2011) contention. Additionally, the presence of mixed DPs with ungendered English articles as in [the casa] ‘the house’ were attested in the Northern Belize context, a switch that Moro’s analysis does not predict.

Winford (2013, p.372) rightly asserts that “contact linguists have paid far more attention to linguistic processes and constraints than to the extra-linguistic factors that might affect their operations and results”. In the case of ‘hacer + V’, semantic category incorporation, and gender assignment in mixed DPs, it is clear that analyzing CS phenomena in relation only to grammatical “constraints” may not be the most informative avenue to understand the emergence and use of innovative forms, given that this analysis yields limited insight into a complex, sociocultural phenomenon like CS.

The use of ‘hacer + V’ and mixed DPs are particularly notable, as we are able to make preliminary comparisons between Spanish/English contact outcomes in Northern Belize, where there has been intense language contact only for about 75 years (Balam, 2014, p.81) versus those in Southwest U.S., where Spanish has been in intense contact with English for at least 150 years

(Torres Cacoullos & Travis, 2010, p.244).

Gardner-Chloros and Edwards (2004, p.122) posit that one of the ways we can evaluate the effects of social and linguistic factors is to examine the outcomes of the same pairs of languages in different sociolinguistic contexts and different pairs in similar social settings. While such a comparative analysis clearly falls outside the scope of the present dissertation, the current findings offer valuable insight into the similarities and differences between Spanish/English

312 contact outcomes in Northern Belize versus communities in the U.S. Hispanophone context, especially Southwestern U.S. bilingual communities where ‘hacer + V’ is also used.

The extensive research conducted on Spanish/English CS in the U.S. has revealed that

U.S. Latinos actively engage in bilingual language practices (e.g., Espinosa, 1914, 1917; Lipski,

2005, 2008; Pfaff, 1979; Silva-Corvalán, 1994; Zentella, 1997, among many others). Torres

Cacoullos and Travis (2010) especially highlight the frequent use of CS in New Mexico.

Furthermore, CS is also employed as an identity marker among U.S. Latinos (e.g., Bailey, 2002;

Jacobson, 1998; Toribio, 2002, 2006; Zentella, 1997).

If Spanish/English CS serves as an identity marker in both the Northern Belize context and in many U.S. Spanish/English communities, how do we explain the seemingly more productive and frequent use of ‘hacer + V’ and mixed DPs in the former context? I argue that a confluence of inter-related historical and sociolinguistic factors allowed Northern Belize

Maya/Mestizos to fully exploit CS as a sociocultural and communicative resource. In particular, relaxed language norms and speakers’ attitudes toward and identification with CS have allowed this language practice to thrive.

We know that the more positive attitudes speakers have toward CS, then the more predisposed they are to the use of CS (Parafita Couto, Deuchar, & Fusser, 2015). In contexts with a history of ultra-normative language attitudes both in academic and non-academic domains, we cannot anticipate CS to flourish (for relevant discussion, see Jacobson, 1998;

Muysken, 2000, 2013). Instead, such contexts are characterized by clear divisions as to where and when language varieties are used. The clear compartmentalization in the use of speakers’ languages is what probably led Jenkins (2003) to propose that BLVCs are used only in certain domains, such as academic and occupational contexts, but not in the home, where only Spanish

313 is used. My view, however, is that the non-use of hacer with high frequency verbs in Jenkins’ data is not necessarily a reflection of linguistic constraints at play, but more so the status of CS and the overall instrumental value afforded to CS by the community itself.

It is important to realize that no matter how much we try to dissociate language and bilingualism from sociopolitical issues, the reality is that the status and use of languages reflect sociopolitical ideologies, movements, and structures (Bucholtz & Hall, 2005). In Zentella’s candid words (1997, p.14), “whether we choose to discuss it or not, there is no language without politics”. Crucial to the status of bilingualism and linguistic variation is the value that governments and academic institutions place on it. If children go to schools where they are discouraged from and/or stigmatized for engaging in bilingual behaviors, they will simply abstain from engaging in such practices. Bhatia and Ritchie (2008) remind us that it is not only prescriptivists that disparage bilingual language practices, and hence, propagate negative attitudes toward these practices. Governments are also part of this endeavor (e.g., implementation of ‘language police’, state bans on , etc.).

In terms of the status of languages, societal attitudes toward bi/multilingualism, and processes of identity formation, the experiences of Maya/Mestizos in Northern Belize and

Latinos in the U.S have been dramatically different. Historically, as pointed out in Chapter 1, an

American identity has always been associated with monolingual English (see Molesky, 1988, p.51; Zentella, 1997, p.266). It is unsurprising, therefore, that since the 1980s, there has been “an anti-immigrant and anti-Spanish fervor that has accompanied the adoption of English-only amendments” in the U.S. context (Zentella, 1997, p.287). This explains why the constant pressure to “Americanize” has resulted in a rapid transition to English monolingualism by the second generation in many Hispanic communities (Bills, Hernández-Chávez & Hudson, 1995,

314 2000; Thompson, 1974; Veltman, 1988; Zentella, 1997). This also accounts for the widespread stigmatization associated with CS and other low prestige varieties (e.g., African American

Vernacular English) in the U.S. context.

In contrast, the sociohistorical experience of bi/multilingualism in Belize is one where political leaders have embraced linguistic variation. Belize’s national language policy recognizes the value of first languages in the classroom. In fact, the use of first languages for classroom instruction seems to have been the norm in many rural areas in colonial Belize (Hagerty, 1979, p.10; Koenig, 1975). Recent work suggests that even today bi/multilingual language practices and BK are not marginalized from Northern Belize classrooms; instead, they are often incorporated as pedagogical resources (Balam & Prada Pérez, in press). Thus, the socio- linguistic and political milieu in Belize is clearly one that has allowed bi/multilingual language practices to flourish.

Thus, external rather than language-internal factors account for some of the differences in the use of BLVCs and mixed DPs in Northern Belize versus the Southwest U.S. In terms of social conditions in Northern Belize and the Southwest U.S. context, what primarily differentiates both contexts is the degree of normativity (but see Martínez, 2010). The New

Mexican context, for instance, is characterized by “strong normative pressures” (Torres

Cacoullos & Travis, 2010) and a history of linguistic repression (Guerra, 2015; MacGregor-

Mendoza, 2000) and suppression (Jenkins, 2003; Villa, 2002) in a wider society where monolingualism is valued.

In contrast, Northern Belize is characterized by a negative attitude toward standard varieties and greater acceptance of cultural diversity and linguistic variation in schools. Thus, degree of normativity at a societal level has played a crucial role in the dynamism or lack thereof

315 in which CS has manifested itself in both contact situations. The different contact outcomes evinced in Northern Belize versus the Southwest U.S. are in line with Muysken’s (2013) contention that a low degree of normativity stimulates or favors convergence.

Importantly, the powerful effect of societal attitudes toward prescriptivism is also evidenced in cases where although CS may be sanctioned as acceptable, this acceptability is maintained within strict confines. For instance, in her analysis of CS and social identities among elite, educated bilinguals in , Pérez Casas (2008) found that her consultants revealed a degree of critical awareness vis-à-vis the social connotations that CS has in this context.

Although CS is perceived as acceptable, there is a limit as to how much a speaker can switch in informal contexts, as it can have negative connotations. Alberto, one of Pérez Casas’ consultants, notes that “It’s okay to switch languages, but you also have to be careful not to switch too much.

I don’t know where the percentage lies: 75% Spanish and 25% English maybe?” (p. 15).

As this comment illustrates, CS in Puerto Rico is socially restricted. Highly frequent CS may be interpreted as sending a message of “superiority” or as a deliberate attempt to “seem superior” (p. 270) since this practice is equated with “a high socio-economic status and closer cultural ties with the U.S.” (p. 271). In addition, Pérez-Casas underscores that whereas ‘limited’

CS is indicative of prestige among Island , excessive language mixing is negatively perceived as it may be indexical of not knowing any language well. This attests to the overall prestige that monolingual varieties have at a social level, and to the powerful effect that attitudes toward CS and the degree of linguistic prescriptivism can have on the frequency of CS in a contact situation.

Previous and current work on CS in Northern Belize and the Southwest U.S. suggest that even prolonged societal bilingualism and intense contact is not enough to allow certain

316 innovations to thrive and further evolve. In her work, Gardner-Chloros (1995, p.69) points out that the emergence of new linguistic forms are just “a matter of time”; thus, in twenty or thirty years’ time or roughly one generation after, one can potentially expect innovations to emerge. If high levels of linguistic prescriptivism prevail in a community, however, CS as a dynamic entity is simply stifled. It may still assume an identity function and many speakers will be positively predisposed to it, but its use will nonetheless be both grammatically and socially restrained.

Thus, while innovation may be attested (e.g., ‘hacer + V’ in Southwest U.S.), the diffusion of these innovations and the subsequent creation of novel structures among bilinguals will be hindered (for relevant data, see Balam & Prada Pérez, 2013).

Overall, findings from this dissertation support the argument that in contact situations, CS outcomes are ultimately socially determined (Gardner-Chloros & Edwards, 2004; Thomason,

2008), and this applies not only to cases with more extraordinary outcomes such as Creoles or mixed languages. This is not to say, however, that there are no constraints or that structural factors do not play an important role in contact situations. They do. My view, and the view of other scholars with similar theoretical leanings, is that CS does not result in “grammatical monstrosities…nor [does] it [depart] from widely accepted universals of language structure and function” (Gardner-Chloros & Edwards, 2004, p.1437). The mere fact that the gender defaults in

Spanish contact situations differ according to the languages in contact (i.e., Spanish/English contact = masculine default gender; Basque/Spanish contact = feminine default gender) confirms that structural factors are indeed influential.

Importantly, however, they do not dictate the way in which speakers will use and manipulate their linguistic resources. Purely linguistic factors also cannot shape the sociopolitical

317 underpinnings and values that are ascribed to language varieties in social and academic domains.

People can and do in everyday discourse.

While the search for universal CS constraints continues to be relevant to the refinement of CS models, there are larger theoretical issues that must be addressed first, in particular the monolingual conceptualization of what CS entails. Canagarajah (2007, p.935) aptly contends that

“insight from non-Western communities should inform the current efforts for alternate theory building in our field” (cited in Norton & McKinney, 2011, p.88). It is vital that theory building in relation to intra-sentential CS takes into greater consideration the study of this dynamic phenomenon in multilingual contexts like Northern Belize where this practice is not restricted and/or hindered by dominant ideologies that privilege monolingualism over bi/multilingualism.

It is in contexts like these that we must conduct more research to examine how convergence and creolization relate to the common practice of CS or translanguaging among bi/multilinguals (also see Lakshmanan et al., 2016).

Of primary importance is comparative research that seeks to further unravel how social conditions shape linguistic outcomes in contexts with the same language pair but different sociolinguistic conditions. Gardner-Chloros (1995, p.126) aptly posits that “[u]ntil more comparative studies have been done, we should avoid making the leap from descriptive to predictive”. Traditionally, social and historical factors have been relegated to background information, when in many cases, they should be brought to the foreground and more thoroughly examined to provide a more contextualized analysis of contact outcomes. This seminal dissertation on Northern Belize, while certainly not an exhaustive analysis of social factors, has made a preliminary step in this direction.

318 6.6 Final Remarks

Endeavoring to fill a gap in the contact Spanish literature, I examined three linguistic phenomena in order to better understand the contact outcomes in Northern Belize, where NBS is in intense contact with English and BK. The analysis of rhotics, BLVCs and mixed DPs revealed that contact outcomes in Northern Belize are characterized not only by stability, but by innovation as well. These findings support the view that CS is a dynamic and complex process that cannot be restricted merely to the combination of two lexicons and/or linguistic systems.

While such conceptualization is well-suited for contexts with a strong monoglossic tradition, it is inadequate for post-colonial contexts where bi/multilingualism has been a way of life for generations. CS has multiple manifestations (i.e., optimization strategies), and the way this complex phenomenon manifests itself in a given community depends on other interrelated factors such as the languages involved, the nature of bilingualism, the status of languages, the degree of normativity, and the level of bilingual proficiency (Muysken, 2013, p.710).

I argue that in the Northern Belize context, social factors override linguistic factors, and this accounts for the patterns observed (i.e., variable intervocalic rhotic neutralization, productive use of BLVCs, and neutralization of the Spanish gender distinction in Spanish/English CS). A confluence of historical and social conditions in the Northern Belize context have allowed

Northern Belize bi/multilinguals to fully capitalize on CS as a linguistic and communicative resource. This explains explain why switching in determiner and verb phrases is more frequent in this context than in other Spanish/English contexts (e.g., Spanish/English CS in Southwest U.S.:

Pfaff, 1979; Spanish/English CS in Gibraltar: Weston, 2013).

This dissertation advocates for further research on understudied, multilingual communities where CS in the norm, as this work will contribute to the refinement of current linguistic theories and typologies. For example, a step in the right direction is Muysken’s (2013)

319 framework for the analysis of language contact, which illustrates an important shift away from binarity. Recall Muysken’s typology is a four-dimensional typology that formally incorporates both convergence and universal processes. Today, Jacobson’s (1998, p.59) contention that “not enough language contexts are yet known to make any far reaching predictions about the principles that govern ‘all’ language mixing events” still remains true. The fact that Belize, where Spanish was in intense contact with English by the 1980s, was completely overlooked reveals how much has to be done yet in terms of exploring similar contexts across the world where CS is tacitly accepted and embraced.

Most importantly, findings from this dissertation advocate for a re-evaluation of CS as phenomenon that specifically refers to the combination of two lexicons or monolingual grammars. This dissertation has shown that classic CS can be characterized by interlingual phenomena such as convergence and creolization. In line with recent scholarship that advocates for a shift away from a monoglossic view of bilingualism (Gardner-Chloros & Edwards, 2004;

García & Kleifgen, 2010; García & Wei, 2014; Otheguy et al., 2015), the present sociolinguistic work advocates for the recognition and formal inclusion of hybridity and innovation as an integral and normal aspect of bi/multilingual speakers’ language practices. For translingual and translanguaging to be accepted and incorporated as valuable instructional tools in the future, it is of vital importance that we first re-evaluate mainstream notions of what bilingualism and CS entails.

6.7 Limitations and Areas for Future Research

Although this dissertation makes a notable contribution to the contact Spanish literature, several limitations need to be addressed in future studies on Northern Belize contact Spanish.

First and foremost, although an in-depth analysis of linguistic factors (e.g., stativity, frequency, masculine default gender criterion, etc.) is provided, social factors such as degree of normativity

320 or language attitudes are not systematically examined vis-à-vis the use of rhotics and mixed constructions. Future work on the deterministic effects of sociolinguistic factors on Northern

Belize language contact outcomes, therefore, need to develop ways to systematically examine these social factors and hence substantiate or refute claims made in this dissertation. In particular, there needs to be more comparative work on degree of normativity and language attitudes across sociolinguistic contexts where Spanish is in contact with English. This will allow us to better understand how linguistic outcomes are influenced by these social factors, which are often undermined in the literature (see, for example, Haspelmath’s (2009) view against the use of attitudinal factors and puristic attitudes as a means of explaining differential borrowing behavior).

Secondly, future studies need to better balance the subject pool in order to have a more comprehensive insight into Kriol-dominant bi/multilinguals. While the present dissertation focused on Spanish-dominant consultants, phonological data from the two Kriol-dominant adolescent speakers in Chapter 3 suggest that a particularly promising avenue of research is language use among these consultants. Given that Kriol use is becoming increasingly prevalent among adolescents, the examination of language practices among Kriol-dominant bi/multilinguals could further elucidate our understanding of language shift and language change, particularly when a Creole plays a more prominent role.

In ongoing work on BLVCs, for example, a few consultants who reported frequently using BK at home employed light verb variants such as hicía for hacía ‘doimperfect’ and hicíamos for hacíamos ‘doimperfect’. In fact, even one case of haciba for hacía ‘doimperfect’ was attested.

Future investigation may reveal the emergence of different phonolongical phenomena and novel forms of hacer, particularly among Kriol-dominant bi/multilinguals. Another phenomenon that

321 merits further investigation is the use of feminine- versus masculine-marked stative and eventive passive BLVCs. As in the case of feminine-marked mixed DPs, it may be that certain factors and/or grammatical contexts favor the use of the feminine-marked light verb hecha ‘done’.

Another limitation in this dissertation was the small sample (n = 10) of the older generation (ages 55 +). Future investigation needs to more thoroughly examine a larger sample, given that this is the generation that bridges the gap between Maya/Spanish bilingualism to

Spanish/English/Kriol bi/multilingualism. Thus, the analysis of their speech patterns holds crucial information regarding features in NBS that were pre-existing versus those that more than likely constitute legitimate innovations that emerged during the last four to five decades. A more detailed analysis of this speaker group will certainly allow us to better understand cross- generational patterns of phonological and morphosyntactic innovation. For instance, research on

(suprasegmental) features (e.g., aspiration in voiceless stops, the use of creaky voice, etc.) could explore whether younger speakers’ naturalistic speech reveals less influence from Yucatec Maya than the speech of the older generation; hence, revealing an approximation to more pan-Hispanic norms (i.e., normative patterns) across time.

In addition, future work on grammatical gender can examine the phenomenon of diachronic convergence. This research can particularly address whether grammatical gender agreement in verbal contexts is being diachronically lost in NBS. One could surmise that with a radical increase in CS among the younger generation, gender agreement may not have had the same fate as gender assignment patterns in terms of its stability in stretches of monolingual

Spanish. While the variable use of gender agreement was attested among consultants from all generations, this pattern may be more pronounced among the youngest speakers, who have much more exposure to BK than previous generations. Alternatively, it could also be that formal

322 instruction in Spanish at the high school level may manifest itself in an increased use of canonical patterns of gender agreement.

It must be highlighted, however, that given students’ negative attitudes toward standard

Spanish, it remains unclear what future work on gender agreement will reveal about the cross- generational use of gender agreement in NBS. If previous scholars are correct in assuming that local forms of convergence (i.e., synchronic) and a more global process of convergence (i.e., diachronic) are driven by CS (Backus, 2004; Sebba, 1998, 2009; Thomason & Kaufmann, 1988), then future work should demonstrate a greater degree of morphosyntactic convergence toward

English among younger generations who code-switch significantly more than older consultants in Northern Belize.

Last but not least, it is important for future work to examine NBS in relation to the

Yucatec Maya language, especially in rural areas where the decline of Yucatec Maya was slower. This work could examine if and how the Spanish variety spoken in these villages that were predominantly populated by native speakers of Yucatec Maya (e.g., Xaibé) reveals influence from Mayan phonology, syntax, etc. Furthermore, it is essential to explore current language revitalization efforts in Northern Belize that seek to promote proficiency in Yucatec

Maya and the subsequent impact such programs have on the younger generations’ ethnolinguistic attitudes and language practices.

323 APPENDIX A READING PASSAGE

The Pink Crocodile

Once upon a time, there lived a pink crocodile that used to be very quiet. He didn’t say much, and he was very big, and all the inhabitants of the forest were afraid of him. On his tenth birthday, he decided to throw a party for all the animals in the jungle. But he didn’t have friends, so he decided to come out of the river to go and personally invite the animals. When he came out of the river, he came across a purple frog that was sitting on a shell. The pink crocodile opened his big snout to speak, but seeing his huge, sharp teeth, the poor frog got scared and it fainted. The crocodile was confused. Then, he continued his path, and he came across a little mouse. Once again, the crocodile tried to speak to him, but the little mouse, without thinking twice, scampered away and hid on top of a tree. The crocodile got sad. But he continued. And then he saw a black dog. And the crocodile was surprised because the dog did not run away when he tried talking to him. The dog said, “Yes, tell me, my friend.” And the pink crocodile replied, “I will have a big birthday party for my birthday. There will be ham, turkey, rice, lots to drink and lots to eat. I want you to come and celebrate my birthday.” Then, the dog started laughing like a madman. “I am not foolish. I want to live,” the dog told him. And the crocodile felt so sad because he asked himself what a party would be without invited guests. So, he decided to go back to the river. When he came close to the river bank, he saw a beautiful, yellow bird on a tree branch. He sadly told the little bird what had happened. And the wise little yellow bird told him, “The heart is like a book. Had you not closed it so many years ago, they wouldn’t be so terrified of you.”

324 APPENDIX B EL COCODRILO ROSADO

Había una vez que vivía un cocodrilo rosado muy callado. Decía pocas palabras y era muy grande, y todos los habitantes del bosque le tenían miedo. Cuando iba a cumplir diez años, decidió que iba a organizar una fiesta para todos los animales de la selva. Pero no tenía amigos así que decidió salir del río para ir a invitar a los animales personalmente. Al salir del río, se encontró con una rana morada que estaba sentada en un caracol. El cocodrilo rosado abrió su gran hocico para hablar, pero al ver sus grandes dientes filosos, la pobre rana se espantó y se desmayó. El cocodrilo quedó confundido. Luego siguió su camino y se encontró con un ratoncito. El cocodrilo una vez más intento hablarle, pero el ratoncito sin pensarlo dos veces, se escabulló como una flecha y se escondió arriba de un árbol. El cocodrilo se quedó triste. Pero siguió. Y entonces miró un perro negro. Y el cocodrilo se asombró porque el perro no corrió cuando intento hablarle. Dijo el perro: “si, dime amigo”. Y respondió el cocodrilo rosado: “Voy a tener una gran fiesta para mi cumpleaños. Habrá jamón, pavo, arroz, mucho que beber y mucho que comer. Quiero que vengas a celebrar conmigo.” Entonces se empezó a reír el perro como loco. “No, no nací ayer. Quiero seguir vivo”: le respondió el perro. Y el cocodrilo se sintió muy triste porque se preguntaba que era una fiesta sin invitados. Y decidió regresar al río. Cuando ya llegaba cerca de la orilla del río, miró a un pajarito amarillo muy bonito en una rama. Tristemente, le contó lo que había sucedido. Y el sabio pajarito amarillo le dijo: “el corazón es como un libro. Si no lo hubieras cerrado hace tantos años, no te tendrían tanto miedo.”

325 APPENDIX C LANGUAGE BACKGROUND QUESTIONNAIRE

Initials/Gender: ______

PART I Personal Information Date of Birth Where were you born? How long have you lived in Orange Walk? Specify Have you lived in another district/country? Where? How long?

Language Background

Circle the languages that you speak: Belizean Spanish Belizean Kriol English Yucatec Maya Other:______

√ Indicate the language that is applicable Belizean Other(s) Spanish Specify What is/are your first language(s)?

What languages were spoken at your home when you were a child (0 - 8 years)?

326 A. Use the scale below to answer the following questions. 1. Everyday, most of the time during the day 2. Everyday, sometimes during the day 3. Sometimes during the week 4. Once a week 5. Once or twice per month 6. Once or twice per year 7. Every two/three years 8. Never

in in in in a mixture % per Belizean Belizean English of the three day Spanish? Kriol? ? languages?

¿How often do you write ¿How often do you speak ¿How often do you listen to music ¿How often do you read newspapers, magazines and books ¿How often do you see televisión programs and movies ¿How often do you visit officially Spanish-speaking countries or communities?

B. Use the scale below to answer the following questions. 1. Only Belizean Spanish 2. Only English 3. Only Belizean Kriol 4. A mix of the three languages 5. Only Belizean Spanish and English 6. Only Belizean Spanish and Belizean Kriol

Number % per day of (in case of 2 Response or 3 languages) At home, what language is used the most? What language(s) do you speak when you talk to your closest friends? What language(s) do you speak when you talk to your classmates?

327 What language do you speak when you address your teachers? What language do you speak when you address your parents?

C. Use the scale below to estimate your proficiency in Standard Spanish, Belizean Spanish, Belizean Kriol, English, and a mixture of languages. (1 is minimum, 4 is moderate, 7 is native-like proficiency).

in in in in A mixture Standard Belizean Belizean English of Belizean Spanish Spanish Kriol Spanish, Belizean Kriol and English Your ability to SPEAK …UNDERSTAND … WRITE … READ PRONOUNCIATION GRAMMAR OVERALL ABILITY TOTAL

328 APPENDIX D ACCEPTABILITY JUDGMENT TASK

Instructions: Carefully read the following sentences and use the scale below to indicate the acceptability of word combinations for every sentence. Circle your answer.

Totally Unacceptable Unacceptable Acceptable Totally Acceptable I don’t know 1 2 3 4

1. Mrs. Blanco se molestó porque sus hijos no quieren limpiar______. i. la baño sucio 1 2 3 4 I don’t know ii. el baño sucia 1 2 3 4 I don’t know iii. el baño sucio 1 2 3 4 I don’t know iv. la baño sucia 1 2 3 4 I don’t know

2. Mi abuelo dice que en su childhood, él nunca ______.

i. hizo need una computadora portátil 1 2 3 4 I don’t know ii. need un lap top 1 2 3 4 I don’t know iii. need una computadora portátil 1 2 3 4 I don’t know iv. hizo need un lap top 1 2 3 4 I don’t know

3. Ahora que Marlon acabó sixth form, ______. i. él hace hope find un job 1 2 3 4 I don’t know ii. él hace hope hacer find un job 1 2 3 4 I don’t know iii. él espera hacer find un job 1 2 3 4 I don’t know iv. él espera find un job 1 2 3 4 I don’t know

4. Mi tío me contó que ______costó casi diez mil dólares.

i. la máquina nuevo 1 2 3 4 I don’t know ii. el máquina nueva 1 2 3 4 I don’t know iii. la máquina nueva 1 2 3 4 I don’t know iv. el máquina nuevo 1 2 3 4 I don’t know

5. Cuando se trata de hacer un presentation, Fernando siempre ______. i. hace surprise a sus compañeros de clase 1 2 3 4 I don’t know ii. surprise a sus classmates 1 2 3 4 I don’t know iii. surprise a sus compañeros de clase 1 2 3 4 I don’t know iv. hace surprise a sus classmates 1 2 3 4 I don’t know

6. Cuando hizo el emprestamo, Mr. Castillo decidió ______. i. comprar una new car 1 2 3 4 I don’t know

329 ii. comprar un car new 1 2 3 4 I don’t know iii. comprar una car new 1 2 3 4 I don’t know iv. comprar un new car 1 2 3 4 I don’t know

7. Danny siempre ______de su novia en su wallet. i. hace carry el small photograph 1 2 3 4 I don’t know ii. hace carry la fotografía pequeña 1 2 3 4 I don’t know iii. carry el small photograph 1 2 3 4 I don’t know iv. lleva la small photograph 1 2 3 4 I don’t know

8. Gianni le contó a su mamá que last night ______. i. tuvo un dream weird 1 2 3 4 I don’t know ii. tuvo un weird dream 1 2 3 4 I don’t know iii. tuvo una weird dream 1 2 3 4 I don’t know iv. tuvo una dream weird 1 2 3 4 I don’t know

9. Todos se rieron cuando el perrito ______. i. hide behind el door 1 2 3 4 I don’t know ii. se hizo hide behind el door 1 2 3 4 I don’t know iii. se hizo hide behind la door 1 2 3 4 I don’t know iv. se hizo hide detrás de la puerta 1 2 3 4 I don’t know

10. El muchacho le dijo a su jefe, “El espacio en que tenemos que trabajar ______así que será difícil.”

i. es oscuro 1 2 3 4 I don’t know ii. era oscura 1 2 3 4 I don’t know iii. es oscura 1 2 3 4 I don’t know iv. era oscuro 1 2 3 4 I don’t know

11. Yadira le dijo a su hermana, “Puedes bañarte. ______.” i. No me importa wait 1 2 3 4 I don’t know ii. No hago mind esperar 1 2 3 4 I don’t know iii. No me importa hacer wait 1 2 3 4 I don’t know iv. No hago mind hacer wait 1 2 3 4 I don’t know

12. Everyone está conciente de que el turismo, como hemos notado durante los últimos años, ______para la economía.

i. era bueno 1 2 3 4 I don’t know ii. es buena 1 2 3 4 I don’t know iii. era buena 1 2 3 4 I don’t know iv. es bueno 1 2 3 4 I don’t know

13. La gente no se da cuenta de que ______es malo para la salud.

330

i. el grasa procesada 1 2 3 4 I don’t know ii. la grasa procesado 1 2 3 4 I don’t know iii. el grasa procesado 1 2 3 4 I don’t know iv. la grasa procesada 1 2 3 4 I don’t know

14. Un día, Mrs. Carbajal le dijo a Mr. Torres, “Cesar no ______. “ i. appear un responsible student 1 2 3 4 I don’t know ii. hace appear un student responsible 1 2 3 4 I don’t know iii. hace appear una responsible student 1 2 3 4 I don’t know iv. hace appear un responsible student 1 2 3 4 I don’t know

15. Cuando se dió cuenta que era 9 o’ clock, la chiquita rápidamente ______. i. se hizo get up de la silla 1 2 3 4 I don’t know ii. se get up del chair 1 2 3 4 I don’t know iii. se hizo get up de la chair 1 2 3 4 I don’t know iv. se hizo get up del chair 1 2 3 4 I don’t know

16. En la tienda, Mrs. Blanco ______para su esposo. i. vio la gift 1 2 3 4 I don’t know ii. vio el gift perfect 1 2 3 4 I don’t know iii. vio la gift perfect 1 2 3 4 I don’t know iv. vio el perfect gift 1 2 3 4 I don’t know

17. Carola le dijo a su novio, “Tú solo a veces ______.” i. haces sound como una sincere person1 2 3 4 I don’t know ii. haces sound como un person sincere 1 2 3 4 I don’t know iii. haces sound como un sincere person 1 2 3 4 I don’t know iv. sound como un sincere person 1 2 3 4 I don’t know

18. Jason está excited porque la cueva que van a explorar la próxima semana ______.

i. era misteriosa 1 2 3 4 I don’t know ii. es misterioso 1 2 3 4 I don’t know iii. es misteriosa 1 2 3 4 I don’t know iv. era misterioso 1 2 3 4 I don’t know

19. Pedro no estaba worried porque ______.

i. entregó la homework 1 2 3 4 I don’t know ii. hizo submit la tarea 1 2 3 4 I don’t know iii. hizo submit el homework 1 2 3 4 I don’t know iv. submit el homework 1 2 3 4 I don’t know

331

20. Carla y Zair no creen que tienen ______para comprar una casa.

i. la sueldo adecuado 1 2 3 4 I don’t know ii. el sueldo adecuada 1 2 3 4 I don’t know iii. la sueldo adecuada 1 2 3 4 I don’t know iv. el sueldo adecuado 1 2 3 4 I don’t know

21. La muchacha estaba triste porque su novio le ______. i. hizo promise cosas imposibles 1 2 3 4 I don’t know ii. promise impossible things 1 2 3 4 I don’t know iii. promise cosas imposibles 1 2 3 4 I don’t know iv. hizo promise impossible things 1 2 3 4 I don’t know

22. Los estudiantes piensan que Wuthering Heights ______. i. es un complicated book 1 2 3 4 I don’t know ii. es una book complicated 1 2 3 4 I don’t know iii. es un book complicated 1 2 3 4 I don’t know iv. es una complicated book 1 2 3 4 I don’t know

23. La gente se reía porque el cuento______. i. hacía sound como una big lie 1 2 3 4 I don’t know ii. hacía sound como un big lie 1 2 3 4 I don’t know iii. hacía sound como un lie big 1 2 3 4 I don’t know iv. sound como un big lie 1 2 3 4 I don’t know

24. Para los refugees, la guerra civil en El Salvador______.

i. fue una traumatizing experience 1 2 3 4 I don’t know ii. fue un experience traumatizing 1 2 3 4 I don’t know iii. fue un traumatizing experience 1 2 3 4 I don’t know iv. fue una experience traumatizing 1 2 3 4 I don’t know

25. Last week, Mrs. Tun ______. i. rifó una gold chain 1 2 3 4 I don’t know ii. raffle un gold chain 1 2 3 4 I don’t know iii. hizo raffle una cadena de oro 1 2 3 4 I don’t know iv. hizo raffle un gold chain 1 2 3 4 I don’t know

26. En town, todos dicen que Vanessa ______. i. tiene una smile beautiful 1 2 3 4 I don’t know ii. tiene un beautiful smile 1 2 3 4 I don’t know iii. tiene un smile beautiful 1 2 3 4 I don’t know

332 iv. tiene una beautiful smile 1 2 3 4 I don’t know

27. Kira estaba tan feliz porque ______. i. hizo watch un romantic movie 1 2 3 4 I don’t know ii. hizo watch una película romántica 1 2 3 4 I don’t know iii. vio una romantic movie 1 2 3 4 I don’t know iv. watch romantic movie 1 2 3 4 I don’t know

28. Mrs. Pollard, after eating a slice of chocolate cake, se sentó, y dijo, “El pastel que nos dieron hoy para dessert si ______.

i. era sabrosa 1 2 3 4 I don’t know ii. estuvo sabroso 1 2 3 4 I don’t know iii. era sabroso 1 2 3 4 I don’t know iv. estuvo sabrosa 1 2 3 4 I don’t know

29. Last week, cuando María fue a Chetumal, ella ______.

i. compró un necklace expensive 1 2 3 4 I don’t know ii. compró una expensive necklace 1 2 3 4 I don’t know iii. compró una necklace expensive 1 2 3 4 I don’t know iv. compró un expensive necklace 1 2 3 4 I don’t know

30. La muchacha le dijo a Carlos, “El servicio que esta compañía le ofrece a sus clientes no ______pero es de calidad.”

i. es barata 1 2 3 4 I don’t know ii. era barato 1 2 3 4 I don’t know iii. es barato 1 2 3 4 I don’t know iv. era barata 1 2 3 4 I don’t know

31. Como ya no tenía dinero, Emir ______que compró en Cancún. i. hizo pawn el expensive bracelet 1 2 3 4 I don’t know ii. hizo pawn la pulsera cara 1 2 3 4 I don’t know iii. empeño la expensive bracelet 1 2 3 4 I don’t know iv. pawn el expensive bracelet 1 2 3 4 I don’t know

32. Hacía tanta calor que Mr. William ______.

i. se hizo sit down under la shade 1 2 3 4 I don’t know ii. sit down bajo el shade 1 2 3 4 I don’t know iii. se hizo sit down bajo la sombra 1 2 3 4 I don’t know iv. se hizo sit down under el shade 1 2 3 4 I don’t know

33. Mi abuela dice que ahora la ciudad ______pero antes era tan diferente.

333

i. era moderno 1 2 3 4 I don’t know ii. es moderna 1 2 3 4 I don’t know iii. era moderna 1 2 3 4 I don’t know iv. es moderno 1 2 3 4 I don’t know

34. Jaime dice que él no______.

i. espera win el essay competition 1 2 3 4 I don’t know ii. hace expect win el essay competition 1 2 3 4 I don’t know iii. espera hacer win el essay competition 1 2 3 4 I don’t know iv. hace expect hacer win el essay competition 1 2 3 4 I don’t know

35. Mrs. Ramirez se dió cuenta que la tarea que les había dado a los second formers ______.

i. era larga 1 2 3 4 I don’t know ii. es larga 1 2 3 4 I don’t know iii. era largo 1 2 3 4 I don’t know iv. es largo 1 2 3 4 I don’t know

36. Mrs. García piensa que su esposo está loco porque ayer ______.

i. se hizo shave en la sala 1 2 3 4 I don’t know ii. se hizo shave en la living room 1 2 3 4 I don’t know iii. se hizo shave en el living room 1 2 3 4 I don’t know iv. se shave en el living room 1 2 3 4 I don’t know

37. María le dijo a Marcos, “La pulsera de oro que compré en Cancun ______.

i. es cara 1 2 3 4 I don’t know ii. era caro 1 2 3 4 I don’t know iii. es caro 1 2 3 4 I don’t know iv. era cara 1 2 3 4 I don’t know

38. Desde el primer día en high school, Juan ______. i. hizo impress a los maestros 1 2 3 4 I don’t know ii. impress a los teachers 1 2 3 4 I don’t know iii. impress a los maestros 1 2 3 4 I don’t know iv. hizo impress a los teachers 1 2 3 4 I don’t know

39. Después de la entrevista, Samantha le dijo a su amiga, “La oferta que me han dado ______pero no sé si quiero trabajar allí.

334 i. es atractivo 1 2 3 4 I don’t know ii. es atractiva 1 2 3 4 I don’t know iii. era atractiva 1 2 3 4 I don’t know iv. era atractivo 1 2 3 4 I don’t know

40. Cuando acabó de limpiar la sala, Gina ______. i. hizo cook la comida 1 2 3 4 I don’t know ii. cook el food 1 2 3 4 I don’t know iii. cocinó la food 1 2 3 4 I don’t know iv. hizo cook el food 1 2 3 4 I don’t know

41. Alex le dijo a su tía, “Prefiero que la camisa ______porque me gusta ese color.

i. sea blanco 1 2 3 4 I don’t know ii. es blanco 1 2 3 4 I don’t know iii. sea blanca 1 2 3 4 I don’t know iv. es blanca 1 2 3 4 I don’t know

42. Dina le dijo a su novio, “Cuando me enamoré de ti, sentí que ______para todos mis problemas. i. hice find la solución perfecta 1 2 3 4 I don’t know ii. encontré la perfect solution 1 2 3 4 I don’t know iii. hice find el perfect solution 1 2 3 4 I don’t know iv. find el perfect solution 1 2 3 4 I don’t know

43. Después del examen, María ______.

i. hizo recognize sus errores negligentes1 2 3 4 I don’t know ii. recognize sus careless errors 1 2 3 4 I don’t know iii. recognize sus errores negligentes 1 2 3 4 I don’t know iv. hizo recognize sus careless errors 1 2 3 4 I don’t know

44. Un hombre independiente nunca ______.

i. hace depend en sus papás 1 2 3 4 I don’t know ii. depend en sus parents 1 2 3 4 I don’t know iii. depend en sus papás 1 2 3 4 I don’t know iv. hace depend en sus parents 1 2 3 4 I don’t know

45. Pablo le dijo a Evlyn que él piensa que ella se enfermó porque ______. i. hizo eat la carne de puerco 1 2 3 4 I don’t know ii. comió la pork meat 1 2 3 4 I don’t know iii. hizo eat el pork meat 1 2 3 4 I don’t know iv. eat el pork meat 1 2 3 4 I don’t know

335 46. Carlos me contó que sus amigos le dijeron que la fiesta ______.

i. fue aburrido 1 2 3 4 I don’t know ii. estaba aburrida 1 2 3 4 I don’t know iii. fue aburrida 1 2 3 4 I don’t know iv. estaba aburrido 1 2 3 4 I don’t know

47. Carla le dijo a su mamá, “No______.

i. hace seem que a Odalis le gusten apples yellow.”1 2 3 4 I don’t know ii. hace seem que a Odalis le gusten yellow apples.”1 2 3 4 I don’t know iii. hace seem que a Odalis le gusten yellow apples.”1 2 3 4 I don’t know iv. seem que a Odalis le gusten yellow apples.” 1 2 3 4 I don’t know

48. No entiendo por qué quieren que la carne ______en aceite de maíz.

i. esté frita 1 2 3 4 I don’t know ii. estaba frita 1 2 3 4 I don’t know iii. estaba frito 1 2 3 4 I don’t know iv. esté frito 1 2 3 4 I don’t know

49. Todo mundo ya sabe que Claudia no ______.

i. hace deserve el premio 1 2 3 4 I don’t know ii. deserve el premio 1 2 3 4 I don’t know iii. hace deserve el prize 1 2 3 4 I don’t know iv. deserve the prize 1 2 3 4 I don’t know

50. Lo único que Pablo recuerda del viaje a Cancún es que la cama en donde él durmió esa noche ______.

i. fue cómodo 1 2 3 4 I don’t know ii. era cómoda 1 2 3 4 I don’t know iii. fue cómoda 1 2 3 4 I don’t know iv. era cómodo 1 2 3 4 I don’t know

51. Para breakfast, ______.

i. yo prefiero hacer drink orange juice 1 2 3 4 I don’t know ii. yo hago prefer drink orange juice 1 2 3 4 I don’t know iii. yo hago prefer hacer drink orange juice 1 2 3 4 I don’t know iv. yo prefiero drink orange juice 1 2 3 4 I don’t know

53. Mientras caminaba en St. Vernon St., Pedro ______.

336 i. encontró una coin old 1 2 3 4 I don’t know ii. encontró un old coin 1 2 3 4 I don’t know iii. encontró un coin old 1 2 3 4 I don’t know iv. encontró una old coin 1 2 3 4 I don’t know

54. Cuando ______, la muchacha empezó a llorar.

i. find out el sad truth 1 2 3 4 I don’t know ii. find out la sad truth 1 2 3 4 I don’t know iii. hizo find out el sad truth 1 2 3 4 I don’t know iv. hizo find out la triste verdad 1 2 3 4 I don’t know

55. At the beginning, Rina ______.

v. hizo appear como una honest friend 1 2 3 4 I don’t know vi. hizo appear como un friend honest 1 2 3 4 I don’t know vii. appear como un honest friend 1 2 3 4 I don’t know viii. hizo appear como un honest friend 1 2 3 4 I don’t know

Thank you so much!

337 APPENDIX E LANGUAGE BACKGROUND QUESTIONNAIRE

Initials/Gender: ______

Personal Information Age/Place of Birth How long have you lived in Orange Walk or elsewhere? Specify Have you lived in another district/country? Where? How long? What is the highest level of education you have completed?

Language Background

Underline the languages that you speak: Belizean Spanish Belizean Kriol English Yucatec Maya Other:______

1. What is/are your first language(s)? 2. What languages were spoken at home when you were a child (0 - 8 years)? 3. What language(s) do your parents speak? 4. Which language would you consider your dominant language (the language you use the most)?

5. On a scale of 1 to 7 (where 1 – 3 = rarely, 4 – 5 = sometimes, 6 – 7 = very often), how frequently do you speak Spanish?

6. On a scale of 1 to 7 (where 1 – 3 = rarely, 4 – 5 = sometimes, 6 – 7 = very often), how frequently do you speak English?

7. On a scale of 1 to 7 (where 1 – 3 = rarely, 4 – 5 = sometimes, 6 – 7 = very often), how frequently do you speak Belizean Kriol?

338 8. On a scale of 1 to 7 (where 1 – 3 = rarely, 4 – 5 = sometimes, 6 – 7 = very often), how frequently do you mix your languages?

Use the scale below to estimate your abilities in Standard Spanish, Belizean Spanish, Belizean Kriol, English, and a mixture of languages. (1-3 is poor, 4-5 is average, 6-7 is excellent).

in in in in A mixture of Standard Belizean Belizean English Belizean Spanish Spanish Kriol Spanish, Belizean Kriol and English Your ability to SPEAK Your ability to UNDERSTAND PRONOUNCIATION OVERALL ABILITY

339 APPENDIX F ACCEPTABILITY JUDGMENT TASK

Instructions: Carefully read the following sentences and use the scale below to indicate the acceptability of word combinations for every sentence. Rate each sentence according to how acceptable or familiar it sounds to you. Underline your answer.

Totally Unacceptable Unnacceptable Acceptable Totally Acceptable 1 2 3 4

1. Gabriel cayó de las escaleras hoy because he lost his balance. 1 2 3 4

2. Ismael se molestó porque la carta que mandó 1 2 3 4 la semana pasada no estaba hecho delivered.

3. Verónica aplaudió a su hermana when she received her diploma. 1 2 3 4

4. Carmen no se quiere bañar late hoy because it’s really cold. 1 2 3 4

5. Juan no puede cepillar sus dientes hoy because he has a toothache. 1 2 3 4

6. Durante el fin de semana, Hector hizo find a wallet at the bus stop. 1 2 3 4

7. Ivan se molestó porque la mesa que compró 1 2 3 4 la semana pasada no estaba hecha varnished.

8. Perla se lavó la cara because she started to feel sleepy. 1 2 3 4

9. Francisco frayó la carne in olive oil because it tastes better. 1 2 3 4

10. Maria drayó los platos with a dish towel. 1 2 3 4

11. Carlos se molestó porque la casa que compró 1 2 3 4 la semana pasada no estaba hecho painted.

12. Karina rayó el papel because she failed the exam. 1 2 3 4

13. Kimberly spelió la palabra mal in the Spelling Bee contest. 1 2 3 4

14. Durante el fin de semana, Carlos hizo eat pepperoni pizza at the mall.1 2 3 4

15. Julio se molestó porque la cuenta que pagó 1 2 3 4 la semana pasada no estaba hecho cleared.

16. Ella no se quiere quitar el abrigo because it’s too cold outside. 1 2 3 4

340 17. Perla lavó su cara because she started to feel sleepy. 1 2 3 4

18. Hector se molestó porque la máquina no fue hecha fixed bien. 1 2 3 4

19. Myrna no se quiso poner la blusa because she thinks it is ugly. 1 2 3 4

20. Claudia copió en el examen and that is why she failed the semester. 1 2 3 4

21. Patricia se molestó porque la cocina no estaba hecho cleaned bien. 1 2 3 4

22. Jenny se golpeó la cabeza when she fell off her bicycle. 1 2 3 4

23. Enrique se molestó porque la camioneta que compró 1 2 3 4 la semana pasada no estaba hecha insured.

24. Claudia chitió en el examen and that is why she failed the semester. 1 2 3 4

25. Francisco frió la carne in olive oil because it tastes better. 1 2 3 4

26. Cristina se molestó porque la casa no fue hecho decorated bien. 1 2 3 4

27. Antonio le puso betún a sus zapatos before he put on his black pants.1 2 3 4

28. William se molestó porque la colcha que sacó 1 2 3 4 la semana pasada no estaba hecho washed.

29. Jenny naqueó su cabeza when she fell off her bicycle. 1 2 3 4

30. Mayra se molestó porque la cadena que perdió 1 2 3 4 la semana pasada no fue hecha recovered.

31. Maria secó los platos with a dish towel. 1 2 3 4

32. Claudia rushió hacia la estación so that she catches her bus. 1 2 3 4

33. Nancy amontonó los libros on top of her computer desk. 1 2 3 4

34. Durante el fin de semana, Ana hizo read chapter one of her textbook.1 2 3 4

35. Carlos se molestó porque la casa que compró 1 2 3 4 la semana pasada no estaba hecha painted.

36. Gabriel se molestó porque la boda no fue hecha organized bien. 1 2 3 4

37. Ana quiere cortar su pelo esta semana because it’s too hot. 1 2 3 4

341 38. Yanira se molestó porque la falda que usó 1 2 3 4 la semana pasada no fue hecha washed.

39. Carmen se molestó porque la cena no fue hecho served temprano. 1 2 3 4

40. María se molestó porque la ventana no estaba hecha closed bien. 1 2 3 4

41. Enrique se molestó porque la camioneta que compró 1 2 3 4 la semana pasada no estaba hecho insured.

42. Antonio shineó sus zapatos before he put on his black pants. 1 2 3 4

43. Gabriel se molestó porque la boda no fue hecho organized bien. 1 2 3 4

44. Carmen se molestó porque la cena no fue hecha served temprano. 1 2 3 4

45. Durante el fin de semana, Jennifer hizo watch Titanic for the tenth time.1 2 3 4

46. Carmen se molestó porque la música no fue hecha played. 1 2 3 4

47. Giselle se molestó porque la cortina que compró 1 2 3 4 la semana pasada no fue hecho washed.

48. Karina scratchió el papel because she failed the exam. 1 2 3 4

49. Patricia se molestó porque la mesa que devolvió 1 2 3 4 la semana pasada no fue hecho replaced.

50. Ana se quiere cortar el pelo esta semana because it’s too hot. 1 2 3 4

51. Karen quiere levantar temprano mañana because she is commuting. 1 2 3 4

52. Giselle se molestó porque la cortina que compró 1 2 3 4 la semana pasada no fue hecha washed.

53. Joe se lastimó la mano ayer because he fell down from a mango tree.1 2 3 4

54. Myrna no quiso poner la blusa because she thinks it is ugly. 1 2 3 4

55. Cristina se molestó porque la casa no fue hecha decorated bien. 1 2 3 4

56. Marlon no pudo quedar hoy because he has a lot of things to do. 1 2 3 4

57. Carlos se molestó porque la plancha no fue hecho unplugged ayer. 1 2 3 4

58. Juan se molestó porque la caja que mandó 1 2 3 4

342 la semana pasada no fue hecha delivered.

59. Ernesto se quiere dormir temprano hoy because tomorrow he has a big exam. 1 2 3 4

60. Pedro se molestó porque la camisa no estaba hecho ironed bien. 1 2 3 4

61. Juan no se puede cepillar los dientes hoy because he has a toothache.1 2 3 4

62. Carlos quiere quitar el tatuaje because his new girlfriend hates tattoos. 1 2 3 4

63. Julio se molestó porque la cuenta que pagó 1 2 3 4 la semana pasada no estaba hecha cleared.

64. Karen se quiere levantar temprano mañana because she is commuting. 1 2 3 4

65. Joe lastimó su mano ayer because he fell down from a mango tree. 1 2 3 4

66. Durante el fin de semana, Gabriela hizo decorate her room with posters.1 2 3 4

67. Mayra se molestó porque la cadena que perdió 1 2 3 4 la semana pasada no fue hecho recovered.

68. Raquel piqueó los zapatos negros instead of the red ones. 1 2 3 4

69. Verónica clapeó para su hermana when she received her diploma. 1 2 3 4

70. Ismael se molestó porque la carta que mandó 1 2 3 4 la semana pasada no estaba hecha delivered.

71. Ernesto quiere dormir temprano hoy because tomorrow he has a big exam. 1 2 3 4

72. Carlos se molestó porque la plancha no fue hecha unplugged ayer. 1 2 3 4

73. Marlon no se pudo quedar hoy because he has a lot of things to do. 1 2 3 4

74. Raúl se molestó porque la carta que escribió 1 2 3 4 la semana pasada no fue hecho mailed.

75. Carlos se quiere quitar el tatuaje because his new girlfriend hates tattoos. 1 2 3 4

76. Aracely crayó en su cuarto because she felt so heartbroken. 1 2 3 4

77. William se molestó porque la colcha que sacó 1 2 3 4 la semana pasada no estaba hecha washed.

343 78. Kimberly deletreó la palabra mal in the Spelling Bee contest. 1 2 3 4

79. José se slipió en la calle because the road was wet and slippery. 1 2 3 4

80. Yanira se molestó porque la falda que usó 1 2 3 4 la semana pasada no fue hecho washed.

81. Patricia se molestó porque la cocina no estaba hecha cleaned bien. 1 2 3 4

82. Carmen se molestó porque la música no fue hecho played. 1 2 3 4

83. Raúl se molestó porque la carta que escribió 1 2 3 4 la semana pasada no fue hecha mailed.

84. Claudia corrió hacia la estación so that she catches her bus. 1 2 3 4

85. Pedro se molestó porque la camisa no estaba hecha ironed bien. 1 2 3 4

86. Lorena se quedó en estado de pánico en frente del policía 1 2 3 4 and she didn’t know what to say.

87. María se molestó porque la ventana no estaba hecho closed bien. 1 2 3 4

88. Jessica se molestó porque la comida no estaba hecha cooked bien. 1 2 3 4

89. Rafael engañó a los policías and they let him go. 1 2 3 4

90. Carmen no quiere bañar late hoy because it’s really cold. 1 2 3 4

91. Juan se molestó porque la caja que mandó 1 2 3 4

la semana pasada no fue hecho delivered.

92. Nancy staqueó los libros on top of her computer desk. 1 2 3 4

93. Antonio se molestó porque la portada no estaba hecho locked bien. 1 2 3 4

94. Hector se molestó porque la máquina no fue hecho fixed bien. 1 2 3 4

95. Aracely lloró en su cuarto because she felt so heartbroken. 1 2 3 4

96. Patricia se molestó porque la mesa que devolvió 1 2 3 4 la semana pasada no fue hecha replaced.

97. Gabriel se cayó de las escaleras hoy because he lost his balance. 1 2 3 4

344 98. Antonio se molestó porque la portada no estaba hecha locked bien. 1 2 3 4

99. José se resbaló en la calle because the road was wet and slippery. 1 2 3 4

100. Ella no quiere quitar el abrigo because it’s too cold outside. 1 2 3 4

101. Paulina se molestó porque la carne no estaba hecha seasoned bien. 1 2 3 4

102. Durante el fin de semana, Karen hizo iron her uniform for school. 1 2 3 4

103. Jessica se molestó porque la comida no estaba hecho cooked bien. 1 2 3 4

104. Raquel escogió los zapatos negros instead of the red ones. 1 2 3 4

105. Ivan se molestó porque la mesa que compró 1 2 3 4 la semana pasada no estaba hecho varnished bien.

106. Rafael triqueó a los policias and they let him go. 1 2 3 4

107. Lorena paniqueó en frente del policía and she didn’t know what to say.1 2 3 4

108. Paulina se molestó porque la carne no estaba hecho seasoned bien. 1 2 3 4

Thank you for participating!

345 APPENDIX G LANGUAGE BACKGROUND QUESTIONNAIRE

First Name: ______

All information provided in this questionnaire is confidential.

PART I: LANGUAGE BACKGROUND

A. Personal Information

Age/Gender/Date of Birth E-mail address Where are you from? Specify the district and town or village Are both of your parents Mestizo? Where are they from? What is the highest level of education you have completed? Have you lived in another district/country? Where? How long?

B. Language Background

At what age did you start learning Spanish? Where? At what age did you start learning English? Where? At what age did you start learning Kriol? Where?

C. Use the scale below to answer the following questions regarding language use.

1. Only Belizean Spanish 2. Only standard Spanish 3. Only English 4. Only Belizean Kriol 5. A mixture of the three languages 6. Only Belizean Spanish and English 7. Only Belizean Spanish and Belizean Kriol 8. Only English and Belizean Kriol

346 Number % per day (in of case of 2 or 3 Response languages) What language(s) do you use the most at home? What language(s) do you use the most with your siblings? What language(s) do you use the most with your cousins? What languages(s) do you use the most with your closest friends? What language(s) do you use the most with your classmates? What language(s) do you use the most with your teachers/students? What language(s) do you use the most with your parents/sons or daughters?

D. For the following items, circle the number (NOT the range) that best reflects your language use. Use the scale below. SCALE: 1-3 = rarely, 4-5 = sometimes, 6-7 = very often

1. How frequently do you speak Belizean Spanish? 1 2 3 4 5 6 7 2. How frequently do you speak standard Spanish? 1 2 3 4 5 6 7 3. How frequently do you speak English? 1 2 3 4 5 6 7 4. How frequently do you speak Belizean Kriol? 1 2 3 4 5 6 7 5. How frequently do you mix Spanish and English? 1 2 3 4 5 6 7 6. How frequently do you mix Spanish and Belizean Kriol? 1 2 3 4 5 6 7 7. How frequently do you mix Spanish, English, and Belizean Kriol? 1 2 3 4 5 6 7

347

E. For the following items, write down the number (NOT the range) that best reflects your ability/proficiency. Use the scale below.

SCALE: 1-3 = poor, 4-5 = average, 6-7 = excellent

in in in in A mixture of Belizean standard English Belizean Belizean Spanish Spanish Kriol Spanish, English and/or Belizean Kriol OVERALL ABILITY SPEAKING WRITING PRONOUNCIATION

F. How do you identify yourself? Latino, Hispanic, Mestizo, Creole, Mayan-Mestizo, Central American, Orangewalkeño, Belizean, or Other (specify). Write down the three labels that best identify who you are (1= most important, 2 = important, 3 = somewhat important). You can put two or more labels for any given number to show that they are equally important to you. 1. ______2. ______3. ______

G. Do you have relatives who live in the surrounding villages of Orange Walk Town?

Yes No

How often do you spend time with relatives who live in rural areas? Briefly explain. ______

348 REFERENCES

Aaron, J. E. (2014). Lone English-origin Nouns in Spanish: The Precedence of Community Norms. International Journal of Bilingualism. DOI: 10.1177/1367006913516021.

Adams, C. (2002). Strong assibilation and prestige: A sociolinguistic study in the Central Valley of Costa Rica. PhD dissertation, University of California

Agnihotri, R. K. (1987). Crisis of identity: A sociolinguistic study of Sikh children in Leeds. Delhi, India: Bahri.

Allsopp, S. R. R. (1965). British Honduras - The Linguistic Dilemma. Caribbean Quarterly, 2, 54–61.

Alvarez-Caccamo, C. (1998). From ‘switching code’ to ‘code-switching’: towards a reconceptualization of communicative codes. In P. Auer (Ed.), Codeswitching in Conversation: Language, Interaction and Identity (pp. 29-50). New York, NY: Routledge.

Anderson, T., & A. J. Toribio. 2007. Attitudes Towards Lexical Borrowing And Intra- sentential Code-switching among Spanish-English Bilinguals. Spanish in Context, 4, 217-240.

Anderson, T. (2010). Attitudes toward Spanish language variation among Latinas living in Western Colorado. In S. Rivera-Mills and D. Villa (Eds.), Spanish of the US Southwest (pp. 291-305). , Spain: Iberoamericana.

Annamalai, E. (1989). The language factor in codemixing. Internationl Journal of the Sociology of Language, 75, 47–54.

Arguello, F. M. (1978). El dialecto žeista del español en el : un estudio fonético y fonológico. Ph.D dissertation, Pennsylvania State University.

Arguello, F. M. (1980). El rehilamiento en el español hablado en la región andina del Ecuador. Lexis: Revista de Lingüística y Literatura, 4, 151–156.

Armstrong, G. (2009). On copular sentences in Yucatec Maya. In Proceedings of the Conference on Indigenous Languages of Latin America – IV. Online publication hosted by the University of Texas at Austin.

Backus, A. (1996). Two in One. Bilingual Speech of Turkish Immigrants in the Netherlands. Tilburg: Tilbury University Press.

Backus, A. (2004). Convergence as a mechanism of change. Bilingualism: Language and Cognition, 7, 179-181.

349 Backus, A. (2005). Codeswitching and language change: one things leads to another. International Journal of Bilingualism, 9, 307-340.

Backus, A., & Boeschoten, H. (1996). Turkish-Dutch code-switching and levels of lexical structure. Paper presented at Linguistic Society of America Annual Conference, San Diego.

Bailey, B. (2002). Language, race and negotiation of identity: A study of , New York, NY: LFB Scholarly Publishing LLC.

Baker, P. (2000). Theories of creolization and the degree and nature of restructuring. In Degrees of Restructuring in Creole Languages, I. Neumann-Holzschuh & E. W. Schneider (Eds.), 41–63. Amsterdam, Netherlands: John Benjamins.

Balam, O. (2013). Overt language attitudes and linguistic identities among multilingual speakers in Northern Belize. Studies in Hispanic and Lusophone Linguistics, 6, 247-277.

Balam, O., & Prada Pérez, A. (2013). Light verb constructions in stative and eventive passives in contact Spanish. Paper presented at the Hispanic Linguistics Symposium, Ottawa, Canada, October 2013.

Balam, O. (2014). Notes on the History and Morphosyntactic Characteristics of Spanish in Northern Belize. Kansas Working Papers in Linguistics, 35, 79-94.

Balam, O., Prada Pérez, A., & Mayans, R. (2014). A Congruence Approach to the Study of Bilingual Compound Verbs in Northern Belize Contact Spanish. Spanish in Context, 11, 243-265.

Balam, O. (2015). Code-switching and linguistic evolution: The case of ‘Hacer + V’ in Orange Walk, Northern Belize. Lengua y Migración, 7, 83-109.

Balam, O. (2016a). Mixed Verbs in Contact Spanish: Patterns of Use among Emergent and Dynamic Bi/multilinguals. Languages, 1, 3.

Balam, O. (2016b). Semantic Categories and Gender Assignment in Contact Spanish: Type of Code-switching and its Relevance to Lingusitic Outcomes. Journal of Language Contact, 9. doi: 10.1163/19552629-00903001

Balam, O., & López Alonso, K. (2015). A sociocultural linguistic analysis of identity among Mestizos in Orange Walk, Belize and Bluefields, . Paper presented at the Ohio State University Congress on Hispanic and Lusophone Linguistics (OSUCHiLL), Columbus, Ohio, April 2015.

Balam, O. & Prada Pérez, A. (In press). Attitudes towards Spanish and Code-Switching in Belize: Stigmatization and Innovation in the Spanish Classroom. Journal of Language, Identity and Education.

350

Balam, O., & Prada Pérez, A. (2016). On the Productive Use of ‘Hacer + V’ in Northern Belize Bilingual/Trilingual Code-switching. In R. E. Guzzardo Tamargo, C. Mazak, and M. C. Parafita Couto (Eds.), Spanish-English Codeswitching in the Caribbean and the U.S. (pp. 261-279). Amsterdam: John Benjamins.

Baptista, M. & Guéron, J. (2007). Noun Phrases in Creole Languages: A Multi-faceted Approach. Amsterdam, Netherlands: John Benjamins.

Barry, T., & Vernon, D. (1995). Inside Belize. Albuquerque, NM: Resource Center Press.

Batty, S., García, A., & Cucul, V. (2011). The and its contribution to social cohesion of Belizeans in San Ignacio and Santa Elena since Independence. Belmopan, Cayo: Institute of Social and Cultural Research.

Belize Population and Housing Census, 2010. Retrieved from http://www.sib.org.bz/Portals/0/docs/publications/census/2010_Census_Report.pdf (25 November, 2014).

Belazi, H. M., Rubin, E. J., & Toribio, A. J. (1994). Code switching and X-Bar Theory: The functional head constraint. Linguistic Inquiry, 25, 221–237.

Benson, E. (2001). The Neglected Early History of Code Switching Research in the United States. Language & Communication, 21, 23-36.

Bentahila, A. (1995). Review of Duelling languages: Grammatical structure in codeswitching. Language, 71, 135– 140.

Berk-Seligson, S. (1986). Linguistic constraints on intrasentential code-switching: A study of Spanish/Hebrew bilingualism. Language in Society, 15, 313-348.

Bernsten, J. (2000). Creative Construction: Shona/English codeswitches. Paper presented at First International Conference on Linguistics in Southern , Cape Town, 1/2000.

Bhatia, T. K., & Ritchie, W. C. (1996). Bilingual Language Mixing, Universal Grammar, and Second Language Acquisition. In W. C. Ritchie & T. K. Bhatia (Eds.), Handbook of Second Language Acquisition, (pp. 627–82). San Diego, CA: Academic Press.

Bhatia T. K., & Ritchie, W. C. (2008). The Bilingual Mind and Linguistic Ceativity. Journal of Creative Communications, 3, 5-21.

Bhatia, T K., & Ritchie, W. C. (2016). Multilingual language mixing, optimization and creativity. Languages, 1, 1-14.

351 Bills, G., E., Hernandez Chavez, E., & Hudson, A. (1995). The geography of language shift: distance from the Mexican border and Spanish language claiming in the southwestern United States. International Journal of the Sociology of Language, 114, 9-27.

Bills, G., Hernandez Chavez, E., & Hudson, A. (2000). Spanish home language home use and English proficiency as differential measures of language maintenance and shift. Southwest Journal of Linguistics, 19, 11-27.

Blecua Falgueras, B. (1999). Características acústicas de la vibrante múltiple del español en habla espontánea. Paper presented at Actas del I Congreso de Fonética Experimental, Tarragona, Spain.

Blecua Falgueras, B. (2001). Las Vibrantes del Español: Manifestaciones Acústicas y Procesos Fonéticos. Unpublished PhD dissertation, Universidad Autónoma de Barcelona.

Blom, J. P., & Gumperz, J.J. (1972). Social meaning in linguistic structure: Code-switching in Norway. In John J. Gumperz & Dell Hymes (Eds.), Directions in sociolinguistics: The ethnography of communication (pp.407-434). New York, NY: Holt, Rinehart & Winston.

Boersma, P. & D. Weenink. (2008). Praat: Doing phonetics by computer. (Version 5.1.08) [Computer Program]. Retrieved from http://www.fon.hum.uva.nl/praat/

Bohnemeyer, J. (2009). Linking without grammatical relations in Yucatec: alignment, extraction, and control. In J. Helmbrecht, Y. Nishina, Y-M. Shin, S. Skopeteas, & E. Verhoeven (Eds.), Form and Function in Language Research (pp.185-214). Berlin, Germany: Mouton de Gruyter.

Bolland, O. N. (1977). The Formation of a Colonial Society. Baltimore: Johns Hopkins University Press.

Bolland, O. N. (2003). Colonialism and resistance in Belize: Essays in historical sociology. Benque Viejo, Belize: Cubola Productions.

Boumans, L. (1998). The Syntax of Codeswitchin: Analyzing Moroccan Arabic/Dutch Conversation. Tilburg University Press.

Bradley, T. G. (In Press). Consonantes róticas: Descripción fonética. In Fonética y fonología descriptivas de la lengua española, J. Gil and J. Llisterri (Eds.). Madrid, Spain: Consejo Superior de Investigaciones Científicas.

Bradley, T. G. (1999). Assibilation in Ecuadorian Spanish. In J. Authier, B. Bullock & L. Reed (Eds.), Formal Perspectives on Romance Linguistics. Selected Papers from the 28th Linguistics Symposium on Romance Languages (pp. 57-71). Amsterdam, Netherlands: John Benjamins Publishing Company.

352 Bradley, T. G. (2006). Phonetic realizations of /sr/ clusters in Latin American Spanish. In Manuel Diaz-Campos (ed.), Selected proceedings of the 2nd Conference on Laboratory Approaches to Spanish Phonetics and Phonology (pp. 1-13). Somerville, MA: Cascadilla Press.

Bradley, G. T. & Willis, E. (2012). Rhotic Variation and Contrast in Veracruz Mexican Spanish. Estudios de Fonética 51 Experimental, 21, 43-74.

Brockmann, T. (1979). Language, communication and ethnicity in British Honduras. In W. F. Mackey & J. Ornstein (Eds.), Sociolinguistic Studies in Language Contact: Methods and Cases. The Hague: Mouton Publishers.

Brown, R. W. (1958). Words and Things. Glencoe, IL: Free Press.

Bruhn de Garavito, J., & Valenzuela, E. (2008). Eventive and stative passives in Spanish L2 acquisition: a matter of aspect. Bilingualism: Language and Cognition, 11, 323-336.

Bucholtz, M., & Hall, K. (2005). Identity and Interaction: A Sociocultural Linguistic Approach. Discourse Studies, 7, 584-614.

Bullock, B., & Toribio, A. J. (2004). Introduction: Convergence as an emergent property in bilingual speech. Bilingualism: Language and Cognition, 7, 91-93.

Bullock, B. E., & Toribio, A. J. (2009). Cambridge handbook of linguistic code-switching. Cambridge: Cambridge University Press.

Bullock, B., Hinrichs, L., & Toribio, J. (2014). World Englishes, Code-switching and Convergence. In The Oxford Handbook of World Englishes, M. Filppula, J. Klemola & D. Sharma (Eds.). Online. Available: DOI: 10.1093/oxfordhb/9780199777716.013.009.

Bullock, B. E., & Gerfen, C. (2004). Phonological convergence in a contracting variety of language variety. Bilingualism: Language and Cognition, 7, 95-104.

Calvo Shadid, A. & Portilla Chaves, M. (1998). Variantes retroflejas de /r/ and /r/ en el habla culta de San José. Revista Káñina, 22, 81-86.

Camacho, J., & Sánchez, L. (2002). Explaining Clitic Variation in Spanish. In M. Amberber and P.Collins (Eds.), Formal Approaches to Language Universals and Language Variation (pp. 21-40). New Jersey: Ablex.

Camille, M. (1996). Population and Ethnicity of Belize, 1861. In M. D. Phillips (Ed.), Belize: Selected Proceedings from the Second Interdisciplinary Conference. Lanham, MD: University Press of America.

353 Canagarajah, A. S. (2004). Subversive identities, pedagogical safe houses, and critical learning. In B. Norton and K. Toohey (Eds.), Critical Pedagogies and Language Learning (pp. 116-137). New York: Cambridge University Press.

Canagarajah, A. S. (2007). Lingua franca English, multilingual communities, and language acquisition. Modern Language Journal, 91, 923-939.

Canfield, D.L. (1940). What Spanish sounds are most difficult for North Americans? Hispania, 23, 154-159.

Cantone, K. F. & MacSwan, J. (2009). The syntax of DP-internal codeswitching. In L. Isurin, D. Winford, and K. de Bot (Eds.), Multidisciplinary Approaches to Codeswitching (pp.243- 278). Amsterdam, Netherlands: John Benjamins.

Caritas Lawrence, M. (2001). A Nation in Search of Her Soul: Toward Building a National Belizean Identity through Cultural Understanding. Ann Arbor: UMI.

Carranza, M. A., & Ryan, E. B. (1975). Evaluative reactions of bilingual Anglo and Mexican American adolescents towards speakers of English and Spanish. Linguistics, 166, 83-104.

Carstens, V. (2000). Concord in minimalist theory. Linguistic Inquiry, 31, 319-355.

Cassano, P. V. (1977). Problems in language borrowing and lending exemplified by American Spanish phonology. Orbis, 36, 149-163.

Chambers, J. (1992). Dialect acquisition. Language, 68, 673-705.

Chan, B. H-S. (1998). Functional heads, Cantonese phrase structure and Cantonese-English code-switching. UCL Working Papers in Linguistics, 10, 253-284.

Chan, B. H-S. (2008). Code-switching, word order and the lexical/functional category distinction, Lingua, 118, 777-809. Chaston, J. (1996). Sociolinguistic analysis of gender agreement in article/noun combinations in Mexican American Spanish in Texas. Bilingual Review, 21, 195-202.

Chatterjee, T. (2014). Bilingual Complex Verbs: So what’s new about them? In Proceedings of the 38th Annual Meeting of the Berkeley Linguistics Society (pp. 47-62).

Chatterjee, T. (2016). Structural changes in Bengali-English bilingual verbs through the exploration of Bengali films. Languages, 1, 1-17.

Cheshire, J., & Gardner-Chloros, P. (1997). Communicating Gender in Two Languages. In H. Kotthoff & R. Wodak, (Eds.), Communicating Gender in Context. Amsterdam, Netherlands: John Benjamins.

Chomsky, N. (1965). Aspects of the theory of Syntax. Cambridge, MA: MIT Press.

354

Chomsky, N. (1981). Lectures of Government and Binding. Dordrecht: Foris.

Chomsky, N. (1991). Linguistics and cognitive science: Problems and mysteries. In A. Kasher (Ed.), The Chomskyan turn (pp. 26–54). Cambridge, MA: Blackwell.

Chomsky, N. (1995). The Minimalist Program. Cambridge, MA: The M.I.T. Press.

Chomsky, N. (2000). Minimalist inquiries: The framework. In R. Martin, D. Michaels & J. (Eds.), Step by step: Essays on Minimalist syntax in honor of Howard Lasnik (pp. 89– 155). Cambridge, MA: MIT Press.

Chomsky, N. (2001). Derivation by phase. In M. Kenstowicz (Ed.), Ken Hale: A life in language (pp. 1–51). Cambridge, MA: MIT Press

Church, M. C., Yaeger, J., & Dornan, J. L. (2011). The San Pedro Maya and the British Colonial Enterprise in British Honduras. In M. Liebmann & M. S. Murphy (Eds.), Enduring Conquests: Rethinking the Archaeology of Resistance to Spanish Colonialism in the . Santa Fe, NM: School for Advanced Research Press.

Clegg, J. (2010). An analysis of the motivations for borrowing in the Spanish of New Mexico. In S. Riviera-Mills & D. Villa (Eds.), Spanish of the U.S. Southwest: A Language in Transition (pp. 223-237). Iberoamericana: Vervuert.

Clegg, J. (2006). Lone English-origin nouns in the Spanish of New Mexico: A Variationist analysis of phonological and morphological adaptation. PhD Dissertation, University of New Mexico.

Clegg, J., & Waltermire, M. (2009). Gender assignment to English-origin nouns in the Spanish of the Southwestern United States. Southwest Journal of Linguistics, 28, 1-17.

Clegg, J., & Waltermire, M. (2009). Gender assignment to English-origin nouns in the Spanish of the Southwestern United States. Southwest Journal of Linguistics, 28, 1-17.

Clyne, M. G. (2003). Dynamics of language contact. Cambridge: Cambridge University Press.

Clyne, M. G. (1987). Constraints on code switching: How universal are they? Linguistics, 25, 739-764.

Colantoni, L. (2006). Increasing periodicity to reduce similarity. An acoustic account of deassibilation in rhotics. In M. Díaz-Campos (Ed.), Selected proceedings of the second conference on laboratory approaches to Spanish phonology (pp. 22-34). Somerville, Cascadilla Press.

Corbett, G. G. (1991). Gender. Cambridge, MA: Cambridge University Press.

355 Cornips, L. (2008). Loosing grammatical gender in Dutch: The result of bilingual acquisition and/or an act of identity? International Journal of Bilingualism, 12, 105-124.

Coulmas, F. (2013). Sociolinguistics: The Study of Speakers’ Choices. New York: Cambridge University Press.

Craig, D. R. (1983). Teaching Standard English to nonstandard speakers: some methodological issues. The Journal of Negro Education, 25, 65-74.

Crutchley. A. (2015). Bilingual compound verbs in children’s Panjabi-English code-switched narratives. Linguistic Approaches to Bilingualism, 5, 2-29.

Davies, P., & Deuchar, M. (2010). Using the Matrix Language Frame model to identify word order convergence in Welsh-English bilingual speech. In A. Breitbarth, C. Lucas, S. Watts, and D. Willis (Eds.), Continuity and Change in Grammar (pp. 77-96). Amsterdam, Netherlands: John Benjamins.

Davies, M. (2002). Corpus del Español. 100 Million Words, 1200’s to 1900’s. Retrieved from www.corpusdelespanol.org.

Decker, K. (2013). Belize Kriol. In World English Volume III: Central America, T. Hopkins & K. Decker (Eds.) (pp. 35-108). London, England: Bloomsbury Publishing.

Decker, K. (2005). The Song of Kriol: A Grammar of the Kriol Language of Belize. Belmopan, Belize: The National Kriol Council of Belize, House of Culture.

De Graff, M. (2009). Language Acquisition in Creolization, and thus, Language Change: Some Castesian-Uniformitarian Boundary Conditions. Language and Linguistics Compass, 3:4, 888-971.

De Graff, M. (2005). Linguists' most dangerous myth. The fallacy of Creole Exceptionalism. Language in Society, 34, 533-591.

De la Cruz Cabanillas, I., T. Martínez, C., M. Díez Prados, and E. Cerdá Redondo. (2007). English Loanwords in Spanish Computer Language. English for Specific Purposes, 26, 52-78.

Del Valle, J. (2014). The Politics of Normativity and Globalization: Which Spanish in the Classroom? The Modern Language Journal, 98, 358-372.

Deuchar, M., Epelde, I., Oyharçabal, B., & Parafita Couto, M. C. (2010). Gender agreement in Spanish–Basque and Spanish–English nominal constructions. Presented at the XXXIX Simposio Internacional de la SEL, Santiago de Compostela, Spain.

Deuchar, M. (2005). Congruence and code-switching in Welsh. Bilingualism: Language and Cognition, 8, 1-15.

356

Deuchar, M., & Stammers, J. (2016). English-origin verbs in Welsh: Adjudicating between Two Theoretical Approaches. Languages, 1, 1-17.

Dewaele, J., & Wei, L. (2014). Attitudes toward code-switching among adult mono- and multilingual language users. Journal of Multilingual and Multicultural Development, 35, 235-251.

Díaz-Campos, M. (2008). Variable production of the trill in spontaneous speech: sociolinguitic Implications, in L. Colantoni & J. Steele (Eds.), Selected Proceedings of the 3rd Conference on Laboratory Approaches to Spanish Phonology (pp. 47-58). MA, Cascadilla Proceedings Project.

Di Scuillo, A.-M., Muysken, P., & Singh, R. (1986). Government and code-mixing. Journal of linguistics, 22, 1-24.

Dobson, N. (1973). A History of Belize. London, UK: Longman Caribbean.

Doğruöz, A. S., & Backus, A. (2009). Innovative constructions in Dutch-Turkish: An assessment of on-going contact-induced change. Bilingualism: Language and Cognition, 12, 41–63.

DuBord, E. M. (2004). Gender assignment to English words in the Spanish of Southern Arizona. Divergencias: Revista de estudios lingüísticos y literarios, 2, 27-39.

Eckert, P. (1997a). Why ethnography? In U. Kotsinas, A. Stenstrom abd A. Karlsson (Eds.), Ungdomssprak i Norden (pp.52-62). Stockholm: Stockholm University.

Eckert, P. (1997b). Age as a sociolinguistic variable. In F. Coulmas (Ed.), The Handbook of Sociolinguistics (pp. 151-167). Oxford, UK: Blackwell.

Eckert, P. (1988). Adolescent social structure and the spread of linguistic change. Language in Society, 17, 183-207.

Eckert, P. (2000). Language variation as social practice. Oxford, UK: Blackwell Publishers.

Eckert. P. (2004). Adolescent language. In E. Finegan and J. R. Rickford (Eds.), Language in the USA: Themes for the Twenty-first Century (pp. 361-374). Cambridge: Cambridge University Press.

Eddington, D. & Hualde, J. I. (2008). El abundante agua fría: hermaphroditic Spanish nouns. Studies in Hispanic and Lusophone Linguistics, 1, 5-31.

Edwards, W. (1983). Code selection and shifting in Guyana. Language in Society, 12, 295–311.

Edwards, M., & Gardner-Chloros, P. (2007). Compound Verbs in Codeswitching: Bilinguals Making Do? International Journal of Bilingualism, 11, 73-91.

357 Elliott, L. S. (1995). National identity and media system dependency in Belize. PhD Dissertation, University of Florida, Gainesville.

Elliott, A. R. (2003). Staking out the territory at the turn of the century: Integrating phonological theory, research, and the effect of formal instruction on pronunciation in the acquisition of Spanish as a second language. In B. A. Lafford & R. Salaberry (Eds.), Spanish second language acquisition: State of the science (pp. 19-46). Washington, DC, Georgetown University Press.

Erker, D., & Gregory, R. G. (2012). The Role of Lexical Frequency in Syntactic Variability: Variable Subject Personal Pronoun Expression in Spanish. Language, 88, 526-557.

Escamilla, K. (2006). Semilingualism applied to the literacy behaviors of Spanish-speaking emerging bilinguals: Bi-illiteracy or emerging biliteracy? The Teachers College Record, 108, 2329-2353

Escobar, A. M. (2010). El español de los Estados Unidos. In J. I. Hualde, A. Olarrea, A. M. Escobar and C.E. Travis (Eds.), Introducción a la lingüística hispánica (pp. 445-502). Cambridge: Cambridge University Press.

Escure, G. (1982). Contrastive patterns of intragroup and intergroup interaction in the Creole continuum of Belize. Language in Society, 11, 239-264.

Escure, G. (1983). . In J. Holm (Ed.), Central American English (pp. 28-70). Heidelberg: Groos.

Espinosa, A. M. (1914). Studies in New Mexican Spanish, Part III: The English elements. Revue de Dialectologie Romane 6 (pp. 241-317), Part of doctoral dissertation: University of Chicago, 1909. Espinosa, A. M. (1917). Speech mixture in New Mexico: the influence of the on New Mexican Spanish. In: Stephens, H. M., and Bolton, H. E. (Eds.), The Pacific Ocean in History (pp. 408-428). The Macmillan Company: New York.

Face, T. L. (2006). Intervocalic Rhotic Pronunciation by Adult Learners of Spanish as a Second Language, in C. A. Klee and T. L. Face (Eds.), Selected Proceedings of the 7th Conference on the Acquisition of Spanish and Portuguese as First and Second Languages (pp. 47-58). Somerville, MA: Cascadilla Proceedings Project.

Fasold, R. (1984). The sociolinguistics of society. New York: Basil Blackwell.

Field, F. W. (2002). Linguistic borrowing in bilingual contexts. Amsterdam: Benjamins.

Field, F. (2005). Long-term effects of CS: Clues to structural borrowing. In International Journal of Bilingualism, 9, 341-360.

Fitts, S. (2006). Reconstructing the status quo: Linguistic interaction in a dual-language school. Bilingual Research Journal, 30, 337–365.

358

Flores, N., & Hopper, R. (1975). Mexican-Americans’ evaluations of spoken Spanish and English. Speech Monographs, 42, 91-98.

Fotiou, C. (2012). Are The Greek Verbs no Longer Enough? Bilingual Compound Verbs in Cypriot Greek. In Linguists of Tomorrow: Selected Papers from the 1st Cyprus Postgraduate Conference in Theoretical and Applied Linguistics, Kleanthes K. Grohmann, Aljona Shelkovaya, and Dionysios Zoumpalidis (Eds.), 22-50. Newcastle- upon-Tyne: Cambridge Scholars Publishing.

Folli, R., Harley, H., & Karimi, S. (2004). Determinants of event type in Persian complex predicates. Lingua, 115, 1365-1401.

Franceschina, F. (2001). Morphological or syntactic deficits in near-native speakers? An assessment of some current proposals. Second Language Research, 17, 213-247.

Fuller, J., & Lehnert, H. (2000). Noun phrase structure in German–English code-switching: variation in gender assignment and article use. International Journal of Bilingualism, 4, 399-420.

Fuller Medina, N. (2005). Spanish-English Contact in Belize: The Case of Hacer + V. In Proceedings of the 2005 Canadian Linguistics Association Annual Conference, C. Gurski (Ed.), 1-9.

Gabbert, W. (2007). In the shadow of the empire: The emergence of Afro-Creole societies in Belize and Nicaragua. Indiana, 24, 39-66.

Galindo, D. L. (1995). Language attitudes toward Spanish and English varieties: A Chicano perspective. Hispanic Journal of Behavioral Sciences, 17, 77-99.

García, M. E. (1998). Gender marking in a dialect of Southwest Spanish. Southwest Journal of Linguistics, 17, 49-58.

García, O. (2009). Bilingual education in the 21st century: A global perspective. Malden, MA: Wiley–Blackwell.

García, O. (2014). U.S. Spanish and Education: Global and local intersections. Review of Research in Education, 38, 58-80.

García, O., & Kleifgen, J. (2010). Educating emergent bilinguals. Policies, programs and practices for English Language Learners. New York, NY: Teachers College Press.

García, O., & Wei, L. (2014). Translanguaging: Implications for language, bilingualism and education. Basingstoke, England: Palgrave McMillan.

359 García, O., & Sylvan, C. E. (2011). Pedagogies and practices in classrooms: Singularities in pluralities. Modern Language Journal, 95, 385–400.

García Tesoro, A. I. (2010). Español en contacto con tzutujil en Guatemala: cambios en el Sistema pronominal átono de tercera persona. Revista Internacional de Lingüistica Iberoamericana, 8, 133-156.

Gardner-Chloros, P. (1987). Code-switching in relation to language contact and convergence. In G. Lüdi. (Ed.), Devenir bilingue-parler bilingue: Actes du 2e coloque sur le bilinguisme, (pp. 99-111). Tübingen: Niemeyer.

Gardner-Chloros, P. (1991). Language selection and switching in Strasbourg. Oxford: Clarendon Press.

Gardner-Chloros, P. (1995). Codeswitching in Community, Regional and National Repertoires: The Myth of the Discreteness of Linguistic Systems. In L. Milroy & P. Muysken (Eds.), One Speaker Two Languages: Cross-disciplinary Perspectives on Codeswitching (pp. 68- 89). Cambridge: Cambridge University Press.

Gardner-Chloros, P. (2009). Code-switching. Cambridge: Cambridge University Press.

Gardner-Chloros, P. (2010). Contact and Code-switching. In R. Hickey (Ed.), The Handbook of Language Contact (pp. 188-208). Malden, MA: Blackwell Riley.

Gardner-Chloros, P., & Malcolm, E. (2004). Assumption behind grammatical approaches to code-switching: When the blueprint is a red herring. Transactions of the Philological Society, 102, 103-129.

Gardner-Chloros, P., & Finnis, K. (2006). Using Lides to correlate compound verbs with other factors. Talk presented at Sociolinguistics Symposium 16, University of Limerick, 6 – 8 July.

Giles, H. (1973). Accent mobility: a model and some data. Anthropological Linguistics, 15, 87- 105.

Giles, H., Coupland, N., & Coupland, J. (1991). Accommodation theory: Communication, context, and consequence, in H. Giles, J. Coupland & N. Coupland (Eds), Contexts of Accommodation (pp. 1–68).

González-Vilbazo, K., & López, L. (2011). Some properties of light verbs in codeswitching. Lingua, 121, 832-850.

González-Vilbazo, K., & López, L. (2012). Little v and parametric variation. Natural Language and Linguistic Theory, 30, 33-77.

360 Government of Belize Language Policy. (2000). Handbook of Policies and Procedures for School Services. Ministry of Education, Youth and Sports.

Granda, G. de. (1980). Algunos rasgos fonéticos del español paraguayo atribuibles a interferencia guaraní. Revista Española de Lingüística, 10, 339-349.

Granda, G. de. (1984). Fenómenos de interferencia fonética del fang sobre el español de Guinea Ecuatorial: consonantismo. Anuario de Lingüística Hispánica, 1, 95-114.

Greene, L. (1999). A Grammar of Belizean Creole. New York, Peter Lang.

Grosjean, F. (1989). Neurolinguists, beware! The bilingual is not two monolinguals in one person. Brain and Language, 36, 3-15.

Grosjean, F. (1982). Life with two languages. Cambridge, MA: Harvard University Press.

Grosjean, F. (1998). Studying bilinguals: methodological and conceptual issues. Bilingualism: Language and Cognition, 1, 131-149.

Grosjean, F. (2001). The bilingual’s language modes. In J. Nicol (Ed.), One Mind Two Languages, (pp. 1-22). Oxford, UK: Blackwell.

Guerra, J. C. (2015). Language, Culture, Identity, and Citizenship in College Classrooms and Communities. New York, NY: Routledge.

Guion, S. G. (2003). The Vowel Systems of Quichua-Spanish Bilinguals: Age of Acquisition Effects on the Mutual Influence of the First and Second Languages. Phonetica, 60, 98- 128.

Gumperz, J. J. (1976). The Sociolinguistic Significance of Conversational Code-switching, In Language and Context Working Papers 46. Berkeley: University of California, Language Behaviour Research Laboratory.

Gumperz, J. J. (1982). Conversational code-switching. Discourse strategies (pp. 233–274). Cambridge: Cambridge University Press.

Gumperz, J. J., & Hernández-Chávez, E. (1971). Bilingualism, Bidialectalism, and classroom interaction. In A. Dil (Ed.), Language in Social Group: Essays by John J. Gumperz (pp. 311-350). Stanford: Stanford University Press.

Gumperz, J., & Wilson, R. (1971). Convergence and creolization: A case from the Indo- Aryan/Dravidian border. In D. Hymes (Ed.), Pidginization and creolization of languages (pp. 151–169). Cambridge: Cambridge University Press.

Hagerty, T. W. (1979). Phonological Analysis of the Spanish of Belize. Unpublished doctoral dissertation, University of California, Los Angeles.

361

Hagerty, T. W. (1996). The influence of English on the Spanish language of Belize. In M. D. Phillips (Ed.), Belize: Selected Proceedings from the Second Interdisciplinary Conference (pp. 131-142). Lanham, MD, University Press of America.

Hagiwara, R. (1995). Acoustic realizations of American /r/ as produced by women and men. UCLA Working Papers in Phonetics, 90, 1-187.

Hammond, R. (1999). On the non-occurrence of the phone [r] in the Spanish sound system, in J. Gutierrez-Rexach and F. Martínez-Gil (Eds.), Advances in Hispanic Linguistics (pp. 135- 151. Somerville, MA: Cascadilla Press.

Haspelmath, M. (2009). Lexical borrowing: Concepts and issues. In M. Haspelmath & U. Tadmor (Eds.), Loanwords in the World's Languages: A Comparative Handbook (pp.35- 54). Berlin: De Gruyter Mouton.

Haugen, E. (1950). Problems of bilingualism. Lingua, 2, 271-290.

Haugen, E. (1972). The Analysis of Linguistic Borrowing. In E. Scherabon Firchow, K. Grimstad, N. Hasselmo & W. A. O’Neil (Eds.), Studies by E. Haugen, (pp. 161-185). The Hague: Mouton.

Haust, D. (1995). Codeswitching in Gambia: eine soziolinguistische Untersuchung von Mandinka, Wolof und Englisch in Kontakt. Cologne: Köppe Verlag.

Heath, J. (1989). From code-switching to borrowing: A case study of Moroccan Arabic. London: Kegan Paul International.

Hebblethwaite, B. (2007). Intrasentential code-switching among Miami Haitian Creole–English bilinguals. Ph.D dissertation, Indiana University.

Hebblethwaite, B. (2010). Adverb code-switching among Miami's Haitian Creole–English second generation. Bilingualism: Language & Cognition, 13, 409-428.

Heller, M. (1988). Codeswitching: Anthropological and sociolinguistic perspectives. Berlin, Germany: Mouton de Gruyter Heller.

Heller, M. (1999). Linguistic minorities and modernity: A sociolinguistic ethnography. London: Longman.

Hellinger, M. (1974). The future of Belizean Creole. National Studies, 2, 11-15.

Henriksen, N. C., & Willis, E. W. (2010). Acoustic characterization of phonemic trill production in Jerezano . In M. Ortega-Llebaria (Ed.), Selected Proceedings of the Fourth Conference on Laboratory Approaches to Spanish Phonology (pp. 115-127). Somerville, MA: Cascadilla Proceedings Project.

362 Henriksen, N. C. (2015). Acoustic analysis of the rhotic contrast in Chicagoland Spanish: An intergenerational study. Linguistic Approaches to Bilingualism, 5, 285-321.

Herring, J. R., Deuchar, M., Parafita-Couto, M. C., & Quintanilla, M. M. (2010). I saw the madre: evaluating predictions about codeswitched determiner-noun sequences using Spanish-English and Welsh-English data. International Journal of Bilingualism, 13, 557.

Hidalgo, M. (1986). Language contact, language loyalty, and language prejudice on the Mexican border. Language and Society, 15, 193-220.

Hirson, A., & Sohail, N. (2007). Variability of rhotics in Punjabi-English bilinguals, in J. Trouvain & W. Barry (Eds.), Proceedings of the 16th International Congress of Phonetic Sciences (pp. 1501-1504). Saarbrucken.

Hock, H. H., & Joseph, B. D. (1996). Language history, language change, and language relationship. Berlin: Mouton de Gruyter.

Holm, J. (1977). Miskito words in Belize Creole. Belizean Studies, 5, 1-19.

Holm, J. (2000). An introduction to pidgins and creoles. Cambridge: Cambridge University Press.

Holm, J. (2006). Portuguese- and Spanish-based Creoles and Typologies. Papia, 16, 53-61.

Hualde, J. I. (2005). The sounds of Spanish. Cambridge, Cambridge University Press.

Hualde, J. I., Olarrea, A., Escobar, A. M. & Travis, C. E. (2010). Introducción a la lingüística hispánica, 2nd Ed. Cambridge, Cambridge University Press.

Hurtado, L. M., & Estrada, C. (2010). Language acquisition of Spanish vibrants. The Modern Language Journal, 94, 74-86.

Jacobson, R. (1998). Conveying a broader message through bilingual discourse: An attempt at Constrastive Codeswitching Research. In R. Jacobson (Ed.), Codeswitching Worldwide (pp. 51-76). Berlin, Germany: Mouton de Gruyter.

Jake, J., Myers-Scotton, C. & Gross, S. (2002). Making a minimalist approach to code-switching work: Adding the Matrix Language. Bilingualism: Language and Cognition, 5, 69–91.

Jake, J., & Myers-Scotton, C. (2009). Which language? Participation potentials across lexical categories in codeswitching. In L. Isurin, D. Winford and K. de Bot (Eds.), Multidisciplinary approaches to code-switching (pp. 207– 242). Amsterdam, Netherlands: John Benjamins.

Jaworski, S., & Gillian, E. (2011). On the phonetic instability of the Polish rhotic /r/. Poznan Studies in Contemporary Linguistics, 47, 380-398.

363 Jenkins, D. L. (2003). Bilingual Verb Constructions in Southwestern Spanish. Bilingual Review, 27, 195-204.

Johanson, L. (1999). The dynamics of code-copying in language encounters. In B Brendemoen, E. Lanza and E. Ryen (Eds.), Language Encounters across Time and Space (pp. 37-62). Oslo: Novus Press.

Johanson, L. (2002). Structural factors in Turkic language contacts. Richmond: Curzon Press.

Johnson, K. (2008). Second language acquisition of the Spanish multiple vibrant consonant. Doctoral Dissertation, University of Arizona.

Joshi, A. K. (1985). Processing of sentences with intrasentential code-switching. In D. R. Dowty, L. Karttunen and A. M. Zwicky, (Eds.), Natural language parsing: Psychological, computational and theoretical perspectives. Cambridge: Cambridge University Press.

Kachru, B.B. (1977). Code-switching as a communicative strategy in India. In M. Saville-Troike (Ed.), Linguistics and Anthropology, Georgetown University Round Table on Languages and Linguistics. Washington D.C.: Georgetown University Press.

Kernan, K., Sodergren, J. & French, R. (1976). Speech and Social Prestige in the Belizean Speech Community. In B. B. Blound & M. Sanches (Eds.), Socio-cultural Dimensions of Language Change. New York: Academic Press.

Khattab, G. (2009). Phonetic Acommodation in children’s code-switching. In B. E. Bullock & A. J. Toribio (Eds.), The Cambridge Handbook of Linguistic Codeswitching (pp. 142-159). Cambridge, Cambridge University Press.

King, R. (2002). Crossing grammatical borders: tracing the path of contact-induced change. Paper presented at the Eleventh International Conference on Methods in Dialectology, 5- 9 August 2002, University of Joensuu, Finland.

Koenig, E. L. (1975). Ethnicity and Language in , Belize: An analysis of code- switching. Unpublished doctoral dissertation, University of Texas at Austin, Texas.

Koenig, E. L. (1980). Ethnicity: the Key Variable in a Case Study of Language Maintenance and Language Shift. Ethnicity, 7, 1-14.

Kouznetsov, V., & Pamies, A. (2008). Trill with one closure, still a trill or a tap? Data from Russian and Spanish, in Proceedings of the XX Session of the Russian Acoustical Society, (pp. 672-675). Moscow: Russian Acoustical Society.

Knowles-Berry, S. (1988). Linguistic decay in Chontal Mayan: the speech of semi speakers. Anthropological Linguistics, 29, 332-341.

364 Kray, C., Church, M., & Yaegar, J. (In Press). Designs on/of the Land: Competing Visions, Displacement, and Landscape Memory in British Colonial Honduras. In F. Armstrong- Fumero (Ed.), Landscape, Memory and the Politics of Place: Archaeologists, Stakeholders, and the Intangible Heritage of Landscape. Colorado: University of Colorado Press.

Labov, W. (1966). The of English in . Washington, DC: Center for Applied Linguistics.

Labov, W. (1972). Sociolinguistic patterns. Oxford: Basil Blackwell.

Labov, W. (2001). Principles of linguistic change: Social factors, Vol. 2. Oxford, UK: Blackwell.

Ladefoged, P., & Maddieson, I. (1996). The sounds of the world’s languages. Malden, MA, Blackwell Publishers.

Lakshmanan, U., Balam, O., & Bhatia, T. K. (2016). Introducing the Special Issue: Mixed Verbs and Linguistic Creativity in Bi/Multilingual Communities. Languages, 1, 9.

Lance, D. M. (1975). Spanish-English Code-Switching. In Hernandez-Chavez, Cohen and Beltramo (Eds.), La Lengua De Los (pp. 138-153). Centre for Applied Linguistics: Arlington, Virginia.

Leeman, J. (2005). Engaging critical : Spanish for native Speakers. Foreign Language Annals, 38, 35–45.

Lehmann, C. (2015). Valency classes in Yucatec Maya. In Malchukov, A. and Comrie, B. (Eds.), Valency Classes in the World’s Languages (pp. 1427-1480). De Gruyter Mouton: Berlin, Germany.

Lent, J. A. (1989). Country of no return: Belize since television. Belizean Studies, 14-36.

Lenneberg, E. (1967). Biological foundations of language. New York: John Wiley.

Le Page, R. B., & Tabouret-Keller, A. (1985). Acts of identity: Creole-based approaches to language and ethnicity. New York, NY: Cambridge University Press.

Le Page, R. (1992). You Can Never Tell Where a Word Comes From: Language Contact in a Diffuse Setting. In Ernst Jahr, (Ed.), Language Contact: Theoretical and Empirical Studies (pp. 70-101). The Hague: Mouton.

Levin, B., & Rappaport, M. (1986). The formation of adjectival passives. Linguistic Inquiry, 17, 623–661.

365 Lewis, A. M. (2004). Coarticulatory Effects on Spanish Trill Production, in A. Agwuele, W. Warren & S. Park (Eds.), Proceedings of the 2003 Texas Linguistics Society Conference, (pp. 116-127). Somerville, MA: Cascadilla Proceedings Project.

Liceras, J. M., Martínez, C., Perez-Tattam, R., Perales, S., and R. Fernandez-Fuertes. (2006). L2 Acquisition as a process of Creolization: Insights from child and adult code-mixing. In C. Lefebvre, L. White, and C. Jourdan (Eds.), L2 Acquistion and Creole Genesis: Dialogues (pp. 113-144). Amsterdam, Netherlands: John Benjamins.

Liceras, J. M., Fernández Fuentes, R., Perales, S., Pérez-Tattam, R. & Spradlin, K. T. (2008). Gender and number agreement in bilingual native and non-native grammars: A view from child and adult functional-lexical mixings. Lingua, 118, 761–852.

Liljencrants, J., & Lindblom, B. (1972). Numerical simulation of vowel quality systems: the role of perceptual contrast. Language, 48, 839–862.

Lipski, J. (1977). Code-switching and the problem of bilingual competence. In M. Paradis (Ed.), Aspects of Bilingualism (pp. 250-264). Columbia, SC: Hornbeam Press.

Lipski, J. (1978). Code-switching and the problem of bilingual competence. In M. Paradis (Ed.), Aspects of bilingualism (pp. 250–264). Columbia, SC: Hornbeam Press.

Lipski, J. (1985). /s/ in Central American Spanish. Hispania, 68, 143-49.

Lipski, J. (1986). English-Spanish contact in the United States and Central America: sociolinguistic mirror images? In M. Görlach and J. Holm (Eds.), Focus on the Caribbean (pp. 191-208). Amsterdam, Netherlands: John Benjamins.

Lipski, J. M. (2005). Code-switching or borrowing? No sé so no puedo decir, you know. In L. Sayahi and M. Westmoreland (Eds.), Selected Proceedings of the Second Workshop on Spanish Sociolinguistics (pp. 1-15). Somerville, MA: Cascadilla Proceedings Project.

Lipski, J. (2006). Afro- and Helvécia Portuguese: Semi-creole parallels. Papia, 16, 96-116.

Lipski, J. M. (2008). Varieties of Spanish in the United States. Washinton, DC: Georgetown University Press.

Lipski, J., & Schwegler, A. (1993). Creole: Creole Spanish and Afro-Hispanic. In J. Green, R. Posner (Eds.), Trends in Romance linguistics and philology 5: bilingualism and linguistic conflict in Romance (pp. 407-432). Berlin, Germany: Mouton De Gruyter.

Lipski, J. M. (1993). Creoloid phenomena in the Spanish of transitional bilinguals. In A. Roca and J. M. Lipski (Eds.), Spanish in the United States: Linguistic Contact and Diversity (pp. 155-173). New York, NY: Mouton de Gruyter.

366 Lope Blanch, J. M. (1975). Un caso de posible influencia maya en el español mexicano. Nueva Revista de Filología Hispánica, 24, 89-100.

Luján, M. E. (1981). The Spanish copulas as aspectual indicators. Lingua, 54, 165-210.

McCormick, K. (2002). Code-switching, Mixing, and Convergence in Cape Town. In R. Mesthrie (Ed.), Language in South Africa (pp. 216-234). West Nyack, NY: Cambridge University Press.

MacGregor-Mendoza, P. (2000). Aquí no se habla español: Stories of linguistic repression in Southwest schools. Bilingual Research Journal, 24, 333–345.

MacSwan, J. (1999). A minimalist approach to intrasentential code switching. New York, NY: Garland Press.

MacSwan, J. (2000). The architecture of the bilingual language faculty: Evidence from code- switching. Bilingualism: Language and Cognition, 3, 37–54.

MacSwan, J. (2005). Codeswitching and generative grammar: A critique of the MLF model and some remarks on: modified minimalism. Bilingualism: Language and Cognition, 8, 1-22.

MacSwan, J. (2014). Programs and proposals in codeswitching research: Unconstraining theories of bilingual language mixing. In J. MacSwan (Ed.), Grammatical Theory and Bilingual Codeswitching (pp. 1-34). Cambridge, MA: MIT Press.

MacSwan, J., & Colina, S. (2014). Some consequences of Language Design: Codeswitching and the PF Interface. In J. MacSwan (Ed.), Grammatical Theory and Bilingual Codeswitching (pp.185-210). Cambridge: MIT Press.

MacSwan, J., & McAlister, K. T. (2010). Naturalistic and Elicited Data in Grammatical Studies of Codeswitching. Studies in Hispanic and Lusophone linguistics, 3, 521-532.

Makihara, M. (2005). Rapa Nui ways of speaking Spanish: Language shift and socialization on Easter Island. Language in Society, 34, 727-762.

Mahootian, S. (2003). Sending a message: Codeswitching and the bilingual identity. Proceedings of the Second University of Vigo International Symposium on Bilingualism, Vigo, Spain, October 2002, (pp. 1493-1499).

Mahootian, S., & Santorini, B. (1996). Code switching and the complement/adjunct distinction: A reply to Belazi, Rubin and Toribio. Linguistic Inquiry, 27, 465–466.

Major, R. C. (1986). The ontogeny model: Evidence from L2 acquisition of Spanish r. Language Learning, 36, 453-504.

367 Martínez, R. A. (2010). Spanglish as literacy tool: Toward an understanding of the potential role of Spanish-English code-switching in the development of academic literacy. Research in the Teaching of English, 45, 124-149.

Martínez-Gibson, E. A. (2011). A comparative study on gender agreement errors in the spoken Spanish of heritage speakers and second language learners. Porta Linguarium, 15, 187.

Matras, Y. (2010). Contact, convergence and typology. In R. Hickey (Ed.), Handbook of Language Contact (pp. 66-85). Oxford, UK: Blackwell.

Mayer, M. (1969). Frog, Where Are You? New York, NY: Dial Press.

Meakins, F. (2008). Land, language and identity: The socio-political origins of Gurindji Kriol. In M. Meyerhoff & N. Nagy (Eds.), Social Lives in Language – Sociolinguistics and multilingual speech communities: Celebrating the work of Gillian Sankoff (pp. 69-94). Amsterdam, Netherlands: John Benjamins.

Meeuwis, M., & Blommaert, J. (1998). A monolectal view of CS. In P. Auer (Ed.), Codeswitching in conversation: Language, interaction and identity (pp. 76-101). London, UK: Routledge.

McClure, E., & Wentz, J. (1976). Code-switching in children's narratives. In 1975 Mid-America linguistics conference papers. Lawrence, Kansas, the Linguistics Department, University of Kansas.

Michnowicz, J. C. (2006). Linguistic and Social Variables in Yucatan Spanish. PhD dissertation. The Pennsylvania State University, State College, PA.

Michnowicz, J. (2012). The Standardization of Yucatan Spanish: Family Case Studies in Izamal and Merida. In K. Geeslin and M. Díaz-Campos (Eds.), Selected Proceedings of the 14th Hispanic Linguistic Symposium (pp. 102-115). Somerville, MA: Cascadilla Proceedings Project.

Moberg, M. (1997). Myths of ethnicity and nation: Immigration, work, and identity in the Belize banana industry. Knoxville: University of Tennessee Press.

Moinzadeh, A. (1999). Bilingual Phenomena: Towards the Fateful Triangle of Language Mixture. Cahier Linguistiques d’Ottawa, 27, 31-63.

Moore, Z., & Abrikian, J. (1995). Español para la vida 2 Workbook. Cheltenham, UK: Nelson Thornes Ltd.

Molesky, J. (1988). Understanding the American linguistic mosaic: A historical overview of language maintenance and language shift. In S. L. McKay and S. C. Wong (Eds.), Language diversity: Problem or resource? (pp. 29-68). Cambridge, MA: Newbury House.

368

Montes-Alcalá, C. (2009). in the United States: More than Spanglish. Camino Real, 1, 97-115.

Montes-Alcalá, C., & Lapidus Shin, N. (2011). Las keys vs. el key: Feminine gender assignment in mixed- language texts. Spanish in Context, 8, 119-143.

Montrul, S. (2004). Subject expression in Spanish heritage speakers: A case of morphosyntactic convergence. Bilingualism: Language & Cognition, 7, 125-142.

Montrul, S., & Potowski, K. (2007). Command of gender agreement in school-age Spanish bilingual children. International Journal of Bilingualism, 11, 301-328.

Morin, R. (2006). Spanish gender assignment in computer and internet related loanwords. Revista di Linguistica, 18, 325-354.

Moro, A. (2001). The semantic interpretation and syntactic distribution of determiner phrases in Spanish/English code-switching. Paper presented at the 2001 International Symposium on Bilingualism.

Moyer, M. (1992). Analysis of code-switching in Gibraltar. Unpublished doctoral thesis, Universitat Autònoma de Barcelona, Barcelona.

Moyne, J. (1974). The so-called passive in Persian. Foundations of Language, 12, 249–267.

Mrak, N. A. (2011). Heritage speakers and the standard: Fighting linguistic hegemony. In Luis A. Ortiz-López (ed.), Selected Proceedings of the 13th Hispanic Linguistics Symposium, pp. 161-168. Somerville, MA: Cascadilla Proceedings Project.

Mufwene, S. (2008). Language evolution: Contact, competition and change. London, UK: Continuum.

Muysken, P. (1997). Code-switching processes: Alternation, insertion, congruent lexicalization. In Martin Pütz (Ed.), Language Choices (pp. 361-381). Amsterdam: John Benjamins.

Muysken, P. (2000). Bilingual speech: A Typology of Code-Switching. Cambridge: Cambridge University Press.

Muysken, P. (2013). Language contact outcomes as the result of bilingual optimization strategies. Bilingualism: Language and Cognition, 16, 714.

Muysken, P. (2014). Déjà voodoo or new trails ahead? Re-evaluating the mixing typology model. In R. Torres Cacoullos, N. Dion, & A. Lapierre (Eds.), Linguistic Variation: Confronting Fact and Theory (pp. 243-262). New York: Francis & Taylor.

369 Muysken, P., & de Rooij, V. (1995). Review of Social motivations for code-switching: Evidence from Africa and Duelling languages: Grammatical structure in code-switching. Linguistics, 33, 1043–1066.

Myers-Scotton, C. (1993). Duelling Languages: Grammatical structure in codeswitching. Oxford, UK: Clarendon Press.

Myers-Scotton, C. (1997). Duelling languages: Grammatical structure in code-switching. Oxford: Oxford University Press.

Myers-Scotton, C. (1998). Structural uniformities versus community differences in codeswitching. In Jacobson, R. (Ed.), Codeswitching Worldwide. Trends in Linguistics: Studies and Monographs 106 (pp. 91-108). Berlin: Mouton de Gruyter.

Myers-Scotton, C. (2002). Contact Linguistics: Bilingual Encounters and Grammatical Outcomes.Oxford: Oxford University Press.

Myers-Scotton, C. (2006). Multiple voices: An introduction to bilingualism. Malden: Blackwell Publishers.

Myers-Scotton, C. M., & Jake, J. L. (1995). Matching lemmas in a bilingual language competence and production model: Evidence from intrasentential codeswitching. Linguistics, 33, 981–1024.

Myers-Scotton, C., & Jake, J. L. (2000). Four types of morpheme: Evidence from Apahasia, code switching and second language acquisition. Linguistics, 38, 1064.

Myers-Scotton, C., & Jake, J. L. (2009). A universal model of code-switching and bilingual language processing of production. In B. E. Bullock and A. J. Toribio (Eds.), The Camebridge Handbook of Lingusitic Code-switching (pp. 336–357). New York, NY: Cambridge University Press.

Myers-Scotton, C., & Jake, J. L. (2013). Nonfinite verbs and negotiating bilingualism in code- switching: implications for a language production model. Bilingualism: Language and Cognition, 17, 511-525.

Naseh, L. (1997). Codeswitching between Persian and Swedish, Eurosla 7 Proceedings, 201- 211.

Norton, B., & McKinney, C. (2011). An identity approach to second language acquisition. In D. Atkinson (Ed.), Alternative Approaches to Second Language Acquisition (pp. 73-94). New York, NY: Routledge.

Norton, B., & Toohey, K. (2001). Changing perspectives on good language learners. TESOL Quarterly, 35, 307–22.

370 Norton, B., & Toohey, K. (Eds.). (2004). Critical Pedagogies and Language Learning. New York, NY: Cambridge University Press.

Neufeld, G. G. (1980). On the adult's ability to acquire phonology. TESOL Quarterly, 14, 285- 298.

O'Brien, V. (2013). Rhotic Production and Contrast in Equatoguinean Spanish. The Ohio State University, MA thesis.

Oliveira, O. S. (1986). Effects of transborder television in Corozal town and surrounding villages. Belizean Studies, 14, 31-51.

Oliveira, O. S. (1990). Mass media in Belize. In S. H. Surlin and W. C. Soderlund (Eds.), Mass media and the Caribbean (pp. 115-128). New York, NY: Gordon A Breach.

Olsen, M. K. (2012). The L2 acquisition of Spanish rhotics by L1 English speakers: The effect of L1 articulatory routines and phonetic context for allophonic variation. Hispania, 95, 65- 82.

Olszanski, F. (2007). Odón Betanzos-Palacios: Lengua de diálogo y lengua de cultura. Interview. Web. 17 January 2011.

Ornstein, J. (1974). Mexican American sociolinguistics: a well-kept scholarly and public secret. In B. Hoffer & J. Ornstein (Eds.), Sociolinguistics in the Southwest (pp. 91-121). San Antonio, TX: Trinity University Press.

Otheguy, R., & Lapidus, N. (2003). An Adaptive Approach to Noun Gender in New York contact Spanish. In R. Nunez-Cedeno, L. Lopez, & R. Cameron (Eds.), A Romance Perspective on Language Knowledge and Use (pp. 209-229). Amersterdam, Netherlands: John Benjamins.

Otheguy, R. & Zentella, A. C. (2007). Apuntes preliminares sobre el contacto lingüístico y dialectal en el uso pronominal del español en Nueva York. In R. Cameron & K. Potowski, Spanish in contact: Policy, social and linguistic inquiries (pp. 275-95). Amsterdam, Netherlands: John Benjamins.

Otheguy, R., García, O., & Reid, W. (2015). Clarifying translanguaging and deconstructing named languages: A perspective from linguistics. Applied Linguistics Review, 6, 281-307.

Oyama, S. (1976). A sensitive period for the acquisition of a phonological system. Journal of Psycholinguistic Research, 5, 261-283.

Palmer, D., & Martínez, R. A. (2013). Teacher agency in bilingual spaces: A fresh look at preparing teachers to educate Latino/a bilingual children. Review of Research in Education, 37, 269-297.

371 Palmer, D. K., Martínez, R. A., Mateus, S. G., & Henderson, K. (2014). Reframing the Debate on Language Separation: Toward a Vision for Translanguaging Pedagogies in the Dual Language Classroom. Modern Language Journal, 98, 757-772.

Pandit, I. (1990).Grammaticality in code switching.In Rodolfo Jakobson (Ed.), Codeswitching as a world-wide phenomenon. New York, NY: Peter Lang.

Parafita, M.C., Deuchar, M., & Fusser, M. (2015). How do Welsh-English bilinguals deal with conflict? Adjective-noun order resolution. In Stell G., & Yakpo K.(Eds.), Code-switching at the crossroads between structural and sociolinguistic perspective (pp. 65–84). Berlin, Germany: Mouton de Gruyter.

Parafita Couto, M.C., Munarriz, A., Epelde, I., Deuchar, M., & Oyharçabal, B. (2014). Gender conflict resolution in Spanish-Basque mixed DPs. Bilingualism: Language & Cognition, 1-20. DOI: 10.1017/S136672891400011X

Patkowski, M. S. (1990). Age and accent in a second language: a reply to James Emil Flege. Applied Linguistics, 11, 73-89.

Pérez Casas, M. (2008). Codeswitching and identity among Island Puerto Rican bilinguals. Washington, DC: ProQuest LLC.

Pennycook, A. (2010). Language as a local practice. NewYork: Routledge.

Penny, R. (1969). El habla pasiega: ensayo de dialectología montañesa. London, UK: Tamesis Books.

Phillips, R. (1967). Los Angeles Spanish: a descriptive analysis. Ph.D dissertation, University of Wisconsin.

Pfaff, C. W. (1979). Constraints on language mixing: Intrasentential code-switching and borrowing in Spanish/English. Language, 55, 296-316.

Poplack, S. (1980). Sometimes I’ll start a sentence in English y termino en español: Toward a typology of code-switching. Linguistics, 18, 581–618.

Poplack, S. (1982). Competing influences on gender assignment: Variable processes, stable outcome. Lingua, 57, 1–28.

Poplack, S. (1988). Contrasting patterns of code-switching in two communities. In M. Heller (Ed.), Codeswitching: Anthropological and sociolinguistic perspectives (pp. 215–245). Berlin, Germany: Mouton de Gruyter.

Poplack, S. (2004). Code-switching. In U. Ammon, N. Dittmar, K.J. Mattheier, & P. Trudgill. (Eds.), Soziolinguistik. An international handbook of the science of language, 2nd ed. (pp. 589-596). Berlin, Germany: Walter de Gruyter.

372

Poplack, S. (2012). What does the Nonce Borrowing Hypothesis hypothesize? Bilingualism: Language and Cognition, 15, 644-648.

Poplack, S. and D. Sankoff. (1984). Borrowing: the synchrony of integration. Linguistics, 22, 99- 135.

Poplack, S., & Meechan, M. (1998). Introduction: How languages fit together in codemixing. International Journal of Bilingualism, 2, 127–138.

Poplack, S., & Dion, N. (2012). Myths and facts about loanword development. Language variation and change, 24, 279-315.

Poplack, S., Pousada, A., & Sankoff, D. (1982). Competing influences on gender assignment: Variable processes, stable outcome. Lingua, 57, 1–28.

Poplack, S., Sankoff, D. & Miller, C. (1988). The social correlates and linguistic processes of lexical borrowing and assimilation. Linguistics, 26, 47–104.

Porras, J. E. (2013). Noun Phrase Marking in Chabacano. Philippine Creole Spanish: A Comparative Perspective. California Linguistic Notes, 38, 122-142.

Quilis, A. (1993). Tratado de fonología y fonética españolas. Madrid, Arco.

Quintana, R. A. (2006). Geografía lingüística del judeoespañol: estudio sincrónico y diacrónico. Berne, Peter Lang.

Ramirez, A., & Milk, R. D. (1986). Notions of grammaticality among teachers of bilingual pupils. TESOL Quarterly, 20, 495-513.

Ramirez, Arnulfo G., Milk, R. D., & Sapiens, A. (1983). Intragroup differences and attitudes toward varieties of Spanish among bilingual pupils from California and Texas. Hispanic Journal of Behavioral Sciences, 5, 417-429.

Ramos-Pellicia, M. F. (2007). Lorain Puerto Rican Spanish and ‘r’ in Three Generations. In J. Holmquist, A. Lorenzino & L. Sayahi (Eds.), Selected Proceedings of the Third Workshop on Spanish Sociolinguistics (pp. 53-60). Somerville, MA: Cascadilla Proceedings Project.

Ravindranath, M. (2009). Language shift and the speech community: sociolinguistic change in a Garifuna community in Belize. Unpublished Dissertation.

Reed, N. (1964). The Caste War of Yucatan. Stanford: Stanford University Press.

Reeder, J. T. (1998). English speakers’ acquisition of voiceless stops and trills in L2 Spanish. Texas Papers in Foreign Language Education, 3, 101-108.

373

Reyes, R. (1982). Language Mixing in Chicano Spanish. In J. Amastae & L. Elías-Olivares (Eds.), Spanish in the United States: Sociolinguistics Aspects (pp. 154-65). New York, NY: Cambridge University Press.

Ritchie, W. C. & Bhatia, T. K. (2004). Social and psychological factors in language mixing. In T. K. Bhatia & W. C. Ritchie (Eds.), The handbook of bilingualism (pp. 336-352). Oxford, UK: Blackwell Publishing.

Rodriguez-Gonzalez, E., & Parafita-Couto, M. C. (2012). Calling for Interdisciplinary Approaches to the Study of Spanglish and its Linguistic Manifestations. Hispania, 95, 461-480.

Romaine, S. (1989). Bilingualism. United Kingdom: Basil Blackwell.

Romaine, S. (1995). Bilingualism. Cambridge, MA: Blackwell Publishers Incorporation.

Romero, R. (2012). Spanish in the Bosphorus: A sociolinguistic study on the Judeo-Spanish dialect spoken in Istanbul. Libra: Istanbul, Turkey.

Rose. M. (2010). Intervocalic tap and trill production in the acquisition of Spanish as a second language. Studies in Hispanic and Lusophone Linguistics, 3, 379-420.

Ryan, E. B., & Carranza, M. (1977). In group and outgroup reactions toward Mexican-American language varieties. In H. Giles (Ed.), Language, ethnicity, and intergroup relations (pp. 59-82). London, UK: Academic Press.

Salmon, W. (2015). Language Ideology, Gender, and Varieties of Belizean Kriol. Journal of Black Studies, 46, 1-21.

Salmon, W., & Gómez Menjívar, J. (2014). Whose Kriol is Moa Beta? Prestige and Dialects of Kriol in Belize. Proceedings of the Berkeley Linguistics Society, 40, 456 – 479.

Samar, R. G., & Meechan, M. (1998). The Null Theory of Code-Switching versus the Nonce Borrowing Hypothesis: Testing the Fit in Persian-English Bilingual Discourse. The International Journal of Bilingualism, 2, 203-219.

Sánchez Corrales. V. (1986). Escisión fonológica de /ɹ/ en el español de Costa Rica. Revista de Filología y Linguística, 12, 129-133.

Sánchez, R. (1973). Nuestra circunstancia lingüística. Voices. Readings from El Grito. A Journal of Mexican American Thought 1967-1973. Berkeley, Quinto Sol Publications (pp. 420- 449).

Sánchez, M. F. (1995). Clasificación y análisis de préstamos del inglés en la prensa de España y México. Lewiston, NY: Mellen University Press.

374

Sánchez, T. S. (2005). Constraints on Structural Borrowing in a Multilingual Contact Situation. PhD dissertation, University of Pennsylvania.

Sankoff, G. (2008). Linguistic Outcomes of Language Contact. In J. K. Chambers, P. Trudgill, and N. Schilling-Estes (Eds.), The Handbook of Language Variation and Change (pp. 638-668). Oxford: Blackwell.

Sankoff, D., & Poplack, S. (1981). A formal grammar for code-switching. Papers in Linguistics, 14, 3–45.

Sankoff, D., Poplack, S., & Vanniarajan, S. (1990). The Case of the Nonce Loan in Tamil. Language Variation and Change, 2, 71-101.

Savic, J. M. (1995). Structural Convergence and Language Change: Evidence from Serbian/English Code Switching. Language in Society, 24, 475-492.

Sawyer, J. B. (1975). Spanish-English bilingualism in San Antonio, Texas. In E. Hernández- Chavez, A. D. Cohen and A. Beltramo (Eds.), El lenguaje de los Chicanos The language of Chicanos (pp. 77-98). Arlington, VA: Center for Applied Linguistics.

Scovel, T. (1969). Foreign accents, language acquisition, and cerebral dominance. Language Learning, 19, 245-253.

Scovel, T. (1988). A time to speak: A psycholinguistic inquiry into the critical period for human speech. Rowley, MA: Newbury House.

Sessarego, S. (2013). Some Remarks on the Origins of Chota Valley Spanish. In A. M. Carvalho and S. Beaudrie (Eds.), Selected Proceedings of the 6th Workshop on Spanish Sociolinguistics (pp. 87-96). Somerville, MA: Cascadilla Proceedings Project.

Scholes, F. V., & Thompson, E. (1977). The Francisco Pérez Probanza of 1654-1656 and the Matricula of Tipu (Belize). In G. Jones (Ed.), Anthropology and History in Yucatan (pp. 43-68). Austin, TX: University of Texas Press.

Sebba, M. (1998). A congruence approach to the syntax of codeswitching. International Journal of Bilingualism, 2, 1–19.

Sebba, M. (2009). On the notions of congruence and convergence in code-switching. In Barbara E. Bullock and Almeida Jacqueline Toribio (Eds.), The Cambridge Handbook of Linguistic Code-Switching (pp. 40-57). Cambridge: Cambridge University Press.

Seitz, K. S. (2005). Migration, demographic change, and the enigma of identity in Belize. Unpublished Master’s Thesis. Arizona State University.

375 Silverman, D. (2012). Neutralization: Key Topics in Phonology. Cambridge University Press: Cambridge.

Shoman, A. (2010). Reflections on Ethnicity and Nation in Belize. In Documento de Trabajo. No. 9 / Document de Travail No. 9, Mexico: Proyecto AFRODESC/EURESCL (pp.1-61).

Siegel, J. (2005). Creolization outside Creolistics. Journal of Pidgin and Creole Languages, 20, 141-167.

Silva-Corvalan, C. (1993). On the permeability of grammars: evidence from Spanish and English contact. In Ashby, W., Mithun M., Perissinotto, G. and Raposo E. (Eds.), Selected Papers from the 21st Symposium on Romance Languages (pp. 19-43). Amsterdam, Netherlands: John Benjamins.

Silva-Corvalán, C. (1994). Language contact and change. New York, NY: Oxford University Press.

Silva-Corvalán, C. (1995). The Study of Language Contact: An Overview of the Issues. In C. Silva-Corvalan (Ed.), Spanish in Four Continents: Studies in Language Contact and Bilingualism (pp. 3-14). Washington: Georgetown University Press.

Silva-Corvalán, C. (2004). Spanish in the Southwest. In E. Finegan & J. R. Rickford (Eds.), Language in the USA: Themes for the Twenty-First Century (pp. 205–29). Cambridge: Cambridge University Press.

Silva-Corvalan, C. (2008). The limits of convergence in language contact. Journal of Language Contact, Thema 2, 213-224.

Simonet, M. (2010). Dark and clear laterals in Catalan and Spanish: Interaction of native categories in early bilinguals. Journal of Phonetics, 38, 664-679.

Slabakova, R., & Montrul, S. (2003). Genericity and Aspect in L2 Acquisition. Language Acquisition, 11, 165-196.

Solé, Y. (1977). Language attitudes towards Spanish among Mexican-American college students. Journal of the Linguistic Association of the Southwest, 2, 37-46.

Solé, M. (2002). Aerodynamic characteristics of trills and phonological patterning. Journal of Phonetics, 30, 655-688.

Smead, R. (2000). On the assignment of gender to Chicano anglicisms: Processes and results. Bilingual Review, 25, 277-297.

Specker, E. (2008). The use of bilingual discourse markers: Identity in mediated learning. Arizona Working Papers in SLA & Teaching, 15, 97–120.

376 Stell, G. (2015). Towards an integrated approach to structural and conversational code-switching through macro-sociolinguistic factors. In G. Stell & K. Yapko (Eds.), Code-switching Between Structural and Sociolinguistic Perspectives (pp. 117-138). Berlin, Germany: Walter de Gruyter.

Stepick, A., & Stepick, C. D. (2002). Power and Identity: Miami Cubans. In M. M. Suarez- Orozco & M. Paez (Eds.), Latinos: Remaking America (pp. 75-92). University of California Press: Los Angeles, CA.

Stevens, K., & Blumstein, S. E. (1975). Quantal aspects of consonant production and perception: A study of retroflex consonants. Journal of Phonetics, 3, 215-233.

Suarez Molina, V. (1996). El español que se habla en Yucatán: Apuntamientos filológicos, 3rd ed. Merida, Yucatan: Ediciones de la Universidad Autonoma de Yucatan.

Tabouret-Keller, A. (1980). They don’t fool around with the Creole much, as with the Spanish: A family case in San Ignacio, Cayo District, Belize. York Papers in Linguistics, 9, 241- 59.

Tagliamonte, S. A. (2006). Analysing Sociolinguistic Variation. Cambridge: Cambridge University Press.

Tamis, A. (1986). The state of modern Greek language as spoken in Victoria. Doctoral dissertation, University of Melbourne, Australia.

Tarone, E. (1983). On the variability of interlanguage systems. Applied Linguistics, 4, 142-163.

Teschner, R. V. (1974). A critical annotated bibliography of Anglicisms in Spanish. Hispania, 57, 631–678.

Thomason, S. (1995). Language Mixture: Ordinary Processes, Extraordinary Results. In C. Silva-Corvalan (Ed.), Spanish in Four Continents: Studies in Language Contact and Bilingualism, (pp.15-33). Washington, DC: Georgetown University Press.

Thomason, S. (2001). Language contact: An introduction. Washington, DC: Georgetown University Press.

Thomason, S. (2003). Social factors and linguistic processes in the emergence of stable mixed languages. In P. Bakker and Y. Matras (Eds.), The Debate. Theoretical and Empirical Advances (pp. 21-40). Berlin: Mouton De Gruyter.

Thomason, S. G. (2008). Social and linguistic factors as predictors of contact-induced change. Journal of Language Contact, Thema 2, 42-55.

Thomason, S. G., & Kaufman, T. (1988). Language contact, creolization and genetic linguistics. Los Angeles: University of California Press.

377 Thompson, J. E. (1972). The Maya of Belize: Historical Chapters Since Columbus. Belize: Benex Press.

Thompson, J. E. (1963). Mayan Archaeologist. Norman, OK: University of Oklahoma Press.

Thompson, R. (1974). Mexican American language loyalty and the validity of the 1970 census. In Garland Bills (Ed.), Southwest area l linguistics (pp. 65-78). San Diego, CA: Institute for Cultural Pluralism.

Thompson, I. (1991). Foreign accents revisited: the English pronunciation of Russian immigrants. Language Learning, 41, 177-204.

Timms, W. W., & Eccott, A. C. (1972). International picture stories: guided composition in English, French, German and Spanish. London: University of London Press.

Timm, L. (1975). Spanish-English code-switching: el porque y how-not-to. Romance Philology, 28, 473-482.

Toohey, K. (2000). Learning English at School: Identity, Social Relations and Classroom Practice. Clevedon: Multilingual Matters.

Toribio, A. J. (2001). On the emergence of code-switching competence. Bilingualism: Language and Cognition, 4, 203-231.

Toribio, A. J. (2002). Spanish-English code-switching among US Latinos. International Journal of the Sociology of Language, 158, 89–119.

Toribio, A. J. (2004). Convergence as an optimization strategy of bilingual speech: Evidence from code-switching. Bilingualism: Language and Cognition, 7, 172.

Toribio, A. J. (2006). Linguistic displays of identity among Dominicans in national and diasporic settlements. In C. Davies & J. Brutt-Griffler (Eds.), English and Ethnicity. New York, NY: Palgrave.

Toribio, J. A., Bullock, B., & Greaser. C. (2012). The Bilingual Compound Verb [hacer + VE] in Texas Spanish: The Value of Corpus Data. Paper presented at the Hispanic Linguistics Symposium, Gainesville, FL, October 2012.

Torres Cacoullos, R., & Travis, C. E. (2010). Variable yo expression in New Mexico: English influence? In S. Rivera-Mills and D. J. Villa (Eds.), Spanish of the U.S. Southwest: A language in transition (pp. 185-206). Madrid, Spain: Iberoamericana.

Torres Cacoullos, R., & Travis, C. E. (In preparation). New Mexico Spanish-English Bilingual (NMSEB) corpus, National Science Foundation 1019112/1019122. http://nmcode- switching.la.psu.edu/

378 Torres Cacoullos, R., & Aaron, J. E. (2003). Bare English-origin nouns in Spanish: Rates, constraints and discourse functions. Language Variation and Change, 15, 289-328.

Treffers-Daller, J. (1994). Mixing two languages: French-Dutch contact in a comparative perspective. Berlin, Germany: Mouton de Gruyter.

Treffers-Daller, J. (2005). Evidence for insertional codemixing: Mixed compounds and French nominal groups in Brussels Dutch. International Journal of Bilingualism, 9, 477–508.

Trudgill, P. (1981). Linguistic accommodation: Sociolinguistic observations on a sociopsychological theory. In R. Hendrick, C. Masek & M. F. Miller (Eds.), Papers from the Parasession on Language and Behavior (pp. 218–237). Chicago: Chicago Linguistics Society.

Trudgill, P. (1986). Dialects in contact. Oxford, UK: Blackwell Publishing.

Umaña Aguilar, J. (1981). Variable vibrants in middle-class Costa Rican Spanish. Unpublished Master’s Thesis, Georgetown University.

Urioste, J. (1966). Transcripciones Quechuas. Cochabamba: Instituto de Cultura Indígena.

Valdés, G. (2001). Heritage language students: Profiles and possibilities. In J. Kreeft Peyton, D. A. Ranard & S. McGinnis (Eds.), Heritage languages in America: Preserving a national resource (pp. 37-77). McHenry, IL: Center for Applied Linguistics.

Valdés, G., González, S. V., López García, D., & Márquez, P. (2003). Language ideology: The case of Spanish in departments of foreign languages. Anthropology and Education Quarterly, 34, 3-26.

Valdés Kroff, J. (2016). Mixed NPs in Spanish-English bilingual speech: Using a corpus-based approach to inform models of sentence processing. In R. Guzzardo Tamargo, C. Mazak, & M. C. Parafita Couto (Eds.), Spanish-English Codeswitching in the Caribbean and the U.S., (pp. 281-300). Amsterdam, Netherlands: John Benjamins.

Varela, S. (1992). Verbal and adjectival participles in Spanish. In C. Laeufer & T. A. Morgan (Eds.), Theoretical analyses in Romance linguistics (pp. 219-234). Amsterdam, Netherlands: John Benjamins.

Vásquez Carranza, L. M. (2006). On the phonetic realization and distribution of Costa Rican Rhotics. Filología y Lingüística, 32, 291-309.

Veltman, C. (1988). The Future of the Spanish Language in the United States. Berlin, Germany: Mouton

Vergara Wilson, D. (2013). One Construction, Two Source Languages: Hacer with an English Infinitive in Bilingual Discourse. In A. M. Carvalho & S. Beaudrie (Eds.), Selected

379 Proceedings of the 6th Workshop on Spanish Sociolinguistics (pp.123-134). Somerville, MA: Cascadilla Proceedings Project.

Vergara Wilson, D., & Dumont, J. (2015). The emergent grammar of bilinguals: The Spanish verb hacer ‘do’ with a bare English infinitive. International Journal of Bilingualism, 19, 444-458.

Verschik, A. (2008). Emerging bilingual speech: From monolingualism to Code-copying. London, UK: Continuum International Publishing.

Verteegh, K. (2009). Loan verbs in Arabic and the DO-construction. In E. Al-Wer & R. de Yong (Eds.), Arabic dialectology: In honor of Clive Homes on the occasion of his sixtieth birthday (pp. 187-200). E.J. Brill: Leiden.

Villa, D. J. (2002). The Sanitizing of U.S. Spanish in Academia. Foreign Language Annals, 35, 222-230.

Waddell, A. G. (1961). British Honduras. Oxford, UK: Oxford University Press.

Wardhaugh, R. (2006). An introduction to sociolinguistics. Oxford, UK: Blackwell Publishing

Wassink, B. A. (1999). Historic low prestige and seeds of change: Attitudes toward Jamaican Creole. Language in Society, 28, 57-92.

Wasow, T. (1977). Transformations and the Lexicon. In P. Culicover, T. Wasow, and A. Akmajian (Eds.), Formal Syntax (pp. 327-360). New York, NY: Academic Press.

Weston, D. (2013). Code-switching variation in Gibraltar. International Journal of Bilingualism, 17, 3-22.

Wichmann, S., & Wohlgemuth, J. (2008). Loan verbs in a typological perspective. In Stolz, T., Palomo, R., Bakker, D., (Eds.), Aspects of Langauge Contact. New Theoretical, Methodological and Empirical Findings with Special Focus on Romancisation Processes (pp. 89-121). Mouton de Gruyter: New York, NY.

Widdison, K. (1998). Phonetic motivation in Spanish trills. Orbis: bulletin international de documentation linguistique, 140, 51-61.

Willis, E. (2006). Trill variation in : an acoustic examination and comparative analysis. In N. Sagarra & A. J. Toribio (Eds.), Selected Proceedings of the 9th Hispanic Linguistics Symposium (pp. 121-131). Somerville, MA: Cascadilla Proceedings Project.

Willis, E. (2007). An acoustic study of the pre-aspirated trill in narrative Cibaeño Dominican Spanish. Journal of the International Phonetic Association, 37, 33-49.

380 Willis, E.W., & Bradley, T.G. (2008). Contrast Maintenance of Taps and Trills in Dominican Spanish: Data and Analysis. In L. Colantoni & J. Steele (Eds.), Selected Proceedings of the 3rd Conference on Laboratory Approaches to Spanish Phonology (pp. 87-100). Somerville, MA: Cascadilla Proceedings Project.

Wilson Vergara, D. (2013). One Construction, Two Source Languages: Hacer with an English Infinitive in Bilingual Discourse. In A. M. Carvalho, and S. Beaudrie (Eds.), Selected Proceedings of the 6th Workshop on Spanish Sociolinguistics (pp. 123-134). Somerville, MA: Cascadilla Proceedings Project.

Winford, D. (2003). An introduction to contact linguistics. Malden, MA: Blackwell Publishers.

Winford, D. (2010). Contact and Borrowing. In R.Hickey (Ed.), The Handbook of Language Contact (pp. 188-208). Malden, MA: Blackwell Riley.

Winford, D. (2009). On the unity of contact phenomena and their underlying mechanisms. In L. Isurin, D. Winford, and K. de Bot (Eds.), Multidisciplinary Approaches to Code- switching (pp. 79-306). Amsterdam, Netherlands: John Benjamins.

Winford, D. (2013). Social factors in contact languages. In P. Bakker and Y. Matras (Eds.), Contact Languages: A comprehensive Guide (pp. 362-416). Berlin: De Gruyter Mouton.

Woods, L. A., Perry, J., & Steagall, J. W. (1997). The Composition and Distribution Of Ethnic Groups In Belize: Immigration And Emigration Patterns, 1980-1991. Latin American Research Review, 32, 63-88.

Yiakoumetti, A., & Esch, E. (2010). Educational complexities inherent in bidialectal communities and the potential contribution of the Common European Framework of Reference to second-dialect development. In B. O’Rourke & L. Carson (Eds.), Language learner autonomy: Policy, curriculum, classroom (pp. 291-312). Oxford, UK: Peter Lang.

Young, C. (1973). Belize Creole: A Study of the Creolized English Spoken in the City of Belize in its Cultural and Social Setting. Unpublished doctoral dissertation, University of York, England.

Zentella, A. C. (1997). Growing up bilingual. Malden, MA: Blackwell.

Zentella, A.C. (1981). Hablamos los dos. We speak both: Growing up bilingual in el Barrio. PhD thesis, University of Pennsylvania.

Zhou, X., Espy-Wilson, C. Y., Boyce, S., Tiede, M., Holland, C., & Choe, A. (2008). A Magnetic Resonance Imaging-Based Articulatory and Acoustic Study of ‘Retroflex’ and ‘Bunched’ American English /r/. Journal of the Acoustic Society of America, 123, 4466- 4481.

381 Zimmer, T. (2011). El español hablado por los afrocostarricenses: estudio lingüístico y sociolingüístico. Hessen, Germany: Kassel University Press.

382 BIOGRAPHICAL SKETCH

Osmer Balam’s research program focuses on the sociolinguistic examination of language contact phenomena and bi/multilingualism. Osmer’s scholarly work on Northern Belizean

Spanish and bi/multilingual CS has appeared in several scholarly journals and edited volumes.

After receiving a Bachelor of Arts in English from the University of Belize and a Master of Arts in TESOL from Southern Illinois University-Carbondale, Osmer completed his Ph.D. in

Romance Languages from the University of Florida in the summer of 2016.

383