Aspects of the Syntax of the Afro-Bolivian Spanish Determiner Phrase

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of

Philosophy in the Graduate School of The Ohio State University

By

Sandro Sessarego, M.A.

Graduate Program in Spanish and Portuguese

The Ohio State University

2010

Dissertation Committee:

Terrell A. Morgan, Advisor

Javier Gutiérrez-Rexach, Advisor

Donald Winford

Abstract

During the last decade, syntactic microvariation analysis, the study of grammatical differences across closely related dialects (cf. Kayne 1996), has received increasing attention. Syntactic microvariation is primarily concerned with the possibility of testing syntactic hypotheses and potential correlations between syntactic variables

(Barbiers & Cornips 2001). So far little research has been carried out on microvariation across Spanish dialects, especially across Afro-Hispanic contact vernaculars. From a theoretical point of view, what is fascinating about these languages is their richness in constructions, that would be considered ungrammatical in standard Spanish, but which form the core grammar of these less-prestigious, but equally efficient, syntactic systems.

The present study aims to explore some of the microsyntactic differences encountered between standard Spanish and ABS; in particular, it focuses on their nominal domain (their Determiner Phrase). The methodology adopted combines generative and sociolinguistic insights. The collection of the data included sociolinguistic interviews

(Labov 1972) and grammaticality judgments. These two different strategies were combined in order to obtain more fine-grained generalizations on the structure of this language.

The dissertation has two main objectives. The first one is to offer a syntactic description of the Afro-Bolivian Spanish (ABS) Determiner Phrase (DP) in order to

ii provide a testing ground for verifying the feasibility of current linguistic hypotheses and, when appropriate, provide new solutions in light of the empirical data collected. The second goal is to shed light on the origin of ABS by analyzing the available socio- historical and linguistic evidence.

This work has theoretical and methodological significance. From the theoretical point of view, it is important because it analyzes Determiner Phrase structures from a microparametric perspective. On the other hand, it is also methodologically groundbreaking in that it combines quantitative sociolinguistics and generative syntax to explore the nature of syntactic phenomena in a little studied Afro-Hispanic variety.

iii

Dedication

Dedicated to Marino, Gianna, Simona, Amanda, Marina, Lorenzo, Terrell, Don, Javier,

Manuel and Maclobia.

iv

Acknowledgments

This work would not have been possible without the support of several people. I owe my gratitude to all of them.

My deepest gratitude is to my advisors: Terrell A. Morgan and Javier Gutiérrez-

Rexach. They have supported me from the very beginning of my studies at OSU. Their patience and help provided me with the means to teach and do research in the academic environment. I have been very fortunate to work with them. I bothered them endlessly with questions, recommendation requests, meetings and telephone calls for several years now. They have been always there to listen and give advice. I hope to one day be as good of an advisor to my students as Terrell and Javier have been to me.

I am also extremely grateful to Donald Winford. This professor introduced me to Contact Linguistics and made me love this subject from the first day of class. Don provided me with insightful comments, great advice and constructive criticism on several drafts of this dissertation. Even though I have never been officially one of his advisees, he made me always feel as if I were one.

Another professor who I would like to thank is Fernando Martínez-Gil. Besides teaching highly instructive courses on Phonology and Language Change, he was always willing to help me both personally and professionally.

v Special thanks go also to Professor Scott Schwenter who motivated me to do research from the very first course I took with him, Quantitative Sociolinguistics. I am grateful to Scott for the help and advice he gave me since my very first conference presentations and talks.

Professor Rebeka Campos Astorkiza supported me equally and her courses on

Phonetics and Phonology were extremely clear and interesting. Moreover, I literally occupied for almost two years the Phonology Lab, where she teaches, to write this dissertation. Rebeka has always been very patient and able to deal with my books, my papers and my constantly-running coffee machine for all this time!

Professor John Grinstead has been a great professor too. His classes on theoretical linguistics have been very useful and helped me formulate several ideas for this dissertation. In particular, his advice and the articles he gave me provided me with theoretical frameworks able to account for some of the grammatical differences encountered between Afro-Bolivian Spanish and standard Spanish.

I am obviously also indebted to Professor John Lipski. I bothered him with several e-mails concerning Afro-Bolivian Spanish from the very beginning of my

Master’s. He has always been willing to provide me with important data and information.

I am thankful to Professor Lisa Voigt who patiently assisted me in several application steps for scholarships and grants. I am grateful to Professor Fernando

Unzueta for his advice on academia and his knowledgeable insights into Bolivian culture and society.

vi My experience at OSU would have been different and probably not even possible at the PhD level, if I had not received the support of Professor Laura Podalsky, who helped me apply for the Hispanic Linguistic program.

I would like to thank several other professors from which I could learn a lot and receive personal and professional support during the time spent here, namely: Charles

Klopp, Janice Aski, Peter Culicover, Luciano Farina, Cynthia Clopper, Wayne

Redenbarger, Mary Rose, Kathryn Campbell-Kibler, Anton Ryttin, Felix Julca Guerrero, and Laura Wagner.

I must thank Lorenzo Sangiacomo, my tireless travel companion and great friend, who accompanied me during several Bolivian journeys. Without his help I would never have been able to collect the data on which the present work is based.

I am also grateful to José Luis Delgado (Pulga) and Sara Busdiecker, who offered me accommodation during my time spent in Los Yungas, and all the inhabitants of Tocaña, Mururata and Chijchipa, who kindly welcomed me into their communities and let me bother them with questions and interviews for years. In particular, I would like to acknowledge Manuel, Maclobia, Reyna, Maricruz, Juana, Desiderio, Raymunda, Diego,

Daniel, Daiana, Meagerly, Fabricio, Persi, Marta, Saturnine, Angelica, Angélia, José,

Juan, Francisco, Johnson, Dainor, Rafael, Víctor, Wilmer and Sandra.

Many friends in Columbus have helped me during these years too. I must thank

Olin, Mike, Wendy, Pablo, Angela, L’André, Manuel, Melvin, Lina, Lorena, Meghan,

Jennifer, Mary, Patrick, Douglas, Pilar, Whitney, Samuel, Diego, Marita, María, María

Teresa, Kristin, Kathleen, Alejandro, Jason, Maribel, Caitlin, Ana, Christine, Kristina,

Freda, Carmen, Fernando, Eliana, Thomas, Juliana, Christina, Monica, Alessia,

vii Gerardina, Rachael, Anna, Lynn, Concetta, Kimberly, Mariana, Isabel, Mario, Jared,

Amanda, Michael, David, Nash, Patricia, Laura, Alondra, Robey, Raquel, Nyanda, Petra,

Anita, Marissa, Asela, Daniela, Catherine, Ian, Hanna, Andy, Bethany, John, Josh,

Antoine, Tatiana, Evelyn, Jorge, Sarah, Madalena, André, Kyle, Carter and Isaí.

Also, I would like to thank the Ohio State University Office of International

Affairs, The College of Humanities, the Department of Spanish and Portuguese the Chu family and the Kubayanda family for the financial support I received for this research.

Finally, none of this would have been possible without the support of my family

(Simona, Marina, Gianna and Marino), my girlfriend (Amanda) and some people who have been very close to me and supported me in unbelievable ways: Rául Diego Rivera

Hernández, Sarah Finley, Mario Rodríguez Polo, Isabel Arranz, Patricia González,

Ignacio Sanz-Valero, Ivano Fulgaro, Mariano Escobedo Ávila, Juliana De la Mora,

Daniela Salcedo, Giovanni Cristina, Carlo Forte, Magda Davoli, Morena Lanieri, Claudio

Ferrari, Guido Borghi, Sara Rossini, Carlos and Midori Pimentel, Terry and Brian

Talbert, Teresa Bombrys, Yassir Shekaldin, Marta Tallone, Pablo Chignoli and Mariana

Lacunza. Thank you!

viii

Vita

2005...... B.A. Languages for Business, Università

degli studi di Genova

2007...... M.A. Languages for International

Communication, Università degli studi di

Genova

2008...... M.A. Hispanic Linguistics, The Ohio State

University

2008- present ...... Graduate Teaching Assistant, The Ohio State

University

Publications

Books

Sessarego, Sandro. In press. Introducción al idioma afro-boliviano: Una conversación con el awicho Manuel Barra. Cochabamba/ La Paz: Plural Editores.

Papers

Gutiérrez-Rexach, Javier. & Sessarego, Sandro. In press. Nominal reference and dialect variation in Spanish. In Selected proceedings of Res per Nomen: La référence, la conscience et le sujet énonciateur . Université de Reims, Reims. Nº 3.

Sessarego, Sandro. In press (a). On the status of Afro-Bolivian Spanish features: Decreolization or vernacular universals? In Selected proceedings of the 5th International Workshop on Spanish Sociolinguistics (WSS5). Somerville, MA: Cascadilla.

ix Sessarego, Sandro. In press (b). On gender and number agreement in the Determiner Phrase: The Afro-Bolivian Spanish Case. In Sintagma . Nº 22.

Sessarego, Sandro. In press (c). On the supposed creole origin for Yungueño Spanish: Linguistic and sociohistorical considerations. In Selected proceedings of the XXVII Biennial Louisiana Conference on Hispanic Languages and Literatures. Louisiana State University, Baton Rouge, Louisiana.

Sessarego, Sandro. In press (d). A phonetic analysis of /sr/ clusters in Cochabambino Spanish. In Selected proceedings of The 12th Hispanic Linguistic Symposium . Somerville, MA: Cascadilla.

Sessarego, Sandro. & Gutiérrez-Rexach, Javier. In press (a). Gender agreement in Afro- Bolivian Spanish: syntax, variation and evolution. Cuadernos de Lingüística de la Universidad de Puerto Rico. Universidad de Puerto Rico, Rio Piedra.

Delicado-Cantero Manuel. & Sessarego, Sandro. In press. Variation and syntax in number expression in Afro-Bolivian Spanish. In Selected proceedings of the 12th Hispanic Linguistic Symposium. Somerville, MA: Cascadilla.

Sessarego, Sandro. & Gutiérrez-Rexach, Javier. In press (b). A minimalist approach to gender agreement in the Afro-Bolivian DP: Variation and the specification of uninterpretable features. In Folia Linguistica .

Sessarego, Sandro. 2010. Temporal Concord and Latin American Spanish Dialects: A Genetic Blueprint. In Revista Iberoamericana de Lingüística . N.º 5, 137-169.

Sessarego, Sandro. 2009 (a). On the evolution of Afro-Bolivian Spanish subject-verb agreement: Variation and change. In Sintagma . Nº 21, 107-119.

Sessarego, Sandro. 2009 (b). Gender agreement variation in the Afro-Bolivian Determiner Phrase: The Interplay of Social and Linguistic Factors. In Revista Iberoamericana de Lingüística . Nº 4, 135-146.

Sessarego, Sandro. 2008 (a). La concordantia temporum y el español andino: Soluciones cuantitativas. In Página y signos , Nº 4, 65-84.

Sessarego, Sandro. 2008 (b). Spanish concordantia temporum : An old issue, new solutions. In Selected proceedings of the 4th International Workshop on Spanish Sociolinguistics (WSS4). Somerville, MA: Cascadilla. 91-99.

x

Field of Study

Major Field: Spanish and Portuguese.

xi

Table of Contents

Abstract………………………………………………………………………………...….ii

Dedication………………………………………………………………………………...iv

Acknowledgments………………………………………………………………………...v

Vita………………………………………………………………………………………. ix

List of Tables……………………………………………………………………...…...xviii

List of Figures………………………………………………………..………………….xix

Chapter 1: Introduction……………………………………………………………………1

1.0. General introduction………………………………………………………….1

1.1. Objectives of this study………………………………………………...... 3

1.2. Organization of this volume…………………………………………………..3

1.3. Theoretical framework………………………………………………………..5

Chapter 2: Language Variation and the Minimalist Program……………………………13

2.0. Introduction………………………………………………………………….13

2.1. Accounting for variation…………………………………………………….14

2.2. Approaches to the study of language variation………………………...... 15

2.2.1. The formal approach………………………………………………15

2.2.2. The sociolinguistic approach………………………………...... 17

2.2.3. The diachronic approach………………………………………..…19

xii 2.2.4. The microparametric approach……………………………………21

2.2.5. Variability in the Minimalist Program…………………………….23

2.3. Variability across closely related dialects: The Afro-Hispanic case………..27

2.4. Data collection and methodology……………………………………...... 28

2.5. Conclusions………………………………………………………………….30

Chapter 3: A Sociohistorical Account……………………………………………...... 31

3.0. Introduction…………………………………………………………………31

3.1. Afro-Hispanic contact varieties and the relative paucity of Spanish

creoles……..……………………………………………………………...... 32

3.2. On the origin of Afro-Bolivian Spanish…………………………………….39

3.2.1. A sociohistorical sketch of slavery in Bolivia…………………….40

3.2.2. Slavery in Bolivia: the 16 th and mid-17 th centuries……………….43

3.2.2.1. Routes of arrivals and origin of the slaves………………49

3.2.2.2. The role of slaves in the Bolivian economy…………..…52

3.2.3. Slavery in Bolivia: from the mid-17 th to the mid 20 th

century……………………………………………………………………59

3.2.3.1. New routes of arrivals and origin of the slaves………….60

3.2.3.2. The role of slaves in the Bolivian economy……………..62

3.3. La Villa de Mizque………………………………………………………….63

3.4. Los Yungas………………………………………………………………….66

3.4.1. The nature of Slavery in the Yungan hacienda and in

Bolivia……………………………………………………………………71

3.5. Conclusions………………………………………………..………………...74

xiii Chapter 4: Accounting for the Afro-Bolivian “creole-like” features…………………….76

4.0. Introduction………………………………………………………………….76

4.1. Afro-Bolivian Spanish “creole-like” features……………………………….79

4.2. The Interlanguage Hypothesis………………………………………………93

4.2.1. Processability Theory…………………………………………..….96

4.2.2. Accounting for the presence of ABS “creole-like”

features……..…………………………………………………………….98

4.3. Vernacular Universals…………………………………………………...…107

4.4. Conclusions……………………………………………………………...…111

Chapter 5: From NP to DP……………………………………………………………...113

5.0. Introduction………………………………………………………………..113

5.1. The Determiner Phrase Hypothesis………………………………………..114

5.1.1. Abney (1987)…………………………………………………….114

5.1.2. Szalbolcsi (1983; 1987; 1994)………………………………...…118

5.2. In between N and D……………………………………………………….121

5.2.1 Ritter (1991; 1993)……………………………………………….122

5.2.2. Picallo (1991)………………………………………………….…126

5.2.3. Bernstein (1993)…………………………………………………127

5.2.4. Cinque (1993)………………………………...……………….…128

5.2.5 Giusti (1993)…………………………...... …..………………..129

5.2.6. Mallén (1997)…………………………………………….………130

5.2.7. Giorgi and Longobardi (1990)……………..………………...... 132

5.2.8. Valois (1991)……………………………………………….……137

xiv 5.2.9.Longobardi (1994)……...... ………………………...... 140

5.3. Quantificational phrase…………………………………………………….143

5.3.1. Zamparelli (2000)………………………………………………..144

5.3.2. Ishane (2008)…………………………………………………….148

5.4. Light ‘n’ (Radford 2000)……………….………………………………….150

5.5. DP prolific domains (Ticio 2003)………………………………………….151

5.6. Conclusions………………………………………………………………...154

Chapter 6: Phi-features: The Syntax of Person, Number and Gender………………….155

6.0. Introduction………………………………………………………………...155

6.1. Default values, Markedness, Agreement and Agree Operation……………156

6.1.1. Agree vs. Agreement…………………………………………….156

6.1.2. Default values and Markedness………………………………….163

6.2. Applying the theory to the data……………………………………………166

6.3. Phi-features distribution across the Afro-Bolivian Spanish and the Highland

Bolivian Spanish DPs………………………………………………………..…174

6.3.1. Nouns…………………………………………………………….176

6.3.2. Num………………………………………………………………178

6.3.3. Subject pronouns…………………………………………………179

6.3.4. Object Pronouns………………………………………………….182

6.3.5. Indefinite and negative pronouns……….………………………..186

6.3.6. Possessive determiners……………....…………………………...187

6.3.7. Possessive Adjectives……………………………………………191

6.3.8. Other Adjectives and Quantifiers………………………………..194

xv 6.3.9. Indefinite articles………………………………………………...196

6.3. 10.Definite articles…………………………………………………197

6.3.11. Demonstratives…………………………………………………199

6.3.12. Summary…………………………………………..……………201

6.4. Microparametric research and phi-feature variability ………………..……205

6.4.1. The grammar of gender and number agreement…………………206

6.4.2. Syntax matters: quantitative evidence for number

variability……………………………………………………………….212

6.5. Previous analysis………………………………………………………...…214

6.6. A Minimalist Analysis…………………………………………………..…219

6.7. Conclusion……………………………………...………………….………225

Chapter 7: Nominal Ellipsis………………………………………..…………………...227

7.0. Introduction……………………………………………………………...…227

7.1. The Data……………………………………………………………………229

7.2. Prepositions uses in ABS and stSp……………………………………...…245

7.3. Previous analyses……………………………………………………..……250

7.3.1. Brucart and Gràcia (1986)…………………………………….…250

7.3.2. Brucart (1987)……………………………………………………254

7.3.3. Torrego (1988)………………………………………………...…261

7.3.4. Ticio (2005)…………………………………………………...…265

7.3.5. Eguren (2010; to appear)………………………………………...274

7.3.6. Braver (2009)……………………………………………………280

7.4. Toward a new proposal……………………………………………………286

xvi 7.5. Conclusions………………………………………………………………..286

Chapter 8: Bare Nouns: Semantic and Syntactic Properties………………………..297

8.0. Introduction……………………………………………………….………..298

8.1. The Nominal Mapping Parameter……………………………………….…298

8.2. Longobardi’s proposal…………………………………………………..…298

8.3. Where to place ABS in the Nominal Mapping Parameter typology……….301

8.4. Bare singulars in ABS………………………………………………...……306

8.5. The semantics of indefinite and definite articles………………………..…310

8.5.1. Indefinite articles……………………………………………...…314

8.5.2. Definite articles………………………………………………..…315

8.6. Number and mass interpretations………………………………………..…322

8.7. The syntax of bare nouns…………………………………………..………328

8.8. Bare Nouns in ABS: A Unified Account ……………………………….…331

8.9. Conclusions……………………………………………………………...…335

Chapter 9: Conclusions…………………………………………………………………337

References………………………………………………………………………………348

xvii

List of Tables

Table 1. Demographic figures for the city of Potosí (Crespo1977)…………………..….53

Table 2. Bolivian population 1650-1950……………………………………………...…60

Table 3. Racial distribution of the population of the provinces of Cochabamba and Santa

Cruz in 1788…………………………………………………………………...... 65

Table 4. Examples of the Afro-Bolivian features salient for the creole hypothesis…..…80

Table 5. Key structural features of Afro-Bolivian Spanish and selected creole languages

(Lipski 2009:184)…………………………………………………………………...……81

Table 6. Developmental stages in English interlanguage syntax……………………...…97

Table 7. Processing procedures for English………………………………………….…..97

Table 8. Articles production per type by Heike (Leonini 2006:112)…………………...100

Table 9. Phi-features and values in the ABS DP………………………………….……163

Table 10. Percentages of /s/ deletion by grammatical status in ABS……………..……213

Table 11. Contribution of Grammatical Category, Following Phonological Segment and

Following Stress to Deletion of Plural /s/ in ABS………………………………...... …213

xviii

List of Figures

Figure 1. Map of Bolivia…………………………………………………………………45

Figure 2. A royal official helped by his slave punishing an Indian community leader

(Poma de Ayala 1615)…………………………………………………………………...47

Figure 3. Routes of slave arrival…………………………………………………………51

Figure 4. Accuracy of gender agreement on definite and indefinite articles (Bruhn de

Garavito & White 2000:172)…………………………………………………...………104

Figure 5. Accuracy of gender agreement on determiners and adjectives (Bruhn de

Garavito & White 2000:172)……………………………………………………...……105

xix

CHAPTER 1: INTRODUCTION

“Comparative work on the syntax of a large number of

closely related languages can be thought of as a new

research tool, one that is capable of providing results of

an unusually fine-grained and particularly solid

character ” (Kayne 1996:xii).

1.0. General introduction

During the last decade, the study of syntactic microvariation across closely related dialects has received increasing attention. Syntactic microvariation is primarily concerned with the possibility of testing syntactic hypotheses and potential correlations between syntactic variables (Barbiers & Cornips 2001). In fact, microcomparative analysis provides researchers with an almost ideal language laboratory (Kayne 1996), by bringing to the forefront previously disregarded phenomena - considered as belonging to

E(xternal)- language.

The study of language variability seems at first to be at odds with formal linguistic analyses that strive to hypothesize principles and generalizations based on

Universal Grammar allowing a very limited range of cross-linguistic variation (Chomsky

1957; 1965). Nevertheless, in the Principles and Parameters/Minimalist framework

1 (Chomsky 1995), several attempts were made to capture dialectal and inter-speaker variation, exploring the notion of parametric variation and, in the last decade, fine-tuning it to include so-called “micro parameters” (Benincà 1989; Kayne 2000; etc.).

Recent syntactic dialect atlas projects, in particular, have undertaken such a

challenge. Two examples of current European projects directly aimed to study

microvariation are the Northern Italian syntactic dialect atlas (ASIS) and the syntactic

atlas of Dutch dialects (SAND) (see Barbiers and Cornips 2001) 1. So far, little attention has been paid to the study of microvariation across Spanish dialects, especially to microparametric syntax across Afro-Hispanic contact varieties. From a theoretical point of view, what is fascinating about these languages is their richness in constructions which would be considered ungrammatical in standard Spanish, but which form the core grammar of these less-prestigious, but equally efficient, syntactic systems.

The majority of the dialects which emerged in Latin America from the contact of

African languages and Spanish at the time of slavery are not “radical creoles”, languages highly influenced by substrate patterns which would be unintelligible for a standard

Spanish (stSp) speaker; rather, they often consist of comprehensible vernaculars with a comparatively reduced inflectional morphology. Besides, from a purely linguistic point of view, these dialects might be considered even more natural systems, as they contain constructions which survived leveling phenomena imposed elsewhere by standardization processes (Weiß 2001).

To indicate the nature of their situation, half-way between ‘radical-creoles’ and standard systems, the terms ‘semi-creoles’ (Holm 1992) and ‘partially restructured

1 More information about the ASIS and SAND projects can be found at: http://asis-cnr.unipd.it and http://meertens.nl/projecten/sand/sandeng.html , respectively. 2 languages’ (Holm 2004) have been proposed in the literature on contact linguistics. Their close relatedness to the standard can provide linguists with a great empirically-based testing ground for formal hypotheses.

1.1. Objectives of this study

This work has a two-fold aim. The first objective is to provide a syntactic

description of the Afro-Bolivian Spanish (ABS) Determiner Phrase (DP). This dialect

presents phenomena which offer a real challenge to current linguistic theory. For this

reason, the main aim of this dissertation is to accurately explore such structures to

provide a testing ground for verifying the feasibility of current linguistic hypotheses, and

when appropriate, to propose new solutions in light of the empirical data collected.

The second goal is to shed light on the origin of ABS by analyzing the available

socio-historical data as well as the linguistic evidence found in this language.

Particularly, I will test whether a creole hypothesis (Lipski 2009) can be proposed for this

language or a different explanation should be provided instead.

1.2. Organization of the volume

Chapter 1 consists of a general overview of the goals of this work. It introduces

the present study’s objectives, as well as the theoretical framework assumed for

analyzing the data.

Chapter 2 illustrates how such a framework is capable of accounting for language

variability. This chapter elaborates on the importance of the present work in linking

formal theory to sociolinguistic methodology, in line with recent micro-parametric

3 studies (see for example Poletto 2000; Benincà 1989; Adger and Smith 2005; Barbiers and Cornips 2001). In doing this, chapter 2 provides a description of the methodology employed for the data collection.

Chapter 3 is an analysis of the socio-historical conditions under which Afro-

Hispanic contact varieties developed in Latin America, it includes the main debates and argumentations concerning the relatively reduced number of Spanish creoles. In this context, it places in space and time Afro-Bolivian Spanish by focusing on the socio- historical scenario which characterized the evolution of this Afro-Hispanic contact variety.

Chapter 4 discusses the linguistic features of Afro-Bolivian Spanish in hopes of clarifying which processes might have been responsible for their presence in this language. It also attempts to combine the linguistic and the sociohistorical data previously analyzed to shed light on the origin of the dialect.

Chapter 5 offers a description of the most important issues concerning the generative study of the nominal domain during the last thirty years. It departs from the main assumptions behind the stipulation of the DP Hypothesis to touch on several studies which have contributed to the evolution of syntactic theories of DP. In doing this,

Chapter 5 provides an outline of frameworks and ideas which will be evaluated based on the empirical data encountered for ABS.

Chapter 6 is an in-depth descriptive and analytical work on the realization of phi- features across ABS and stSp DPs 2. It provides quantitative and qualitative data on their

2 Throughout this work I will use the terms standard Spanish (stSp) and Highland Bolivian Spanish (HBS) interchangeably. 4 distribution, concluding with a formal proposal to account for the specific phenomena encountered in this Afro-Hispanic vernacular.

Chapter 7 surveys NP-ellipsis in stSp and ABS. It analyzes the differences and similarities encountered in these two languages. Results will provide a testing ground for previous theoretical models developed exclusively on stSp data. Microparametric findings will also allow us to build novel generalizations to account for the cross- linguistic facts.

Chapter 8 provides a detailed account of bare nouns in subject and object position.

It tests Chierchia’s (1998) Nominal Mapping Parameter and speculates as to the nature of the ABS nominal domain in light of Longobardi’s (1994) generalization on the structure of DPs.

Chapter 9 consists of the final conclusion.

1.3. Theoretical framework

The primary theoretical framework adopted in this study is the one provided by

the Minimalist Program (Chomsky 1995, 2000, 2001). The central concept of the

minimalist framework is Economy. A theory of grammar has been built around this

notion. The Minimalist Program tries to achieve descriptive and empirical adequacy

while keeping to a minimum theoretical abstractions and stipulations.

According to this model, the component of the human mind devoted to language,

the language faculty, is optimal; it is defined by a small number of very general syntactic

operations (Merge, Move and Agree) and it is common to all human beings. The cyclical

application of the Merge and Move builds constituent structure.

5 The operation Merge selects two elements from the collection of lexical items

(Numeration) and assembles them. The operation Move consists of the operations Merge and Copy. When Move applies, the Computational System (CS) creates a copy of a certain element and merges it in a different part of the syntactic structure.

The syntactic constituent must receive an overt form; this overt realization occurs at Spell-Out, where computations split and derive two independent representations, Logic

Form (LF) and Phonetic Form (PF). LF and PF belong to the Articulatory-Perceptual (A-

P) and the Conceptual-Intentional (C-I) systems respectively. Restrictions on linguistic well-formed objects are conceived as restrictions on these two interfaces.

Agree, on the other hand, does not create constituent structure. The operation

Agree is a formal mechanism for valuation of certain features (unvalued) and deletion of others (uninterpretable) in the narrow syntax (in line with Pesetsky and Torrego 2007).

In fact, in the most recent formulations of the Minimalist Program (Chomsky

2001, 2002, 2006), syntactic derivations are viewed as strictly dependent on feature valuation and checking. The distinction between interpretable and non-interpretable features, though controversial, has proven very useful. Several features have an interpretation at Logic Form (LF), thus they are semantically-interpretable features. Other features, however, lack such semantic import and are present to trigger the necessary merger or agreement operations during the derivation. Said uninterpretable features have to be matched via Agree and are finally deleted before Spell-Out.

6 (1) CS

Merge, Move, Agree LF C-I

Lexicon Spell Out

PF A-P

Therefore, once a collection of lexical items (Numeration) is selected from the

Lexicon, the Computational System leads the computation till Spell-Out. It creates each linguistic expression via two operations: Merge and Move. The syntactic items created in this way undergo Agree (when needed); subsequently they are shipped to the two interfaces. Once Spell-Out has applied, to the Computational System does not have access to the information contained in the structure and the derivation follows unconnected paths.

Along the A-P path, several processes (e.g. morphological processes) apply, leading to a PF representation. On the other hand, the processes applied along the C-I path contribute to the semantic interpretation of the syntactic structure and lead to a LF representation. A well-formed linguistic object is an object which meets the requirements imposed at these two interfaces. In case such requirements are not met, the derivation crashes resulting in an ill-formed linguistic object.

The framework assumes that syntactic features come in two types: those with a

semantic interpretation (such as the feature [past]), and those with a purely syntactic

function (for example, nominative case on a nominal, or agreement marking on a verb).

All features whether interpretable or uninterpretable must be checked (or valued) during

the syntactic derivation.

7 An element drawn from the lexicon with a valued feature will be specified as

[u(ninterpetable)F(eature):val] if the feature is uninterpretable and as

[i(nterpetable)F(eature):val] if the feature has an interpretation at LF 3. A feature of the

same kind that has not participated in a valuation process and is not already valued is

annotated (where relevant) by an empty pair of brackets: [uF: ] (or [iF: ] ) and, after valuation takes place, it turns into [uF :val] (or [iF :val] ).

Before presenting how the system is supposed to work, I should lay out some general assumptions. It is generally accepted that number interpretation is located in the

Num head (Ritter 1991) and that N-raising to Num derives the noun+adjective order typical of Spanish and other Romance languages. To account for the fact that not all adjectives are post-nominal in Spanish and Romance, it has been suggested that some may be generated in the specifier of Num or other functional projections between D and

Num (Cinque 1993). For now, let us consider the derivation of a simple Spanish DP (2).

3This is in line with Pesetsky & Torrego (2007) who argue against Chomsky’s (2001) Valuation/Interpretability Biconditional assumption, here reported as (a):

(a) Valuation/Interpretability Biconditional (Chomsky (2001:5)) A feature F is uninterpretable iff F is unvalued.

The elimination of (a) allows lexical items to come from the lexicon with features that display two combinations of properties not contemplated by the Minimalist Inquiries (Chomsky 2000) / Derivation by Phase (Chomsky 2001) (MI/DbP) theory: (i) uninterpretable but valued; and (ii) interpretable but unvalued. Therefore, the types of features encountered in the system are now those reported in (b):

(b) Types of features (boldface = disallowed in MI/DbP) uF val uninterpretable, valued iF val interpretable, valued uF [ ] uninterpretable, unvalued iF [ ] interpretable, unvalued

8 (2)

Esas pequeñas casas rojas

This-F-Pl small-F-Pl house-F-Pl red-F-Pl

‘These small red houses’

In order to simplify the illustration of this case, I assume that at the beginning of the syntactic computation we have six elements entering the Numeration. The entries selected from the Lexicon are a Determiner ( Es-), a Noun ( casa-), two Adjectives ( roj-,

pequeñ-), and the syntactic entries Num and n.

(3)

{D, N, A1, A2, Num, n }

The elements in (3) will be combined by the Computational System by using the

only two operations available to create syntactic constituents: Merge and Move. Agree

will also apply to value unvalued features and to erase uninterpretable ones (not readable

at LF) before Spell-Out.

At the point of merge, the noun casa- ‘house’ carries an interpretable valued

gender feature, an uninterpretable valued number feature and an uninterpretable valued

person feature. N bares also an unvalued Case feature; however, as Case is not relevant to

our discussion, this last feature will not be included in the present representation. Note

that the symbol “+” next to the value indicates that the item is entering the derivation

with a certain feature valued e.g. [ igen:+F].

9 (4)

n’

n NP casa- [igen:+F] [unum:+Pl] [upers:+3 rd ]

The noun will raise and adjoin to the n, probably prompted by the strength of its

categorial feature. In this way, N leaves an invisible copy behind. Subsequently, n

projects a specifier in which the first AP is merged. The adjective roj - ‘red’, which bears uninterpretable unvalued gender and number features, represents the AP’s head. The uninterpretable unvalued features on A act as probes looking for a goal in their local c- command. The noun values the adjective’s gender and number features, which can now act as a goal (Frampton and Gutmann 2000) and which will be deleted before Spell-Out.

(5)

After this step, the nP will merge with the Num head, which carries an interpretable and unvalued plural number feature, able to probe for the value contained in

N. N raises to Num, Num projects a specifier where the second AP can be merged. Its 10 head ( pequeñ -) undergoes the Agree operation previously applied to roj -, so that also its

uninterpretable unvalued gender and number features will be valued and deleted before

Spell-Out.

(6)

Eventually, also the D head is merged and its unvalued phi-features acquire a

value. Its uninterpretable gender and number instances are finally erased, while the

interpretable person one survives.

11 (7)

After the point of Spell-Out, further phonological and logic operations (at PF and

LF respectively) apply to the construction to achieve the final overt message.

12

CHAPTER 2: LANGUAGE VARIATION AND THE MINIMALIST

PROGRAM

“Granting the presence of widespread structured

variation in the use of language (e.g. Labov 1972),

the following question arises for theoretical

linguistics: how is the mental grammar (the I-

language of Chomsky 1986) organized so that such

variation arises?” (Adger & Smith 2005:149).

2.0. Introduction

Traditionally, generative syntactic models have mainly been built on standard language data, on the basis of well-formedness judgments of a limited set of informants.

This approach has proven very powerful in producing an impressive number of generalizations, exactly because it could exclude from its analyses all variability complications due to performance (Barbiers 2009:1608). On the other hand, such a methodology has often been criticized by sociolinguists, who instead base their observations on much larger corpora of naturalistic production data, and have developed several techniques to study the ‘real vernacular’, the language spoken by people when paying no metalinguistic attention to their speech (Labov 1972). In the last decade, works

13 on microvariation attempted to combine these previously contrasting approaches to compare a speaker’s intuitions with real production data, with the goal of developing more fine-grained, empirically-testable generalizations (Cornips and Poletto 2005).

Recent works within the minimalist framework entailed a derivational approach that was inconsistent with parameter-based accounts. Thus, an extension of the parameter/micro- parameter idea to individual variation seems to not be a straightforward matter (Adger &

Smith 2005).

This chapter provides an overview of the major approaches proposed in the literature to deal with inter- and intra-speaker variability. In doing this, it shows the importance of the present work in combining the formal and the sociolinguistic methods to obtain a more fine-grained account of the syntactic phenomena characterizing the

Afro-Bolivian Spanish Determiner Phrase.

2.1 Accounting for variation

Given the presence of widespread language variation, it is legitimate to wonder: how is the mental grammar (the I-language of Chomsky 1986a) organized so that such variation arises? (Adger & Smith 2005:149).

Several answers have been proposed. Traditionally, formal linguists have considered such variability as an instance of E-language, thus not concerning the core syntactic competence of the speaker and therefore, for the most part, ignorable. On the other hand, a different response, radically contrasting with the former, grew within the variationist/sociolinguistic paradigm. It consisted of positing variable rules (e.g. Labov

1972; Cedergren & Sankoff 1974), where probabilities would be built into the notion of

14 grammar. Other models stipulated multiple grammars, where speakers have several parametric configurations (Henry 2005; Kroch 1989; 1994). Finally, a more recent account (Adger & Smith 2005; Adger 2006; Parrott 2007), developed within the architecture provided by the Minimalist Program (Chomsky 1995; 2001; 2006) and the study of micro-parametric variation (Kayne 1996, 2000; Barbiers & Cornips 2001; etc.), postulates that variation is the overt result of covert lexical selections. Such a model claims that there is a reduced available number of syntactic operations which are universal and constant (merge, move and agree), while the elements entering the syntactic numeration can vary in feature specification. The following sessions will provide a brief overview of these competing accounts.

2.2. Approaches to the study of language variation

2.2.1. The formal approach

Traditionally, mainstream syntactic theories (Chomsky 1957; 1965; 1986a) have given little room to the study of inter- and intra- speaker speech variability. Within the formal approach of Principles and Parameters (P&P) (Chomsky & Lasnik 1993), languages can essentially be seen as combinations of a finite set of fundamental principles, which are common to all varieties, and a set of binary parameters that determine syntactic variability among languages. Within this framework, principles are

“language-invariant statements” (Chomsky 1995:25), whereas parameters must be set for certain values. Principles are part of a genetically innate Universal Grammar (UG) which all humans possess. As such, they do not need to be learned by exposure to language.

Rather, exposure to language merely triggers the parameters to adopt the correct setting.

15 The final goal of linguistics is to discover this set of paradigms, which can combine differently to form the I(nternal)-language of each human being.

Chomsky (2000) compares the language faculty to a switch box. This box consists of two components: a fixed network, which are the innate principles of language, and several switches, which are options determined by experience, binary parameters that can be set on or off. Different parametric combinations lead to different grammars. In

Chomsky’s (2000: 8) words:

“When the switches are set one way, we have Swahili; when they are

set another way, we have Japanese. Each possible human language is

identified as a particular setting of the switches—a setting of parameters, in

technical terminology".

Within the field of generative syntax, cases of language internal variation have often been disregarded as instances of E(xternal)-language, not interesting from the perspective of scholars aimed to unveil the secrets of the I-language. As a result, formal syntactic theories have traditionally been built on partly-idealized standard languages, on the basis of well-formedness judgments of a limited set of informants. As van Gelderen

(2005:180-1) pointed out, the use of vast corpora has never flourished among traditional syntactic studies. Wasow (2002) reviews Chomsky’s negative attitudes on the employment of quantitative data, which in his view would not provide any useful insight into the knowledge of the I-language. Here is a quote:

16 “It seems that probabilistic considerations have nothing to do with

grammar, e.g. surely it is not a matter of concern for the grammar of

English that ‘New York’ is more probable than ‘Nevada’ in the context ‘I

come from–’ ” (Chomsky 1962:128).

Needless to say, as we will see, sociolinguistic models of language variation have a radically different opinion on the issue.

2.2.2. The sociolinguistic approach

Formal methodology produced an impressive amount of data, generalizations and insights, exactly because it could avoid all sorts of empirical complications. Yet, for these same reasons, such a model has often been criticized by sociolinguists, who instead based their observations on bigger corpora of naturalistic production data, and developed several techniques to study the ‘real vernacular’, the real language spoken by people when paying no metalinguistic attention to their speech (Labov 1972). In the earliest transformational model (Chomsky 1957), obligatory and optional rules were conceived as transformations that linked deep structure with surface structure. Optional rules, for example, derived passive, negative or question sentences from declarative sentences.

Labov (1969) extended the concept of optional rule to the notion of variable rule to account for extralinguistic factors (e.g. social class, style, etc.). In this way, Labov was able to account for interspeaker and intraspeaker variation (Cornips 2006).

Labov’s ‘variable rules’ are based on the concept of ‘orderly heterogeneity’

(Weinreich et al. 1968:100), the idea that variation in language is not random or free, but

17 systematic and rule-governed (Tagliamonte 2006:129). This notion of variable rules was first developed from the observation that speakers make choices when they use language and that this choice is systematic (Labov 1969). Due to such systematicity, statistical models can be developed, thus introducing a probabilistic component into the theory of language.

The sociolinguistic approach takes the principle of accountability as basic

(Sankoff 1990:296). This principle states that all variants belonging to the same syntactic

variable must be accounted for in the variable environment. In order to determine what

should be considered as a variant of a specific variable, the synonymy principle (principle

of sameness) must be followed. In other words, variants of the same variable are only

those tokens which consists of “alternate ways of saying ‘the same’ thing”.

Sociolinguists collect the speech they study from people engaged in ordinary

activities. Among the practical methods employed to collect data, it is worth pointing out

at least two of them, because these are particularly suitable to the study of

(morpho)syntactic and phonological variations. The Anonymous and Rapid Survey

method (Labov 1966) is indicated for the study of specific content-limited phonological

variation, because the ‘observer paradox’ is almost completely eliminated. The observer

paradox refers to the difficulty of observing how people speak when they are not being

observed (Labov 1972:113), so that they will pay minimum attention to the way they

speak. On the other hand, the sociolinguistic interview is recommended for the analysis

of (morpho)syntactic variation as, in this way, a larger body of data can be collected.

Sociolinguists, like professional researchers in all empirical disciplines, also developed refined sets of research tools that allow them to cope with the demands of their

18 data (Chambers: 2003). One of the most famous statistical instruments employed to analyze sociolinguistic variation is VARBRUL (by David Sankoff), which evolved in several more powerful software programs during the last decades (e.g. GoldVarb X by

Sankoff, Tagliamonte, & Smith 2005). VARBRUL requires the establishment of a basic variant and directionality between variants in order to perform analysis. It calculates the degree to which a given independent factor (or factor group) favors (>.50) or disfavors

(<.50) the process in question, the surface realization of some variable (Tagliamonte

2006).

Within the Labovian/sociolinguistic approach, variation has been at the center of attention for at least thirty years; on the other hand, within the traditional

Chomskyian/generative framework it has generally been excluded from the research agenda, because it was considered to be a case of performance rather that competence.

Nevertheless, in more recent years, scholars working within the generative framework have attempted to account for diachronic variation (Kroch 1989) and also to capture dialectal and inter-speaker variation, exploring the notion of parametric variation and, in the last decade, fine-graining it to include so-called “micro parameters” (Benincà 1989;

Kayne 2000; etc.).

2.2.3. The diachronic approach

Dating from the 1980s, the study of diachronic syntax has seen the rise of a particularly rich tradition of combining quantitative and formal perspectives (see Kroch

1989; Kroch 1994 for an overview). In Kroch’s view the historical evolution of competing variants in syntactic change is similar to the evolution of morphological

19 doublets. In both cases, the coexistence of the variant forms is diachronically unstable:

One form tends to drive the other out of use and thus out of the language (Kroch 1994:7).

The parallelism between the cases would be explained by the existence of the so- called “Blocking Effect” (Aronoff 1976), which excludes morphological doublets, and more generally, it seems, any coexisting formatives that are not functionally differentiated (see Kiparsky 1982). The blocking effect does not prevent doublets from arising. In fact, doublets can enter a language by sociolinguistic means such as dialect and language contact. However, this effect seems to be a global economy constraint on the storage of formatives. According to Kroch, this effect would be inviolable in primary language acquisition by young children; but it can clearly be overridden as speakers learn a wider range of styles and social dialects in the course of maturation. Nevertheless, over long periods of time, the constant pressure of economy on acquisition would win out over sociolinguistic variation in the history of doublet forms. Therefore, morphological doublets occur quite frequently, but they are diachronically unstable so that languages always evolve in such a way that one or the other variant becomes extinct.

Kroch supports his analysis of syntactic change by suggesting the general validity of the “Constant Rate Hypothesis” (Kroch 1989). The “Constant Rate Hypothesis” claims that usage frequencies change rates mirror the gradual replacement of one abstract grammatical option by another. The formulation of the “Constant Rate Hypothesis” represents a key step toward the analysis of E-data. In fact, it clearly states that statistical tendencies and patterns encountered in vast corpora can unveil the principle and parameters belonging to speakers’ competence. See also van Gendeleren (2005) for an account of how the use of corpora has been integrated to syntactic theory.

20 Kroch proposes that variation in the course of syntactic change is between options that are grammatically incompatible and that variation reflects grammar competition. He admits that the use of the notion grammar competition in formal theorizing is usually not well accepted, as it would introduce learnability problems. However, he objects to this claim. In his view, speakers would learn either one or the other form in the course of basic language acquisition. Only later in life, they may be exposed to the competing form, which, at that point, would represent a foreign form. They may borrow this element into their own language for its sociolinguistic value or even just because it is frequent in their language environment. Over time, however, as dialects and registers level out through prolonged contact, the doublets tend to disappear or to differentiate in meaning, thereby ceasing to be doublets (Kroch 1994:6).

2.2.4. The microparametric approach

Besides these approaches to diachronic variation and change, there is also a recent

and growing body of research, which, over the past decade or so, has attempted to

combine insight of generative linguistics with variationist analysis in the study of

synchronic variation, especially across dialects of the same language (cf. Kayne 1996,

2000; Auger 1998; Cornips & Corrigan 2005; Barbiers 2005; Heap 2001; King &

Nadasdi 1997; etc.).

Kayne (1996) was the first one to propose the notion of ‘microparameter’.

Microparameters represent the ‘switches’ which distinguish closely related languages. In

Kayne’s (1996: xii) words:

21 “Comparative work on the syntax of a large number of closely

related languages can be thought of as a new research tool, one that is

capable of providing results of an unusually fine-grained and particularly

solid character. If it were possible to experiment in languages, a

syntactician would construct an experiment of the following type: take a

language, alter a single one of its observable syntactic properties, examine

the result to see what, if any, other property has changed as a consequence

of the original manipulation. If one has, interpret that result as indicating

that it and the original property that was altered are linked to one another

by some abstract parameter”.

Obviously, such an experiment is impossible to carry out; however, by examining closely related languages, one can begin to approximate the results. Recent syntactic dialect atlas projects, such as the ASIS and the SAND have undertaken such a challenge.

These research projects have various objectives. They not only explore the geographic distribution of syntactic variables, they also correlate them to the investigation of language change and to the more broad study of universal properties of the human language, to understand the loci and limits of syntactic variation within linguistic systems

(Barbiers 2009). The great improvement of this new approach on previous ones is that, whereas past formal studies were primarily concerned with standard varieties, microvariation research includes non-standard languages. “This does not only enhance the empirical basis of syntactic theory, but it also reduces the influence of prescriptive rules and makes it possible to test potential correlations between syntactic variables while

22 keeping other, possibly interfering factors constant” (Barbiers and Cornips 2001:2). Weiß

(2001), in fact, argues that standard languages, learned by special instruction, e.g. in schools, have properties which may not entirely be due to Universal Grammar. They may in some cases reflect unnatural prescriptive habits which have been imposed by processes of standardization. Researchers working on this track acknowledge that every dialect is a heterogeneous system, on the grounds that a dialect does not exist in isolation; it is in constant interaction with one or more standard languages and with other dialects. Each individual may speak in a slightly different way, which allows the model to postulate the existence of slightly different grammars and the existence of individual micro- parameters.

2.2.5. Variability in the Minimalist Program

Finally, a model which combines aspects of the micro-parametric framework, the sociolinguistic approach to data collection and recent formal proposals concerning the architecture of language (the Minimalist Program) is the one by Adger & Smith (2005).

This model, in contrast with previous formal accounts, seems particularly well suited for the study of intra-speaker variation. In fact, even in the most recent approaches to the study of dialectal variation, work on variability between the idiolects of individuals has been largely lacking. On the other hand, in Adger & Smith (2005) and related studies

(e.g. Adger 2006; Parrott 2007), intra-speaker variability becomes the core of linguistic research, bringing previously disregarded phenomena –considered as belonging to E- language - to the fore (Adger & Trousdale 2007).

23 The minimalist framework admits several phonological outputs for a given semantic interpretation. Adger & Smith (2005) argue for characterizing syntactic variation in terms of (un)interpretable features. Certain uninterpretable features may be present in one category but absent in another. Since they are uninterpretable, they would have no semantic repercussion, thus being equally legitimate for a convergent derivation.

Therefore, variation is reduced to the specification of the uninterpretable features in a derivation (Adger & Smith 2005:161). As expected, syntax per se remains invariable or

“perfect” (Brody 2003), given that variation is located only in the lexical component.

Variation will occur when one item or other enters the numeration and takes part in a syntactic derivation. Several (social) factors may affect the outcome: ease of lexical access (probably linked to frequency of use), speaker-hearer relationships, social identity, etc. (Adger & Smith 2005:164).

Notice that this approach is different from Kroch’s or Labov’s alternatives in that variation is located in the selection of lexical entries. On the other hand, Kroch’s model suggests that there is more than one system of grammatical knowledge available to the speaker, and variation depends on the speaker’s selection of one or the other grammatical output. The Labovian variable rules approach contrasts with this in that it states the variation is part of the rule, tying it in more directly to grammatical competence.

Within the framework proposed by Adger and Smith there is no notion of a probability tied to a particular rule; the only possible rules, (Merge, Move and Agree) are invariant and apply categorically in particular cases across languages. Adger and Smith’s perspective differs from the previous approaches in that it assumes only one invariant grammatical system, containing universal mechanisms, rather than a range of systems.

24 Each speaker, however, has a lexicon, a memorized store of pairings of syntactic features and lexical meanings, and it is the choice of the lexical items that is the source of variation. This choice may be influenced by various internal and external factors. This may be considered as a ‘very minimal theory’ (Adger & Smith 2005:16), since the idea that speakers have to choose lexical items is one which cannot be avoided. Localizing morphosyntactic variation in the choice of lexical items means that no other special mechanism must be posited to deal with variation.

A concrete example of this proposal is the analysis of alternation between was

and were in Buckie, a small fishing town near Aberdeen (UK). After providing accurate

information on the methodology used to collect natural data and presenting quantitative

analyses on internal and external factors affecting the alternation, Adger & Smith (2005)

propose a formal model to account for the data.

Consider the following examples (Adger & Smith 2005:154):

(8)

a. He says ‘I thocht you were a diver or somethin’

He said ‘I thought you were a diver or something’

b. Aye, I thocht you was a scuba diver

‘Yes, I thought you were a scuba diver’

Was instead of standard were appears 54% of the time, it can happen with all

persons except they . In terms of syntactic derivation, pronouns bear certain interpretable

features for person and number. In the authors’ view, 1st and 2nd person bear [+person],

while 3rd person is [-person], assuming that third person is lack of a positive

25 specification for person (see, for example, Harley & Ritter (2002) and references therein).

T bears unvalued features for number and person: T [tense:past, ucase:nom, unum:, upers:]. After merging with T, the pronoun we values for plural number and first person in T, and the result is we were . If T is valued as third person, singular, the result is he/she was .

Variation occurs once there is an alternative lexical item which could appear in

the same syntactic context, with the same interpretable features, but with different

uninterpretable features. The authors propose the following alternative: T2 [tense:past,

ucase:nom, upers:] (Adger & Smith 2005:166), which lacks [ unum:]. If T2 is selected,

the derivation will remain the same except for the absence of specification for number.

Since there is no specification for number in T, the verb will be was even when the

pronoun is 1 st person singular/plural (further feature valuation applies in order to account for the ungrammaticality of they was in this dialect). Therefore, this model provides a reasonable account for the variable grammatical phenomena encountered in Buckie by combining quantitative approaches to data collection and formal generative theories.

The postulation that overt variability is due to different lexical specifications, while syntax per se remains invariable, will be one of the assumptions of the present dissertation. This will allow us to account for important cross-dialectal, intra-dialectal and intra-personal instances of variation, while testing syntactic hypotheses and potential correlations between variables. In chapter 6, I will provide a concrete example of how inter- and intra-speaker variation can be accounted for by Adger & Smith’s approach. In particular, I will postulate the presence or absence of certain uninterpretble gender and

26 number features across ABS DP elements to capture the agreement variability encountered in the speech of my informants.

2.3. Variability across closely related dialects: the Afro-Hispanic case

In recent years, syntactic studies have devoted more and more attention to dialects and substandard varieties of different languages e.g. Italian (Poletto 2000; Benincà 1989;

1994), English (Adger and Smith 2005), Dutch (Barbiers and Cornips 2001), etc. So far, little attention has been paid to the study of microvariation across Spanish dialects; to my knowledge, no article has been published yet on microparametric syntax concerning

Afro-Hispanic contact varieties.

The majority of the dialects which emerged in Latin America from the contact of

African languages and Spanish dialects at the time of slavery consists of vernacular varieties. The grammars of these languages are relatively close to that of standard

Spanish. In fact, for the most part, they do not present African substrate structures, which would make them unintelligible to monolingual speakers of the standard variety.

Nevertheless, they clearly present the traces of second language acquisition strategies and contain a variety of constructions which would be considered ungrammatical in standard

Spanish. The nature of their relative close relatedness to the standard can provide linguists with a great empirically-based testing ground for formal hypotheses, in line with recent micro-parametric investigation (Kayne 1996).

The present work represents the first attempt to carry out a micro-parametric analysis of an Afro-Hispanic language. It will combine sociolinguistic methodology and formal theory to explore the syntax of the Afro-Bolivian Spanish Determiner Phrase. In

27 particular, it will analyze phi-features distribution across the DP elements (Chapter 6), nominal ellipsis (Chapter 7) and bare nouns constructions in subject and object position

(Chapter 8).

2.4. Data collection and methodology

The fieldwork was conducted during July-August 2008, December 2009 and July-

August 2010. More than 50 speakers participated in this study. All of them were residents in the communities of Tocaña, Mururata and Chijchipa, three villages in the municipality of Coroico, North Yungas. The informants were native speakers of the Afro-Bolivian dialect; they did not speak any other language spoken in Bolivia, such as Quechua or

Aymara.

The interviews were conducted by allowing the speakers to talk about any topic of

their liking and asking them follow-up questions, in line with the principle of Tangential

Shift (Labov 1984:37)4. The goal was therefore to reduce the Observer’s Paradox (Labov

1972) as much as possible.

Only later, usually after one or two days from the time of the interview, the same informant was asked for grammaticality judgments. This was done in order not to affect the results of the interview by telling the speaker the nature of the phenomena under analysis in advance.

Responses on acceptability judgment tasks rely at least in part on explicit, prescriptive notions held by the speakers (Cornips and Poletto 2005). One way of

4 According to the Principle of Tangential Shift, interviews may be arranged into a network of topics which do not need to be followed according to a prescribed sequence. The conversation between the interviewer and the informant should start with the least personal questions and progress, step by step, towards more intimate topics. The shift between topics should be as smooth as possible. It should be based on follow-up questions to what has just been said by the informant. 28 diminishing this effect which proved successful according to experimental methods described in Labov (1984) is to ask for grammaticality judgments in an indirect way.

Thus, to discover whether or not a variable was present in the community, not only the direct intuitions were elicited: ‘Do you judge X a grammatical/better sentence than Y?;

‘Can you say X?’; also indirect questions were asked: ‘Is variant X present in this community?’; ‘Do you know anybody who can say X?’

The comparison of these two different sources of data resulted in the interesting - but not unexpected- finding that almost everybody who claimed not to say X, but to know people who could say it, were found using an X structure several times during the naturalistic interview, thus indicating that such a structure was indeed part of their grammar.

Such a finding confirmed the importance of relying on different sources. While grammaticality judgments can give us good insight into the abstract idealized language of the informant, only a comparison of such information with empirical data can help us build a robust, fine-grained generalization.

A comparison of grammatical intuitions and sociolinguistic interviews can be very helpful in studying socially stigmatized dialects like Afro-Bolivian Spanish; nevertheless, not all linguistic phenomena analyzed are equally suitable to exactly the same methodology. In fact, while the syntactic variability of high-frequency phenomena can efficiently be studied by using statistical programs such as GoldVarbX (Sankoff,

Tagliamonte, & Smith 2005), less frequent constructions are better analyzed by asking informants for grammaticality judgments. Therefore, while a comparison between natural data and inducted elicitations has been a prerogative in the study of Afro-Bolivian

29 Spanish constructions, this dissertation also considers necessary methodological adjustments imposed by the nature of the different phenomena under study.

2.5. Conclusion

To summarize, this chapter has provided an overview of the main approaches proposed to account for language variability. I focused on recent developments within the field of micro-parametric syntax (Barbiers and Cornips 2001). In particular, I presented the framework suggested by Adger and Smith (2005), who combines quantitative analysis and formal theory. The present dissertation will adopt such model to study several syntactic aspects of the Afro-Bolivian Spanish Determiner Phrase.

Microparametric syntax is a growing field of research which can be employed to test formal theories, mainly designed on standard language data. While several micro- parametric generalizations have been provided for Italian, English and Dutch dialects (see for example Poletto 2000; Benincà 1989; Adger & Smith 2005; Barbiers & Cornips

2001), such research has not been conducted yet on Afro-Hispanic contact varieties. This opens up a new field of investigation, which, if addressed in a methodological and systematic way, could lead to interesting discoveries. One of the main goals of this dissertation is to place the first stone in building such a research program.

30

CHAPTER 3: A SOCIOHISTORICAL ACCOUNT OF SLAVERY IN BOLIVIA

“La historia es siempre una aproximación”

‘History is always an approximation’ (Crespo

1977:11)

3.0. Introduction

Before exploring the possible origins of Afro-Bolivian Spanish (ABS), I would like to point out the sentence directly quoted by Crespo (1977) in his book Esclavos

negros en Bolivia : ‘History is always an approximation’. This statement is particularly

significant when one has to approach such a delicate topic as the Atlantic slave trade,

especially when the region under study is Bolivia, and one is confronted with data which

are scarce at best, and provide only partial information on certain locations at certain

moments.

In this chapter, the available sociohistorical evidence is examined to provide a

description of the sociohistorical setting in which ABS developed in relation to other

Afro-Hispanic contact varieties. The study consists of two main parts. The first section is

an overview of the sociohistorical conditions under which Afro-Hispanic contact

languages developed in Latin America. It includes the main debates and argumentations

concerning the relative reduced number of Spanish creoles. The second part focuses

31 exclusively on the Bolivian scenario. It attempts to shed light on the genesis and evolution of ABS, whose exact origins are still a topic of much controversy.

3.1. Afro-Hispanic contact varieties and the relative paucity of Spanish creoles

Due to several historical reasons, which I will try to pinpoint throughout this work, Spanish creoles did not evolve in the Americas to the extent to which we find them for several other European lexifiers (e.g. French, English). The only two language that are generally identified as Latin American Spanish creoles are Palenquero, spoken in San

Basilio de Palenque, Colombia, and Papiamentu, spoken in Aruba, Bonaire and Curacao,

Dutch Antilles. However, also for these two contact varieties, several doubts have been cast on whether they should really be called Spanish creoles or instead be identified as

Portuguese contact languages which were subsequently relexified with a Spanish lexicon

(for a detailed account see Goodman 1987; Martinus 1989; Schwegler 1993; McWhorter

2000).

The rest of the linguistic varieties that emerged from the contact of African slaves and the Spaniards involved in the conquest of the Americas have not traditionally been classified as creoles. In fact, while presenting certain phonological and morphological simplifications, as well as African lexical borrowings, these varieties clearly lack the radical grammatical restructuring found in creoles such as Cape Verdian Creole, Sranan

Creole, Haitian Creole, etc. On the other hand, these languages closely resemble their superstrate and, even if highly stigmatized from the social point of view, they are perfectly intelligible by someone speaking only standard Spanish.

32 Different hypotheses have been suggested to explain why Spanish creoles are not found in former Spanish American colonies, which at first blush seem to closely resemble other European territories where such languages developed. Certain scholars accounted for the scarcity of Spanish creoles by claiming that Spaniards were more ‘mild’ to their slaves than other Europeans (Mintz 1971). It is sometimes suggested that interracial relations and manumission were common practices so that the acquisition of the lexifier language would have been facilitated (Mintz 1971; Laurence 1974).

In the literature, it has also been argued that the absence of Spanish creoles, at least in the Spanish Caribbean, is due to the fact that these colonies were devoted to small farms, on which the Blacks to Whites ratio was relatively low, until conversion to sugar plantation economies in the nineteenth century (Mintz 1971; Megenney 1985;

Chaudenson 1992). Presumably, the small farm environment would have led Africans to acquire a good approximation of the language spoken by their masters, so that when in the 19 th century major waves of slaves arrived in the islands, the newcomers could

acquire the language in place in the colony from the slaves who were already there.

Laurence (1974) provides parallel analyses between the Spanish colonies in the

Caribbean and the territories under English and French control. She suggests that before the 19 th century, the social condition of slaves in the Spainish Caribbean differed significantly from that of the Africans in the rest of the European colonies.

She claims that the following factors would have been the main sociohistorical conditions characterizing the evolution of Afro-Hispanic contact varieties in the

Caribbean: (i) Africans did not outnumber Spaniards; (ii) in the Spanish colonies the hierarchical structure of society was not as rigid as in the English and French Caribbean,

33 often leading to coartición (self-purchased manumission) and mixed-race marriages; and

(iii) plantations were introduced later in the Spanish colonies; agricultural production was predominantly based on minifundios , small farms where free blacks and slaves used to

work side by side with poor Spaniards.

Mintz (1971) essentially shared Laurence’s opinion. In his view, the

sociohistorical differences between French and English colonies on the one hand and the

Spanish Caribbean on the other played a key role in determining the non-creolization of

the Spanish language. He stated (1971:488):

“In such colonies as Cuba, where one may suppose that a pidgin

language did exist, at least briefly and in those periods when the influx

of multilingual slave shipments was considerable, Spanish would

provide a continuing medium of communication for culturally creolized

slaves and freemen of all physical types. For the greater part of Cuba’s

and Puerto Rico’s post-conquest history –that is, from the Discovery

until at least the eighteenth century – the relative proportions of slaves

to freemen were low, and the rates of manumission apparently high. In

such colonies as Saint Domingue, where the importation of slaves after

1697 was both massive and rapid, the stabilization of a pidgin and the

emergence of a creole language thereafter would be expectable, even

though manumission was common, and the growth of an intermediate

and economically influential free mulatto class – probably bilingual –

was swift. Revolution and independence at the close of the eighteenth

34 century, and the substantial elimination of the French colonists, may

have contributed powerfully to the full stabilization of the Haitian

Creole thereafter. […] In the British possessions, rapid slave

importations and the substantial lack of a firm creole culture, a

numerous intermediate group, or insular institutions that could unify

the colonial population, probably contributed to the particular linguistic

situations typical of these colonies.”

Lipski (1987, 1993, 1998, 2000) examines the linguistic and social aspects of

Bozal Spanish 5 spoken in 19th-century Spanish Caribbean. He proposes that an Afro-

Hispanic bozal language never underwent complete restructuring due to its limited status as a native language. However, in 19 th century Cuba, after the sugar plantation boom, the conditions favoring restructuring briefly existed in some of the larger haciendas but the

new waves of slaves speaking second language varieties of Spanish did not creolize the

language already in place on the island so that subsequent generations would have

acquired Spanish natively, thus explaining why no other linguistic traces are left except in

the vocabulary.

The above explanations have been generally used to account for the lack of full- fledged Spanish creole languages in the Caribbean. McWhorter (2000) acknowledges that the aforementioned reasons could actually be valid for Cuba, Puerto Rico and the

Dominican Republic; however, he claims that such an account may not provide an explanation for the rest of the Spanish colonies in the Americas.

5 Bozal Spanish is the variety of Spanish spoken by Afican-bornslaves. These people, called bozales , spoke Spanish as L2. 35 In his view, for several mainland colonies like Mexico, Colombia, Peru, Ecuador and Venezuela, large-scale agriculture should have created the optimal conditions for a creole to emerge. More specifically, the long period of intimate contact – that is, the société d’habitacion stage (Chaudenson 2001) familiar to certain English and French creoles – simply did not exist in the mainland colonies (McWhorther 2000:34); besides, several historical reports clearly indicate that Spaniards were not by any means more kind to their slaves as is often suggested (Blackburn 1997).

In his book, The Missing Spanish Creole , McWhorter (2000) backs the idea that the lack of access to the substrate language could not have been the reason for the development of creoles in the New World, as in many mainland Spanish colonies such a factor was in place and creoles did not evolve. He provides partial data for the Chocó

Region (Colombia), Chota Valley (Ecuador), Veracruz (Mexico), Lima (Peru) and the

Mocundo hacienda (Venezuela) and claims that a different factor must have played a crucial role. According to McWhorter, such a factor would be the lack of a Spanish pidgin spoken on the West African coast from which a Spanish creole could subsequently develop in the New World.

He believes that plantation creoles were once pidgins which expanded into fully referential languages by either children or adults (McWhorther 1997; 2000). As a result, he does not subscribe to the view of Chaudenson (1979; 1992) and Mufwene (1996) that plantation creoles are transformed varieties of their lexifiers, with the distance from the lexifier due to widespread second-language acquisition, as in their account there would be no room for the pidgin stage.

36 However, even assuming that the sociohistorical and demographic conditions reported by McWhorter (2000) for the mainland colonies were detailed and correspond to reality, his account does not completely explain why a pidgin did not evolve in such territories, or why a Portuguese pidgin previously formed in Africa did not relexify and develop into a Spanish creole, as supposedly happened in the case of Papiamentu and

Palenquero (Lipski, MS).

As far as Venezuela is concerned, Díaz-Campos and Clements (2008) show that

the picture presented by McWhorter does not accurately reflect the social and historical

reality of the colony. In fact, McWhorter use the term “Africans” to describe mixed-race

segments of the population who were born in the Spanish colony and spoke Spanish

natively; moreover, he does not account for the fact that in some parts of the colony the

Spanish Crown was interested in exploiting mineral resources rather that developing

agricultural commerce (2008:358). In this article, the authors clearly discuss the role of

the Spanish Crown’s monopolization of the slave trade, which kept the Black/White ratio

relatively low in certain Spanish colonies until the end of the 18th century, thus indirectly

reducing the likelihood of a Spanish creole language developing in the Americas.

On the other hand, the birth and development of Spanish creoles in the New

World has been suggested by some scholars (Bickerton and Escalante 1970; Granda

1978; Schwegler 1993, 1996). In the literature, in fact, it has also been argued that a

Spanish-based creole was once spoken throughout the Spanish Caribbean and Latin

America, but that it now survives only in Palenque and in the Dutch Antilles, having

everywhere else decreolized and disappeared due to normative pressure from standard

Spanish. However, as has been pointed out, it seems unlikely that Spanish creoles did in

37 fact exist and disappear so completely without leaving behind some residual traces

(Laurence 1974; McWhorter 2000).

Nevertheless, it is worth mentioning some cases in which a possible decreolization path for former Spanish creoles had been suggested. Granda (1978) indicates that a now-extinct creole was once spoken in the inner Colombian town of Uré by descendants of slaves who had escaped from gold mines in the province of Antioquia.

Unfortunately, however, as McWhorter (2000) noticed, Granda gives no data and was unable to actually visit Uré.

Possible creole origins for another Afro-Hispanic contact variety have been suggested also by Álvarez and Obediente (1998) for Barlovento Spanish, a dialect spoken

on the Caribbean coast of Venezuela. In this work, the authors claim that some linguistic

features encountered in the language (e.g. deletion of the copula verb, non-inverted order

in questions, etc.) would be the result of a process of decreolization, of a language that

was once a creole and that gradually approximated to the superstrate. The basis for this

claim would be that this kind of phenomena is not generally encountered in Spanish sub-

standard varieties, and that they should therefore be attributed to decreolization.

Such a claim has been proven to be misleading by Díaz-Campos and Clements

(2008), who provide an account of the sociohistorical situation in which this language

developed. First of all, they show that blacks were not sufficiently numerous in the area

as for a radical creole to emerge; secondly, they provide an alternative account for all the

features classified by Álvarez and Obediente as potentially inherited from the creole, by indicating that they are actually better explained in terms of imperfect second language

38 acquisition , in a context in which the superstrate language was relatively available to the

African slaves.

A third case of possible de-creolization of a pre-existing creole has been suggested more recently by Lipski (2006a,b; 2007; 2008a,b; 2009) for Afro-Bolivian

Spanish (ABS). In the next section, I will analyze Lipski’s claim and provide a socio- historical scenario in which to test the feasibility of a possible de-creolization for Afro-

Bolivian Spanish.

3.2. On the origin of Afro-Bolivian Spanish

The exact origin of ABS is not yet completely clear. Lipski (2008b) suggests that traditional ABS might have been once a full-fledged creole language and adds that “in absence of any other viable scenario, Afro-Yungueño Spanish must be viewed as the descendant of a colonial Afro-Hispanic pidgin (2009:186)”. In fact, by looking at the

“radically simplified VP and DP of the basilectal Afro-Yungueño dialect”, Lipski

(2008a:37) hypothesizes a possible ‘radical’ creole origin for this vernacular, which after undergoing a process of decreolization due to contact, would now be in one of its final stages, closer to more prestigious regional Bolivian Spanish.

Such a process would have taken place during the last 50-60 years, immediately after the Land Reform of 1952 which freed Afro-Bolivians from forced peonage 6. Lipski,

however, recognizes the lack of reliable sociodemographic data for Afro-Bolivian; for

6 Whitin such working system, Afro-bolivians were forced to work three days a week for the owner and three days a week for themselwes. They were assigned a piece of land on which they could grow their oun products. They were not allowd to leave the hacienda without the owner’s permission and they could not attend public schools. 39 this reason he does not discard the hypothesis of a stable but not creolized variety of

Spanish, which co-existed with highland Bolivian Spanish since its inception.

As I will argue in more detail below, the radical creole hypothesis suggested by

Lipski does not seem to be backed by historical evidence. On the contrary, social and demographic records concerning black Bolivia seem to indicate that a case of radical decreolization would be rather unlikely for ABS.

The present chapter will try to locate Afro-Bolivian Spanish in space and time with respect to the colonization of Spanish America. In doing so, it will also provide a detailed account of the socio-historical conditions which led to the development of this

Afro-Hispanic contact variety.

3.2.1. A sociohistorical sketch of slavery in Bolivia

African slavery persisted in Bolivia from the very beginning of its colonization, in the first decades of the 16 th century, until 1826, when in the new constitution, immediately after the country’s independence from Spain, slaves were declared free.

However, in a law of December 19 of the same year, they were classified as debtors to their owners, which practically meant they were still slaves. On April 10, 1827 slave trading was abolished and those who did not respect the law could be charged 500 pesos as punishment. In 1830, the new president Andrés Santa Cruz re-established slavery, claiming that giving freedom to slaves had “damaged the agriculture of the Republic, and blacks’ lives too, as it pushed them towards being lazy, begging, and other vices”

(Portugal-Ortíz 1977:86). In an article of 1837 a slave entering the Bolivian territory became a free man. In 1840 there was a ratification of the English treaty on slavery

40 abolition. In 1844, a person who was caught trafficking in slaves risked up to 10 years in prison. During Isidoro Belzú’s Government (1848-1855) the traffic is completely erased.

However, Afro-Bolivians will have to wait for the Land Reform (1952) to be completely free, which means not having to work for the owner and acquire the right to receive an education and to vote.

Brockington (2006) divides the black Diaspora to Bolivia in two separate waves.

The first one concerns the Africans who arrived in the early times of colonization

– during the 16 th and mid-17 th centuries (roughly 1530-1650) – with the Spanish settlers.

These Africans included both slaves and free men, who served as military personnel or servants in the many campaigns of discovery, pacification and settlement.7 They came mainly from Latin American territories already colonized and presumably could speak good approximations of Spanish.

Brockington claims that:

“The African slaves and people of African descent were active

participants – voluntary or otherwise – in a militaristic,

conquering/pacifying, horse-and-gun culture here, as elsewhere in the

Spanish Americas at that time. I am convinced that some of them

remained […] as, among other things, slave and free cowboys and

ranch hands” (2006:130).

7 Browser (1974:3-5) gives an explanation of the role played by blacks in the first military expeditions led by Pizarro and Almagro. 41 Over time the situation changed. Brockington (2006) suggests that probably those

Africans already present in the territory around the mid-17 th century formed unions with

more recently arrived slaves. The newcomers were Africans brought to the colony – by

legal means or through smuggling – to substitute the now reduced Indian workforce

available, and a significant number of them might have been bozales . This was the

second wave of the African Diaspora in Bolivia, which would last for almost two

centuries, and whose members were mainly employed as domestic servants or farmers

with a much reduced percentage in the mining industry.

The following parts of this chapter will try to analyze these two waves, at times in

general terms, at times focusing on specific regions, with the objective of shedding light

on the currently unclear origins of ABS. Even though Afro-Bolivian slavery appears to

have been a highly controversial and variable business, at least from a legal point of

view, certain features of this tragic historical episode seem to have been fairly constant

across the centuries in all of the Bolivian territory.

In fact, importation and employment of Africans in Bolivia has been conditioned by certain characteristics, which indirectly give us important information about the possible means of communication adopted by the slaves. All of these considerations seem to indicate that an Afro-Bolivian creole did not develop and that the black population encountered in Bolivia could speak Spanish – or a good approximation to it – as a result of a relatively good access to the language spoken by their masters. Such features can be briefly summarized in the following points:

a) The slave trade was monopolized by the Spanish Crown, which kept the

Black/White ratio relatively low (Díaz-Campos & Clements 2008).

42 b) The location of Bolivia, in the heart of South America, made the

introduction of slaves into this region a particularly difficult endeavor. The higher cost of

slaves, deriving from such a geographical barrier, severely limited their importation

(Klein 1999; Wolff 1981).

c) Spaniards in Bolivia did not need as much of an African workforce as

elsewhere in the Americas. In fact, the indigenous population could be enslaved or forced

to work for a minimal wage, a scenario that was economically more profitable for the

owner than investing in expensive African slaves (Mellafe 1984; Browser 1974).

d) The combination of (a), (b) and (c) strongly constrained the introduction of

Africans to the territory. As a consequence, massive importation was never reported and rare were the purchases of more than 10 salves at a time (Bridikhina 1995a; Lockhart

1994).

These were the general conditions which characterized African slavery in the

Bolivian region for almost four centuries (first and second waves of African slaves in

Brockington’s terms), making therefore highly unlikely the development of an Afro-

Bolivian creole spoken across the country. The following sections will analyze in more detail the two different waves which characterized the black Diaspora to Bolivia.

3.2.2. Slavery in Bolivia: the 16 th and early-17 th centuries

The territory of the Andean region now known as Bolivia was first settled by Spanish conquerors during the first decades of the 16 th century. The colonization proceeded from the Peruvian coast up to the Bolivian Highlands, as well as from the Paraguay River to the Bolivian tropical regions. Colonization, from the Peruvian side, proceeded in the

43 following way: in 1528, the Spanish conqueror Francisco Pizarro arrived by boat in

Tumbes. By 1532 Spaniards reached Cajamarca and killed the Inka Emperor Atahualpa.

In 1533 they took Cuzco, the capital of the Inka Empire (Lockhart 1994). In 1538 they founded La Plata de la Nueva Toledo, now known as Sucre. In 1542 they settled

Cochabamba, and by 1546 Potosí, the richest silver mines of Latin America. In 1548 they founded La Paz. From the Paraguayan side the colonization proceeded in a similar manner: Spaniards wanted to find a way to connect the rest of the Latin American colonies to Perú. By 1537, Asunción, the actual Paraguayan capital, was settled by the conqueror Juan de Ayola. Several explorations followed; in 1548 Spaniards settled

Guapay, and a year later the Spanish captain Ñuflo de Chávez reached the already settled city of La Plata de la Nueva Toledo. The same captain founded the city of Santa Cruz in

1561 and in 1565 the explorer Diego de Villaroel founded Tucumán, in present Argentina

(Klein 2003) (see Figure 1 below).

By 1559 La Plata de la Nueva Toledo (Sucre) became the capital of the Audiencia of

Charcas, which had jurisdiction over Bolivia, Paraguay and the Rio de la Plata, today the region roughly corresponds to Uruguay and Northern Argentina. This was the provincial court that oversaw the huge silver output of the mines at Potosí. It belonged to the

Viceroyalty of Peru until 1776, when it became part of the newly created Viceroyalty of

Río de la Plata.

44 Figure 1. Map of Bolivia.

Los Yungas

In conquering the Andean region, Spaniards brought with them black slaves from the very beginning of the colonial enterprise (Browser 1974).

According to Browser:

“Pizarro was one of the many conquistadores who gravitated to the

Isthmus of Panama and achieved a measure of wealth, yet were urged on by

the news of the spectacular success of their compatriots in Mexico. Hearing

persistent tales of wealthy native kingdoms to the south of Panama, Pizarro

formed a company for the purpose of Pacific exploration. African slaves 45 were relatively numerous in Panama by this date, at least in part because of

the decline of the native population, and blacks figured in all the

expeditions undertaken by the company between 1524 and 1528. An

African slave helped to save the life of Pizarro’s most prominent associate,

Diego de Almagro, for example, and later, in 1528, when Pizarro

discovered the city of Tumbes, a black sent along to observe the place was

seized by the incredulous natives and vigorously scrubbed in an effort to

remove what was presumed to be the dye on his skin” (Browser 1974:4).

Browser shows that, in fact, Africans played a very significant role in the conquest and settlement of the region. Soon after the discovery of Tumbes, Pizarro went back to Spain to enlist additional support for the military expedition; among other things, he was given the right to import 52 African slaves (Mellafe 1984).

For several years during early Bolivian colonization, the Crown granted licenses to trustworthy servants to import black slaves for military purposes and for the building of roads, bridges and public infrastructure. These first blacks to enter the Andean region were generally not directly shipped from Africa, but rather imported from other

American colonies (e.g. Panama) or Spain. Only Africans born in the power of

Christianity, or who had at least resided with their masters long enough to be baptized, were employed. They had to be docile and obedient; besides, the Spanish Crown and the

Roman Catholic Church were committed to Christianizing the New World and concerned with the introduction of potential “enemies of the Faith” into it. Therefore, as elsewhere in Spanish America (e.g. Venezuela; see Díaz-Campos & Clements 2008), the Spanish

46 Crown’s monopolization of the slave trade played an important role in constraining the introduction of black slaves into the region, thus keeping the black/white ratio relatively low.

Africans soon came to occupy an intermediate position between Spaniards and

Indians. Often they were used to help suppress native rebellions. Many priests and corregidores de indios, royal officials who exercised judicial and political authority in

Indian areas , had one or more black servants helping them carrying out their work activities. This factor indirectly suggests a better acquisition of the language spoken by the masters.

Figure 2. A royal official helped by his slave punishing an Indian community leader (Poma de Ayala 1615).

47 Differently from other European powers, the Spanish Crown generally did not encourage the massive importation of slaves to the American colonies, especially to

Bolivia. The Spanish monarchy had a monopoly on this activity and was very careful about releasing trading licenses and contracts ( asientos ). Spaniards could not risk carrying with them potentially dangerous Africans in their colonial enterprise in the

Andes, especially given that the first rebellion had taken place on the Caribbean island of

Hispaniola, where on December 26, 1522 a group of twenty Wolofs escaped Diego

Colón’s sugar refinery and killed several Spaniards (Troconis de Veracoechea 1969:20; see also Díaz-Campos & Clements 2008).

No official Bolivian census reporting the percentages of Blacks, Whites and

Indians exists for the 16 th century; however, from the legal documents dating back to those years we can understand that several serious constraints were posed on the introduction of African population into this region. In fact, many cases of demand by

Spanish settlers to the royal crown for African slaves were completely or partially denied in Bolivia.

Pedro Cornejo de Estrella, a Spanish citizen from Potosí, called the attention of the Court to the number of mines that could not be worked because of ventilation and drainage problems (Browser 1974:121). In 1587 and later in 1588, he offered to demonstrate for the Crown a mechanism he had invented to solve these problems in the gold deposits of Cabraya province. In return, he asked for permission to import 150

African slaves free of royal duties to assist in the operation and for the exclusive use of his invention for a period of twelve years. The members of the joint Indies-Treasury committee, responsible for the evaluation of the project, decided in February of 1588 that

48 only 50 slaves could be imported free of charge, and that Pedro Cornejo de Estrella would be allowed to acquire another 50 for the standard fee payable within four years.

In other instances, the permits were never given; as in the case of Viceroy

Marqués de Cañete, who in 1556 urged the Crown to send three thousand blacks to the colony. A part of the Africans would have been sold to private persons to recover the expenses, while the rest were to be employed in gold deposits in the Chachapoyas area.

The Crown rejected the petition because this venture would have required large amounts of capital to be invested in the first place.

Africans in fact were much more expensive than the local indigenous labor force.

This factor is crucial for understanding the dimensions of the slave trade in Spanish

America, and in particular in Bolivia. This territory, in fact, was further away from Africa

than the Caribbean region or the other costal colonies, and it was reasonably populated by

Indians, who could be exploited as well.

Summarizing, the Spanish Crown’s monopolization of the slave trade seems to

have highly constrained the introduction of a black workforce into the territory. Africans

were not usually employed on plantations, but rather in militaristic enterprises or public

works. Transportation difficulties due to the long journey from Africa to Bolivia favored

the use of blacks coming from already settled colonies and incentivized the exploitation

of the native workforce.

3.2.2.1. Routes of arrivals and origin of the slaves

Those blacks who did end up in Bolivia were taken to the Andes through several

routes. Until the last decades of the 16th century, slaves were imported from Africa and

49 from the Iberian Peninsula. They were shipped to the port of Cartagena on the Caribbean coast of Colombia. From there, they were taken to the Caribbean side of the Isthmus of

Panama; after traveling over land to the Pacific, they were shipped again to the port of

Callao, present-day Peru, or to Arica in what is now Chile. In order to achieve the

Bolivian lands, slaves had to make their final trek, walking from Callao or from Arica across the Andes (Klein 1999; Wolff 1981; Busdiecker 2006) (see Figure 3). Such a complicated route implied that the blacks taken to the Bolivian region had to embark in a significantly longer trip than those transported to the Latin American costal regions. This factor probably resulted in a higher number of casualties during the trip (Klein 1999;

Busdiecker 2006).

By 1580, after the colonization of Tucumán and the Spanish expansion into the region of La Plata, a new route became available, linking Bolivia to the port of Buenos

Aires which received slaves proceeding from Africa and Brazil. This new route significantly reduced the sea voyage but implied a longer overland journey (Wolff 1981;

Busdiecker 2006).

By whatever route, the introduction of Africans into the Andes was not an easy task; such a transportation barrier was inevitably reflected in a higher price for the slaves sold in Bolivia. For this reason, unlike elsewhere in the New World, in the Bolivian region, it is rare to find massive acquisitions of blacks by a single owner (Busdiecker

2006). Legal documents confirm that a record of a purchase of more than ten slaves at a time was very unusual (Lockhart 1994). Those who had significant numbers of slaves acquired them gradually, buying no more than two or three in a single transaction

(Bridikhina 1995a).

50 Figure 3. Routes of slave arrival.

It is difficult to establish with certainty the origin of the slaves transported to

Bolivia. Among other things, usually, the information available, when reported, was the one of the port of departure rather than the actual birthplace (Busdiecker 2006). Curtin

(1969), Crespo (1977) and Wolff (1981) indicate that Wolofs from Senegal were the principal ethnic group among the slaves imported to Spanish America in the first half of the 16 th century, while Bight from Biafra were the dominant one during the second half and blacks from Angola were the majority from the beginning of the 17 th century.

51 In sum, we cannot say with certainty the exact origin of the slaves imported to

Bolivia; what we can instead be sure about is that the geographic location of this colony played a role in constraining the number of Africans introduced into the territory.

3.2.2.2. The role of slaves in the Bolivian economy

From the beginning of its colonization, the economic structure of the Bolivian region was mainly based on the mining industry, and only partially on the production of agricultural products, especially coca, to support the exploitation of mines. Coca leaves, in fact, are traditionally chewed by Bolivian miners to resist fatigue at high altitudes.

Spaniards mainly relied on the Indian workforce, organized through the mita , a

system already employed during the Inca Empire by which certain work tasks were to be

assigned to different native families on the basis of rotation. No one was required to serve

again until all the rest had had a turn (Browser 1974). Indians forced into the mita were

given a minimum wage; in fact; after the New Laws of 1542-43, designed to attain the

moral and political goals of the Crown, enslavement of Indians was regarded as illegal.

Behind the proclamation of these regulations, which conceded a slightly fairer condition

to the aboriginal population, were the increasing complaints of religious men, such as

Bartolomé de las Casas, who were disgusted by the excess of violence with which Spain

had treated the natives.

As applied by Spaniards, the mita system set very harsh working conditions for

the Indian population. Many official attempts were made to abolish it, or at least reduce

the workload of Indians, with black labor proposed as an alternative; however, almost all

of these attempts failed. The massive employment of natives in the mines was

52 implemented during Viceroy Toledo’s government (1569-80). In the case of Potosí, for example, workers were drawn from the vast highlands from Cuzco to Tarija. In 1578, it was established that 14,248 men would have served each year in the cerro rico mines for a period of 12 months (Browser 1974). It is unlikely that within the mita system a creole language spoken by African slaves could have emerged, at least in the mining centers of the colony. Records of the time, in fact, show that in such areas blacks were never a significant percentage of the population as shown by the figures of Table 4 (Crespo

1977).

Table 1. Demographic figures for the city of Potosí (Crespo 1977)

Year Afro- Total Afro- descendant population descendant population population (percentage) 1611 6,000 160,000 3.74 % 1719 3,206 70,000 4.58 % 1832 1,142 224,000 0.51 %

Additional cues suggesting a limited black presence in the frigid highlands are current Peruvian sayings such as el gallinazo no canta en puna 8, contemporary idiomatic expressions indicating that Africans were not suitable for working in such an environment. Blacks imported into Potosí were primarily employed as domestic servants or in the Casa de la Moneda , melting, fragmenting, and minting silver; while Indians were put to work in mines. The work reserved for natives was much harsher than the work given to Africans; this is because blacks were expensive and owners wanted to

8 In literal English, ‘the turkey buzzard doesn’t sing in the highlands’. The gallinazo , a black-colored vulture with a bare, reddish head, is common along the Peruvian coast, but is not seen in the cold tablelands of the Andes (Bowser 1974:14). 53 protect their costly investments. As data show, Africans were never significantly employed in highland mines; however, it must be acknowledged that in other parts of the

Viceroyalty, both on the Peruvian coast, as well as in the tropical regions of Bolivia, black laborers could have been used to replace and complement the Indian workforce, decimated by overwork and disease.

Lockhart (1994) and Brower (1974) have shown how African slaves and their

descendants came to populate the warmer costal areas of Peru, where they were often

devoted principally to agriculture and livestock. In the urban centers, especially in Lima,

they were employed as lackeys and domestic servants. The authors were able to provide

accurate data on Africans’ occupations as well as general figures concerning the

percentage of black population in Lima at different points in time. Such a detailed work

has yet to be done for Bolivia, even though several authors (Brockington 2006;

Busdiecker 2006; Crespo 1977) have clearly indicated that wealthy families in La Paz

and other cities employed Africans for domestic purposes too, in order to show their

higher social status.

Along with these occupations, to a much reduced extent, Africans have also been

reported to do agricultural and farming work. Brockington (2006) was able to provide

new and interesting data on Mizque, in the Province of Cochabamba, where as early as

1573 the first African presence was attested (2006:139). It was the case of Cristóbal

Roldán, a mulato owing cacique Don Pedro Arapa 29 silver pesos for a colt and a

stallion. Cristóbal was listed among other ‘goods’ on the list of the cacique’s personal

properties.

54 Brockington reports several other cases; for example, by 1629 she found six men and two women, three of whom were identified as bozales , in the Álvarez Meléndez

hacienda , supporting up to seventy-two people, including the wives and children of his

yanaconas (tribute-paying Indians) (2006:141). Another document described the Oloy

hacienda in 1603. The complex had two houses, two vineyards, fruit-bearing orchards,

wine cellars, a mill, a forge, and the necessary equipment for wine production. It

supported a large crew of yanaconas – 179 all told, 49 of them working adults, and some

mulatos (2006:239).

Among the most interesting discoveries, Brockington points out the presence in

Don Rivero’s will (1629) of sixty-eight slaves, revealing the significant role of African

labor in this territory. Of these negros , mulatos and zambos , 26.5 percent were of African

origin (Angola, Congo, Anchico, Bran and Xoxo or Xhosa 9). The ratio of adult males to

adult females was twenty eight to thirteen (68 percent male), while there were equal

numbers of male and female children. There were about twice as many adults as children

(forty-five to twenty-three). Subtracting from the adults the elderly (ten) and the mentally

or physically incapacitated (four), only thirty-one were full-time, able-bodied workers.

The identifiable occupation assigned to slaves consisted of irrigator, carpenter, team

driver and barber. General jobs in agriculture and animal farming are assumed for the rest

of the slaves, as thousands of heads of livestock, as well as extensive records of

agricultural and sugar-refining equipment, are also present in the document. Slave prices

changed according to their age, skill, and physical and mental condition. The records

confirm that the cost of buying a slave was much higher in Chacras if compared with

9 Brockington was not able to determine the origin of one slave who was labeled Blas Soxa. She attributes this denomination to Xoxo of Western Africa or possibly to Xhosa, from Southern Africa. 55 Buenos Aires. For example, a small slave child was sold at a price of 200 to 300 pesos in

Mizque, while for the same amount of money, an adult slave could be bought in

Argentina (2006:144).

Even though Don Rivero’s will provides us with a better idea of a possible scenario for black slaves in early colonial Bolivia, it is still hard to understand whether or not a creole-like variety could have developed in his hacienda . In fact, no data concerning white and indigenous populations are reported. However, if attention is paid to the proportions between black and Indian workers in other haciendas (e.g. the Álvarez

Meléndez one), or to the demographic situation of Mizque at the time, it is likely that a much higher number of natives and Spaniards would have been present. In fact,

Brockington quotes a letter of the archbishop of the Mizque-Santa Cruz diocese, Antonio de Barranca, reporting the population statistics for baptismal of six of the nine parishes under jurisdiction in 1618, with 250 negros , 150 mulatos and zambos , 2,600 yanaconas ,

8,500 Indians and 1,800 Spaniards, indicating an overall 3.8 percent African descendant population (2006:176).

By analyzing Don Rivero’s will, it is already possible to try to get an idea of the socio-demographic scenario of his hacienda . No data referring to white or indigenous population are reported; however, it is hard to believe that a hacienda including a big vineyard, a sugar refinery, and several thousand heads of livestock could be run by only the thirty-one able-bodied workers calculated by Brockington in the will, and maybe also by some of the less-able ones.

Figuring out the exact proportions between Spaniards, blacks and Indians is not possible, and therefore I will not try to do it; what can be said is that probably the main

56 ethnic group employed was native, as encountered in other haciendas and records.

Besides, if Don Rivero’s hacienda was in line with the census data of the time, whites probably outnumbered blacks.

In Don Rivero’s will the ratio bozales : criollos 10 was almost 1:3. This piece of information can be useful to understand the kind of means of communication used by the slaves. If we assume, as Brockington does, that until the early 17 th century the employment of Africans, mainly proceeding from already settled colonies, was primarily limited to serve Spaniards in their conquering and Christianizing enterprises, we would assume that blacks present in Bolivia by that time would have mastered the Spanish language quite proficiently. Slaves reported as criollos were probably the descendents of

the first Africans who arrived in the colony, or at most the descendents of bozales who

arrived soon afterwards, where massive importation – if it ever existed in Bolivia – was

not present. This would suggest that the language spoken by the criollos owned by Don

Rivero was likely to be a good approximation to Spanish, closer to it than the contact

variety spoken by the vast Indian population. The rest of the slaves, 17 bozales , had

probably closer contact with the criollos than with the free Indians; it is therefore feasible

to think that with time they would have gotten closer and closer to the language spoken

by the rest of the slaves. Such a process of approximation did not restructure the language

radically as happened, for example, in different American colonies such as Haiti

(Lefebvre 1998) or Surinam (Migge 2003), where massive waves of new slaves came to

replace seasoned ones resulting in a further separation of the contact language spoken in

the plantation from the European lexifier. As reported by Bridikhina (1995a) and

10 The term criollos indicates people born in the Latin American colonies. In this case, it reffers to the black slaves born in Bolivia, in contrast to those born in Africa ( bozales ). 57 Lockhart (1994), slave transactions in Bolivia were usually limited to a couple of

Africans at the time, and rare were those involving more than ten. These small purchases, in contrast with the massive ones seen elsewhere in Latin America, would have facilitated the process of language acquisition among the newcomers.

It is important to say that slaves seemed to be quite old in Don Rivero’s hacienda , ten being described as elderly and recorded as over 50 years old. This indicates that probably the working conditions were not so harsh as for example in Haiti or Surinam, where the average age was 30, and the time spent working on the plantation before dying was usually not longer than seven years (Migge 2003).

To describe this aproximación idiomática entre españoles y africanos Crespo

(1977) says that Africans and Spaniards were somehow closer than Indians. In fact, this last group perceived both blacks and whites as invaders; natives were more reluctant to learn Spanish as they could communicate with each other by using their own language.

On the other hand, Africans were more willing to use Spanish with their owners and among themselves, because they were a minority and the probabilities for a slave to find other Africans with the same linguistic background were low (Crespo 1977: 20).

In sum, this section has treated the economic structure of the colony and the role played by slaves in it. We have observed that both in the mining and the agricultural sectors, Spaniards relaied mainly on the employment of native workforce through the use of the mita system, while Africans tended to be employed as domestic servants. Overall, the demographic data available for the Potosí Province and the Mizque-Santa Cruz diocese indicate that Africans and their descendents were a small minority when compared to the Spanish population. This piece of information, in addition to the points

58 highlighted in the previous sections, suggests that a full-fledged creole language did not likely developed in Bolivia during the first wave of importation (from the 16 th to the mid-

17 th centuries).

3.2.3. Slavery in Bolivia: from the mid-17 th to the mid 20 th century

As mentioned at the beginning of this chapter, the Spanish Crown’s monopoly on

the slave trade did not favor the introduction of Africans. Morevover, the transportation

impediments, and the availability of a native workforce did not incentivize the

employment of black workers as their cost was comparatively higher to that of Indians.

These factors persisted and regulated the dimensions of black slavery in Bolivia also

during the “second wave” of forced migration, roughly 1650- 1850.

Even after 1650, when the native population had been presumably reduced by

European diseases and the harsh working conditions, African importation did not achieve

in Bolivia the levels observed elsewhere in the Americas. Nevertheless, forced work

persisted for a very long time: slavery was officially abolished after independence from

Spain (1827); however, forced peonage continued until 1952, year of the Land Reform.

The demographic data reporting the number of blacks in Bolivia are fragmentary

at best. Table 5 provides a rough breakdown of the population of Bolivia from 1650 to

1950 as reported by combining the data indicated by several authors treating this topic

(Rosemblat 1954; Crespo 1977; Pizarroso-Cuenca 1977a,b; Delance 1975).

59 Table 2. Bolivian population 1650-1950 11 :

Blacks Mulattos 12 Mestizos 13 Whites Indians Total 1650 30,000 5,000 15,000 50,000 750,000 850,000 1846 27,941 659,398 701,558 1,388,897 1940 7,800 5,000 870,000 1,595,000 2,900,000 1950 7,000 5,000 907,709 1,660,467 3,019,031

As can be observed, the demographic data presented in Table 2 are not detailed, they cannot provide us with a clear idea of the evolution of the ethnic groups from 1650-

1950. The only idea that can be derived from these data is that overall, in Bolivia, blacks seem to have always been a very small minority during the period considered, and they never outnumbered whites.

Moreover, it has to be remembered that after abolition of slavery in 1851, blacks were no longer officially acknowledged, making the reconstruction of their presence on the territory much more complicated. Recent unofficial estimations of the Afro-Bolivian descendants (Angola-Maconde MS) suggest that the total approximate number should be

15,800, close to 0.18% of the whole Bolivian population (see also Lipski 2008b:30-31).

3.2.3.1. New routes of arrivals and origin of the slaves

Even when in the 18 th century a new route was able to connect Buenos Aires to

Cordoba, Tucumán, Salta, Juyjuny, Potosí, Oruro and La Paz, the price of a slave was

still much higher in Bolivia than in Argentina. Birdikhina (1995a:184) reports that a

bozal could be acquired for 200 pesos in Buenos Aires and sold for 500 pesos in La Paz,

after the journey that lasted on average four months.

11 Sources: (Rosemblat 1954; Crespo 1977; Pizarroso Cuenca 1977a,b for years 1650-1846; Dalence 1975 for years 1940-1950). 12 Offspring of white and black. 13 Offspring of white and Indian. 60 Crespo (1977:36-37) reviews several documents concerning slave trafficking in

La Paz for the years 1650-1710. He fund information about 216 slaves (129 women and

87 men). Crespo could not find the origin of almost half of these individuls (98); the rest was reported as: Angola (33), Congo (9), Banguela (9), criollos from La Paz (22),

mulatos (9), Biafra (1), criollos from other cities (19), and others (16) (Cabo Verde,

Chaloi, Lubolo, Canbunda, Mozambique, Bran, Jolof, Portugal and Mandinga). This

report, even if very fragmentary, indicates that the majority were bozales (175), while the

rest were criollos (41); this may indicate that during this second wave of importation several slaves came directly from Africa. However, in the same pages, the author reminds us that “in La Paz archives the sale of more than ten people at the time is very exceptional” (1977:37), and that, due to the high fragmentation of the ethnic groups taken to Bolivia, “the greatest means of unification was […] the language spoken by their owners, Castilian Spanish” (1977:39).

Brockington (2006) analyzes the records of a Jesuit enterprise in the province of

Mizque by 1767 and found that the ratio bozales : criollos was 1:1. These new data, even if fragmentary, may suggest that during this second wave the importation of slaves coming directly from Africa was higher than before. I will return on this point when considering the slavery conditions in Mizque (section 3.3. below).

Lipski (2007) reports that there are no religious or cultural practices that allow a precise reconstruction of the origin of the various regional African tribes that were introduced to Bolivia, however some surnames that persist among the members of these communities in the Yungas, namely Angola and Maconde (apparently from Congo) can provide geographic cues, as they were typically given by ship captains and slave dealers

61 to approximate the costal African areas from which the slaves were drawn. Pizarroso

Cuenca (1977:67) also suggests that a part of the Bolivian black population may be originally from Senegal, by providing regional tales that refer to this land, furthermore, he describes Zamba, a folkloric dance performed by black Yungueños, presumably from

Angola.

3.2.3.2. The role of slaves in the Bolivian economy

From the mid-17 th century to the mid-20 th century the Bolivian economy

continued to be mainly based on the mining industry and agriculture.

As indicated by the data of table 4 for the city of Potosí, the employment of slaves

never reached significant numbers in the mining industry. However, in all of Bolivia’s

urban centers, by the mid-17 th century, blacks could be found working as domestic

servants. Due to their high value in Bolivia, possessing black slaves was a symbol of

economic wealth. As before, Bolivian elite members used to demonstrate their social

status, by showing in public ceremonies their African slaves (Crespo 1977). Even though

having black domestic servants was a quite common practice among the highest class, the

percentage of Africans in Bolivia never achieved the levels found in other cities

elsewhere in Latin America. A clear case is shown by the 1778 census from Oruro,

reporting 229 black and mulatto residents, just 0.29 percent of the population; all of them

were employed as domestic workers (Busdiecker 2006). Another example consists of the

census of 18 th century La Paz, where only 350 blacks could be found out of a total

population numbering 40,000. These people not only were employed as domestic

62 workers, some of them were free artisans, working as shoemakers, knitters, tailors and carpenters (Bridikhina 1994; Busdiecker 2006).

Not suitable for the cold mining highland and employed usually as domestic servants in cities, African slaves were also used for agricultural and livestock works in several other regions of Bolivia, mainly: Mizque, in vineyards and farms, and the

Yungas, to produce coca, crucial for the support of the flourishing mining industry in the highland. In the next part of this chapter, we will focus on the available sociohistorical evidence to determine the feasibility of a creole hypothesis for these regions.

3.3. La Villa de Mizque

Besides Don Rivero’s will from the early 17 th century Mizque, Brockington

(2006) reveals interesting data concerning a Jesuit enterprise in the same region by 1767.

This time the total number of slaves encountered is 134. They were organized into two groups, the married and the bachelors. The married consisted of 110 slaves, who formed

29 households, made up primarily of young families with growing children. The other group consisted of twenty-four adult men. The Jesuits’ demographics differed from those of Rivero’s hacienda. Here 49.9 percent were bozales in contrast to 26.5 percent found in the Rivero’s will. The highest concentration of bozales was found in the men, while two out of three women were reported as criollas . Also the demographics of their origin were

different. While the majority of Rivero’s slaves hailed from Angola, the bulk of

Chalguani’s (67 percent) came from Sierra Leone, followed by 28 percent from Congo

and central-south Africa. Four slaves were Arara, from west Africa-Dahomey while one

was Fula, of Senigabria.

63 Again, no clear predictions can be done about the language spoken by these slaves as neither the number of Jesuits nor the number of free Indians is reported. Several elderly people are indicated but not as many as in the previous will. Brockington indicates that probably, when slaves became too old to work, they were manumitted to avoid paying property taxes. Interestingly, several slaves were more than 48 and one was recorded as being 71.

What can be said is that the ratio bozales : criollos is almost 1:1 and could indicate

a general tendency toward major importation of Africans toward the end of the 18 th century. The presence of some old individuals, and the existence of many nuclear families, could indicate a not-too harsh treatment.

To get a better idea of the proportions between African descendants and other ethnic groups it is possible to look at the census data provided for this region around that time. Table 3 shows the demographic figures reported by 1788 in the census conducted by the governor of the Province Cochabamba-Santa Cruz, Francisco de Viedma.

Numbers suggests that the overall percentage of Afro-descendents rose from 3.8% in

1618 14 to 6.4% in 1788, but never overcame the white-mestizo population (48.5%).

14 The territory covered by the 1618 baptisimal statistics (by Antonio de Barranco) for the Mizque-Santa Cruz dioceses coincides almost completely with the governamental Province Cochabamba-Santa Cruz. 64 Table 3. Racial distribution of the population of the provinces of Cochabamba and

Santa Cruz in 1788.

Racial Category City Total Spanish Mestizo Cholo 15 Indian Mulat to Black Number Cochabamba 22.305 6,368 12.980 0 1,182 1,600 175 Sacaba 7,614 1,249 2,290 0 3,805 269 1 Tapacarí 26,937 3,277 6,280 1,597 14,770 996 17 Cliza (Mizque) 37,615 6,682 12,192 0 16,355 2,366 20 (3,031) (643) (825) (0) (891) (672) (0) Arque 22,137 1,238 3,936 1,286 15,158 496 23 Ayopapa 8,637 1,275 1,493 0 5,620 247 2 Valle Grande 8,373 2,995 1,918 0 217 3,215 28 Santa Cruz 10,578 4,303 1,376 2,638 2,111 0 150 Total Number 144,250 27,387 42,465 5,521 59,218 9,189 416 Percentage of Total 100.0 19.1 29.4 3.8 41.0 6.4 0.3 Source: Descripcion geografica” AGN, Sala 9, Intendencia, 5.8.5, Aug, 10, 1793. This is the original manuscript on which the subsequent Bolivian publication was based, and where several of the population figures were printed incorrectly (Larson 1998:175).

Valle Grande presented the largest mulato-negro sector, 38.7 percent (3,246);

Cochabamba, 7.9 percent (1,775); Santa Cruz, 6.6 percent (150); Cliza, 6.3 percent

(2,386); Mizque, 22 percent (672); Tapacarí, 3.8 percent (1013); Sacaba, 3.5 percent

(270); Ayopapa, 1.9 percent (249) and Arque, 1.4 percent (519).

Interestingly, in Viedma’s census zambo is not mentioned as a category apart, a

possible explanation could be that the offspring of blacks and Indians is counted as

mulato . As far as this last ethnic group is concerned, it seems too large if compared with

negro . In fact while in the Jesuits’ records (1767) blacks were numerous and 49.9 percent of them were bozales , in the census of Mizque, twenty-one years later, their number appear as zero. This could indicate that some of them were not declared, to avoid taxation, or even more likely, as they could not be hidden, they were declared as mulatos, maybe free from charges or subject to inferior ones.

15 Officialy defined as a person of one-quarter white ancestry and three-quarters Indian ancestry. 65 Unfortunately, Viedma’s census does not provide us with data concerning black population in the Yungas at that time. However, an important piece of information about this tropical region is encountered in the Jesuit document presented by Brockington. The

Yungan valleys are mentioned as a possible destination for three slaves that used to be employed by the Jesuits. There is evidence therefore that by 1767 the Yungan haciendas

already employed African workforce. This piece of information will be usefull in the

following discussion concerning the colonization and settlement of Los Yungas.

In sum, this section was concerned with the potential development of a creole

language in the Mizque area around the 18 th century. Data from the Jesuit enterprise (year

1767) do not imply that such a creole existed. However, they can be taken as evidence

that the ratio bozales : criollos probably augmented during this 2 nd wave of immigration.

On the other hand, Viedma’s census suggests that the overall Black/White ratio for the region was low for a full-fledged creole to emerge.

3.4. Los Yungas

After this broad description of African slavery in Bolivia, we have finally come to talk about the region which interests us the most, Los Yungas, in the Department of La

Paz, where ABS is still spoken. Unfortunately, the date of the introduction of forced

African labor to Los Yungas is not clear. In a document dated 1805 by Francisco Xavier de Bergara, overseer of a Yungan hacienda , the Marquesa de Haro is mentioned, as the first owner having employed Africans in that area for some time (Crespo 1977:105). This might indicate that black slaves started being employed significantly in the Yungas not

66 earlier than the last decades of the 18 th century, and therefore it could potentially coincide with what encountered in the Jesuits’ 1767 records.

Several documents are reported by Portugal Ortíz (1977) concerning the sale, or purchase of one or two slaves at a time by this period. An example is the case of an 18 years old girl, who moved to the Yungas in 1773. In this record it is said that she was originally purchased in Potosí and after residing for a while in Oruro, with her last owner, she was now sold to a new master to work in the tropical valleys. Another case in 1761 is the purchase of a married couple from Angola by a priest living in Chulumani (1977:78).

These small transactions, concerning only a couple of slaves at the time, support

Bridikhina’s (1995a) and Lockhart’s (1994) view, regarding the high price of slaves as a powerful constraint on the massive employment of black workforce. According to them, those who possessed several slaves often accumulated them during a long period. It could be the case of Don Antonio de Tejada, who at the time of his death (1806) owned 41 slaves in a Yungan hacienda (Portugal-Ortíz 1977:80).

On the other hand, in Pizarroso-Cuenca (1977b:74), the story of ‘King Bonifacio’ is told, who sometime around 1600 was recognized by the black population of the

Mururata hacienda as the king and leader of the Afro-community. The story says that at the beginning of the 17 th century a group of slaves taken from Senegal or Congo, after

being transported to Peru and eventually to Los Yungas, arrived to the Mururata

hacienda , belonging to Marqués Pinedo. Soon, some of the slaves already employed in the hacienda realized that one of the newcomers was the king of an African tribe, taken as prisoner and enslaved. They could not tolerate such a humiliation for their king and for this reason they agreed with Marqués Pinedo to work extra hours to free him. The slaves

67 also built him a house and the freed slave came to be known as King Bonifacio Pinedo or

Negro Bonifáz. While the story concerns a fact that had probably happened, as it refers to a monarchy that is still present in the Afro-Bolivian community, the date indicated as

1600 could have been reported wrongly as neither documents nor records of any kind indicate significant African presence in Los Yungas by that time. Besides, Portugal-Ortíz notes that, by the 18 th century, the Yungas was only partially colonized and in the case of

the Coroico area, by 1736 the lands were taken for the first time from the “ enemigos

ynfieles ”, non-christianized Indians, to be given to Don Buena Aventura Joseph

Rodríguez (1977:76). As the Mururata hacienda borders the Coroico one and the distance

between the two villages is no more that one hour walk, it is unlikely that one could have

been colonized more than 136 years before the other. 16

The earliest pertinent reference to Yungan slavery is a Church record of the death

of a mulato in 1703 in the village of Chicaloma; according to Leons (1984) the reduced

mention of Afro-descendents in church records at that time, till the first decades of the

19 th century, does not seem to be due to missing records, but rather to a really limited

number of slaves in the region. The first document providing concrete information on the

black population in the Yungas is from the early 19th century. Crespo (1977) raises a question about the relative proportion of black labor employed in agriculture and provides the following answer:

16 Moreover if the colonization proceeded from La Paz, Mururata would have hardly been conquered before, because it remains after Coroico. 68 “Talking about Yungan haciendas in the Department of La Paz,

not all of them used this kind of workforce, and at least in some regions

of this area, those that employed it were a minority” (1977:109).

He mentions a document written by the Chirca priest in 1802, where he accounted

for the percentages of blacks and Indians in 23 local haciendas . Of these chacras 17 , only

4 were employing black workforce, namely: in Guayraoata 15 negros or mulatos out of

65 workers; in San Agustin 17 out of 28, in Yacata 23 out of 128, and in Collpar 28 out

of 142. Crespo also adds that in 1802, Ocabaya, Yungas had a population of 32 blacks, 80

mestizos , 94 Spaniards, and 643 Indians.

No other data are available for the Yungas till 1883. For this year comprehensive information concerning the demographics of whites, Indians, mestizos , and morenos

(accounting for negros , mulatos and zambos taken together) is disclosed for two

haciendas: Pacallo and Mururata, where the highest concentration of morenos is found.

Pacallo had 67 whites, 63 mestizos , 340 Indians and 56 morenos ; Mururata numbered 55 whites, 183 mestizos , 236 Indians and 324 morenos . Moreover, mentioned without specific ethnic group separation: Chulumani with 14 morenos out of a total of 220 inhabitants; Tajmo, Calupre, Chigno, Chimasi, Tolopala, Suquillo with 49 out of 902;

Coroico with 113 out of 5,335; Impata with 252 out of 2465; Coripata with 315 out of

3,867; Chupe with 240 out of 1212; and Lanza with 102 out of 8,255.

The general data provided for these last haciendas indicate that overall the black population was a very small minority. Even in Pacallo and Mururata, with the highest

17 Small pieces of land used for agriculture. 69 concentration of African descendents, the ratios morenos : whites+ mestizos (who supposedly spoke Spanish proficiently, being offspring of Spanish mother or father), are

56:130 for Pacallo and 324:233 for Mururata. Morenos outnumber whites and mestizos only in Mururata. However even here, the feasibility of the creole origin at this time does not hold as probably the majority of the slaves was made up of criollos 18 , and the morenos not only included blacks, but also mulatos and zambos .

In sum, the historical evidence collected seems to indicate that Los Yungas were

settled and colonized only around the 18 th century. The only clue which may suggest a

previous settlement is the orally-told story of King Bonifacio. In fact, as reported by

Pizarroso-Cuenca (1977b), the legend would indicate that significant African workforce

had already been introduced in the Yungas by 1600. However, no data reporting African

presence in the Yungas by that time is known; on the other hand, historical documents

confirm that by 1736 the region had still to be completely colonized and the employment

of black workers was not massive: small slave transactions and a relative small ratio

between blacks and whites appear to have characterized the African presence in this area.

Moreover, Bonifacio’s coronation indirectly indicates that manumission was quite

possible in the Yungan community. We will explore this and other social issues in the

next section. The identification of codes of social interaction, as well as the hacienda community setting of Yungas will help us get a better understanding of the possible origin of ABS.

18 As the traffic was declared illegal in 1826 but was completely eradicated only by 1848, it is likely that importations decreased significantly during such period of time. 70 3.4.1. The nature of Slavery in the Yungan Hacienda and in Bolivia

Data by the late 18 th century indicate clearly that the Yungan hacienda was not a plantation society; rather, it consisted of small and medium properties mainly dedicated to the production of coca (Busdiecker 2006). As Busdiecker (2006:38) points out, Soux

(1993) describes the organization of labor in the hacienda of Dorado Chico during the early 19 th century. She clearly shows that slaves were not the only source of labor. Soux explains that three kinds of workers were employed: mingas , peones, and slaves. Mingas could be Indians or freed blacks, they were occasional paid laborers. Peones, who were also Indians or freed blacks, were employed in times when extra work was needed; they were given the usufruct on a piece of land, which they could work in order to keep the crop for themselves. They were given free days to work the land assigned. Black slaves were employed for everyday’s work in the hacienda , they received some extra provisions as payment as they could not grow their own food.

Due to the unhealthy climate, there were many casualties among the hacienda workers. For this reason landowners were forced to bring in new workforce. Sometimes slaves but also free Indians, as the price of blacks was high in the colony and the death of a slave represented a big capital loss (Busdiecker 2006).

With respect to slave versus non-slave labor, in the above mentioned document by

Francisco Xavier de Bergara (1805), an economic study called ‘ Demonstracions

matematicas’ ‘Mathematic demonstrations’, the author provides evidence that black slaves were not a convenient business and should have been replaced with free native workers. Bergara enlisted several reasons to convince the hacienda owner, Don Antonio

Sáez de Tejada. Bergara’s observations indirectly provide us with a good idea of the

71 social patterns encountered in the Yungas at that time. Owners had to provide slaves with food, shelter, clothing and Christian education, estimated by Bergara as a cost per-capita of 1.5 daily reales . Besides, the owner also had to pay for Church rights: wedding (12

pesos ) and funeral (12 pesos ). Moreover, Africans were likely to rebel, and therefore extra overseers were needed to control a group of them. To catch a runaway slave could cost from 25 to 100 pesos, while the death of one of them, often for illness, would have represented the loss of the total investment.

In this hacienda , Africans and Indians received the same daily wage, consisting in three reales for the men and two reales for the women and the children. However, while

slaves implied an initial investment of capital (360 pesos a man and 410 a woman),

Indians, being free, did not represent such cost. Busdiecker (2006:38) also highlights that,

according to Soux (1993), in Dorado Chico all workers ( mingas , peones and slaves) were

supposed to receive equal payment; however, in practice, only the mingas received

money; the rest were generally given provisions (e.g., corn, potatoes, etc.).

According to Bergara, part of the problem with black workers, when compared to

Indians, was due to the many days off slaves had: 79 a year, considering Sundays and

Catholic holidays. Besides, periods of pregnancy, illness or just childhood would have

implied a lower production, which did not free the owner from feeding and clothing the

slaves.

Generally, the treatments reserved to slaves in Bolivia were not as harsh as those

encountered in other slaving societies in the Americas. This was due to the high prices of

blacks in the region and to the fact that owners wanted to protect their costly investments.

This said, some cases of mistreatment can be encountered. For example, there are legal

72 documents of the 17 th century recording the case of a group of slaves that escaped from the owner in Potosí. When captured, the tribunal established that the reason for the flight was the hunger and the bad treatments suffered by the slaves. However, the court decided to punish the blacks with 50 whippings, even though it could be proved that they did not steal anything from their master (Crespo 1977:27). Another case is one of a six year old girl from La Paz, whose owner damaged her eye, and broke her arm (Bridikhina

1995b:60).

Collective acts of violent black resistence were rarely encountered in Bolivia. A

slave rebellion took place in Mururata in 1795 and the hacienda owner, Ignacio Pinedo,

had to call army troops to suppress the uprising. A similar act of resistence happened

again in Murutrata in 1805; while in 1809 Santa Cruz experienced another case of

collective revolt (Busdiecker 2006).

Busdiecker (2006:41) indicates that in the absence of prominent collective revolts,

blacks usually tried to pursue freedom in non-violent ways; for example, they registered

legal complaints against owners, they tried to get new owners, and they tried to buy their

own freedom. Crespo (1977) suggests that paying for one’s freedom must not have been

easy due to the reduced salary received by the slaves. However, Portugal-Ortíz shows

that this was a quite widespread practice, an example he mentions is the one of Juan José

Nieto (1795), who paid Antonio Sáez de Tejada (Yungan owner), 400 pesos to free him.

Portugal Ortíz tells that to find the money Nieto asked Don Pedro Oquendo for a loan,

promising him that he would have paid it back in 4 years (100 pesos a year). This and

other evidence provided by Portugal Ortíz, in addition to the story told about Rey Bonifáz

73 (Pizzaroso-Cuenca 1977b), suggest that probably the Bolivian hacienda was quite a flexible system if compared with other enslaving societies elsewhere in America.

In sum, the social conditions encountered in Los Yungans seem to have been of a relatively mild kind. In fact, the Yungan hacienda was not a plantation society, it was rather organized in small farms; blacks not always were slaves, they were also working as mingas and peones ; the costly investment represented by the slave for the owner led -

supposedly- to more human, less brutal treatments; finally, manumission was relatively

common.

3.5. Conclusions

This chapter has provided an overview of the evolution of Afro-Hispanic contact varieties in the New World and placed the case of ABS in space and time within such a context.

Findings strongly undermine the creole hypothesis for ABS (Lipski 2008b).

Several factors have affected the dimension of African slavery in Bolivia and consequently the presence of black population in the territory from the 16 th century through the middle of the 20 th century. In fact, the Spanish Crown’s monopoly on slave

trading, the geographic location of Bolivia and the availability of native workforce

affected the cost of Africans, raising their price and, as a result, reducing the number and

the dimension of slave transactions. The non-massive introduction of black workforce

into the territory favored the acquisition of a closer approximation to Spanish by the

slaves.

74 None of the records analyzed lead to the conclusion of a possible creole development during the time span of four centuries of slavery, including the first and second waves of forced migration (Brockington 2006). Nevertheless, some demographic records seem to suggest an overall higher introduction of bozales during the second period. As far as the Yungan hacienda is concerned, the available sociohistorical evidence strongly discouraged the possibility of nativization of an earlier Afro-Hispanic pidgin (Lipski 2008b:186): low Black/White ratio and relative high social mobility seem to have characterized the nature of slavery in this area.

75

CHAPTER 4: ACCOUNTING FOR THE AFRO-BOLIVIAN SPANISH

‘CREOLE-LIKE’ FEATURES

“There is little awareness of semi-creoles. The term

is not yet generally established, and most such

languages have been little studied as yet, apparently

having been presumed to be less interesting than the

‘true’ creoles. However, they are of interest in that

they are pre-creoles, those that haven’t quite made

it – varieties that have originated in areas and

under circumstances that might have led to the

creation of a creole but did not go all the way. I

believe that studying these varieties and comparing

them to full creoles may provide some interesting

insight into the notion of creoleness itself”

(Schneider 1990:83).

4.0. Introduction

This chapter analyzes the features of ABS which have been suggested as possible indicators of its more radical creole origin, in opposition to the existence of a stable, but

76 not creolized, variety which coexisted with stSp since its inception (Lipski 2008b). In fact, as mentioned earlier, Lipski suggests that ABS might be the result of the nativization of an earlier pidgin by claiming that “in absence of any other viable scenario, Afro-

Yungueno Spanish must be viewed as the descendant of a colonial Afro-Hispanic pidgin

(2008b:186)”.

Before proceeding with the analysis, I have to point out that there is not common consensus among linguists on what a ‘creole language’ is and on how the words ‘creole’ or ‘creoleness’ should be used. In fact, some proposals describe these varieties entirely in term of their cultural and social history (Mufwene 1997; DeGraff 2005). Other accounts focus on their stages of formation, indicating that creoles are the byproduct of a break in transmission of a native language to following generations (Thomson and Kaufman

1988), or highlighting that these vernaculars should be seen as nativized forms of early pidgins (Romaine 1988; Muhlhausler 1997). Some definitions attempt to classify creole languages according to their structural features (Bickerton 1981; Seuren and Wekker

1986; McWhorter 1998). Finally, comparative cross-linguistic analysis based on presence/absence of linguistic features and relative distance from the lexifier language gave birth to new terms such as ‘semi-creoles’ (Schneider 1990; Holms 1992),

‘intermediate creoles’ (Winford 2000) and ‘partially restructured languages’ (Holms

2004), varieties which would be place somewhere on a linguistic continuum going from

‘radical creoles’ to their lexifiers.

Due to this wide range of definitions, I will have to put aside terminological debates and employ for convenience Lipski’s (2008b) structural account of ‘creoleness’.

In fact, even though the author does not deny the importance of social, historical and

77 cultural factors in creating and shaping creole languages, he provides a feature-based account to determine where to place ABS with respect to other contact varieties and speculate on whether ABS might have been linguistically further away from stSp in the past, and in this sense, more radical.

Given that the historical background so far provided does not suggest the presence of a radical creole in the Yungas, this chapter will explore whether the grammatical elements reported by Lipski could be due to a different development, not necessarily linked to a previous Afro-Hispanic pidgin stage.

I will suggest that the presence of these features in ABS is the result of processes of intermediate and advanced second language acquisition, which left room in this dialect for much of the morpho-syntactic patterns encountered in stSp. The development and crystallization of this contact variety took place in isolated rural communities, unaffected by standardization processes imposed by urban society and linguistic norm. The outcome is a vernacular – perfectly understandable by the monolingual standard Spanish speaker – which carries the undeniable trace of second language learners’ strategies.

The model here proposed for ABS not only considers the importance of social factors in patterning the presence of creole-like features into this dialect; it also attempts to explain why certain constructions might have emerged as a result of universal processing constraint on learnability posed by the human mind.

The present chapter deals with concepts which have been largely addressed by the linguistic literature on creole formation, second language acquisition and sociolinguistics; crucial to our discussions are the theoretical frameworks provided by the Interlanguage

Hypothesis (Plag 2008a,b 2009a,b), Processability Theory (Pinamann 1998, 2005), and

78 Vernacular Universals (Chambers 2003; 2004); such paradigms will be explained in the following sections.

4.1. Afro-Bolivian Spanish “creole-like” features

Before considering the aforementioned frameworks, let us have a look at the ABS features which have been suggested as indicators of a possible more radical creole origin for this dialect.

After extensive fieldwork in the Afro-Bolivian communities, John Lipski (2007)

was able to identify the most conservative features characterizing the traditional ABS

dialect. His work must not have been easy, given that, as he says, “full active competence

[in the traditional vernacular] is probably limited to at most a few hundred [elderly]

individuals, possible even fewer” (2007:178).

In fact, from 1952, year of the Land Reform, which freed Afro-Bolivians from

forced peonage, black communities had access to public education, and their members

were able to travel without having to ask owners for permission. As a consequence of

schooling and mobility, ABS has been undergoing a change, consisting of the systematic

substitution of basilectal Afro-Bolivian features with more prestigious stSp ones (Lipski

2007). As a consequence of this recent change, many of the characteristic ABS features

are partially or totally absent from the speech of the youngest generations. In Table 4,

Lipski (2008b) reports the features of ABS which he selected as potentially indicative of

a prior stage of creolization.

79 Table 4. Examples of Afro-Bolivian traditional features, salient for the creole origins hypothesis.

Phenomenon Example Elimination of definite articles in Ø perro ta flojo [ los perros están flojos] `dogs are subject position worthless’; Ø patrón huasquiaba Ø mujé [ los patrones huasqueaban a las mujeres] `the landowners beat the women’; Use of the Spanish 3 rd person Nojotro tiene [nosotros tenemos] jrutita `we have fruit’; singular as invariant verb form for yo no conoció hacienda [yo no conocí hacienda] `I never all persons and numbers knew the haciendas’; yo miró jay [yo miré] `I saw [it]’ Constructions based on invariant ¿Quién ta comprá? [¿quién está comprando? ] `who is ta(ba) + INFINITIVE instead of buying [coca]?’ conjugated verbs Constructions based on invariant ya Furlano ya murí [murió] ‘so and so just died’ + INFINITIVE instead of conjugated Ya viní [vino] temprano tia Francisca ‘Francisca came verbs early’ Constructions based on invariant va Nojotro va trabajá [nosotros vamos a trabajar] ‘We are + INFINITIVE instead of conjugated going to work’ Yo va recogé mi lena [Yo Voy a recoger verbs mi lena] ‘I’m going to get my firewood’ Non-inverted questions ¿Cuánto hijo pue oté tiene? [¿cuántos hijos tiene usted? ] `How many children do you have?’; ¿ande pue oté viví? [¿dónde vive usted? ] `where do you live?’ Suspension of grammatical gender Tudu un [toda una] semana `a whole week’; nuestro in nouns and adjectives cultura antigo [nuestra cultura antigua] `our traditional culture’ Use of nuay [Spanish no hay `there Yo nuay cajué [no tengo café ] `I don’t have any coffee’; is/are not’] and nuabía [Spanish no Ele nuay ningún marido nada [ella no tiene ningún había `there was/were not’] instead marido ] `she does not have any husband at all’ of no tener `to not have’ Use of tener `to have’ instead of Tiene un negrita qui taba aquí [había una mujer negra haber `to exist’ to express existence que estaba aquí ] `there was a black woman who lived here’; Tenía un señora, un negra `[ había una señora, una negra ] `there was a woman, a black woman’ Occasionally, double negation with Ningun misa nada [ninguna misa (para nada)] ‘no mass nada ‘nothing’ at all’ Yo no va i nada [yo no voy a ir (para nada)] ‘I am not going at all’

In Table 2 Lipski compares the same Afro-Bolivian traditional features with those encountered in other Spanish/Portuguese ‘creoles’ (2008b:184), to determine the status of

Yungueño Spanish with respect to them.

80 Table 5. Key structural features of Yungueño Spanish and selected creole languages (Lipski 2008b:184).

Yungueño Helvecia Palenquero Papiamentu Chabacano Gulf of Cape Guinea Verd e Null def. art. Yes Yes Yes Yes No Yes Yes Invariant Yes Some Yes Yes Yes Yes Yes verb form for person and number TMA Maybe No Yes Yes Yes Yes Yes particles Non-inverted Yes Yes Yes Yes Yes Yes Yes questions Pluralizing No No Ma 3 pl. ( nan ) Mga Some 3 No particle/3pl. pl. No gender Yes Some Yes Yes Yes Yes Yes concord in NPs Subject No (except Yes Yes (except Yes No Some No pronouns as yo after (except 1s.) object prep.) 1s.) ‘to have’ as some Yes Yes Yes Yes, in Yes Yes existential affirmative verb Negation Preposed Double Postposed Preposed Preposed Double Prepo (occasional NEG NEG/ sed double postposed NEG with nada ) NEG derived Yes Yes Yes Yes Hendeq and Na ... f(a) Yes from no/não no Postposed No No Yes No No Yes No NP as possesive Serial verbs No No No Some No Yes Few Predicate No No Yes Yes No Yes No clefring

The label ‘creole’ has been applied to a wide variety of contact vernaculars, which often differ significantly in their structure. For this reason, linguists have tried to classify these languages in relation to other outcomes of contact, and to make distinctions among creoles themselves. The byproduct of this effort resulted in new terms such as ‘semi- creoles’ (Holm 1992) or ‘intermediate creoles’(Winford 2000), languages which would

81 be placed somewhere on a continuum, ranging from close approximations to the lexifier varieties to vernaculars structurally further away from them, which would be unintelligible to a speaker of the lexifier.

These languages can be placed on a cline reflecting varying degrees of distance from their lexifier languages, and closeness to their substrates (Parkvall 2000; Winford

2000).

Alleyne (1980:181) argues that:

Afro-American dialects can be plotted on a scale representing

different degrees of transmission of West African elements, and

differentials in degrees of transmission that are to be explained by

differences in sociolinguistic circumstances in each area.

In other words, these languages can be placed on a continuum ranging from close approximations to L2 to radical creoles.

A continuum of outcomes involving degrees of substrate and L2 input.

Slight substrate retention Moderate Extreme L1 retention

Advanced ‘Indigenized’ Intermediate Basilectal Radical Interlanguage varieties creoles creoles creoles

(Winford 2000:216)

82 Based on the information presented in Table 2, Lipski concludes that “at least impressionistically, Afro-Yungueño Spanish is more creole-like than Helvécia (Brazilian)

Portuguese. At the same time, Afro-Yungueño Spanish seems less ‘deep’ than the creoles of Palenque (Colombia), São Tomé or the heavily substrate-influenced

Chabacano varieties of Philippine Creole Spanish” (2009:183).

I will now proceed with a closer analysis of the Afro-Bolivian features presented in Table 2, because even though Yungueño Spanish seems more ‘creole-like’ than

Helvécia, it is worth pointing out that many of the key features mentioned as prototypical of Spanish and Portuguese creoles can often be found in advanced second language varieties and less educated/rural dialects of Spanish and Portuguese, for which a previous pidgin phase or a more ‘radical’ creole stage is not feasible.

Note that by identifying many features also common to SLA, I am not necessarily proving that the creole hypothesis for ABS is wrong. In fact, some creolists suggested that some of the features encountered in creoles are also typical of interlanguages of an earlier stage (Plag 2008a,b; 2009a,b). This said, it is also important to observe that if the features encountered in ABS are not only typical of an interlanguage at an earlier stage, but rather they tend to be encountered also in the speech of very advanced learners, and in some cases in that of less-educated speakers, the bases for a realistic creole hypothesis become less solid; especially if they are the only cues potentially leading to it, and the historical evidence suggests a non-creole origin. Let us have a closer look at the ‘creole- like features’ encountered in ABS.

Null definite article, a feature considered key in the evolution of creoles by some linguists, e.g. Bickerton (1981), can also be encountered in other Spanish/Portuguese

83 dialects, such as Chota Valley Spanish (Lipski 1982) and Brazilian Portuguese (Guy

1981; Munn & Schmitt 2001; Schmitt & Munn 2003), which could hardly be classified as creoles.

According to Bickerton, creole languages have an article system with “a definite article for presupposed-specific NP; an indefinite article for asserted-specific NP; and zero for nonspecific NP” (1981:56). As we will see in more details in chapter 7, this description does not capture the features of ABS article system used by the eldest speakers. In this vernacular, in fact, there are three definite articles ( el , la , lu ) and an indefinite one ( un ). Their distribution resembles standard Spanish with the exception that bare nouns can take on either plural/singular, specific/non-specific/generic readings, given proper syntactic and pragmatic contexts (see chapter 7; cf. Gutiérrez-Rexach and

Sessarego (2010). Moreover, it is well known that second language speakers, coming from a first language with a different article system, or with no article system at all, tend to present bare nouns and article mismatches even at very advanced levels (Sànchez and

Giménez 1998, Leonini 2006).

The use of the Spanish 3 rd person singular as invariant verb form is a quite common feature of second language acquisition and can also be encountered in varieties which present no African influence, such as transitional Spanish-English bilinguals through the United States (Lipski 1990). In addition, ABS verb forms clearly inflect for tense and aspect (e.g. cantó ‘he sung’ and cantaba ‘he was singing’), while more radical

creoles tend to employ an invariant verb form for all persons, numbers, tenses and

aspects.

84 The TMA features claimed for Afro-Bolivian Spanish seems to have been directly derived from Spanish, in which they are used in much the same way. Constructions based on invariant ta(ba) + INFINITIVE instead of the usual gerund ending in –ando or -iendo

can also be incountered in the non-creole Bervolento Spanish (Lipski 2005; Díaz-Campos

& Clements 2008). Besides, Lipski acknowledges that “Afro-Bolivian speakers

effectively regard the use of ta + verb as a performance phenomenon, and indeed a

comparison with the remainder of the Afro-Yungueño corpus points to phonetic erosion

[of the gerundive form] and unguarded speech as the origin of this construction. As such,

it cannot be considered an integral part of the verb system of their dialect” (2009:123).

In addition, the analysis of ta as a creole preverbal marker should always be

considered with much caution. Unless functions significantly divergent from Standard

Spanish está ‘he/she is’ are clearly documented, and for ABS this not the case, the creole

TMA status can be questioned for ta because /s/ in coda position is often aspirated or deleted in this langage.

Constructions with ya as a perfective particle are ambiguous and not systematic.

Lipski could find only few examples. The cases that he provided reported in Table 1 and here in (9-10), correspond to similar sentences in Standard Spanish.

(9)

Furlano ya murí [murio]

‘So and so just died’

(10)

Ya vení [vino] temprano tía Francisca

‘Francisca came early’

85

Lipski suggests that these sentences could correspond to an infinitive verb form preceded by a TMA particle (either ta or ya ), however given the similarities between these constructions and the standard ones and the general tendency toward simplifying verbal paradigms, these few realizations do not prove the status of ya and ta as creole-like

TMA markers.

Besides, in several articles concerning the status of bozal Spanish, Lipski argues

that these particles are somehow problematic and should not be taken as Spanish

pidgin/creole quintessentials.

An example is Lipski (2000), who states:

Another source of ta + V inf constructions in Afro-Caribbean Spanish

is phonetic reduction of the gerund, suggested by such half-reduced

constructions as pavo real TA bucán [

peacock is looking for [a] tree’ (2000:442).

It is difficult to unequivocally assign particle status to these

elements, since ya is a commonly-used sentential adverb in Spanish

(2000:445).

Also the possible use of va as future/irrealis particle is somehow dubious, and

Lispki himself points it out (2009:125-127). Since all dialects of Spanish use periphrastic

future constructions based on ir ‘go’ + a ‘to’ + infinitive with future reference, and since

86 the preposition a is absorbed phonetically by the 3s. verb va (< va a ), the only feature which distinguish Afro-Bolivian Spanish from other non-contact varieties is the lack of subject-verb agreement, which is commonly found, as we saw, in SLA.

After presenting his corpus data on the nature of these particles, Lipski concludes:

“There is no evidence that ta , va and ya once had particle status. Since there are no

written attestation of earlier varieties of Afro-Yungueno speech (nor of contemporary

varieties), and since the communal memories of the speech of earlier generations have all

but disappeared, the evolutionary path of ta , va and ya cannot be reconstructed with

certainty” (Lipski 2008b:127).

Non-inverted questions have also been linked to a possible creole hypothesis.

Lipski (2005) maintains that non-inverted questions seem to appear in Spanish varieties

in contact with creoles, such as the case of the eastern Guiria Peninsula area in contact

with Trinidad creoles; nevertheless, non-inverted questions appear also in non-creolized

Caribbean Spanish dialects such as Cuban, Puerto Rican, Dominican, Panamanian,

Venezuelan and costal Colombian (Díaz-Campos & Clements 2008); moreover, non-

inverted questions seem to appear cross linguistically also in very advanced stages of

SLA (Pienemann 1998, 2005), thus indicating that they are not only indicative of radical

creoles.

Suspension of grammatical gender across the Determiner Phrase may be a

second-language acquisition phenomenon rather than a creole feature. It can be observed

not only in the speech of advanced students learning Spanish, but in the speech of all

advanced students learning a L2, especially if coming from a L1 lacking ‘gender’ as a

feature (Franceschina 2002). It can also be encountered in the native language of many

87 speakers of contact varieties for which a creole hypothesis is not suitable but a crystallized second language varieties seem more plausible, such as Chota Valley Spanish

(Lipski 1982).

Moreover, it must be said that ABS does not lack gender agreement. All ABS speakers, even the eldest ones, can clearly distinguish between masculine and feminine nouns. The main distinction between ABS and standard Spanish concerns the DP elements specified for agreement. In fact, while in standard Spanish adjectives, articles, demonstratives and quantifiers all agree in gender with the noun, the operation Agree for this feature is just limited to definite articles in ABS (Sessarego and Gutiérrez-Rexach

2009a,b). The limitation of nominal gender agreement to these determiners inherently indicates the presence of the feature ‘gender’ in the dialect. In this respect, Yungueño

Spanish diverges quite significantly from the majority of the Romance-based creoles, which are generally supposed to lack gender features.

Distinction between existence and possession is generally maintained between tener ‘to have’ and non-auxiliary haber ‘to exist’:

(11)

Oté tiene coca?

‘Do you have coca?’

(12)

Negros pues había aquí

‘There were black people here’

88 However, sometimes nu hay (Sp. No hay ‘there is not’) and nu había (< Sp. No había ‘there was not’) can be used to express lack of possession:

(13)

Yo nu hay cajué

‘I do not have coffee’

(14)

Yo nu había zapato pa vistí

‘I did not have shoes to wear’

The employment of tener as existential is only occasional, but may occur:

(15)

En la mesa tiene gallina

‘On the table there are some chickens’

Lipski suggests that the presence of these forms indicates that in earlier stages the language presented greater overlap between existential and possessive constructions

(2008b:181). These traces may be indicative of a possible partial restructuring; however, this phenomenon does not seem enough as to claim a more radical creole origin for

Yungueño Spanish 19 .

19 It is also important to notice that some of the slaves who arrived in the Bolivian Yungas might have spent some time in Brazil or might have spoken some Portuguese, the language used by the slave traders dealing with Spanish America. As it is well known, Portuguese employs the verb tener where Spanish haber would be used instead. This hypothesis may be further supported by the fact that the last name Angola is common among Afro-Bolivians and Angola was a former Portuguese colony.

89 Cases of double negation with nada are encountered too, even though they are not very common. Double negation with nada , which in several cases of contact has been ascribed to Kikongo substrate influence (Schwegler 1991), can also be heard in other

Spanish/Portuguese varieties in which they seem to be more likely to be due to recent innovation e.g. Brazilian Portuguese (Schwenter 2005). Moreover, Lipski points out the presence of such structures also in Puerto Rican Spanish and Venezuelan Spanish

(2008b:138):

(16)

No es dificil ná

‘It’s not difficult’ {Puerto Rican Spanish}

(17)

No llegué nada

‘I did not arrive at all’ {Venezuelan Spanish}

Negation derived by no/não is present in all Spanish dialects; while the double negation derived by these particles is restricted to a sub-group of them. According to the results of my grammaticality judgments, Afro-Bolivian Spanish does not exhibit double negation based on no … no pattern, e.g. no me ayuda no ‘he does not help me at all’.

Furthermore, Lipski admits that in his recorded data these cases could rather be of the kind no me ayuda, no , with a ‘ no’ that serves as “afterthought”, outside the main clause.

90 On the other hand, features which are generally assumed to belong to more radical creoles, e.g. postponed NP as possession, serial verbs 20 , and predicate clefting, are

completely absent in Afro-Bolivian Spanish.

Therefore, the ABS features identified by Lipski (Table 2) as potential indicators

of ‘creoleness’ on which we can really make a claim as effectively belonging to the core

grammar of traditional ABS are:

a) Cases of null definite article in subject position.

b) Use of Spanish 3 rd person singular as invariant verb form for all persons and

numbers, but use of verb forms clearly inflected for tense and aspect (e.g. cantó “he

sung”, cantaba “he was singing’).

c) Non-inverted questions.

d) Gender agreement restricted to definite articles.

e) Some cases of ‘to have’ as existential verb.

In addition, it must be remembered that, with exception to these suggestive

features, -which however are not necessarily indicative of prior pidginization- the

majority of the morpho-syntactic patterns encountered in Standard Spanish, is also found

in ABS.

If ranking the features found among contact languages in terms of “creoleness”

cannot provide us with a precise quantitative result, it can at least give us an idea of

where a certain language stands with respect to others. It is clear that ABS is a contact

20 Note that not all creoles have serial verbs. Serial verbs constructions are encountered in creoles which have usually been highly influenced by the substrate languages. Such structures are not considered as typologically ‘unmarked’, therefore, only if the substrate had it they might have appeared in the creole (See Bruyn (1996) for a more detailed account on this issue). 91 variety presenting several key differences from standard Bolivian Spanish; however, even in its most basilectal variety, this dialect would be perfectly intelligible by any standard

Spanish speaker.

Yungueño Spanish is therefore a vernacular which derived much of its structure from Spanish, but which, at the same time, carries on morphological simplifications and regularizations. These elements seem to be the remaining traces of crystallized second language strategies, rather than the evidence of the existence of a more radical creole.

The communalities between traditional ABS and Standard Spanish would hardly be the result of a change that took place in the last 50-60 years, especially because the speakers of this vernacular are old people who were at least 30 years old when the Land

Reform (1952) happened. As a consequence, they did not experience any formal education and spent their entire life in these rural Yungan communities, where they continued to carry out agricultural work. Claiming a radical decreolization, in this specific case, would imply that these people since the Land Reform had such an intensive contact with standard Spanish to where that they could learn a very close approximation to it at an already advanced age and, at the same time, abandon almost completely their creole-like mother tongue. This scenario is not likely.

Note that my claim does not imply that after the Land Reform ABS did not go through a process of approximation to the standard variety. This is something that I am not questioning, and that is clearly visible by looking at the evolution of inflectional morphology across generations (Sessarego 2009a,b; Sessarego and Gutiérrez-Rexach

2009a,b; Sessarego and Delicado-Cantero 2009; Delicado-Cantero and Sessarego 2009).

92 What I am actually arguing against, is the hypothesis that ABS could have been a radical creole before 1952, which underwent a drastic decreolization since that date.

Could it be that the traditional dialect we know was itself the byproduct of a previous process of approximation to standard Spanish? Answering this question is not an easy task, as no written documents of early Yungueño Spanish exist. However, I think that the historical and linguistic evidence so far provided suggest that probably it was not the case. My data seem to indicate that ABS was not a radical creole, it probably was a vernacular which approximated quite closely Spanish from its inception, and in which processes of second language acquisition left an undeniable trace.

Recalling Schneider’s words from the initial quote, if we assume a continuum of

linguistic creoleness, ABS could probably be characterized as one of the “varieties that

have originated in areas and under circumstances that might have led to the creation of a

creole but did not go all the way” (1990:83).

4.2. The Interlanguage Hypothesis

As I mentioned in the introduction to this chapter, I will try to address several

paradigms which have been treated by different sub-fields within linguistics, and which, I

believe, can provide us with interesting insights into the possible presence of certain

features in ABS. I want to discuss the relevance of Plag’s (2008a,b; 2009a,b)

Interlanguage Hypothesis hypothesis of creole formation to the case of Afro-Bolivian

Spanish and determine to what extent it explains the grammatical patterns encountered in

this language.

93 In the last decades the view that sees creoles as particular forms of interlanguage has grown among linguists (Kouwenberg & Patrick 2003; Lefebvre et al. 2006; etc.).

Ingo Plag (2008a,b; 2009a,b) is one of the main promoters of such an approach to the creole genesis. He formulated the Interlanguage Hypotheis, which I report below:

(18) Interlanguage hypothesis

Creoles are conventionalized interlanguages of an early stage.

As he points out, this hypothesis is not completely new, as it had already been suggested by Andersen (1980; 1983). However, what makes Plag’s suggestion attractive as well as relatively innovative is the combination of such a hypothesis with recent theories of SLA processing (Pienemann 1998; 2005). Also Kouwenberg and Patrick

(2003) have argued in favor of the relevance of second language processes in creoles formation, especially in the early stages of contact, characterized by rudimentary acquisition of the (socially) dominant language by the African slaves.

Such a situation would be prior to the stage of target shift (e.g. Baker 1990), consisting of a second stage of creole evolution in which the next generation of speakers acquires the new medium of interethnic communication (cf., e.g., Baker 1994), and no longer the superstrate language. This next generation of speakers may consist of first language learners and second language learners of the new variety, e.g. newly arrived slaves, as in the case of the Caribbean plantations. Hence, at this stage, processes of SLA, first language acquisition and dialect leveling may all be going on at the same time.

By proposing the Interlanguage Hypothesis, Plag (2008a,b; 2009a,b) tries to account for cases of morphological inflection, syntactic structures, word formation and

94 phonological phenomena which seem to be found in both creoles as well as in interlanguages. With his proposal, Plag does not pretend to account for all properties of

present-day creoles; as he acknowledges, many traits of today’s creole languages have

arisen through language-internal changes, and/or have emerged in diglossic or

multilingual situations much after the time of the target shift. The real point of Plag’s

claim consists of finding a psycholinguistic reason for why many creole features seem to

be trackable to SLA processes; he indicates that the SLA Processability Theory

(Pienemann 1998; 2005) might provide creolists with the answer to such a question.

Plag’s focus is therefore on the individual mental processes involved in the

creation of features; however he recognizes that in addition to the individual mental

processes that generate quite variable outputs, i.e. a pool of variants, we need to

acknowledge effects of selection and establishment of particular features from these

outputs across speakers, i.e. in the (emergent) speech community (e.g. Mufwene 2001;

2002). These effects are generally referred to as mixing and leveling, and are

encapsulated by the inclusion of the word ‘conventionalized’ into the formulation of the

interlanguage hypothesis in (18).

Plag (2008a) also points out that SLA processes in creolization do not necessarily

mean substrate transfer, as it has often been suggested in the literature. On the other hand,

there is transfer without SLA as, for example, in diglossic situations or in bilingual first

language acquisition (Kouwenberg 2006), and there are SLA processes at work in

creolization that do not result in transfer, but rather in gradual interlanguage evolutions,

which obey hierarchical chronological steps encountered in the gradual acquisition of a

second language.

95 As none of the (a-e) features encountered in ABS seem to be directly linked to transfer, we will devote the majority of our attention to understanding how the second kind of processes encountered in SLA may have been responsible for the ABS features

4.2.1. Processability Theory

The SLA theory which would be able to account for these phenomena is

Processability Theory (Pienemann 1998; 2005). Processability Theory is a theory that strives to explain one of the fundamental puzzles in SLA research: why is there a well- defined universal path in the development of second languages? In order to provide an answer to this question, the theory recurs to psycholinguistic models of speech production as those developed by Levelt (1989) and Kempen and Hoenkamp (1987).

The central claim of Processability Theory is that the processing procedures reflect their sequence of activation in language generation which drives their acquisition.

The developmental path characteristic of SLA would be reflected in the implicational hierarchy of processing activations. Pienemann (2005) presents the processing hierarchy with data from English. In fact, it has been observed that L2 learners of English follow universally the developmental path illustrated in Figure 4, which correspond to the different stages of activation indicated in Figure 5. They begin with one-word utterances to gradually develop more complex constructions in a hierarchical order. Eventually, the most advanced learners achieve a target-like proficiency and are able to produce very complex structures like the cancelation of inversion in subordinate interrogative clauses.

96 Table 6. Developmental stages in English interlanguage syntax (Pienemann 2005)

Table 7. Processing procedures for English (Pienemann 2005 )

Within the framework provided by the Interlanguage Hypothesis, Plag tries to account for the loss of inflectional morphology, for the allegedly unmarked nature of many syntactic structures across creoles, for the make-up of creole phonological

97 inventories, most prominently the conflation of phonological categories and the emergence of unmarked structure, and for several cases of word-formation, such as cases of circumlocutions, which are also typical of interlanguages (see also Klein & Perdue

1997).

Our present goal is not so broad. For the moment, we will provide an account for the ABS morpho-syntactic (a-e) features presented, as those are the grammatical cues which have been mentioned as potential indicators of previous radical creolization.

Therefore, even though a similar analysis for different levels of representation (e.g. phonology, word-formation, etc.) is certainly of equal interest, it will not be considered in the present inquiry and left for future research.

4.2.2. Accounting for the presence of ABS ‘creole-like’ features

As far as ABS is concerned, I would like to argue that the ‘creole-like’ features encountered in the traditional dialect do not seem to correspond to just an interlanguage of an early stage, they rather seem to occur also in the speech of intermediate and advanced L2 Spanish speakers.

In order to prove this point, I will contrast these suggestive ABS features with the findings of specific SLA studies on advanced L2 speakers of Spanish and other varieties, thus providing potential evidence for their non-creolized origin.

As already suggested, cases of null article in subject position do not seem to belong only to radical creole languages. On the other hand, research on SLA has provided evidence that adult L2 learners may find difficulties in mastering the article system of the target language, even after a long period of exposure. Particularly, studies which have

98 compared article choice among L2-learners with different L1s (e.g., Larsen-Freeman

1975a; 1975b; Thomas 1989; Parodi et al. 1997; Murphy 1997) generally agree that speakers of article-less L1s (e.g., Korean or Japanese) omit English articles in obligatory contexts to a greater extent than speakers whose L1s have articles (e.g., Spanish).

Another work indicating the importance played by L1 in acquiring the L2 article system is the study by Leonini (2006), who focused on the acquisition of Italian determiners by adult L1 German speakers. In her PhD dissertation, the author presents the results for the spontaneous data production of Hike, a German student who spent one year in Italy studying at the University of Siena; at the moment of her arrival her Italian skills were classified as ‘intermediate’.

Hike was recorded two times in one month intervals for a period of 1 year.

Leonini identified 4 stages of language development on the basis of verbal morphology

accuracy (2006:102):

- Stage 1: 6 months after her arrival in Italy / at the beginning of a 3 month-long

Italian language course (verb morphology not productive);

- Stage 2: At the end of the language course; beginning of second term at

University;

- Stage 3: End of second term at University; (verb morphology getting

productive);

- Stage 4: After 1 year at University (verb morphology almost target-like).

Results indicate that obligatory articles in Italian were often missing from Heike’s speech, with a higher use of definite articles over indefinite ones (Table 3).

99

Table 8. Articles production per type by Heike (Leonini 2006:112).

Moreover, when the data were analyzed from a qualitative point of view, she noticed that most of the contexts where determiners omission occurred were possible contexts in German, thus supporting one more time that L1 has an effect on L2 grammar building.

As far as we are concerned, Leonini’s study provides evidence that also in a language with a relatively similar determiner system to the one targeted (L1 German, L2

Italian), the complete mastery of Ds might be hard to obtain.

An investigation by Sànchez and Giménez (1998) on the acquisition of Spanish articles by Quechua speakers suggested that this process is conditioned by the syntactic environment in which DPs appear. Quechua does not have an article system; definiteness and indefiniteness on N are conveyed through context-driven pragmatic strategies. The researchers analyzed the spontaneous speech of 19 Peruvian informants which moved from the country (an environment predominantly Quechua speaking or bilingual in

Quechua and Spanish) to Lima (the capital city, mostly monolingual in Spanish). They found that the percentage of errors was higher in DPs that were complements of prepositions than for DPs in subject position, with DPs in direct object position falling

100 somewhere in between and concluded that this could be due to the availability of licensing of the null determiner by incorporation to the prepositional head (1998:645).

What is most interesting to us is that in this study all subjects had their first contact with

Spanish between the ages of 4 and 16 mainly in naturalistic context but also through formal instruction. All of them were able to communicate in Spanish and claimed to use

Spanish with a range of frequency going from frequently to always, though some did not answer the question. Moreover, some people had more than 30 years of exposure to

Spanish. Nevertheless, the ability of mastering definite articles varied widely among the informants, with percentages of error ranging from 0% to 38.3%, and an overall average of 6%. This fact indicates that also in cases of extensive and frequent exposure to

Spanish, speakers coming from an L1 background with a radically different determiners configuration can fail to proficiently acquire its article system. All these studies on the nature of L2 acquisition are related to the ABS case. In fact, they show that articles are difficult to master not only for speakers of L1s with no articles, but also for advanced speakers with a relatively similar system in their L1. We are not in the position of stating what African languages were introduced in Bolivia; nevertheless, we can say that the

ABS article system features do not necessarily need to be ascribed to a previous stage of radical creolization; rather, they can be seen as the results of advanced processes of SLA.

In chapter 8, we will look at the ABS article system in more details.

The use of 3 rd person singular as invariant verb form is notoriously common

among L2 speakers of Spanish and in child language (Bybee 1985). It has also been

extensively reported for transitional Spanish-English bilinguals throughout the United

States (Lipski 1996). Obviously, if corrected through formal instruction, more advanced

101 students tend to present wider conjugation patterns, as also reported for Heike studying

Italian (Leonetti 2006); nevertheless, in the Spanish of Equatorial Guinea, where Spanish is taught at schools and is widely spoken as a second language- frequent gravitation to the third-person singular as the default unmarked verb form is common, despite the fact that

Guineans routinely attempt to conjugate verbs correctly (Lipski 1985). Given that the sociohistorical scenario in which ABS emerged has never been characterized by formal education, it is not completely surprising to encounter the presence of 3 rd person singular

as invariant verb form. Moreover, the presence of tense and aspect inflection on verbs

further suggests that the stage of acquisition must have been quite advanced.

Non-inverted questions seem to persist in the speech of L2 speakers for a very

long time; indeed, Pienemann (1998) reports them as one of the final developmental steps

of English interlanguages. Learning to form questions involves multiple developmental

steps. In an earlier work, Pienemann, Johnston and Brindley (1988) identified the

following stages for English: Stage 1 is characterized by the use of single words and

formulaic expressions, such as a store? What’s that? In stage 2, the learners use

declarative word order you like pasta?. In Stage 3, fronting of wh-words and do begin to

appear, resulting in expressions such who you are? , do she like the movie? Only by Stage

4 inversion of wh in copular questions appears, questions such as who are you ?. Inversion

of copula and auxiliary is seen also in yes/no questions as learners produce are you a

student? and was she driving the car? Stage 5 is characterized by the appearance of

inversion in questions that require do-support to lexical verbs. Examples include do you

like movies? And who is driving the car? . Stage 6 is characterized by the appearance of

complex forms. Among the complex forms that emerge are question tags, as in she’s

102 French, isn’t she? and negative and embedded questions such as didn’t you like the

movie? and do you know what the answer is?

These developmental stages for non-inverted questions are not only L2 English

specific; rather they seem to be encountered in all languages presenting similar Wh-

constructions (Pienemann 2005). Besides, not only creole languages and L2 varieties

present these structures. They are also extremely frequent in non-creolized Spanish

Caribbean dialects: Puerto Rican, costal Colombian, Dominican, Cuban, Costa Rican,

Panamanian (Díaz-Campos & Clements 2008; Lipski 1994).

As far as grammatical gender is concerned, it is well known that it represents a

major problem in language acquisition, especially for speakers coming from a L1 lacking

such feature. Bruhn de Garavito and While (2000) investigated gender agreement

accuracy in L1 French teenagers with either one (Group 1) or two (Group 2) years of

classroom exposure to L2 Spanish. They used a card description game to elicit

naturalistic data from the subjects, and their descriptions were recorded and transcribed.

Gender agreement accuracy was fairly high, and comparable to the accuracy rates

reported in Hawkins (1998) for much more advanced L2 French speakers whose L1 was

English. This can be interpreted as evidence that there is an advantage in having gender

in the L1, at least for rate of development, since Bruhn de Garavito and White’s subjects

reached similar levels of accuracy to Hawkins’s subjects with considerably less exposure.

Moreover, Bruhn de Garavito and White break down their findings according to

definite and indefinite determiners. They indicate clearly that, as in Hawkins’s (1998)

study, agreement is more likely to occur on definite articles than indefinite ones (Figure

7). This fact is in line with Leonini’s (2006) findings for L2 Italian, with Franceschina’s

103 (2005) for L2 Spanish, and with my results on the evolution of gender agreement features across the ABS DP (Sessarego 2009a; Sessarego and Gutiérrez-Rexach 2009a,b; see also chapter 6).

Figure 4. Accuracy of gender agreement on definite and indefinite articles (Bruhn de Garavito & White 2000:172) 21 .

Also, Bruhn de Garavito and White provide comparisons between the rates of

gender agreement of determiners vs. adjectives (Figure 8), this time without showing

Hawkins’ (1998) results as he did not present such a piece of information in his paper.

21 Bruhn de Garavito and White (2000) compares their two groups with the three groups studied by Hawkins who looked at data from 30 advanced English-speaking learners of French (one group of university students in Canada and two groups in the UK). 104 Figure 5. Accuracy of gender agreement: determiners vs. adjectives (Bruhn de

Garavito & White 2000:72).

As can be observed in Figure 8, the overall rate of agreement is higher for

determiners than adjectives. These findings are again in line with Franceschina’s (2005)

for L2 Spanish, and with my results for the ABS DP. In fact, Franceschina (2005)

provides a detailed account of the acquisition of grammatical gender in L2 Spanish

learners. She states that gender has been shown to be a persistent problem for some adult

L2 learners even though it is a property of Spanish for which there is unambiguous

positive evidence in the input. After testing advanced L2 speakers with a variety of L1

backgrounds (Italian, Portuguese, English, Arabic, German, French, Greek), she

concluded that while +genL1 (Italian, Portuguese, Arabic, German, French, Greek)

informants could perform at native speaker levels in a range of tasks designed to test

knowledge of grammatical gender, –genL1(English) speakers were generally successful

but yet consistently and significantly less target like than +genL1. Moreover, also in this

study, experiments showed that adjectives tended to agree less than determiners and

among Ds, indefinite articles agreed less than definite ones. 105 All these studies on gender agreement also share the common view that masculine is the default gender value in cases of agreement mismatch. On light of the fact that also very advanced L2 Spanish learners may have difficulties in mastering gender, it should come as not surprise if also in ABS we can found traces of this SLA phenomenon.

Besides, the limitation of nominal gender agreement to definite articles in ABS inherently indicates the presence of the feature ‘gender’ in the dialect. In this respect, ABS diverges quite significantly from the majority of the Romance-based creoles, which are generally supposed to lack gender features.

As I will explain in more details in chapter 5, ABS presents exactly the same agreement operations for gender as encountered in standard Spanish, with the crucial distinction that in the latter language the domain of feature specification for Agree is broader (it includes also demonstratives, quantifiers and adjectives). Moreover, the different rates of gender agreement per grammatical category (determiners vs. adjectives vs. quantifiers) as encountered for ABS and L2 Spanish (Bruhn de Garavito and White

2000; Franceschina 2005) are in line with Processability Theory (Pienemann 1998), which predicts that they should pose different levels of difficulty to L2 learners.

As acknowledged previously, the marginal cases of tener ‘to have’ employed as

an existential verb, and more common nu hay ‘do not exist’ used as a negation of

possession, may indicate a previous case of partial restructuring, which might have been

driven by contact with Portuguese, language in which tener is often used where Spanish

would employ haber .

In sum, all the so-called ‘creole-like’ features that could be ascribed with certainty

to ABS seem to be present also in advanced or intermediate L2 Spanish interlanguages.

106 These findings, in addition to the socio-historical evidence provided in chapter 3, reinforce the idea that ABS was probably never a radical creole, which later decreolized, but rather a relatively close approximation of Spanish from the very beginning of its inception.

4.3. Vernacular Universals

A comprehensive account of the ABS creole-like features should not only be limited to the SLA/processing aspects of their development; for this reason, I will also provide a sociolinguistic framework able to explain why these features crystallized and survived in ABS in the way we know them today. A theoretical model, which I think can complement my hypothesis on the nature of these features, is the framework provided by

Chambers’ (2003; 2004) Vernacular Universals. In Chambers’ view, a small number of phonological and grammatical processes seems to recur in vernaculars wherever they are spoken. Chambers (2003:266-70) characterized these recurring natural processes as

“vernacular roots”.

The theory of Vernacular Universals rests squarely on the assumption that basilect forms are in some demonstrable sense more ‘natural’ than standard forms (see also Weiß

2001). Chambers takes a radical position on this issue and claims that they are the natural outgrowth of the language faculty, that is, the species-specific bioprogram (or UG) that allows normal human beings to become homo loquens (Chambers 2004). He provides a

list of the best candidates for English, based on their recurrence (Chambers 2004:129):

- (ng) or alveolar substitution in final unstressed –ing, as in walkin’, talkin’

and runnin’ .

107 - (CC) or morpheme-final consonant cluster simplification, as in pos’office,

han’ful.

- Final obstruent devoicing, as in hundret (for hundred ), cubbert (for

cupboard ).

- Conjugation regularization or leveling of irregular verb forms, as in

Yesterday John seen the eclipse or Mary heared the good news.

- Default singulars, or subject-verb non-concord, as in They was the last ones .

- Multiple negation, or negative concord, as He did not see nothing.

- Copula absence, or copula deletion , as in She smart or we going as soon as

possible.

I do not believe that these vernacular roots should be considered as the principles of UG. The way in which Chambers formulates his claim is reminiscent of Bickerton’s

(1981) Bioprogram Hypothesis, to which, I do not personally adhere. On the other hand, I believe that Chambers’ observation that pidgins, creoles and second language varieties tend to share some common features might be reinterpreted from a SLA perspective by simply saying that constructions requiring less processing are mastered before constructions requiring more processing, so that elements which are hard to process will not be encountered in the initial stages of these developing grammars. The result is that these languages will share structures which may be comparatively easier to learn and, in this sense, more ‘simple’ and ‘natural’.

Demonstrating naturalness seems relatively easy for phonological processes compared to grammatical ones. For example, consonant cluster simplification (CC)

108 represents econimies in articulatory (motor) gestures that are quantifiable (Chambers

2003:258-259). Principles that might underlie natural grammatical processes are never so concrete, and I think they could – and should – be questioned. For example, some doubts might be cast on whether Chamber’s universal “double negation in English” is really more natural than single negation (cf. Chamber 2004:129). It is true that it seems to appear in the majority of English vernaculars, but to claim that it represents a more natural form, some additional linguistic evidence should be provided, e.g. analyzing whether it is really easier to process or easier to acquire in L2.

However, as far as the ‘creole-like’ elements in ABS are concerned, we have found enough evidence indicating that these features seem to be more processable than their standard Spanish counterparts because they can be encountered even in the speech of advanced L2 Spanish speakers, who often fail to master the more complex standard counterparts.

The factors underlying Vernacular Universals have different cognitive and social

aspects. From a sociolinguistic perspective, these elements represent the features

characteristic of basilect, non-prestigiuos speech; from a more functional point of view,

they seem to have to do with principles of motor economy and cognitive overload. As

Chambers points out (2004:130), “unifying the functional principles into a few empirical

defensible cognitive strategies may be too much to ask of any branch of linguistics at this

time, important though is to try”. To this statement, I would like to add that such a goal

could be achieved, at least partially for contact-varieties, by combining Processability

Theory (Pienemann 1998; 2005), the Interlanguage Hypothesis (Plag 2008a,b; 2009a,b)

and Vernacular Universals (Chambers 2003; 2004).

109 The core of Chambers’ theory maintains that “the standard dialect differs from other dialects by resisting certain natural tendencies in the grammar and phonology”

(Chambers 2003:254). Chambers pushes the model so far as to say that “the basilectal form is primitive, part of the innate bioprogram, and the standard form is learned, an experiential excrescence on the bioprogram” (2003:286). I am not sure of whether in cases of standard L1 acquisition such a radical view can account for all the features that he indicates. However, for the case of contact varieties like ABS, we may want to say that certain forms are easier to learn/process than others, and unless social pressure rules them out, there should be no reason why they could not normally crystallize and be convenzionalized by the speech community, becoming in this way part of the core grammar of the language natively acquired by following generations.

Los Yungas provided the perfect place for such a crystallization to take place, as they were isolated, rural valleys far from the social pressure posed by formal education, standardization and the linguistic norm. Note that my claim does not imply that after the

Land Reform Yungueño Spanish did not experience a process of approximation to the standard variety. This is something that I am not questioning, and that is clearly visible by looking at the evolution of inflectional morphology across generations (Sessarego

2009a,b; Sessarego and Gutiérrez-Rexach 2009a,b; Sessarego and Delicado-Cantero

2009; Delicado-Cantero and Sessarego 2009). What I am actually not convinced of is that

Afro-Bolivian Spanish could have been a radical creole before 1952, which underwent a drastic decreolization since that date.

The recent approximation to standard Bolivian Spanish further supports the claim of Vernacular Universals. In fact, when the factors preserving the traditional dialects

110 came to disappear – essentially after 1952 – the pressure, imposed by standardization and the linguistic norm, pushed Afro-Bolivians towards dropping the traditional dialect, in favor of more prestigious Spanish variety. Could it be that such traditional dialect was itself the byproduct of a previous approximation to Spanish which radically decreolized the language? Answering this question is not an easy task, as no written documents of early Yungueño Spanish exist. However, I think that the historical and linguistic evidence so far provided suggest that probably this was not the case. My data seem to indicate that

Afro-Bolivian Spanish was not a radical creole, it probably was a vernacular which approximated quite closely to Spanish from its inception, and in which processes of second language acquisition left an undeniable trace.

In sum, if on one hand a modified version of the Interlanguage Hypothesis and

Processability Theory can account for the development of the ABS ‘creole-like’ elements, the framework of Vernacular Universals appears to be able to explain why these features could crystallize and survive up to our days.

4.4. Conclusions

This chapter provided a closer look at the linguistic features of ABS which have been proposed as potential indicators of prior creolization (Lipski 2008b). Findings suggest that the grammatical elements found in ABS can also be ascribed to a conventionalized advanced/intermediate second language which did not undergo processes of standardization imposed elsewhere by the urban society and the linguistic norm. I decided to discuss the development of such grammatical elements in light of the

Interlanguage Hypothesis (Plag 2008a,b; 2009a,b) and Processability Theory (Pienemann

111 1995; 2000). The crystallization and survival of these features in ABS up to today has been explained by adopting the sociolinguistic framework provided by Vernacular

Universals (Chambers 2003; 2004).

In line with Kayne’s (1996) microparametric approach, the close proximity of

ABS to stSp can provide us with a great “testing lab” for syntactic theories. The following chapters of this work will be focusing on scrutinizing theoretical paradigms and identifying parametric differences between ABS and stSp.

112

CHAPTER 5: FROM NP TO DP

"What's in a name? That which we call a rose by

any other name would smell as sweet" (Shakespeare

1597: Romeo and Juliet (II, ii, 1-2)).

5.0. Introduction

The aim of this chapter is to present the most important issues concerning the generative study of the nominal domain during the last thirty years and to provide an outline of frameworks and ideas which will be evaluated in the following chapters based on the empirical data encountered for ABS.

The internal structure of the Noun Phrase (NP) became a central issue in syntactic investigation from the revised notion of X’-theory to the nominal domain (Chomsky

1986b), and especially after Abney (1987), who provided theoretical and empirical bases for the idea that the DP is the maximal category projected by the class determiner.

This idea came to be known as the DP Hypothesis, and during the last three decades much research has been carried out with the aim to understand the configuration of the syntactic structure contained between D and N (20) (XP and YP stand for any intermediate functional projection as e.g. PossP, NumP, GenP, etc.).

113 (19)

[DP ... [ XP ... [ YP ... [ NP ... ]]]]

The rest of this chapter is organized as follows: Section 4.1. describes Abney’s

(1987) and Szabolcsi’s (1983; 1987; 1994) works which provided the first theoretical and empirical evidence for the study of the DP as maximal category projected by D (DP

Hypothesis). Section 5.2. presents an analysis of some of the main articles which have argued in favor of additional projections between D and N and movement of N to its surrounding layers (Longobardi 1994; Ritter 1991; 1993; Picallo 1991; Bernstein 1993;

Giorgi and Longobardi 1990; Valois 1991; Giusti 1993; Mallén 1997; Cinque 1993).

Section 5.3. focuses on two proposal concerning the status of the quantificational Phrase in a layered DP (Zamparelli 2000; Ishane 2008). Section 5.4. presents Radford’s (2000) claim on light ‘n’. Section 5.5. summarizes the argument in favor of DP-internal prolific domains (Ticio 2003). Finally, Section 5.6. concludes.

5.1. The Determiner Phrase Hypothesis

5.1.1. Abney (1987)

In his PhD dissertation, Abney (1987) develops the idea that a nucleus D, containing agreement features, precedes the nominal nucleus, mirroring in this way the structure of the orational clause, where Infl precedes V. He analyzes several languages in which the noun agrees with the possessive, considered to be the internal NP subject.

In order to show this apparent parallelism between NP and VP, (now DP and CP),

Abney presents data from several languages like Yup’ik, a Central Alaskan Eskimo

114 language, whose nouns carry two different number agreement markers: one indicates concord with the possessing subject, the same one occurring on the verb agreeing with the subject, while the other represents the noun number agreement as in (20) (Abney

1987:39).

(20)

Angute-t kuig-a-t

men-ERG (Pl) river-Sg-Pl

‘The men’s river’

The possessor angute ‘men’ is semantically plural and present ergative case, as subjects of transitive verbs do. The noun kuig ‘river’ has two different number markers, a is its own number morpheme, which is singular; while the morpheme t indicates plural concord with the subject angunte-t. When the noun appears without possessive, it only bares its own semantically motivated number marker. On the other hand, the second number marker, which occurs only in agreement with a possessive, does not express any real semantic content, such as number morphemes on verbs in Romance.

Abney’s suggested structure is represented in (21). In his view, the agreement

marker is a lexical realization of a functional category, D, which can assign case to the

element appearing in its specifier.

115 (21) DP

Spec D’

D° NP

D Agr1 N’

N Agr2

From this perspective, N is a predicate and D picks out a particular extension of it, as Infl does with VP. As in the case of Infl there are two possible realizations of the functional category: D and Agr1. When there is agreement, the nucleus assigns case to its subject, when the possessive is not present, and therefore agreement cannot be encountered either, the nucleus does not assign case, as infinitive verbs do not assign nominative case to their subjects. Ds’ function is to turn Ns into arguments, which become, in this way, referential objects.

What Abney classified as Agr2 in (21) represents the number agreement found in

Romance and Germanic languages, for which, in following works, a functional projection

(NumP) will be postulated (Ritter 1991; Picallo 1991).

In Romance and Germanic languages D never agrees with the possessive in the way we saw in Yuk’pil; the lexical realization of such nucleus is the article. On the other hand, when the noun agrees with the subject in genitive case (e.g. Saxon genitive phenomena), lexical realizations are agreement morphemes which are found in complementary distribution with the article (22).

116 (22)

a. Marino’s book

b. The book

c. *The Marino’s book

Abney proposes a DP structure for Saxon genitive constructions in which a null

AGR morpheme assigns genitive case (the ’s morpheme) to the possessor DP as in (24).

(23)

DP

Spec D’

D° NP

Spec N’

N°

The student’s AGR book

Such a structure is based on the fact that in many languages the possessed N

agrees with its subject the same way the V agrees with its subject, while the possessor

presents the same case as the subject of the sentence. The parallel between CP and DP is

therefore highlighted. Abney suggests that AGR occupies in the DP the position of Infl in

the CP.

117 5.1.2. Szalbolcsi (1983; 1987; 1994)

Additional support to the parallelism between DP and CP is provided by

Szabolcsi, who suggests that both the complementizer and the article are elements responsible for turning the clause or the noun phrase into arguments. In fact, C and D are assumed to act as subordinators capable of receiving a Θ-role assignment.

The case is exemplified in (24), where a category with an overt or covert C (24a) may function as an argument, while, on the other hand, clauses unable to function as arguments (24b), disallow the use of C.

(24)

a. I think [CP (that) you are nice]

b. *[CP That you are nice]

Szabolcsi argues that in Hungarian the possessive morpheme behaves similarly to the tense/mood tense morpheme in sentences: sentences like (25a) are ungrammatical without the tense morpheme t, just as noun phrases like (25b) are ungrammatical without the possessive morpheme e:

(25)

a. Maria-0 alud-t-0

Mary-NOM sleep-Past-3rd -Sg

‘Mary slept’

b. A Mari-o vendeg-e-0

The Mary-NOM guest-Poss-3rd -Sg

‘Mary’s guest’

118 Szabolcsi suggests that in Hungarian the NP has a head like Infl capable of governing the subject and assigning nominative case. Nominal and verbal Infls differ in their feature matrix: Nominal Infl contains the set for [+/- poss], whereas verbal Infl has

[+/- tense] features. Szabolcsi’s Nominal Infl [Poss] corresponds to Abney’s functional category D, whereas Szabolcsi’s verbal Infl [Tense] corresponds to Abney’s verbal functional category Infl.

She postulates two nominal categories in the extended nominal projection.

According to Szalbolcsi, there are two nominal nuclei (26a): a lower nucleus, Nominal

Infl° (NI°), which assigns nominative case to its specifier; and a higher nucleus,

Determiner° (D°) , assigning dative case to its specifier, its lexical realization is the article, which turns nouns into arguments, mirroring in this way the function of CP in the extended verbal projection (26b).

(26a)

DP

Spec D’

D° NI

Spec NI’

NI° NP (26b)

CP

Spec C’

C° IP

Spec I’

I° VP 119 The theoretical assumptions for the stipulation of (26) are mainly based on

Hungarian data. In this language in fact possessives have the form presented in (27) and

(28) (Szabolcsi 1994:180).

(27)

A te kapal-ja-i-d

the you (NOM) hat-Poss-Pl-2nd Sg

‘your hats’

In (27) the possessor te takes the nominative case, the noun denoting the possessed object kalap carries its own plural number marking i as well as the number agreement morpheme d, indicating second person singular, as on verbs. NI assigns the nominative case, while NP presents the article in its specifier.

The structure provided in (26) accounts also for other possessive constructions, those in which the possessive precedes the determiner (28).

(28)

A fiú-k-nak a kalap-ja-i

the boy-Pl-DAT the hat-Poss-Pl

‘The boys’ hats’

Differently from the possessives encountered in constructions like (27), in sentences like (28) they can be extracted, their position is similar to the one of the CP specifier from which it can be extracted too, so that the new structure proposed can

120 account for the similarities encountered between the nominal domain and the orational clause. In (28) the dative case nak would be assigned to fiú by NP in its specifier.

NP° behaves in Szabolcsi (1994) as D does in Abney (1987), where it turns Ns into arguments. Besides, Szabolcsi accounts for the fact that nominal inflexion and articles are not found in complementary distribution in Hungarian by postulating an additional head.

The parallelism pointed out by Abney (1987) and Szabolcsi (1994) between nominal and orational domains will be further investigated in following works

(Longobardi 1994; Ritter 1991; Picallo 1991; Cinque 1993; among others) aimed to demonstrate the movement of N in DP mirroring V movement in CP.

5.2. In between N and D

Several works argued for N to D movement in support of the DP Hypothesis. I will present here only few of them, those which I retained the most relevant in terms of the concepts introduced in the field of formal syntax. I will examine here Ritter (1991;

1993) for Hebrew, proposing the existence of the functional category Number (NumP). I will then summarize Picallo (1991), who postulates the presence of a Gender Phrase

(GenP) in Romance and Bernstein (1993) claiming the existence of a Word Marker

Phrase (WMP) instead, Cinque (1993), who provides a layered structure to account for adjectival projections, Giorgi and Longobardi (1990) and Valois (1991) for an account of nominal Θ–role assignments, Giusti (1993) and Mallén (1997), for a discussion on Kase

Projection, and finally, Longobardi (1994), who claims the existence of N to D

121 movement in all languages, with the difference that some of them present it overtly while others covertly.

5.2.1. Ritter (1991; 1993)

Ritter (1991) was the first to suggest the presence of a functional projection corresponding to numeral marking. She argued that Num(ber)P is the complement of D in

Modern Hebrew and in Romance languages.

The argumentation in favor of a DP-internal functional projection Num(ber) P is based on two types of genitive constructions in Modern Hebrew. The first construction analyzed is the construct state (29), which shows a Noun-Subject-Object-surface order derived by movement. In fact, she shows that the subject must asymmetrically c- command the object, and proposes that the noun raises from N to D, crossing over the possessor (30).

(29)

Ahavat Dan et ift- o

love Dan ACC wife-his

‘Dan’s love of his wife’

(30) DP

D NP

subj N’

N obj

Ahavat i Dan ti et ift- o 122 The second genitive construction is the free genitive one, in which the genitive

phrase is introduced by the dummy case marker šel , and is preceded by the head noun

and its determiner (31).

(31)

Ha-axila šel Dan et tapuax

The-eating of Dan ACC the-apple

‘Dan’s eating of the apple’

This free-state construction with both a subject and an object argument cannot be

accounted by (30): N cannot be raised to D°, as this position is already being occupied by

the overt definite determiner. Moreover, the presence of the overt case marker šel

suggests that genitive case is not assigned by means of the null determiner. If movement

has to obey the head movement constraint (Travis 1984) 22 , only the postulation of another head position between DP and NP can derive NPs like the one in (32).

This category is called Num, it contains the number specification of the NP. The

noun moves from N° to Num° to check its number feature, before landing in SpecNumP.

22 THE HEAD MOVEMENT CONSTRAINT (Travis 1984 in Ritter 1991:39): A head (X°) can move only to the position of the head (Y°) that properly governs it. 123

(32)

DP

Spec D’

D° NumP

Spec Num’

Num° NP

Spec N’

N° DP

Ha axila i t i šel Dan ti et ha-tapuax

Therefore, Ritter suggests that in Hebrew there are two nuclei, one whose realization is the definite article with a [-genitive] feature (31), the other is empty D, with a [+genitive] feature which is assigned to the complement (29). When the article is not present, N rises to D as in (32). D assigns genitive case to the complement resulting, in this way, in a construct state phrase. On the other hand, when the article is present (14), N does not rise to D° and the overt case marker šel is needed, as only empty D can assign

genitive case on its right.

Ritter (1993) elaborates also on gender. She suggests that while in Romance, gender is base-generated as a feature on Num and may be attached to N as a consequence of syntactic head movement; in Hebrew, gender is attached to the noun stem in the lexicon and thus appears on the noun at all levels of syntactic representation.

124 (33)

The claim that gender is a feature on the lexical stem in Hebrew would be supported by the fact that in this language gender switching is a fairly productive strategy for deriving new nouns from existing ones (34), while in Romance this strategy is restricted to nouns with human or animate reference and to only a few with inanimate ones (cf. Harris 1991).

(34)

Additional support to the gender feature distinction between Hebrew and

Romance is provided by the analysis of irregular plurals in these two varieties. While

Hebrew plural affixes are underlying specified for number only, Romance plurals affixes

seem to be specified for both number and gender. This last point would be backed by the

fact that a large number of irregular nouns in Romanian are masculine in the singular and

feminine in the plural, as shown in (35).

125 (35)

5.2.2. Picallo (1991)

The idea that both gender and number features are checked under NumP in

Romance is not shared by Picallo (1991), who suggests the existence of an additional intermediate gender projection to a functional phrase within the DP, Gen(der)P. This functional projection would be located between NP and NumP (36), reflecting in this way the linear order of the morphemens attached to the noun stem, as shown in (37):

(36) [ DP Det [ NumP Num[ GenP Gen [ NP N]]]]

(37) Spanish: sill-a-s

chair-F-Pl

‘chairs’

Since Romance determiners and adjectives agree in gender with N, Picallo (1991)

suggests that gender projects into a syntactic functional category, even when it does not

have semantic content (1991:282). NP is the complement of Gen°, while GenP is the

complement of of Num°. Gender and number morpheme affixations are the result of

126 cyclical movements to Gen° and Num°. Example (38) shows the derivation of the plural feminine noun phrase les gates (‘the cats’).

(38)

DP

Spec D’

D° NumP

Spec Num’

Num° GenP

Spec Gen’

Gen° NP

Spec N’

N° Les gat iejs t j t i t i

5.2.3. Bernstein (1993)

Bernstein’s (1993) proposal is similar to Picallo’s (1991). Bernstein suggests that word markers (Harris 1991) are syntactic elements projecting into a W(ord) M(arker)

P(hrase). The reasoning behind such proposal is the fact that the declention class of

Spanish nouns does not always match with the gender. Agreement in fact involves gender, not declension class, (cf. la mano derecha , ‘the right-F-Sg-WM:A hand-F-Sg-

WM:O’).

Harris (1991) calls the nouns in which gender and declension class converge

‘inner core’ e.g. hijo (‘boy-M-Sg-WM:O’) and hija (‘girl-F-Sg-WM:A’). In his view,

127 masculine ‘inner core’ nouns are not specified for gender and declension class so that they surface with a –o word marker by default. On the other hand, feminine ‘inner core’ nouns are marked in the lexicon as [+FEM] and receives the –a word marker. When gender and declension class do not converge, both must be marked in the lexicon (e.g. programa ‘program-M-Sg-WM:A’). The structure suggested by Bernstein (1993) mirrows Picallo’s, with the only distinction in WMP, which comes to substitute GenP, giving (39).

(39)

DP

Spec D’

D° NumP

Spec Num’

Num° WMP

Spec WM’

WM° NP

Spec N’

N° Les gat iejs t j t i t i

5.2.4. Cinque (1993)

The parallelism between DP and InflP seems to be further supported by the distribution of adjectives with respect to NP, which present striking similarities with the distribution of adverbs with respect to VP (Crisma 1990 and 1993). In line with this view,

128 Cinque (1993) proposes that APs are generated in distinct specifier positions of a functional projection between D and N. The alternative to this hypothesis would be the one that sees adjectives as adjoined to the maximal projection (Picallo 1991; Valois 1991;

Bernstein 1993); however given the existence of a specific unmarked serialization of different AP classes with event nominals (40a) or objects denoting Ns (40b), the first hypothesis would be preferred as adjoined elements are generally supposed to be free.

(40)

a. poss> cardinal> ordinal> speaker oriented> subject oriented> manner> thematic

b. poss> cardinal>ordinal>quality>size>shape>color>nationality

Another point in favor of the first stipulation is that the number of adjective is limited (no more than seven), while adjunction position would be unlimited. In this way the appearance of adjectives on the left of the head-noun follows from the location of the specifiers in which they are base-generated and does not need to be stipulated.

Cinque suggests that the difference in position between Romance and Germanic languages (postnominal vs. prenominal) is not due to a left/right assignment parametric distinction, but rather to partial N rising in Romance.

5.2.5. Giusti (1993)

Also the existence of a Case Projection (KP) has been an important topic of

discussion in the last decades of formal syntactic theorizing. Several authors proposed the

existence of such projection, basing their hypothesis on a vast gamma of languages.

129 Giusti (1993) proposes a KP external to the DP, whose role is to assign case marking to the DP elements. Since only arguments may receive case, KP represents the link between the selected nominal and its syntactic legitimacy (1993:42). The reason for positioning KP in such a high layer would be the fact that morphological case marking in many languages often appear only on articles. K should therefore be higher than D in order for D to be able to incorporate into it. The structure suggested is the one of (41).

(41)

KP

K° DP

D° NP

However, acknowledging that not all languages have articles, and assuming that quantifiers, demonstrative and possessives are located in layers other than D, (41) is reformulated into (42) where FP replace KP and assimilate DP; it bears the case assigned by the syntactic legitimator and realizes the argumental properties of the NP.

(42)

FP

F° …

… DP

D° NP

5.2.6. Mallén (1997)

On the other hand, Mallén (1997) proposes a KP internal to the DP (see also

Valois 1991; Cornilescu 1993; Parodi 1994; etc.) basing his analysis on German and

130 supporting it by providing examples from French, Spanish, Icelandic, Latin and other languages. Adhering to Chomsky’s (1992) proposal that the distribution of genitive elements and categorical heads is conditioned by functional morphology, he claims that morphological Case-matching derives from a relation between functional categories and lexical categories. Thus, the head of the noun phrase must move to a position where its

Case-feature is matched (Mallén 1997:49). Following Sigur δsson (1993) he calls such a position Kase.

The tree structure is schematically represented in (43).

(43)

DP

Det NumP

Num KP

Kase GenP

Gen NP

N

Moreover, taking genitive Case to be an instance of agreement relation between a specifier and its head (Valois 1991; Lattewitz 1995), he assumes that genitive DPs are displaced to specifier positions to establish a proper relation with a nominal functional agreement (Agr). Under minimalist theory, the established raising of the noun to Kase can be explained as a morphologically triggered movement. The N is required to match its Case feature before Spell Out, proper checking being only possible in Kase. Once checked the feature on Kase disappears. In Mallén’s view the Case-feature on the noun survives and reaches Det through an agreement chain (Olsen 1989) which transmits the 131 nominal inflectional feature (case, number, gender according to Olsen) to the elements belonging to the chain by means of selectional coindexation.

In this way, Det selects Num; Num selects Kase (which in turn selects Agr)

(Mallén 1997:55-56), giving the following structure (44), where Agr s represents the

landing place for subjects and Agr o the one for objects.

(44) DP

Det NumP

Num KP

Kase AgrP

Agr s AgrP

Agr o NP

N

5.2.7. Giorgi and Longobardi (1990)

NP θ-role assignments are analyzed in Giorgi and Longobardi (1990), who focus

on Italian structures and back their claims by employing cross-linguistic comparisons.

They build on the hypothesis, already suggested in (Cinque 1980, 1981), that one of the

arguments specified by the θ-grid of an N is generated in a structurally prominent position with respect to the others and behaves like the ‘subject’ of the NP. Their argumentation, supported throughout the book by binding evidence, is based on Cinque’s observation that only a particular type of argument can be expressed by a possessive pronoun (45):

132 (45)

a. la descrizione degli avvenimenti di Gianni

‘Gianni’s descriptionof the events’

b. la sua descrizione degli avvenimenti

‘His description of the events’

c. la loro ( = degli avvenimenti) descrizione di Gianni

‘their ( = of the events) description of Gianni’

Example (45a) (partially) realizes the θ-grid of the Noun descrizione

(‘description’), with the agent Gianni, and the theme gli avvenimenti (‘the events’). The

θ-roles expressed here are the same as those which can be assigned to the arguments of the corresponding verb descrivere (‘describe’). Both the N and the V identify an internal

θ-role (theme) and an external one (agent). The terminology adopted by Giorgi and

Longobardi is the one usually employed for sentences, concerning internal and external arguments, even if the latter in NPs do not seem to occur outside the maximal projection of the head (1990:28-29). In Italian NPs both the argument expressing the internal and the external θ-roles are usually preceded by the preposition di (‘of’), considered as the semantically empty realization of the Genitive Case; but the external θ-role can always be

expressed by the employment of a possessive pronoun. In fact (45b) has the same

meaning of (45a). On the other hand, (45c) is ungrammatical under the interpretation of

(45a-b); the internal θ-role cannot be expressed by a possessive pronoun, unless the argument bearing the external one is overtly realized or appear in a by-phrase (expresses in Italian with the locution da parte di ‘on the part of’), as in (46):

133 (46)

la loro (=degli avvenimenti)descrizione da parte di Gianni

NPs as (46) would have undergone passivization, so that the element identifying the external argument appear in a by-phrase and an internal argument can display the properties previously typical of the external one replacing it as the ‘subject’ of the NP.

Giorgi and Longobardi (1990) also show that excluding a few peculiar instances 23 the phrase di + pronoun is not possible as the expression of an external Θ-role (agent) in

Italian (47):

(47)

a. *la descrizione degli avvenimenti di te

‘The description of the events of you’

b. La tua descrizione degli avvenimenti

‘Your description of the events’

On the other hand, a di + pronoun phrase is permitted, and a possessive is not, when there is an AP, or a relative (or pseudo-relative) clause, modifying the pronoun

(48):

(48)

a. Ricordo il racconto di lui che partiva per l’Africa

‘I remember the story of him who was leaving for Africa’

b. *Ricordo il suo racconto che partiva per l’Africa

23 See Belletti (1978) for this specific cases. 134 ‘I remember his story who was leaving for Africa’

The crucial difference between di + pronoun and the possessive in (47-48) is based on the two different realizations of Case. In order to appear within an NP, a regular pronoun must be Case-marked by means of di , whereas the possessive does not need this support. In fact, a possessive can be assumed to satisfy the Case-requirements imposed by Chomsky’s (1986b) theory by realizing its inherent (Genitive) Case by means of Case- agreement with the head N, as suggested by many languages with overt Case inflection.

The relevant principle can be expressed as follows (49):

(49) A Case-marking Preposition is licensed iff the assigned Case cannot be directly realized.

In this sense, directly realized means without the introduction of a preposition, which is a ‘last resort’. Whithin the framework proposed in Giorgi and Longobardi

(1990), sentences like (50) are analyzed according to structure given in (51):

(50)

a. La mia restiruzione di Maria a se stessa

‘My restoration of Mary to herself’

b. La mia restiruzione di se stessa a Maria

‘My restoration of herself to Mary’

c. La mia restiruzione di lei a se stessa

‘My restoration of her to herself’

135 d. La mia restiruzione di se stessa a lei

‘My restoration of herself to her’

(51)

N’’

N’

N N’’ N’’

La mia restiruzione di Maria a se stessa di se stessa a Maria di lei a se stessa di se stessa a lei

In this view, the distinction between internal and external positions is maintained and a and de are classified as dummy prepositions, which never block c-command.

Based on extraction evidence already pointed out by several authors (Cinque

1980; Milner 1982; Aoun 1985; Torrego 1984; Zubizarreta 1986; etc.) Giorgi and

Longobardi (1990) argue in favor of a hierarchical organization of NPs composed at least by the following layers (52).

(52)

136 Assuming the structure in (52), the Head Subject Parameter is stated as follows

(1990:117) (53):

(53) While internal arguments ( β) are projected to the right of the head both in

Romance and Germanic, external semantic functions (α) are licensed at D-structure on the right in Romance but on the left in Germanic.

Below is given a graphic representation of the parametric variation (54).

(54)

5.2.8. Valois (1991)

Valois (1991) argues in favor of a layered NP in which each Θ-role assignment corresponds to a specific argument position in the structure, mirroring in this way the clause structure. The external Θ-role is generated in SpecNP* and the internal one as a

sister to N°. Head movement of N° to N*° takes place in order to license the external Θ-

role of nouns projecting NP*. Internal and external Θ-roles can be combined with an

additional possessor, giving the structure in (55).

137 (55)

DP

Spec D’

D° PossP

Spec Poss’

Poss° NP*

Spec N*’

N*° NP

Spec N’

N° DP

POSS AGENT THEME

Valois acknowledges Longobardi and Giorgi’s (1990) Thematic Hierarchy and provides French examples to support it. In his view, extraction out of NPs depends on the hierarchical position of the extracted arguments with respect to other DP-internal arguments. Thus, an agent is not extractable in the presence of a possessor and a theme is not extractable in the presence of either a possessor or an agent. He illustrates the case with French examples, where the (a) examples represent the structure before extraction, with the higher complement underlined. The contrast between (b) and (c) examples shows that only extraction of the higher complement yields grammatical results.

138 (56)

(57)

(58)

However, Valois does not agree completely with Giorgi and Longobardi’s generalizations. In fact, one of the main points he makes consists of criticizing the Head

Subject Paramenter. He claims that there is no such asymmetry in the projection of arguments in DP (see 54), and that differences in sentences like (57) (1991:13-14) follow from a distinction which has already been documented between the two languages at the clausal level: the presence versus absence of head-movement to inflectional morphology

(Emonds 1978; Pollock 1989; Chomsky 1990).

139 (59)

5.2.9. Longobardi (1994)

One of the main generalizations in Longobardi’s (1994) proposal is that bare common nouns -at least in Romance- can only be found in a lexically governed position

(object position), where they take on an indefinite/generic reading. This would imply the presence of an empty D that is lexically governed, typically by V, and is also responsible for the existential, indefinite/generic interpretation of bare nouns.

As far as bare nouns are concerned, Longobardi distinguishes between bare common nouns, generic nouns and proper names. He points out that generic and proper names are not subject to the same governing principles as those to which common bare nouns must obey. The reason for this difference would be that generics and proper names can rise from N-to-D overtly or covertly, while common nouns cannot.

The argument that in Italian proper names are base generated in N and move to D

is shown in (60) and (61) (Longobardi 1994:623).

(60)

a. Il mio Gianni

The my Gianni (no contrastive)

‘My Gianni’

140 b. Il Gianni mio (contrastive)

The Gianni mine

‘My Gianni’

(61)

a. Gianni mio

Gianni mine (no contrastive)

‘My Gianni’

b. *Mio Gianni

My Gianni

‘My Gianni’

Crucial to Longobardi’s analysis is the interpretation of prenominal and postnominal possessive. While a postnominal possessive has a contrastive interpretation, such reading is not possible when the possessive proceeds N. In (62) and (63) these two different constructions are given.

(62) DP

D NP

Il Spec N’

mio N

Gianni

141 (63)

DP

D NP

Il Spec N’

N AP

Gianni mio

As (61a) is not contrastive the possessive is supposed to be prenominal, like in

(62). The ungrammaticality of (61b) is due to the fact that in this construction N moves to

D and should therefore appear before the possessive. When the possessive is not contrastive it must appear in SpecNP; in this way it can be accounted for the co- appearance in Italian of article and possessive (64).

(64)

Il mio libro

the my book

‘My book’

Personal pronouns appear without determiner too and act like arguments, they are base generated in D as they do not describe as Ns do, but rather refer.

These elements, in fact, can also be followed by complements NPs like (65)

(Longobardi 1994:653).

(65)

Noi ricchi stiamo diventando ancora più ricchi

‘We rich are becoming even richer’ 142

Therefore, we will assume that Ds have the function of referring to objects described by Ns. Longobardi’s proposal is reported here in (66).

(66)

A ‘nominal expression’ is an argument only if it is introduced by a category D.

(1994:620)

5.3. Quantificational phrase

Several hypotheses have been suggested concerning the status of quantifiers in the

NP. Certain scholars proposed that quantifiers should be treated as a specific kind of adjective modifying the noun (Abney 1987); others claimed that they would be better explained as adjuncts (Sportiche 1988); only later quantifiers started being considered as heads of a quantificational phrase (QP) (Cardinaletti and Giusti 1991; Shlonsky 1991).

Based on this last proposal, several other hypotheses have been made on the structure of the DP (Zamparelli 2000; Ishane 2008) and on the nature of the role played by the quantificational elements that it contains.

For reasons of space, I will not discuss in details all the above mentioned proposals, in this section I will just focus on two of the most recent works dealing with quantificational elements in a layered DP framework. More precisely, I will briefly summarize the main points made by Zamparelli (2000) and Ishane (2008).

143 5.3.1. Zamparelli (2000)

A strict mapping between semantic and syntactic categories is assumed in

Zamparelli’s (2000) proposal. In his view, different functional projections give different semantic readings. Intuitively there are three different types of NPs: referential ( Tom arrived late ), quantificational ( Each student reads a book ) and predicative ( John is a

doctor ). Montague’s grammar assigns to all NPs the same semantics, NPs are considered

generalized quantifiers, <,t>. Partee (1987) suggests that the three NPs types have

three different semantic representations: referential NPs denote individuals ;

quantitative NPs are generalized quantifiers <,t>; and predicative NPs indicate

properties . Zamparelli reduces this tripartite distinction to a binary one: either

or . NPs with strong quantifiers undergo Quantifier Raising and Quantifier Construal;

they leave in their original position a variable and end up being of type.

QR extracts the quantified NP from its original position in the sentence and

adjoins it to IP or VP, while QC extracts the quantifier from the quantified NP. These

operations allow the subdivision of NPs in three parts: the operator, the restrictor and the

variable, within the nuclear scope (Hornstein 1995) (67):

(67)

Each student reads a book

a.[InflP[QP Each student] i [InflP h i reads a book]] (QR)

b.[(operador) Each j [InflP [QP (restrictor)h j student] i [InflP (nuclear scope)h i

(variable) reads a book]] (QC)

144 In this way it is possible to obtain the tripartite formula (operator, restrictor and nuclear scope) employed in semantics to represent operations with quantifiers.

Zamparelli proposes that definite determiners can denote individuals without undergoing

QR and QC at FL. In his view the determiner phrase should be divided in two functional nuclei: Strong Determiner Phrase (SDP) and Predicative Determiner Phrase (PDP), the first denoting individuals and the second properties. Strong and weak determiners sit in different projections in the nominal structure. The former are in the SDP and the latter in the PDP.

PDP represents the noun-phrase internal predicate. One of the functions of this layer is to turn nouns into properties (Zamparelli 2000:16). The determiner system of a language consists of SDP and PDP. PDP also embeds a third functional layer, the Kind

Determiner Phrase (KIP) which hosts the NP-proper and any restrictive adjective.

Zamparelli’s structure is presented in (68) and the structure for strong, weak and predicative noun phrases in (69a), (69b) and (69c) respectively:

(68)

SDP

Strong Quantifiers PDP

Weak Quantifiers KIP

Adjectives NP Noun

Complement

145 (69)

a. SDP

SD PDP Every, both SOME He, John PD KIP

(Adj) N

b. SDP

SD PDP

PD KIP Three, few, some (Adj) N

c. PDP

PD KIP Three, few, some (Adj) N

As this framework assigns referential functions only to SDP elements, they are the only ones which can act as arguments. On the other hand, PDP determiners are conceived as property-denoting elements, which behave like adjectives and cannot appear in argumental positions. Quantificational noun phrases have the same structure as referential ones, except that they undergo Quantifier Raising (QR) at Logic Form (LF).

That some noun phrases can have a strong and a weak reading is due to the position of the determiner, which can sit in SDP or PDP as illustrated in (49 a-b).

All elements that can appear in one of these functional projections are determiners

(articles, demonstratives and quantifiers). To be turned into an argument a NP must be dominated by the highest projection SDP. This requirement forces Zamparelli to 146 postulate the existence of an empty category occupying SDP nucleus every time a weak quantifier appears as an argument.

Definite determiners and strong quantifiers are base-generated in SDP. The latter are not able to turn a nominal projection into a referential one because they are located in

SpecSDP. The presence of SDP implies that the nominal projection indicates an individual. All weak quantifiers are base-generated in PredDP, some in the nucleus

(cardinal numerals and indefinite articles) and other in the specifier (the rest). These quantifiers stay in Pred DP if the nominal projection that they command is a predicate and not an argument as in (70).

(70)

John is a good doctor

In this case SDP is not present, so that the phrase would denote a property and not an individual. If, on the other hand, the SDP is present the phrase is argumental and can be headed by an empty category or by a phonologically expressed one.

Zamparelli’s model is not perfect though. For example Gutiérrez-Rodríguez

(2009) and Ishane (2008) think that the association SDP: argument, PDP: predicate is too strong. In fact, Zamparelli treats weak quantifiers with cardinal or no partitive reading in a way similar to Abney’s, in that the quantifier is seen as a pure predicative or adjectival element.

The referential part of the phrase is assigned to the empty category in SDP; however there is not empirical evidence indicating that the cardinal reading of weak quantifiers has an empty category in SDP. For example we would expect these NPs not to

147 appear as preverbal subjects, position in which the empty category would not adequately be bounded. Zamparelli also hypothesizes the existence of two empty categories to account for bare nouns and this does not seem convincing. The main problem consists of the fact that we lose the intuition by which also in cardinal or adjectival reading the quantifier is still a determiner, in that it closes the projection and converts N into an argument (Gutiérrez-Rodríguez 2009:69).

5.3.2. Ishane (2008)

Ishane (2008) departs from Zamparelli’s (2000) analysis of the DP to focus mainly on French un -NP constructions. She considers Zamparelli’s strong/weak distinction inappropriate in that it assimilates definite NPs to quantificational ones. In fact, even though Milsark’s (1974) test for strength put proper names and pronouns on the side of strong noun phrases, which are quantificational, she considers the former referential. Although proper names and pronouns pattern like strong quantifiers in existential sentences and with individual predicates, other contexts suggest that they should be distinguished 24 . For example, topicalization typically excludes quantificational noun phrases:

(71)

a. John, I saw (him) on Monday

b. *Every boy, I saw (him) on Monday

24 See for example the Law of Excluded Middle and the Law of contradiction reported in Ishane (2008: Chaper 2). 148 Ishane identifies three types of NPs, namely: Referential NPs, Quantificational

NPs and Property-denoting NPs, which, based on their different interpretations, are claimed to have different structures. In fact, if referential and quantificational NPs had to be put under a single structure (as Zamparelli suggested), it would be unclear how QR may distinguish between these two types of NPs when it comes to move ones

(quantificational) and not others (referential).

She refines Longobardi’s distinction between DPs and NPs (46) and proposes that argumental un -NPs minimally involve one layer of the nominal left periphery, the lowest one. She argues that it is this layer which turns predicative noun phrases into arguments.

As it is characteristic of un -NPs interpreted as properties, she labels it Property

Phrase (PropP). This phrase as the function of distinguishing predicates which can act as arguments (nominal predicates), from predicates which cannot (adjectival predicates).

This implies the smallest arguments are property denoting noun phrases.

Ishane’s layered DP can be schematically represented as in (67), with Property

Phrase (PropP), Quantifier Phrase (QP), and Speaker-reference Phrase (SRefP); schematically represented as (72)

(72)

SRefP>QP>PropP> …N

The interpretation of un -NPs depends on the piling up of these different functional layers in the nominal left periphery. No postulation of empty categories has to be made, as PropP transforms predicates into arguments and it is characteristic of property- denoting un -NPs.

149 As for QP, she suggests that it carries a quantificational feature which is responsible for a reading where no entity can be picked out and which is able to trigger

QR at LF. This projection is therefore required for a constituent to undergo QR and have scope properties. SRefP encodes S-reference by adding a projection encoding the relevant features 25 (73).

(73)

SRefP

Spec SRef’

SRef QP

Uni Spec Q’

Q PropP

t i Spec Prop’

Prop NumP

ti Spec Num’

Num ....

ti livre

5.4 Light ‘n’ (Radford 2000)

In line with the supposed parallelism between VP/CP and NP/DP, in the literature it has been proposed the existence of light ‘n’. In fact, several authors suggested that verb phrases have a complex internal structure including an inner VP core headed by a lexical verb V and an outer vP shell headed by an abstract light ‘v’ (Larson 1988; Hale & Keyser

25 The idea that the highest projection contains a feature related to referentiality is not new. Longobardi (1994) proposes that D hosts a (+/- Referential) feature, which differentiates ‘object denoting’ (i.e. pronouns and proper names) from ‘kind-denoting’ (i.e. common nouns) and which must be ‘checked’. 150 1993; Chomsky 1995; cf. Alexiadou et. al. 2007). Radford (2000) builds on previous works (see e.g. Carstens 2000; Sportiche 1990; and Valois 1991) and proposes that noun phrases have a similar shell structure, comprising an outer nP shell headed by a light noun and an inner NP core headed by a lexical noun; the nP shell in turn serves as the complement of higher-level nominal projections. He focuses mainly on the syntax of argumental (i.e. argument-selecting) nouns in English. He suggests that in sentences like

(74a) the agent DP the enemy is merged in SpecnP, a projection strictly outside NP, headed by the noun destruction . Just as V raises to v, N rises to n (74b). The Agent DP the enemy raises to the specifier of D to satisfy the EPP feature (74c); in this way it mirrors the subject of the clause raising out of vP to a higher position (see Alexiadou et. al. 2007: 563-564 for alternative proposals).

(74)

a. The enemy’s distruction of the city

b. [nP the enemy k [n destruction j] [NP [N t j] of the city]]

c. [nP the enemy’s k [D 0][nP tk [n destruction j] [NP [N tj] of the city]]]

5.5. DP prolific domains (Ticio 2003)

In line with the supposed parallelism between DP and CP, it has been proposed

that the structure of the former is as rich as the one of the latter. It has been suggested that

the structure of the DP, like the clause, can be divided in three domains, following the

cartographic approach to syntactic structure (Cinque 2002; Belletti 2004; Rizzi 2004).

151 (75)

a. [DP … [DP [FPadj1 … [FPadj2 … [nP…[NP..]]]]]]

b. [CP … [CP [FPadv1 … [FPadv2 … [vP…[VP..]]]]]]

The NP/VP-shells contain the thematic domain, where the arguments of the noun or verb are merged (Larson 1988; Chomsky 1995; Grimshaw; 1990; Valois 1991).

Agreement, phi-features and case are checked in the inflectional domain, which also contains adverb and adjective modifiers. The left periphery contains topic and focus features (Aboh, et. al. 2009). In the nominal structure, it is often associated with the notions of (in)definiteness, specificity and referenciality (Rizzi 1997).

Ticio (2003) adopts a layered DP based on Grohmann’s (2000) division of clause structure into three domains, and its extension to the DP proposed in Grohmann and

Haegeman (2002). Grohmann (2000) analyzes several cases of ill-formed movement in the clausal domain. He comes to postulate the existence of three Prolific domains into which the clause should be divided: A Thematic domain containing the predicate and its arguments; an Agreement domain, in which arguments can receive Case and Phi-features; and a Discourse domain, where discourse information is encoded.

Adapting Grohmann’s (2000) and Grohmann and Haegeman’s (2002) proposals,

Ticio (2003) assumes (76) as the basic DP structure.

152 (76)

The n P in the nominal domain is the counterpart of the vP in the clausal domain.

Similarly, the presence of FocP parallels the clause structure, in line with Rizzi’s (1997) proposal of split CP. FocP is the equivalent of Szabolsci’s (1983) and Ormazabal’s

(1991) K(om)P and it represents the escape-hatch for the elements extracted out of DPs.

It is projected only if there is need to check off a feature in the structure.

AgrP contains all the agreement-related functional categories like NumP, GenP,

PossP. She assumes, in line with Abney (1987) and Zamparelli (2000), that not all determiners are generated in D. Following Milsark’s (1974) division of Ds, she claims that DP is projected only if a strong determiner is present, while weak Ds are generated lower in the structure, in this case in AgrP.

In line with Grohmann’s (2000) implementation of the Anti-Locality Hypothesis, based on Boškovi ć’s (1997) and Saito and Murasugi’s (1999) proposals on a lower-bound on locality, movements within Prolific domains can be grammatical only if elements move from a Prolific domain to another, without possibility of skipping already occupied

153 landing sites. Therefore, she assumes that phrasal movement within DPs crosses only one maximal projection in each step.

5.6. Conclusion

In this chapter I have briefly summarized some of the main hypotheses concerning the study of the Determiner Phrase during the last thirty years, for a more detailed account I invite the interested reader to consult also Carstens (2000); Coene &

D’hulst (2003a,b); Alexiandou, Haegeman & Stavrou (2007) and Picallo (2008; 2010) among others. In the rest of this work, I will use the data encountered for ABS as a testing ground (Kayne 1996). According to the empirical evidence emerging from the study of this dialect, I will adopt some paradigms while refusing others.

154

CHAPTER 6: PHI-FEATURES: THE SYNTAX OF PERSON, NUMBER

AND GENDER

“Despite extensive research, agreement remains

deeply puzzling. There was a time when it was

treated mainly as a tool for researching other

syntactic phenomena. Yet there has also been a

tradition of recognizing it as a challenging problem

in its own right. Indeed agreement presents serious

problems for all theories of syntax” (Corbett

2001:1).

6.0. Introduction

This chapter consists of a descriptive and analytical work on the distribution of phi-features across the ABS DP. In line with recent works on Minimalism (Adger &

Harbour 2008), I take phi-features to be those involved in predicate-argument agreement, typically person, gender and number.

First, I will provide a brief overview of concepts adopted within the Minimalist

Program, such as: default feature, markedness, agreement, and Agree operation.

Secondly, based on the results from the grammaticality judgments of the oldest

155 informants, I will describe the traditional phi-feature composition of the ABS DP elements. Thirdly, a section will be dedicated to the study of the feature variability encountered in the Afro-Bolivian community, as suggested by discordant grammaticality intuitions and variation in the spontaneous speech of my informants. An additional section will present Lipski’s (2006a) model for variable gender and number agreement in the ABS DP. Finally, I will propose a different formal account to analyze the data.

6.1. Default values, Markedness, Agreement and Agree Operation

The presentation of the phenomena encountered in the ABS DP will make

reference to minimalist terminology and mechanics. A basic overview of the Agree

framework, as proposed by Chomsky (2000; 2001) and subsequently revisited by

Pesetsky and Torrego (2007), is given in the next section for readers who may not be

familiar with it. Moreover, concepts such as default feature and markedness will be

illustrated to facilitate the comprehension of the following parts.

6.1.1. Agree vs. agreement

In this work, I will refer to Agree as a syntactic operation (as opposed to agreement, the surface phenomenon). In the literature several hypotheses have been proposed to try to unveil the nature of Agree and the function of agreement in natural languages. These are fundamental questions, for which I do not believe a definite answer has yet been provided. I will not try to find a solution to this issue here, as this would generate a discussion which would lead us beyond the scope of the present dissertation. However, based on the parametric distinctions encountered between stSp and ABS, I will try to propose an implementation of the framework able to account for the data. 156 Nevertheless, it is worth noticing that different models have been proposed in the literature. For example, it has been suggested that agreement exists to permit freer word orders, to allow cross-reference between elements in the sentence, thus enhancing the power of communication, or to provide speech redundancy, needed to avoid transmission and reception problems due to language noise and to facilitate on-line processing (see e.g.

Corbett 2006; Givón 1976; Keenan 1974; Lehmann 1982; Di Domenico & De Vincenzi

1999; Hagoort and Brown 1999; etc.).

Minimalism approaches this topic by attempting to understand how agreement can contribute to making the computational system optimal. In fact, language is seen as a perfect machine. Economy is the central concept of the minimalist framework, so that every operation must have a specific function. Nevertheless, the specific function of agreement is still not clear. Chomsky (1998) himself has classified agreement as a potential case of ‘imperfection’ in language design; he also suggested that its purpose might be the implementation of Move. On the other hand, Atkinson (2002: Chapter 4) has criticized this argument, because agreement can be found also when movement is not encountered. For example, Franceschina (2005:83) shows an instance of Agree without

Move between finite T and the nominal further down in the structure (77):

(77)

a. There seem to be problems

b. There seems to be problems

Franceschina also suggests that the function of agreement may be related to learnability. In her view, “it is possible that agreement morphology might be one of the

157 properties of human language that makes particular languages readily learnable by providing a fairly direct route to the identification of the grammatical makeup of their lexical items” (Franceshina 2005:83). Given these considerations on the nature of agreement - which, even if intriguing, are not conclusive - I will now examine the proposals for the implementation of Agree.

Chomsky (2000; 2001) argues that Agree consists of a relation between two elements within a syntactic domain: a probe and a goal. He suggests that agreement is the consequence of a situation in which an unvalued instances of a feature F c-commands another instance of F. The probe-goal relation has been compared to a search, where the probe acts as seeker and the goal as object (Béjar 2008:133). A probe consists of an uninterpretable unvalued set of phi-features on a functional head, which will receive a value from some other syntactic constituent during the derivation.

For this reason Agree is seen as a case of feature assignment, which can be summarized in the following steps:

(78)

Agree (Assignment version; following Chomsky (2000; 2001))

(i) An unvalued feature F (a probe ) on a head H scans its c-command domain

for another instance of F (a goal ) with which to agree.

(ii) If the goal has a value, its value is assigned as the value of the probe.

The operation Agree serves the purpose of deleting uninterpretable features which

are unreadable at the interfaces and -if not eliminated- would cause the derivation to

crash. Deletion takes place in a cyclical fashion at the end of each phase.

158 Uninterpretable features, however, cannot be deleted during the syntactic derivation just by virtue of the fact that they cannot be interpreted at LF. The only means that the framework has to eliminate such features is to assume a biconditional relation correlating unvalued features with uninterpretable ones (79):

(79) Valuation/Interpretability Biconditional (Chomsky (2001:5))

A feature F is uninterpretable iff F is unvalued.

By recurring to (79), the model can now delete uninterpretable features because they are unvalued, and therefore acting as probes. Such a stipulation inevitably leads us to the conclusion that once an uninterpretable feature has been valued, it will also get automatically deleted.

Chomsky’s Agree operation is therefore a syntactic mechanism of ‘feature assignment’, triggered during the derivation by an unvalued-valued (probe-goal) relation, which, by virtue of a feature-biconditional requirement, results in the cyclical deletion of uninterpretable features before Spell-Out.

Chomsky’s (2000; 2001) proposal has been revisited and refined by Pesetsky and

Torrego (2007), among others (see for example Frampton and Gutmann (2000). In fact, recent work on Agree advocates a version of this operation which departs from the previous view of ‘feature assignment’ mechanism (Chomsky 2000). Rather, the process is seen as an instance of ‘feature sharing’ an idea in line with the view of agreement as feature unification common in HPSG (Pollard & Sag 1994). Within the probe-goal theory of the syntactic computation, the operation Agree has been reformulated as in (80).

159 (80) Agree ( Pesetsky and Torrego 2007 )

(i) An unvalued feature F (a probe ) on a head H at syntactic location α(F α) scans its c-command domain for another instance of F (a goal ) at location β(F β) with which to

agree.

(ii) Replace F α with F β, so that the same feature is present in both locations.

If a goal is valued for F, replacing the token-value of the probe with the value of

the goal results in an instance of valued F substituting for the specification of the

unvalued probe. A valued F may now serve as the goal for some ulterior operation of

Agree triggered by an unvalued, higher instance of F serving as a new probe. The result is

that a single feature F will be shared by several positions, and the process could iterate

further.

Pesetsky and Torrego’s proposal is different from Chomsky’s approach not only

in its feature-sharing view of Agree, but also in the absence of the

Valuation/Interpretability Biconditional in (79). By removing this last constraint, the

authors postulate the presence of features containing combinations of properties not

available in the model previously suggested by Chomsky: (i) uninterpretable but valued;

and (ii) interpretable but unvalued.

Lexical entries can now enter the derivation with four different kinds of features:

(81) Types of features (boldface = disallowed in Chomsky (2000; 2001))

uF val uninterpretable, valued iF val interpretable, valued

uF [ ] uninterpretable, unvalued iF [ ] interpretable, unvalued

160 This new framework, which stipulates the independence of valuation and interpretability, seems to be validated by several syntactic phenomena: the relationship between Tns and the finite verb, the formation of an interrogative CP, the formation of a declarative CP that supports successive-cyclic wh- movement; etc. For reason of space, I will not enter into details concerning how all these phenomena are accounted for by the model 26 ; on the other hand, I will limit the illustration of this approach to the explanation of how the relationship between Tns and the finite verbs obtains.

In fact, an example of an interpretable unvalued feature acting as a probe is the T feature of the category Tns. In line with Pollock (1989), who posited a distinct Tns node as the locus of semantic tense interpretation, an uninterpretable feature that participates in an Agree relation with the T feature on Tns has been postulated for languages in which finite verbs bear morphological tense markers. Since Tns c-commands the finite verb, its

T feature will act as a probe. For this reason, the T feature on Tns is seen as an interpretable unvalued feature searching for a goal, represented by the T feature on the finite verb, which is uninterpretable but valued:

(82) The relationship between Tns and the finite verb

Agree

... Tns ... [ v walked] ... ... Tns ... [ v walked]

i T[ ] uT +past iT[2] uT +past[2]

26 The interested reader is invited to consult Pesetsky and Torrego (2007) and the relevant bibliography.

161 This situation thus exemplifies precisely the two types of features expected under the current proposal, but disallowed in the previous framework. Nevertheless, the authors do not reject completely Chomsky’s model. They maintain that Agree serves the purpose of deletion to avoid a crash in the derivation. At the same time, they share Brody’s (1997) view on Radical Interpretability , which states as follows:

(83) Thesis of Radical Interpretability (Brody 1997)

Each feature must receive a semantic interpretation in some syntactic location.

Therefore, if all features must have an interpretation at a certain point, it follows

that what is deleted is not the feature itself, but rather its uninterpretable instances.

Radical Interpretability in conjunction with the feature sharing view of Agree also offers

an immediate explanation for the fact that an uninterpretable valued feature (like [ uT val ]

on the finite verb) must enter an Agree relation with an interpretable counterpart ([ iT [ ]] on Tns). If this Agree relation were not established, then the T feature would not receive an interpretation in any syntactic location, in violation of Radical Interpretability.

It is important to point out that both the ‘feature assignment’ and the ‘feature sharing’ models share the idea that agreement is not a projection (contra previous frameworks, e.g. Mallén 1997). On the other hand, agreement is now pictured as the over result of Agree, an operation taking place during the syntactic derivation given a proper c-command relation between a probe and a goal. In fact, recent minimalist works clearly support the elimination of Agreement projections (Chomsky 2002).

162 6.1.2. Default values and Markedness

The notion of markedness is not a new one. It can at least be tracked back to the

1930’s, to the works on phonology by Trubetzkoy (1939). Yeas later, Jakobson elaborated on Trubetzkoy’s notion of ‘distinct oppositions’ and formulated his theory of

‘distinctive features’, with the main difference that while Jackobson adopted a binarist approach, Trubetzkoy’s feature concept included non-binary as well as binary features

(see also Jakobson and Halle (1952) and Chomsky and Halle (1968)).

Subsequently, different scholars have incorporated the concept of markedness to new linguistic theories. A variety of syntactic and lexical ‘unmarked’ universals are discussed in Greenberg (1966), Chomsky (1981, 1982), Comrie (1981), Faingold (2003),

Corbett (1991, 2000, 2006), among many others. Since this chapter focuses on phi- features in the Afro-Bolivian Spanish DP, I will analyze how the concept of markedness is related to person, number and gender.

In the Afro-Bolivian Spanish DP all of the three phi-features are encountered. As in standard Spanish (stSp), such features present the following values, as reported in

Table 1:

Table 9. Phi-features and values in the ABS DP

Feature Values Person 1st 2nd 3rd Gender Masculine Feminine Number Singular Plural

163 As far as phi-features in ABS and stSp are concerned, there is no distinction in their quality; as we will see in the next section, the only difference encountered between the two languages has to do with their distribution across the DP elements.

Demonstrating naturalness or markedness seems relatively easy for phonological processes compared to grammatical ones. For example, consonant cluster simplifications

(CC) represent economies in articulatory (motor) gestures that are quantifiable

(Chambers 2003:258-259). Principles that might underlie natural grammatical processes are never so concrete. For this reason, a comprehensive theory of syntactic markedness has still to be formulated even though markedness has been a major theme in phi-theory

(Harbour, Adger and Béjar 2008).

Dixon (1994) distinguishes between “formal” and “functional” markedness, where the former indicates whether a form is overtly marked, while the latter concerns how grammatical categories can be distinguished from others. As far as formal markedness is concerned, we can exemplify the case by comparing Spanish nouns. The singular noun perro ‘dog’ is unmarked, while its plural counterpart perros ‘dogs’ is marked in that it carries an additional morpheme indicating plurality -s. An example of

functional markedness can be provided by analyzing the English pronominal system,

where nominative and genitive pronouns are employed in specific contexts, while

accusative pronouns are unmarked, in that accusative acts as the default case, appearing

for direct objects, predicates, subjects of gerunds, etc. (Adger & Harbour:22).

Often, formal and functional markedness coincide. The third person aptly

exemplifies this situation. Silverstein (1976) points out a variety of pronominalization

phenomena where third person forms are employed even though no person is actually

164 involved in the action described. From this, he suggests that third person is functionally unmarked (a default case), whereas first and second are marked. Even if in the literature there is not a precise criterion for morpho-syntactic markedness; as far as person, gender and number features are concerned, some generalization have been proposed.

Greenberg (1966) indicates that languages that mark gender in the plural, most likely indicate it also in the singular, but not viceversa. So that gender distinctions increase where number decreases. Moreover, Corbett (1991; 2000) argues that gender and number distinction tend to be marked with respect to person in opposite ways. In fact, often languages which do not mark gender distinctions for all persons tend to do it only for third or only for second and third. On the other hand if number distinctions are made only for some persons, these will most likely be only first, or only first and second.

A generalization that can be made on Greenberg’s and Corbett’s observations is that where language is most likely to make number distinctions (in first, or first and second person), it is least likely to make gender distinctions (Adger & Harbour 2008:24). Even if these generalizations are based on vast typological surveys, it must be said that they are not universals; they rather represent tendencies and exceptions can always been discovered.

As far as person, gender and number features are concerned, the common view in the literature on Spanish and the rest of the Romance languages treats the values third, masculine and singular as the default or the least marked ones, and considers first, second, feminine and plural as their marked counterparts (Corbett 2001).

In light of what was stated in Chapters 3 and 4 on the evolution of Afro-Bolivian

Spanish; if we analyze this language as the result of crystallized processes of second

165 language acquisition, the distribution of phi-features across its grammatical elements further supports the assumption on the status of third, masculine and singular as the unmarked values. In the next section the theoretical framework so far provided on feature markedness and Agree will be applied to the ABS and stSp DP data.

6.2. Applying the theory to the data

In order to account for the presence of plural morphology on English nouns,

Chomsky (2000; 2001) had to postulate the presence of a valued interpretable number feature on this element. By assuming such valued number specification, all DP entries specified for an unvalued uninterpretable number feature would be able to probe for it, in line with the c-command restriction imposed by Agree. On the other hand, if a higher element were bearing the interpretable feature, N would not be able to c-command such a value and its overt morphological marking could not be explained.

However, Chomsky’s position has been criticized because it fails to identify Num as the locus of number interpretability (Carstens 2000; Picallo 2008), contrary to what is generally assumed in the literature. Nevertheless, if we hold to the

Valuation/Interpretability Biconditional and accept that number is interpretable on Num,

Agree cannot account, at least in Spanish, for some crucial morphological facts: First, there is no way to account for plural marking on N; second, postnominal adjectives, which are generally believed to be based generated in projections lower than NumP, should not carry number morphology either.

A way to circumvent such problems would consist of resorting to a different

operation, Concord (Carstens 2000; Demonte 2008), which does not depend on c-

166 command. An additional proposal (Franceschina 2005), has suggested an ad hoc co-

indexation between N and the postnominal A, so that when N moves to Num, the noun

and the adjective will simultaneously agree and get identical number value 27 . Arguably,

such moves are undesirable, since they eliminate any generalization of agreement.

As far as the valuation of number and gender features in the Spanish DP is

concerned, the elimination of the Valuation/Interpretability Biconditional seems to

account perfectly for the data. In fact, if we postulate that N contains an interpretable

gender feature and an uninterpretable valued number feature, while Num contains an

interpretable unvalued number specification, all DP elements become able to probe a

gender and number value from N while obeying the principle of c-command.

Pesetsky and Torrego (2007) do not provide a detailed explanation of how such a

reconfiguration would be implemented in the Spanish DP. They limit themselves to

suggesting that locating [ inum] on Num, and the number value on N would provide an

explanation for Latin pluralia tantum nouns (Pesetsky and Torrego 2007:264, fn 1).

Therefore, to provide a better account of how the syntactic framework here adopted

works in stSp, let us consider the derivation of a simple stSp DP (84).

(84)

stSp : Esas pequeñas casas rojas

This-F-PL small-F-PL house-F-PL red-F-PL

‘These small red houses’

27 Francheschina (2005:87 fn14) remains uncommitted about the exact implementation of this operation. 167 At the point of Merge, the noun casa- ‘house’ carries an interpretable valued gender feature, an uninterpretable valued number feature and an uninterpretable valued person feature. N bears also an unvalued Case feature; however, as Case is not relevant to our discussion, this last feature will not be included in the present representation. Note that the symbol “+” next to the value indicates that the item is entering the derivation with a certain feature valued e.g. [ igen:+F].

(85)

n’

n NP casa- [ igen:+F] [unum:+Pl] [upers:+3 rd ]

The noun will raise and adjoin to the n, probably prompted by the strength of its

categorial feature. In this way, N leaves an invisible copy behind. Subsequently, n

projects a specifier in which the first AP is merged. The adjective roj - ‘red’, which bears uninterpretable unvalued gender and number features, represents the AP’s head. The uninterpretable unvalued features on A act as probes looking for a goal in their local c- command. The noun values the adjective’s gender and number features, which can now act as a goal (Frampton and Gutmann 2000) and which will be deleted before Spell-

Out 28 .

28 To account for the fact that not all adjectives are post-nominal in Spanish and Romance, it has been suggested that some may be generated in the specifier of Num or other functional projections between D and Num (Cinque 1994). The present work is based on these assumptions. 168 (86)

After this step, the nP will merge with the Num head, which carries an interpretable and unvalued plural number feature, able to probe for the value contained in

N. N raises to Num, Num projects a specifier where the second AP can be merged. Its head ( pequeñ -) undergoes the Agree operation previously applied to roj -, so that its uninterpretable unvalued gender and number features will also be valued and deleted before Spell-Out.

(87)

169 Eventually, the D head is also merged and its unvalued phi-features acquire a value. Its uninterpretable gender and number instances are finally erased, while the interpretable person instance survives.

(88)

The system so far provided seems to work perfectly for stSp, where gender and number are marked redundantly across all the DP elements. Nevertheless, this model, given the c-command restriction on Agree and the valued number feature on N, cannot account for the ABS data. In fact, all stSp demonstratives, quantifiers, nouns and articles come from the lexicon with a specification for number and gender features. Such specification, as shown in (84-88) is what will result in overt number and gender morphological marking after all the Agree operations have applied. On the other hand, traditional ABS does not posses the richness in feature specification characteristic of stSp

170 and other Romance languages. In traditional ABS, nouns are specified for gender; this feature is not morphologically marked on the majority of the DP elements (it only appears on definite articles). Also, the morphological distribution of number marking is much more restricted: it is limited to determiners, and it never applies to adjectives, nouns and quantifiers (see Section 6.3 for a full account of the ABS DP feature specifications).

The ABS counterpart of (84) is (89).

(89)

ABS: Ejes pequeño casa rojo

This-M-Pl small-M-Sg house-F-Sg red-M-Sg

‘These small red houses’

As we want to keep syntactic processes constant and universal (Brody 2003) 29 ,

neither ad hoc modifications to the operation Agree nor the introduction of special mechanisms to account for the data are available options. Nevertheless, the theory offers a different solution to this problem. Within the Minimalist Program/Principles and

Parameter framework, an account of cross-linguistic variation can be found in the different distribution of feature specifications between the lexical entries of the varieties under analysis. Therefore, to account for constructions like (89) in ABS, we may postulate that in this language, contrary to stSp, nouns only carry interpretable valued gender features and uninterpretable valued person ones, so that they are not specified for number. On the other hand, Num is the element carrying interpretable valued number

29 Brody refers to that as ‘perfect syntax’. 171 features; D bears uninterpretable unvalued number features and interpretable unvalued person ones; while Adjectives do not have any specifications for phi-features. In other words, traditional ABS DP elements lack many of the unvalued uninterpretable features encountered on their stSp counterparts. Such a deficiency results in the default singular and default masculine morphological realizations, so that the stSp example (88) can be derivationally represented as (90) for ABS.

(90)

ABS: Ejes pequeño casa rojo

This-M-Pl small-M-Sg house-F-Sg red-M-Sg

‘These small red houses’

Besides the difference in feature specification between the two varieties, it is important to state another clear parametric distinction: in ABS the number value enters the derivation as a specification of Num (i.e.[ inum:+Pl/+Sg]); in stSp it is carried by N.

172 Note that this parametric distinction could arguably be postulated also for the contrast in number marking found between standard Brazilian Portuguese (stBP) (redundant plural marking) and popular Brazilian Portuguese (pBP) (non-redundant plural marking), where constructions like (91) are grammatical (cf. Simioni 2007).

(91)

pBP: As casa vermelha

the-F-Pl house-F-Pl red-F-Pl

stBP: As casas vermelhas

the-F-Pl house-F-Pl red-F-Pl

‘The red houses’

Moreover, note that ABS’s poverty of feature specifications does not prevent this

language from presenting the same adjctive+noun and noun+adjective order

combinations encountered in Romance. In fact, as Carstens (2001:54) and Alexiadou

(2001:223), among others, have demonstrated, raising of N to Num is not prompted by

number feature checking, but rather by other mechanisms such as EPP or categorial

features. This indicates, in line with Atkinson’s (2002) observation, that agreement at

least in these clear examples, cannot feed movement.

In the next section, an account of the distribution of person, gender and number

features will be provided for the grammatical elements contained within the ABS and the

stSp Determiner Phrase.

173 6.3. Phi-features distribution across Afro-Bolivian Spanish and Highland

Bolivian Spanish DPs

This section consists of a contrastive description of the phi-feature composition of the traditional ABS and the stSp DP elements. While for stSp I could rely for the most part on manuals and grammars of Latin American and Peninsular Spanish (Lipski 1994;

Zagona 2002; etc.) findings for traditional ABS have been derived from the grammatical intuitions of the oldest informants living in the community of Tocaña, Mururata and

Chijchipa, villages in the municipality of Coroico, La Paz.

Before proceeding with this descriptive task, it is worth restating that, in line with

Pesetsky and Torrego (2007), features come from the lexicon in four sorts: uF [ val ] uninterpretable, valued; iF [ val ] interpretable, valued; uF [ ] uninterpretable, unvalued; iF [ ] interpretable, unvalued. Moreover, in order to represent more clearly the distinction between these two languages, elements coming from the lexicon without a specification for a certain feature will be annotated, when relevant, as NO-F [ ].

As far as phi-features are concerned, Num, nouns, pronouns and possessives are the only elements coming directly from the lexicon with some sort of feature value. The rest of the elements, when specified for a given feature, will act as probes for a value, ending up sharing it with other elements across the DP (Frampton and Gutmann 2000).

Moreover, each feature has a unique locus of interpretation (Brody 1997), so that the rest of the instances of such feature will be uninterpretable.

174 I am assuming the DP structure provided in (92); where the loci of interpretation for person, number and gender are D °, Num ° and N °, respectively (see also Carstens

2000).

(92)

However, it must be kept in mind that due to the elimination of

Valuation/Interpretability Biconditional (Pesetsky and Torrego 2007), I am not claiming that such interpretational loci will always come from the lexicon valued in both languages. In fact, the assumption is that in ABS, N enters the derivation with a value for gender [ γ] and one for person [ α], Num carries a value for number [ β] and D lacks person value [ α]; on the other hand, in stSp N introduces into the derivation all phi-values [ α],

[β] and [ γ], so that Num and D do not introduce valued features into the derivation. Let us now start with a description of the elements carrying valued phi-features in ABS and stSp.

175 6.3.1. Nouns

Nouns in stSp come from the lexicon with an interpretable valued gender feature, an uninterpretable valued 3 rd person feature and an uninterpretable valued number

feature. The ABS N presents a similar phi-configuration; it only differs from the stSp N

in that it does not have a specification for number, as reported below (93).

(93)

Afro-Bolivian Spanish Standard Spanish

uPERSON [3 rd ] uPERSON [3 rd ]

NO-NUMBER [ ] uNUMBER [singular/plural]

iGENDER [masculine/feminine] iGENDER [masculine/feminine]

Of these three features, only gender has interpretation on the noun, while number

and person have their loci of interpretation on Num and D respectively. Following

Longobardi (1994), I assume that only DPs can serve as arguments, NPs are predicates

turned into arguments by D (cf. also Contreras 1986); from this follows that D elements

must contain also referential features. We will explore this issue in more details in

chapter 7.

The application of Agree to the different number specifications of ABS and stSp

has a clear effect on the overt result. Due to the differences between these two languages,

overt morphological realizations on N diverge quite significantly. In fact, while in stSp

plurality is overtly indicated on the noun by the morphological markers -s, -es , such

inflectional elements do not appear on ABS nouns.

176 (94)

Afro-Bolivian Spanish Standard Spanish

Singular Plural Singular Plural

gato gato (cat) gato gatos (cat)

ciudá ciudá (city) ciudad ciudades (city)

mujé mujé (woman) mujer mujeres (woman)

No overt morphological indicators show the presence of the 3 rd person value on

N, neither in ABS nor in stSp. In ABS, As in stSp, nouns ending in –a tend to be

feminine, while those ending in –o tend to be masculine, even if many exceptions are

encountered (see Harris (1991) for a detailed account).

The gender values encountered in stSp nouns are generally maintained in ABS.

Nevertheless, some mismatches can be found, so that the lexical specification for gender

on N does not coincide completely between the two languages. Examples of such

differences are presented in (94), where the agreement patterns with the article indicate

the gender value for the noun.

(95)

ABS: Él dice que es el máximo autoridad

He-M-Sg say-pres that be-pres the-M-Sg highest-M-Sg authority-M-Sg

stSp: Él dice que es la máxima autoridad

He-M-Sg say-pres that be-pres the-F-Sg highest-F-Sg authority-F-Sg

‘He says he is the highest authority’

177

ABS: El serpiente no se come, se mata con escopeta

The-M-Sg snake-M-Sg no reflex eat reflex kill-pres with rifle-F-Sg.

stSp: La serpiente no se come, se mata con escopeta

The-F-Sg snake-F-Sg no reflex eat-pres reflex kill-pres with rifle-F-Sg

‘The snake cannot be eaten, it must be killed with a rifle’

ABS: La problema de Tocaña era lu patrón

The-F-Sg problem-F-Sg of Tocaña be-past the-M-Pl owner-M-Pl

stSp: El problema de Tocaña era lu patrón

The-M-Sg problem-M-Sg of Tocaña be-past the-M-Pl owner-M-Pl

‘The land owners were the problem of Tocaña’

ABS: La sistema de hacienda no sirve pa’ nada

The-F-Sg system-F-Sg of plantation-F-Sg no work-pres for nothing

stSp: El sistema de hacienda no sirve para nada

The-M-Sg system-M-Sg of plantation-F-Sg no work-pres for nothing

‘The plantation system is useless’

6.3.2. Num

Num is a head. Both in ABS and stSp, Num is responsible for the interpretation of number at LF. In ABS Num carries valued [ inum: ] from the Lexical Array/Numeration, while in stSp the value is introduced by N. I assume that for the two dialects the valued

178 Num is affixed to D to ensure the materialization of [+Pl] in the morphology component since there is no lexical item in Num to host the morpheme -s in PF. Note that the symbol

“” indicates that the feature will be valued during the derivation.

(96)

Afro-Bolivian Spanish Standard Spanish

Num Num

iNUMBER [singular/plural] iNUMBER []  singular/plural

6.3.3. Subject pronouns

Pronouns, differently from nouns, are referential elements. In other words, they

come from the lexicon with an already valued referential feature and they do not need to

be introduced by a different determiner. A pronoun can be seen as an element subsuming

all DP’s phi-features in a single entry. For this reason, I suggest that they are generated in

D position. ABS pronouns come from the lexicon with their gender and number features

interpretable and valued. In addition, in stSp, 3 rd person singular, 3 rd person plural and 1 st person plural pronouns carry also a valued and interpretable specification for gender features, which is not found in ABS.

For this reason, the overt distinction between feminine and masculine encountered for those pronouns in stSp is not present in ABS. As a result, ele refers to both ‘he’ and

‘she’, eyu combines the function of Spanish ellos and ellas , while nojotro is used instead of nosotros and nosotras .

179 Moreover, as noted by Lipski (2008b:98), also the familiar-formal distinction ( tú vs. usted ) has been lost in Afro-Yungueño, and in the most traditional speech, regional tú

‘2nd sing.’ and vos ‘2 nd sing’, both of which are used in stSp, are not used.

(97)

Afro-Bolivian Spanish Standard Spanish

Yo Yo

iPERSON [1 st ] iPERSON [1 st ]

iNUMBER [singular] iNUMBER [singular]

Oté Tú ≈ Vos ≈ Usted

iPERSON [2 nd ] iPERSON [2 nd ]

iNUMBER [singular] iNUMBER [singular]

Ele Él/Ella

iPERSON [3 rd ] iPERSON [3 rd ]

iNUMBER [singular] iNUMBER [singular]

NO-GENDER [] iGENDER [masculine/feminine]

Nojotro Nosotros/Nosotras

iPERSON [1 st ] iPERSON [1 st ]

iNUMBER [plural] iNUMBER [plural]

NO-GENDER [] iGENDER [masculine/feminine]

180 Otene Ustedes iPERSON [2 nd ] iPERSON [2 nd ] iNUMBER [plural] iNUMBER [plural]

Eyu Ellos/Ellas iPERSON [3 rd ] iPERSON [3 rd ] iNUMBER [plural] iNUMBER [plural]

NO-GENDER [] iGENDER [masculine/feminine]

Some examples are provided in (98).

(98)

ABS: Yo vive a Mururata. ¿Oté es de La Paz?

I live-pres to Mururata you be-pres of La Paz

stSp: Yo vivo en Mururata. ¿Tú eres de La Paz?

I live-pres in Mururata you be-pres of La Paz

‘I live in Mururata. Are you from La Paz?’

ABS: Eyu come lechón en Navidá, nojotro también.

They eat-pres lechón-M-Sg in Christmas we too

stSp: Ellos comen lechón en Navidad, nosotros también.

They eat-pres lechón-M-Sg in Christmas we too

‘They eat roast suckling pig for Christmas, we eat it too’

181 6.3.4. Object Pronouns

In Afro-Yungueño Spanish, direct and indirect object pronouns are phonologically identical, while in stSp some overt distinction is maintained. Moreover, in stSp direct object pronouns present gender distinction on the third singular and plural persons, while such a gender specification is not found in ABS. Direct object pronouns in

ABS and stSp are presented below.

(99)

Afro-Bolivian Spanish Standard Spanish

Mi Me

iPERSON [1 st ] iPERSON [1 st ]

iNUMBER [singular] iNUMBER [singular]

Ti Te

iPERSON [2 nd ] iPERSON [2 nd ]

iNUMBER [singular] iNUMBER [singular]

Lu Lo ≈ Le /La

iPERSON [3 rd ] iPERSON [3 rd ]

iNUMBER [singular] iNUMBER [singular]

No-GENDER [] iGENDER [masculine/feminine]

Noh Nos

iPERSON [1 st ] iPERSON [1 st ]

iNUMBER [plural] iNUMBER [plural]

182 Luh Los ≈ Les

iPERSON [2 nd ] iPERSON [2 nd ]

iNUMBER [plural] iNUMBER [plural]

Luh Los ≈ Les /Las

iPERSON [3 rd ] iPERSON [3 rd ]

iNUMBER [plural] iNUMBER [plural]

No-GENDER [] iGENDER [masculine/feminine]

The following table provides a list of the indirect object pronouns in ABS and stSp.

(100)

Afro-Bolivian Spanish Standard Spanish

Mi Me

iPERSON [1 st ] iPERSON [1 st ]

iNUMBER [singular] iNUMBER [singular]

Ti Te

iPERSON [2 nd ] iPERSON [2 nd ]

iNUMBER [singular] iNUMBER [singular]

Lu Le

iPERSON [3 rd ] iPERSON [3 rd ]

iNUMBER [singular] iNUMBER [singular]

No-GENDER [] iGENDER [masculine/feminine]

183 Noh Nos

iPERSON [1 st ] iPERSON [1 st ]

iNUMBER [plural] iNUMBER [plural]

Luh Les

iPERSON [2 nd ] iPERSON [2 nd ]

iNUMBER [plural] iNUMBER [plural]

Luh Les

iPERSON [3 rd ] iPERSON [3 rd ]

iNUMBER [plural] iNUMBER [plural]

No-GENDER [] iGENDER [masculine/feminine]

The second person singular clitic ti corresponds to the subject pronoun oté , while the second object plural clitic lu(h) corresponds to otene. ABS object clitics occur in preverbal position in simple sentences with only one verb. In this way they behave like stSp object clitics.

(101)

ABS: Mi ha dició

to-me have said

stSp: Me ha dicho

to-me have said

‘He/she told me’

184 ABS: Yo mi caíu

I reflex fallen

stSp: Yo me he caido

I reflex have fallen

‘I fell down’

When object clitics occur in more complex constructions of the kind modal + infinitive or auxiliary + gerund , they appear after the auxiliary or the modal verb, in a

fashion which resembles verbal patterns found in Portuguese.

(102)

ABS: Yo va ti avisá

I go-pres to-you tell-inf

stSp: Yo te voy a avisar

I to-you go to tell-inF

‘I’m going to tell you’

ABS: Yo quere ti preguntá un cosa

I want-pres to-you ask-inf a-M-Sg thing-F-Sg

stSp: Yo quiero preguntarte una cosa

I want-pres ask-inf-to-you a-F-Sg thing-F-Sg

‘I want to ask you something’

185 6.3.5. Indefinite and negative pronouns

ABS and stSp share the same indefinite and negative pronouns for non-human entities algo ‘something’ and nada ‘nothing’ . On the other hand, their human

counterparts differ. While in stSp alguien ‘somebody’ and nadie ‘no one’ are employed, in traditional ABS algún and ninguno are encountered. These elements derive from their stSp adjectival forms, algún and ningún respectively. Some examples of their usage are

provided in (103).

(103)

ABS: Juan comió algo malo

Juan eat-past something-M-Sg bad-M-Sg

stSp: Juan comió algo malo

Juan eat-past something-M-Sg bad-M-Sg

‘Juan ate something bad’

ABS: Nada tiene sabor allí

nothing have-pres taste-M-Sg there

stSp: Nada tiene sabor allí

nothing have-pres taste-M-Sg there

‘Nothing is tasty there’

186 ABS: Algún mi tomó mi vaso

some to-me take-past my-Sg glass-M-Sg

stSp: Alguien me tomó el vaso

somebody to-me take-past the-M-Sg glass-M-Sg

‘Somebody took my glass’

ABS: Ninguno vinió a la cancha

no come-past to the-F-Sg field-F-Sg

stSp: Nadie vino a la cancha

nobody come-past to the-F-Sg field-F-Sg

‘Nobody came to the soccer field’

In spite of their overt differences, as far as phi-features are concerned, all of these pronouns seem to be endowed with the same interpretable set: number and person.

Gender specification is not present in either language.

(104)

Afro-Bolivian Spanish Standard Spanish

Algo/Nada/Algún/Ninguno Algo/Nada/Alguien/Nadie

iPERSON [3 rd ] iPERSON [3 rd ]

iNUMBER [singular] iNUMBER [singular]

187 6.3.6. Possessive determiners

Possessive determiners in ABS and stSp are basically identical; they may differ slightly in phonological form (see examples below). The role of these grammatical entities is to establish a relation of possession between the possessor and the element possessed (possessum). For this reason, they enter the derivation with a ‘mixed’ feature specification: for the possessor, they carry interpretable valued person and number features; for the possessum, they carry an uninterpretable unvalued number and interpretable unvalued person feature, which will be valued during the derivation. I assume that only in stSp, 1 st person plural possessive determiners carry unvalued gender features, while in ABS these elements are not specified for gender and appear with the default masculine marking (105).

(105)

ABS: Nuestro manzana

our-M-Sg apple-F-Sg

stSp: Nuestra manzana

our-F-Sg apple-F-Sg

‘Our apple’

188

(106)

Afro-Bolivian Spanish Standard Spanish

Mi/Mis Mi/Mis iPERSONpossessor [1 st ] iPERSONpossessor [1 st ] iNUMBERpossessor [singular] iNUMBERpossessor [singular]

uNUMBERpossessum []  singular/plural uNUMBERpossessum []  singular/plural

Tu/Tus Tu/Tus iPERSONpossessor [2 nd ] iPERSONpossessor [2 nd ]

iNUMBERpossessor [singular] iNUMBERpossessor [singular] uNUMBERpossessum []  singular/plural uNUMBERpossessum []  singular/plural

Su/Sus Su/Sus iPERSONpossessor [3 rd ] iPERSONpossessor [3 rd ] iNUMBERpossessor [singular] iNUMBERpossessor [singular] uNUMBERpossessum []  singular/plural uNUMBERpossessum []  singular/plural

189 Nuestru/Nuestros Nuestro/Nuestros

iPERSONpossessor [1 st ] iPERSONpossessor [1 st ]

iNUMBERpossessor [plural] iNUMBERpossessor [plural]

uNUMBERpossessum []  singular/plural uNUMBERpossessum []  singular/plural

No-GENDERpossessum [] uGENDERpossessum [] masculine/feminine

Su/Sus Su/Sus

iPERSONpossessor [2 nd ] iPERSONpossessor [2 nd ]

iNUMBERpossessor [plural] iNUMBERpossessor [plural]

uNUMBERpossessum []  singular/plural uNUMBERpossessum []  singular/plural

Su/Sus Su/Sus

iPERSONpossessor [3 rd ] iPERSONpossessor [3 rd ]

iNUMBERpossessor [plural] iNUMBERpossessor [plural]

uNUMBERpossessum []  singular/plural uNUMBERpossessum []  singular/plural

As nouns do not inflect for number in ABS, possessive determiners are crucial to distinguish between singular and plural DPs (107).

(107)

ABS: Mi campo

My-Sg field-M-Sg

stSp: Mi campo

My-Sg field-M-Sg

‘My field’

190 ABS: Mis campo

My-Pl field-M-Sg

stSp: Mis campos

My-Pl field-M-Pl

‘My fields’

Possessive determiners are not the only elements coming form the lexicon with two different sets of phi-features; a grammatical category similar to determiner in this respect is possessive adjectives, which will be analyzed in the next section.

6.3.7. Possessive Adjectives

In stSp possessive adjectives ( mío , tuyo , suyo , etc.) are also encountered. These elements occur post-nominally and tend to be used to express contrastive possession

(Longobardi 1994) as in (108).

(108)

El libro mío, no el tuyo .

the-M-Sg book-M-Sg my-M-Sg no the yours-M-Sg

‘My book, not yours’

They come from the lexicon with interpretable valued person and number features

(of the possessor) and uninterpretable unvalued gender and number features (of the possessum). After Agree has applied, the possessive adjective’s unvalued features will share the number value introduced in the derivation by Num and the nominal gender value.

191 Possessive adjectives are completely absent in ABS. Nevertheless, this language recurs to periphrastic constructions, consisting of the preposition di (

Contrastive possession (109):

(109)

Afro-Bolivian Spanish Standard Spanish

Di mí Mío/Míos Mía/Mías

iPERSONpossessor [1 st ] iPERSONpossessor [1 st ]

i NUMBERpossessor [singular] iNUMBERpossessor [singular]

uNUMBERpossessum []  singular/plural

uGENDER possessum [] masculine/femenine

Di ti/ Di oté Tuyo/Tuyos Tuya/Tuyas

iPERSONpossessor [2 nd ] iPERSONpossessor [2 nd ]

iNUMBERpossessor [singular] iNUMBERpossessor [singular]

uNUMBERpossessum []  singular/plural

uGENDER possessum [] masculine/femenine

192 Di ele Suyo/Suyos Suya/Suyas iPERSONpossessor [2 nd ] iPERSONpossessor [3 rd ] iNUMBERpossessor [singular] iNUMBERpossessor [singular]

uNUMBERpossessum []  singular/plural

uGENDER possessum [] masculine/feminine

Di nojotro Nuestro/Nuestros Nuestra/Nuestras= iPERSONpossessor [1 st ] iPERSONpossessor [1 st ] iNUMBERpossessor [plural] iNUMBERpossessor [plural]

uNUMBERpossessum []  singular/plural

uGENDER possessum [] masculine/femenine

Di otene Suyo/Suyos Suya/Suyas iPERSONpossessor [2 nd ] iPERSONpossessor [2 nd ]

i NUMBERpossessor [plural] iNUMBERpossessor [plural]

uNUMBERpossessum []  singular/plural

uGENDER possessum [] masculine/femenine

Di eyu Suyo/Suyos Suya/Suyas iPERSONpossessor [3 rd ] iPERSONpossessor [3 rd ] iNUMBERpossessor [plural] iNUMBERpossessor [plura]

uNUMBERpossessum []  singular/plural

uGENDER possessum [] masculine/femenine

193 Some examples of these constructions are presented in (110).

(110)

ABS: La casa di nojotro la hizo yo

the-F-Sg house-F-Sg of we it-F-Sg make-pres I

stSp: La casa nuestra la hice yo

the-F-Sg house-F-Sg ours-F-Sg it-F-Sg make-pres I

‘I built our house’

ABS: La cerveza era di mí, Johnson no tenía

the-F-Sg beer-F-Sg be-past of me Johnson no have-past

stSp: La cerveza era mía, Johnson no tenía

the-F-Sg beer-F-Sg be-past mine-F-Sg Johnson no have-past

‘The beer was mine, Johnson did not have any’

6.3.8. Other Adjectives and Quantifiers

In stSp, with the exception of possessive adjectives, the rest of the adjectives do not come from the lexicon with any valued phi-feature. The same applies to quantifiers.

In fact, all these grammatical elements enter the derivation with uninterpretable, unvalued gender and number features. In stSp, the application of the operation Agree with the valued features of the noun results in surface morphological agreement between the two elements. On the other hand, ABS adjectives and quantifiers are not specified for such features, so they do not enter into an agreement relation with the number value of Num and the gender value of N. For this reason, the overt result is always the default one

194 (singular and masculine). In addition, both in stSp as well as in ABS, quantifiers come from the lexicon with a valued referential feature.

(111)

Afro-Bolivian Spanish Standard Spanish

Adjectives/Quantifiers Adjectives/Quantifiers

No-NUMBER [] uNUMBER []  singular/plural

No-GENDER [] uGENDER []  masculine/feminine

(112)

ABS: Algún casa rojo

Some-M-Sg house-F-Sg red-M-Sg

stSp: Algunas casas rojas

Some-F-Pl house-F-Pl red-F-Pl

‘Some red houses’

ABS: Mucho chica simpático

Much-M-Sg girl-F-Sg nice-M-Sg

stSp: Muchas chicas simpáticas

much-F-Pl girl-F-Pl nice-F-Pl

‘Many nice girls’

ABS: Todo lu fiesta divertido

All-M-Sg the-M-Pl party-F-Sg funny-M-Sg

195 stSp: Todas las fiestas divertidas

All-F-Pl the-F-Pl party-F-Pl funny-F-Pl

‘All the funny parties’

6.3.9. Indefinite articles

In traditional ABS the indefinite articles are only un and unos ; differently from stSp, in ABS indefinite articles do not carry gender feature specification. In both languages they are specified for interpretable and unvalued person features. This specification always probes a 3 rd value from N. Examples (113) and (114) are illustrative.

(113)

Afro-Bolivian Spanish Standard Spanish

Un/Unos Un/Unos/Una/Unas

uNUMBER [singular/plural] uNUMBER []  singular/plural

No-GENDER uGENDER []  masculine/feminine

uPERSON []  3 rd uPERSON []  3 rd

(114)

ABS: Un gato

a-M-Sg cat-M-Sg

stSp: Un gato

a-M-Sg cat-M-Sg

‘A male cat’

196 ABS: Unos gato

a-M-Pl cat-M-Sg

stSp: Unos gatos

a-M-Pl cat-M-Pl

‘Some male cats’

ABS: Un gata

a-M-Sg cat-F-Sg

stSp: Una gata

a-F-Sg cat-F-Sg

‘A female cat’

ABS: Unos gata

a-M-Pl cat-F-Sg

stSp: Unas gatas

a-F-Pl cat-F-Pl

‘Some female cats’

6.3.10. Definite articles

Definite articles in ABS are el and la in the singular and lu in the plural. In stSp, on the other hand, definite articles come in four forms el, la, los, and las . In both languages these elements are specified for unvalued gender, number and person features

197 and are base-generated in D °. During the derivation they enter into an agreement relation with the gender and person value of N and the number value of Num.

(115)

Afro-Bolivian Spanish Standard Spanish

El/ Lu La/Lu El/ Los La/Las

uNUMBER[]  singular/plural uNUMBER []  singular/plural

uGENDER []  masculine/feminine uGENDER []  masculine/feminine

iPERSON []  3 rd uPERSON []  3 rd

(116)

ABS: El amigo boliviano

the-M-Sg friend-M-Sg Bolivian-M-Sg

stSp: El amigo boliviano

the-M-Sg friend-M-Sg Bolivian-M-Sg

‘The Bolivian male friend’

ABS: La amiga boliviano

the-F-Sg friend-F-Sg Bolivian-M-Sg

stSp: La amiga boliviana

the-F-Sg friend-F-Sg Bolivian-F-Sg

‘The Bolivian female friend’

198 ABS: Lu amigo boliviano

the-M-Pl friend-M-Sg Bolivian-M-Sg

stSp: Los amigos bolivianos

the-M-Pl friend-M-Pl Bolivian-M-Pl

‘The Bolivian male friends’

ABS: Lu amiga boliviano

the-M-Pl friend-F-Sg Bolivian-M-Sg

stSp: Las amigas bolivianas

the-F-Pl friend-F-Pl Bolivian-F-Pl

‘The Bolivian female friends’

6.3.11. Demonstratives

Lipski (2008b) reports the existence of only two demonstratives in traditional

ABS: a singular eje [‘e-he] < ese (in Mururata and Chijchipa; pronounced ese in most

other Afro-Bolivian sub-dialects), and a plural one ese lu /eje lu . Spanish este , estos , esta , estas , aquel , aquella , aquellos , aquellas would not be used in the traditional dialect.

The author does not indicate whether ABS speakers lacking these demonstratives recsort to other strategies to convey deixis. For example, in French, where demonstratives like ‘this’ and ‘that’ are not present, other constructions are available, consisting of a demonstrative + noun + locative adverb (117).

199 (117)

French: Ce pomme-ci

The apple-here

‘This apple’

French: Ce pomme-là

The apple-there

‘That apple’

Lipski does not provide data of this kind for the ABS determiner system. The results of my interviews do not suggest the presence of such constructions either.

However, in partial contrast with Lipski’s findings, my eldest informants (mainly from the Tocaña community, not far from Mururata and Chijchipa) indicated that, at least in their dialect, both ese and aquel exist, and that their plural counterparts are esos and aqueyos. Therefore, based on my informants’ intuitions, ABS demonstratives seem to be specified for deictic distinction. This differentiation however is just a binary one ( ese

‘this’ vs. aquel ‘that’), not a ternary one as in stSp ( este ‘this’ vs. ese ‘that’ vs. aquel ‘that over there’). In both languages demonstratives carry referential features.

As far as phi-features are concerned, ABS determiners come from the lexicon

with an unvalued uninterpretable number feature and an interpretable unvalued person

one, while, in contrast with stSp, they lack gender specification.

200 (118)

Afro-Bolivian Spanish Standard Spanish

Este/Estos/Esta/Estas

Ese/Esos Ese/Esos/Esa/Esas

Aquel/Aqueyos Aquel/Aquellos/Aquella/Aquellas

uNUMBER []  singular/plural uNUMBER []  singular/plural

No-GENDER uGENDER []  masculine/feminine

uPERSON []  3 rd uPERSON []  3 rd

6.3.9.12. Summary

As we have noticed, the distribution of phi-specifications between the two languages is significantly divergent. While there is no distinction in person feature specification between ABS and stSp DP elements, a clear contrast can be found when number and gender specifications are considered.

Summarizing, differently stSp, ABS pronouns, adjectives, quantifiers, indefinite articles and demonstratives are not specified for gender features; in addition, the ABS plural definite article ( lu ) is invariant and does not present different phonological realizations depending on the gender feature of N, in contrast with masculine los and feminine las for stSp. Moreover stSp adjectives, quantifiers and nouns inflect for number while their ABS counterparts do not. The number value is introduced into the derivation by Num in ABS and by N in stSp. Therefore, as opposed to stSp, in ABS only singular definite articles show overt gender agreement (119).

201 (119)

ABS: Todo la comida delicioso

all-M-Sg the-F-Sg food-F-Sg delicious-M-Sg

stSp: Toda la comida deliciosa

all-F-Sg the-F-Sg food-F-Sg delicious-F-Sg

‘All the delicious food’

ABS: Todo lu comida delicioso

all-M-Sg the-M-Sg food-F-Sg delicious-M-Sg

stSp: Todas las comidas deliciosas

all-F-Pl the-F-Pl food-F-Pl delicious-F-Pl

‘All the delicious food’

As far as number features are concerned, the lack of overt plural marking on nouns, quantifiers and adjectives resulted in a variety of strategies employed in ABS to indicate plurality. One of the ways to mark number in ABS is via possessives (a), demonstratives (b), articles (c), weak (d) and strong (e) quantifiers and numerals (f), while the nominal stem, as a rule, remains bare, so that plural marking is non-redundant.

A plural reading may also obtain with bare nouns (g), when preceded by no overt D element. More details on constructions like (g) will be provided in chapter 7.

202 (120)

ABS: Mis buen amigo mayó

My-Pl good-M-Sg friend-M-Sg old-M-Sg.

stSp: Mis buenos amigos mayores

My-Pl good-M-Pl friend-M-Pl old-M-Pl

‘My old good friends’

ABS: Ejes buen amigo mayó

This-M-Pl good-M-Sg friend-M-Sg old-M-Sg

stSp: Esos buenos amigos mayores

This-M-Pl good-M-Pl friend-M-Pl old-M-Pl

‘These old good friends’

ABS: Lu buen amigo mayó

The-M-Pl good-M-Sg friend-M-Sg old-M-Sg

stSp: Los buenos amigos mayores

The-M-Pl good-M-Pl friend-M-Pl old-M-Pl

‘The old good friends’

ABS: Mucho buen amigo mayó

Much-M-Pl good-M-Sg friend-M-Sg old-M-Sg

203 StSp: Muchos buenos amigos mayores

Much-M-Pl good-M-Pl friend-M-P old-M-Pl

‘Many old good friends’

ABS: Todo lu buen amigo mayó

All-M-Sg the-Pl good-M-Sg friend-M-Sg old-M-Sg

stSp: Todos los buenos amigos mayores

All-M-Pl the-M-Pl good-M-Pl friend-M-Pl old-M-Pl

‘All the old good friends’

ABS: Cuatro buen amigo mayó

Four good-M-Sg friend-M-Sg old-M-Sg

stSp: Cuatro buenos amigos mayores

Four good-M-Pl friend-M-Pl old-M-Pl

‘Four old friends’

ABS: Amigo mayó

friend-M-S gold-M-Sg

stSp: * Amigos mayores 30

friend-M-Pl old-M-Pl

‘Old friends’

30 Note that in stSp nouns in subject position cannot occur without an overt D. 204 6.4. Microparametric research and phi-feature variability

Now that we have benefited from the preceding overview of the phi-feature distribution across the DP in traditional ABS speech, it is worth discussing the fact that not all the speakers who participated in my research spoke exactly the same variety.

Several of them, in fact, had different grammatical intuitions.

As we will see in this section, due to the recent contact between ABS and stSp, many of the people interviewed show a distribution of features and values which appear to be in between the traditional ABS and the stSp extremes. As the specification and the values of person are identical for the two languages, no variation of this kind can be encountered for this feature. For example, nobody would use a third person subject pronoun instead of a second person one; on the other hand, in my oral corpus, I encounter alternation between the traditional ABS second person oté and its stSp counterpart tú .

Such alternation, however, is not relevant to the study of phi-feature variability.

Nevertheless, an interesting phenomenon, relevant to variable phi-specification, consists of the variability found for gender and number features. In fact, recall that certain elements in traditional ABS have no specification for these features, while they have it in stSp.

The methodology employed to collect the data is the one described in chapter 2, section 2.4. In fact, both grammaticality judgments and sociolinguistic interviews were considered to build a more robust and fine-grained generalization.

205 6.4.1. The grammar of gender and number agreement

With respect to what pertains to DP gender-agreement operations, grammaticality judgments and oral questionnaires indicated the presence of several possible patterns. As far as subject pronouns are concerned, several speakers seemed to be aware of the fact that some people use ele when referring both to a man or a woman, however all my informants claimed to distinguish el from ella and ele was never encountered in the oral corpus. Different informants indicated that in the most traditional ABS variety gender agreement occurs only on singular definite articles, while the rest of the DP elements show default-masculine concord (121). These findings are in line with the data presented by Lipski (2008b) for the most traditional dialect. However, examples in (121) must not be considered as an observed pattern but rather as a reported one; in fact, none of my informants recognized (121) as a product of their own grammar; nevertheless, the Afro-

Bolivians interviewed indicated that some people in their community use these more traditional constructions.

(121)

a. Todo la comida delcioso

all-M-Sg the-F-Sg food-F-Sg delicious-M-Sg

‘All the delicious food’

b. Todo lu comida delicioso

all-M-Sg the-M-Pl food-F-Sg delicious-M-Sg

‘All the delicious foods’

206 c. Este/ese comida delicioso

This/that-M-Sg food-F-Sg delicious-M-Sg

‘This/that delicious food’

d. Mucho/un comida delicioso

Much/a-M-Sg food-F-Sg delicious-M-Sg

‘Much/a delicious food’

Some subjects presented the grammatical intuitions exemplified in (122). They showed gender agreement on plural and singular definite articles and demonstratives

(122), but not on other categories.

(122)

a. Todo la comida delcioso

all-M-Sg the-F-Sg food-F-Sg delicious-M-Sg

‘All the delicious food’

b. Todo las comida delicioso

all-M-Sg the-F-Pl food-F-Sg delicious-M-Sg

‘All the delicious foods’

c. Esta/esa comida delicioso

This/that-F-Sg food-F-Sg delicious-M-Sg

‘This/that delicious food’

207 A third group of informants used agreement on plural and singular definite

articles, demonstratives, pre-nominal adjectives, and also on weak quantifiers 31 (123):

(123)

Mucha/ una comida delicioso

Much/a-F-Sg food-F-Sg delicious-M-Sg

‘Much/a delicious food’

The last group claimed to use gender agreement for all the elements, including strong quantifiers and post-nominal adjectives (124):

(124)

Toda la comida deliciosa

all-F-Sg the-F-Sg food-F-Sg delicious-F-Sg

‘All that delicious food’

This would intuitively lead us to argue in favor of four different grammars; however, a closer look at the empirical data from the oral interviews complicates the picture. In fact, it was common for somebody to claim to speak a certain grammar and then be found using patterns belonging to another. Sometimes, speakers would freely alternate between forms in within the same sentence, like the following informant (125), whose grammaticality judgments indicated a grammar of type (124):

31 Since Milsark’s dissertation (1974) natural-language quantifiers have been classified into two different groups called weak and strong quantifiers. Weak quantifiers (e.g. some , many , etc.) can occur in existential (there ) sentences. On the other hand, strong quantifiers (e.g. all , most , etc.) can not. Milsark suggests that this phenomenon would be due to the fact that strong quantifiers refer to subsets of previously established sets, while weak quantifiers establish such sets and for this reason can occur in existential constructions. 208 (125)

Todo la comunidad participaba;

all-M-Sg the-F-Sg community-F-Sg participate-Past

mucha gente venía, mucho gente

much-F-Sg. people-F-Sg come-Past much-M-Sg people-F-Sg

venía desde lejos. Todas, toditas las

come-past from far. All-F-Pl all-F-Pl the-F-Pl

personas se reunían. Muy bonito la

people-F-Sg reflex meet-Past. very nice-M-Sg the-F-Sg

fiesta era...

party-F-Sg be-past

“All the community used to participate, many people used to come, many people

used to come from far away. All, all the people gathered. The party was very

nice...”

Speakers like this present gender agreement on strong quantifiers in 50-60% of instances (cf. Sessarego and Gutiérrez-Rexach 2009a,b), thus indicating that cases of agreement mismatches are very common and should not be regarded as just E(xternal)-

209 language phenomena. A closer analysis of the data, also seriously undermines hypotheses claiming the existence of several parallel grammars, accessible to the speaker at the same time (Kroch 1989; Henry 2005). Henry (2005), for example, argues for the existence of a

‘formal’ and an ‘informal’ grammar in Belfast English. In formal situations speakers would employ a grammar for which the forms there is/ there are require agreement,

while for informal situations a different grammar would allow the verb not to agree with

the subject, thus resulting in the default singular construction. See also den Dikken et. al.

(2006) for some reflections on individual data and abstract grammar.

As far as Afro-Bolivian Spanish gender agreement is concerned, such a claim is

not likely. Within a one-hour interview period, the switches between the four potentially

parallel grammars could be so many that no formal/informal style alternation might serve

as a reasonable justification.

On the other hand, as far as grammaticality judgments for number features are

concerned, less variability was reported; nevertheless, much alternation could still be

found when analyzing the transcripts from the naturalistic interviews. In fact, all

informants, even the oldest ones, seem to be aware of the fact that in traditional ABS

number is marked non-redundantly, while ‘at school’ or more generally in stSp, it is

marked on all the DP elements. No one, for example, claimed to inflect nouns while

keeping post-nominal adjectives bare or vice versa. Nevertheless, besides the clear

grammatical intuitions distinguishing ABS from stSp number marking, my corpus reveals

several additional patterns, which are neither limited to the traditional marking of

plurality on one single element in the DP, nor to the marking of all DP elements as in

stSp.

210 Sometimes, speakers would freely alternate between forms within the same sentence, like the following informant:

(126)

Lu pequeño se ha muerto, mis hijas

The-M-Pl small-M-Sg reflex have-pres died-Sg, my-Pl daughter-F-Pl

jovena también se ha muerto. Mis hijo se ha muerto.

youngF-Pl too reflex have died my-Pl son-M-Sg reflex have died-Sg

uno vivía aquí, uno a la Argentina.

one-M-Sg live-past here one-M-Sg to the Argentina

‘The kids died, my daughters died too. My sons died. One lived here, one in

Argentina.”

Speakers like the one reported in (126) show number agreement on nouns in 40% of the cases, thus indicating that also for this phenomenon many agreement mismatches can be encountered and should not just be regarded as performance phenomena.

While in the case of gender variability phonological factors cannot be invoked, some skeptical readers may suggest that number variability could be just the result of phonological processes of –s elision, external to syntax. In the next section, I will briefly show that several reasons seem to indicate that we are not just dealing with phonology.

211 6.4.2. Syntax matters: quantitative evidence for number variability

Following the model in Poplack (1980) on Puerto Rican Spanish (PRS), my data were analyzed using GoldVarbX program (Sankoff, et. al. 2005), which calculates probabilities for the application of a given rule. In order to obtain a reasonable number of tokens, I coded data extracted from four one-hour conversations with the eldest informants (cf. Sessarego and Delicado-Cantero 2009; Delicado-Cantero and Sessarego

2009). Here, as in Poplack (1980) the different phonological realizations of final –s encountered [s, z, h, ħ] were considered together and contrasted with the null realization

[Ø] to investigate the constraints on plural –s realization. I must acknowledge, though,

that the tokens presenting [h, ħ] realizations (only 21 out of 944 tokens) were scarce, indicating that the phenomenon is more constrained by syntactic factors than just phonological ones.

I differentiated between inflectional and morphemic instances of final –s. In fact, in a dialect like ABS in which final -s might be weakened and deleted due to phonological processes (Lipski 2008b:65), it is crucial to understand to which extent phonology is contributing to –s omission on the DP elements.

According to the Functional Hypothesis, developed from Kiparsky’s distinctness condition (Kiparsky 1972:195), phonological elements loaded with a morphological plural value (e.g. casa -s ‘house-s’) should be more resistant to deletion than elements free from this charge, like cases of final -s in monomorphemic words (e.g. mes ‘mounth’). My results indicate that the rate of –s omission is higher for inflectional than for morphemic instances of –s (see Table 2), thus confirming that phonological factors are playing only a limited role. Such a claim is also clearly supported by the occurrence of plural forms like

212 lu dictador [los dictador es] ‘the dictators’, lu varón [los varon es] ‘the males’ without the

–e of the Spanish plural morpheme.

Table 10. Percentage of /s/ deletion by grammatical status in ABS

Grammatical status Percentage of deletion Inflection 64% (N=532) Morpheme 30% (N=412)

After excluding from the analysis all the morphemic tokens, I ran, as in Poplack’s work, a variable analysis of the following factor groups: Grammatical Category,

Following Phonological Segment and Following Stress (Table 3).

Table 11. Contribution of Grammatical Category, Following Phonological Segment

and Following Stress to the Deletion of Plural (s) in ABS (Log likelihood = -287.688,

Significance = 0.007, N=532).

Grammatical category Following phonological segment Following stress Adjective .65 (N=47) Vowel [.53](N=121) Weak [.51](N=230) Noun .55 (N=313) Pause [.49] (N=61) Heavy [.50](N=302) Determiner .38 (N=172) Consonant [.47](N=350)

Grammatical Category was the most significant factor group, with a Range of .27.

Adjectives seem to be the grammatical category showing the highest rate of deletion, followed by nouns and determiner 32 . As for Following Phonological Segment and Stress, neither stress nor phonological context was selected as significant factors. Moreover,

32 In this case, following Poplack (1980), I included strong and weak quantifiers under the label ‘determiners’. 213 while a weak following stress seems to slightly favor deletion, following phonological segments present a different order of factors from the one reported by Poplack, with vowels favoring rather than disfavoring the phenomenon. These findings are a further evidence of the fact that, differently from other Spanish dialects, in ABS –s plural marking has more to do with syntactic rules than phonological ones.

For this reason, it can be said that number and gender variability in ABS should not be seen as just performance phenomena, sporadic E-language errors external to the syntax; rather, I would like to suggest that, within the minimalist framework, we can find a logic formal analysis to account for these cases in terms of I-language. In the next section, I will provide Lipski’s account for such variability. In section 6.6. a novel model will be suggested.

6.5. Previous analysis

Based on the reduced inflectional morphology encountered in ABS, this dialect has been classified by Lipski (2006:9) as a case of “DP impoverished agreement”. While in standard spec-head agreement processes, a certain feature should percolate to all elements m-commanded (Chomsky 1986) by the head, in cases of “impoverished agreement”, concord can be limited to some of them. Within the GB framework, DP features are claimed to percolate up from the noun to the determiner (Grimshaw 1991;

1997). Thus, in this agreeing relation, the relevant feature is transmitted from the head to all the elements m-commanded by it.

Lipski (2006a) adopts this model to explain ABS gender agreement processes.

Looking at data from several speakers of different ages and levels of education presenting

214 variable gender-agreement configurations, he noticed that no case of post-nominal gender/number concord is found unless pre-nominal elements agree, as shown in

(127,128).

(127)

ABS:

a. una curva ancha

a-F-Sg curve-F-Sg large-F-Sg

b. una curva ancho

a-F-Sg curve-F-Sg large-M-Sg

c. un curva ancho

a-M-Sg curve-F-Sg large-M-Sg

d. * un curva ancha

a-M-Sg curve-F-Sg large-F-Sg

stSp: una curva ancha

a-F-Sg curve-F-Sg large-F-Sg

‘A large curve’

(128)

ABS:

a. lu(s) guagua joven

the-M-Pl kid-F-Sg young-M-Sg

b. lu(s) guaguas joven

the-M-Pl kid-F-Pl young-M-Sg

215 c. lu(s) guaguas jóvenes

the-M-Pl kid-F-Pl young-M-Pl

d. *El guaguas jóvenes

the-M-Sg kid-F-Pl young-M-Pl

stSp: Las guaguas jóvenes

the-F-Pl kid-F-Pl young-F-Pl

‘The young kids’

Lipski concludes that a generational change is taking place, where ABS is approximating more and more to stSp. My data, when analyzed cross generationally, are generally in line with those reported by Lipski (Sessarego & Gutiérrez-Rexach

2009a,b) 33 . However, an important difference can be noticed in the speech of several informants. Many speakers present complete lack of gender/number agreement on quantifiers, as shown by (129) and also by examples (121-124):

(129)

ABS: Todo las casa bonito

all-M-Sg the-F-Pl thing-F-Sg nice-M-Sg

stSp: Todas las casas bonitas

all-F-Pl the-F-Pl thing-F-Pl nice-F-Pl

‘All the nice houses’

33 More considerations on how the generational factor is affecting morphological evolution in ABS can also be encountered in Sessarego (2009a,b), Sessarego and Delicado-Cantero (2009) and Delicado-Cantero and Sessarego (2009). 216 ABS: Todo la chica de Tocaña

all-M-Sg the-F-Sg girl-F-Sg of Tocaña

stSp: Todas las chicas de Tocaña

all-F-Pl the-F-Pl girl-F-Pl of Tocaña

‘All girls from Tocana’

ABS: Todo la comunidad

all-M-Sg the-F-Sg community-F-Sg

stSp: Toda la comunidad

all-F-Sg the-F-Sg community-F-Sg

‘All the community’

ABS: Mucho viajero viene

much-M-Sg traveler-M-Sg come-pres

stSp: Muchos viajeros vienen

much-M-Pl traveler-M-Pl come-pres

‘Many travelers come’

ABS: Había harto caballo

have-past much-M-Sg hourse-M-Sg stSp: Había hartos caballos

have-past much-M-Pl hourse-M-Pl

‘There were many horses’

217 ABS: Todo lu chica de ciudá

all-M-Sg the-M-Pl girl-F-Sg of city-F-Sg

stSp: Todas las chicas de ciudad

all-F-Pl the-F-Pl girl-F-Pl of city-F-Sg

‘All the girls from the city’

Cases like these seem to violate the pre-nominal to post-nominal percolation order (Grimshaw 1991; 1997), unless we postulate that quantifiers are elements external to the DP, and then we argue, independently, in favor of a different mechanism for the checking of phi-features in languages where they agree in gender and number with N, like stSp. Additionally, the feature-percolation account of gender agreement runs into problems when compared with data from other Romance varieties in which post-nominal adjectives may agree with N and disagree with D (cf. Pomino and Stark 2008 for Fassano

Ladin). Alternatively, one might propose a system with one or more agreement projections inside DP and with the relevant displacement operations applied to agreeing elements so that they enter into a spec-head relation (Sportiche 2002; Koopman 1997).

This type of approach would be problematic too in that the number of internal agreement projections required would not be limited, and would most likely be established on an ad hoc basis. It also runs against recent minimalist ideas supporting the elimination of

Agreement projections (Chomsky 2002).

218 6.6. A Minimalist Analysis

The study of language variation seems at first to be at odds with formal linguistic analyses that strive to hypothesize principles and generalizations based on Universal

Grammar. Nevertheless, in the Principles and Parameters paradigm, several attempts were made to capture dialectal and inter-speaker variation, exploring the notion of parametric variation and, in the last decade, fine-tuning it to include so-called “micro parameters” (Benincà 1989; Kayne 1996; etc). Language internal variation, on the other hand, has long been excluded from formal linguistic analyses. However, during the last few years, several scholars have taken into account variation beyond the usual parametric inter-language domain (Adger & Smith 2005; Adger 2006). Individual variation becomes the core of linguistic research, bringing previously disregarded phenomena - considered as belonging to E-language - to the fore (Adger & Trousdale 2007).

Recent minimalist research entailed a derivational approach that was inconsistent with parameter-based accounts. Thus, an extension of the parameter/micro-parameter idea to intra-speaker variation seems to not be a straightforward matter (Adger & Smith

2005). In the most recent formulations of the Minimalist Program (Chomsky 2001; 2002;

2006), syntactic derivations are viewed as strictly dependent on feature valuation and checking. The distinction between interpretable and non-interpretable features, though controversial, has proven very useful. Several features have an interpretation at LF, thus they are semantically-interpretable features. Other features, however, lack such semantic import and are present to trigger the necessary merger or agreement operations during the derivation. Said uninterpretable features have to be matched via Agree and are finally deleted before Spell-Out.

219 Given these theoretical assumptions, the Minimalist Program seems to show enough promise to account for important aspects of variation because it admits several phonological outputs for a given semantic interpretation. Adger and Smith (2005) argue for characterizing syntactic variation in terms of (un)interpretable features. Certain uninterpretable features may be present in one category but absent in another. Since they are uninterpretable, they would have no semantic repercussion, thus being equally legitimate for a convergent derivation. Therefore, variation is reduced to the specification of the uninterpretable features in a derivation (Adger & Smith 2005: 161). As expected, syntax per se remains invariable or “perfect” (Brody 1997), given that variation is located only in the lexical component. Variation will occur when one item or other enters the numeration and takes part in a syntactic derivation. Several (social) factors may affect the outcome: ease of lexical access (probably linked to frequency of use), speaker-hearer relationships, social identity, etc. (Adger & Smith 2005:164).

Such (variable) underspecification for gender and number is what we can observe when analyzing the ABS DP. In fact, within the minimalist framework, agreement is conceived as the result of valuation processes which do not necessarily require movement but just a configurational feature checking mechanism (Picallo 2008). DP agreement, in fact, involves the transmission or sharing of features across lexical items (adjectives, nouns) and functional elements (Num, determiners, quantifiers).

Neither demonstratives nor adjectives come from the lexicon with a value for number and gender. Their features are lexically unvalued (Chomsky 2001), and get

220 valued as a consequence of a syntactic process of agreement with the gender and number features of N and Num 34 .

An element drawn from the lexicon with a uninterpretable valued feature will be

specified as u(ninterpetable)F(eature)[val]; a feature of the same kind that has not participated in Agree and is not already valued is annotated by an empty pair of brackets: uF[ ] and, after the Agree operation takes place, it turns into uF[val] . On the other hand, an element coming from the lexicon without a specification for such a feature, will be annotated (where relevant) as no-F[ ], and it will not be able to act as a probe for Agree

operations of that particular kind.

(130)

uF[val]...... uF[val].…uF[val]…...uF[val]… …no-F[ ]

Therefore, if we postulate that an uninterpretable instance of the feature gender

and number may be present in certain DP elements but absent in others, and that variation

is the result of lexical differences in the feature specification of certain items, it follows

that contrasts in overt syntax will be the result of differences in the computation of

varying specifications. I propose an account of the different agreement configurations

across DP strings in ABS that can be summarized in the following fashion, where +val

indicates the feature value introduced into the derivation:

34 I am assuming that in stSp both gender and number values are in N; while, in ABS, the gender value is in N but the number value is in Num. 221 (131)

[DP una [NP curva ancha]] uG[fem]……… iG[+fem]………………uG[fem]

[DP una [NP curva ancho]] uG[fem]…………iG[+fem]……………… No-G[]

[DP un [NP curva ancho]] no-G[]………… iG [+fem]……………… No-G[]

[DP a [NP curve large]]

‘A large curve’

(132)

[DP lus [Num [NP guaguas jóvenes]]] uN[pl]…………. iN[pl]…………… uN[+pl]…………………uN[pl]

[DP lus [Num [NP guaguas joven]]] uN[pl]…………. iN[pl]…………… uN[+pl]…………………No-N[]

[DP lus [Num [NP guagua joven]]] uN[pl]………iN[+pl]………… No-N[pl]…………………No-N[l]

[DP the [Num [NP kids young ]]]

‘The young kids’

222 Therefore, this minimalist approach can account for all the gender/number agreement configurations encountered in the ABS Determiner Phrase by postulating the presence/absence of gender/number specifications on the different DP components. The differences between ABS and stSp reported in (127-128) can be explained without recurring to different mechanisms of feature percolation or different movement operations involving one or more agreement heads. Rather, we just need to assume that in the speech of my informants certain DP elements often do not carry gender/number specification while they normally do in stSp.

Having clarified this point, we can now proceed to solve a different empirical issue, namely why no instances of post-nominal gender/number concord can be found on adjectives unless pre-nominal articles agree, in both Lipski’s (2006a) and my corpora.

In minimalistic terms, this phenomenon can be seen as the emergence and development of unvalued features on elements which previously were not specified for them. This is in line with current approaches to grammaticalization in the minimalist framework (Roberts & Roussou 2003; Van Gelderen 2004). The non-occurrence of strings like (133) might indicate that cross-generationally indefinite articles developed an unvalued gender/number-feature specification before post-nominal adjectives. Therefore, speakers mastering post-nominal agreement would also present concord on indefinite articles.

223 (133)

*[ DP un [NP curva ancha]

no-G[]…………………..uG[fem]…....…………… uG[fem]

[DP a [NP curve large]

‘A large curve’

Given these data and the results from the grammaticality judgments, we may hypothesize that in ABS unvalued gender/number features developed gradually: first on certain elements (e.g. articles) and only later on others (e.g. strong quantifiers, adjectives). However, even though this is the general tendency, different linguistic and social factors may affect an item’s selection, and therefore the overt syntax result (Adger and Smith 2005) 35 .

Moreover, the common lack of overt plural marking on post-nominal adjectives and the impossibility of finding constructions like (134), where As carry plural –s while

the preceding elements do not, indirectly suggest that two conditions have to be satisfied

for As to be overtly marked:

(a) Adjectives must be specified for uninterpretable unvalued features.

(b) Nouns must enter the derivation with a value for number.

35 While sociolinguistic factors are undoubtedly at play, a more fine-grained variationist study is outside the scope of this dissertation. Preliminary analyzes on the role of social factors in the development of nominal and verbal inflection have been conducted by Sessarego (2009a,b) and Sessarego & Gutiérrez-Rexach (2009a,b).

224 (134)

ABS:

a. * El guagua jóvenes

the-M-Sg kid-F-Pl young-M-Pl

b. * Lu guagua jóvenes

the-M-Pl kid-F-Sg young-M-Pl

‘The young kids’

In fact, the violation of (a) prevents post-nominal adjectives from probing for a value; while the violation of (b) eliminates the only possible goal from which they could get valued. These considerations, in addition to the ABS evidence, further back the model proposed. Moreover, recall that the same architecture proposed to account for the distinction between ABS and stSp number marking can also explain the differences encountered between popular Brazilian Portuguese and standard Brazilian Portuguese (cf.

Simioni 2007). More cross-linguistic research is definitely needed to make a broader generalization. Nevertheless, the framework and the data seem highly promising.

6.7. Conclusions

This chapter provided an overview of the distribution of phi-features across the

ABS and the stSp DPs. Traditional ABS DP elements present a much reduced number of phi-feature specifications than their stSp counterparts. Such a deficiency results in overt default singular and masculine morphological marking. Nevertheless, it must be acknowledged that recent contact with stSp has caused ABS to approximate to this more

225 prestigious Spanish dialect (Lipski 2006a,b; Sessarego & Gutiérrez-Rexach 2009a,b). For this reason, my findings do not pretend to be completely categorical. Variation is a component of human languages, and my results confirm this.

The Minimalist Program offers the theoretical tools to characterize the elements of variation in a systematic fashion, as computationally determined by differences in the specification of lexical and functional items and by restrictions on syntactic operations: more specifically, on Agree. The great improvement of the Minimalist Program over previous theoretical frameworks is its capability of accounting for alternation between forms (Adger and Smith 2005). From a theoretical perspective, this chapter sheds some light on the linguistic constraints regulating gender and number agreement in an Afro-

Hispanic vernacular approximating to a more prestigious Spanish dialect. As stated in previous chapters, this process is driven by social factors; nevertheless, it clearly obeys specific syntactic mechanisms.

226

CHAPTER 7: NOMINAL ELLIPSIS

“The true mystery of the world is the visible, not the

invisible” (Wilde 1890: The Picture of Dorian Gray,

Chapter 2, 32).

7.0. Introduction

This chapter is a survey of NP-ellipsis phenomena in stSp and ABS. The

generative literature on Spanish NP-ellipsis is particularly rich, so that different

frameworks have been designed to account for the processes responsible for their

distribution in the standard variety.

Some scholars have ascribed the possibility of nominal ellipsis to the richness of

Spanish inflectional morphology and the presence of pro in the language (Torrego 1988), some have justified their presence by recurring to post-syntactic movement operations and post-syntactic insertion processes (Ticio 2003; 2005), others have suggested nominal properties for certain prepositions and complementizers to account for cliticization phenomena (Brucart and Gràcia 1986; Brucart 1987; 1999; Torrego 1988). A more recent account (Braver 2009), partially based on semantic principles (Eguren 2010), has proposed a new clicitization algorithm, which accounts for the majority of the data, but

227 does not consider important thematic correspondences between the structure of the pronounced and the elided noun.

Even though each of these proposals provides a different explanation for NP- ellipsis, all of them have something in common, in that they attribute “special” features to the stSp preposition de ‘of’ to account for the fact that differently from other

prepositions, it can follow elided nouns without compromising the grammaticality of the

enunciate (135).

(135) stSp

a. El papel blanco y el [e] de manchas negras

the paper white and the [e] of spots black

b. *El papel blanco y el [e] con manchas negras

the paper white and the [e] with spots black

‘The white paper and the black-spotted one’

Interestingly, a comparison of the ABS and the stSp data will indicate that even if

ABS is not inflectionally rich and does not have pro , it allows all the elliptical configurations encountered in stSp. Moreover, in this Afro-Hispanic dialect, nouns can often be elided also in contexts for which an ungrammatical construction would obtain in the standard language, namely when the elided noun is followed by cun ‘with’.

(136) ABS

a. El papel blanco y el [e] di mancha negra

the paper white and the [e] of spots black

228 b. El papel blanco y el [e] cun mancha negra

the paper white and the [e] with spots black

‘The white paper and the black-spotted one’

By examining the differences and similarities between stSp and ABS nominal ellipsis, this chapter will try to shed light on the phenomenon by proposing a crosslinguistic generalization able to account for the data in the two varieties.

7.1. The Data

As far as standard Spanish is concerned, Depiante and Masullo (2001) and later

Saab (2004) have pointed out that in cases of nominal ellipsis it is possible to have sloppy identity for number, but not for gender:

(137) stSp

a. El gato negro y los [e] blancos

the-M-Sg cat-M-Sg black-M-Sg and the-M-Pl [e] white-M-PL

‘The black cat and the white ones’

b. Los gatos negros y el [e] blanco

the-M-Pl cat-M-Pl black-M-Pl and the-M-Sg [e] white-M-Sg

‘The black cats and the white one’

c. * El gato negro y la [e] blanca

the-M-Sg cat-M-Sg black-M-Sg and the-F-Sg [e] white-F-Pl

‘The black cat and the white one’

229 d. * La gata negra y el [e] blanco

the-F-Sg cat-F-Sg black-F-Sg and the-M-Sg [e] white-M-Sg

‘The black cat and the white one’

Saab (2004) also notes that the constructions get even more ungrammatical when

a switch in gender concerns nouns that are derivational couples (138 a,b,c,d) or

suppletive couples (138 e,f,g,h)36 .

(138) stSp

a. **El manzano de Juan y la [e] de María

the-M-Sg apple-tree-M-Sg of Juan and the-F-Sg [e] of María

‘Juan’s apple tree and Mary’s apple’

b. **La manzana de María y el [e] de Juan

the-F-Sg apple-F-Sg of María and the-M-Sg [e] of Juan

‘Juan’s apple and Mary’s apple tree’

c. **El duque de York y la [e] de Holanda

the-M-Sg duke-M-Sg of York and the-F-Sg [e] of Holland

‘The duke of York and the duchess of Holland’

d. **La duquesa de Holanda y el [e] de York

the-F-Sg duchess-F-Sg of Holland and the-M-SG [e] of York

‘The duchess of Holland and the duke of York’

36 I follow Saab (2004) in adopting the ‘**’ marking to indicate that a construction is highly ungrammatical. 230 e. **El padre de Juan y la [e] de María

the-M-Sg father-M-Sg of Juan and the-F-Sg [e] of María

‘Juan’s father and María’s mother’

f. **La madre de Juan y el [e] de María

the-F-Sg mother-F-Sg of Juan and the-M-Sg [e] of María

‘Juan’s mother and María’s father’

g. **El toro de Juan y la [e] de María

the-M-Sg bull-M-Sg of Juan and the-F-Sg [e] of María

‘Juan’s bull and María’s cow’

h. **La vaca de María y el [e] de Juan

the-F-Sg cow-F-Sg of María and the-M-Sg [e] of Juan

‘María’s cow and Juan’s bull’

The same exact configurations -with the due differences in phi- specification reported in chapter 6- are encountered in traditional ABS:

(139) ABS

a. El gato negro y los [e] blanco

the-M-Sg cat-M-Sg black-M-Sg and the-M-Pl [e] white-M-Sg

‘The black cat and the white ones’

b. Lu gato negro y el [e] blanco

the-M-Pl cat-M-Pl black-M-Pl and the-M-Sg [e] white-M-Sg

‘The black cats and the white one’

c. * El gato negro y la [e] blanco

231 the-M-Sg cat-M-Sg black-M-Sg and the-F-Sg [e] white-M-Pl

‘The black cat and the white one’ d. * La gata negro y el [e] blanco

the-F-Sg cat-F-Sg black-M-Sg and the-M-Sg [e] white-M-Sg

‘The black cat and the white one’

(140) ABS a. **El manzano di Juan y la [e] di María

the-M-Sg apple-tree-M-Sg of Juan and the-F-Sg [e] of María

‘Juan’s apple tree and Mary’s apple’ b. **La manzana di María y el [e] di Juan

the-F-Sg apple-F-Sg of María and the-M-Sg [e] of Juan

‘Juan’s apple and Mary’s apple tree’ c. **El duque di York y la [e] di Holanda

the-M-Sg duke-M-Sg of York and the-F-Sg [e] of Holland

‘The duke of York and the duchess of Holland’ d. **La duquesa di Holanda y el [e] di York

the-F-Sg duchess-F-Sg of Holland and the-M-Sg [e] of York

‘The duchess of Holland and the duke of York’

232 e. **El padre di Juan y la [e] di María

the-M-Sg father-M-Sg of Juan and the-F-Sg [e] of María

‘Juan’s father and María’s mother’

f. **La madre di Juan y el [e] di María

the-F-Sg mother-F-Sg of Juan and the-M-Sg [e] of María

‘Juan’s mother and María’s father’

g. **El toro di Juan y la [e] di María

the-M-Sg bull-M-Sg of Juan and the-F-Sg [e] of María

‘Juan’s bull and María’s cow’

h. **La vaca di María y el [e] di Juan

the-F-Sg cow-F-Sg of María and the-M-Sg [e] of Juan

‘María’s cow and Juan’s bull’

Examples (137-140) indicate that, in ABS as well as in stSp, gender, differently from number, is a feature of the noun, which gets deleted in the process of elision; this fact strongly undermines approaches which postulate a unique projection for number and gender (e.g. Ritter 1991), as well as those that argue in favor of two separate projections

NumP and a GenP (e.g. Picallo 1991; Bernstein 1993). In fact, masculine and feminine nouns do not seem to be derivable from the same lexical entry. Data, on the other hand, back a framework in which lexical entries are clearly different in gender specifications before entering the syntactic numeration. Therefore, only NumP is a licit projection while

‘gender’ represents an intrepetable feature of N.

233 The nominal ellipsis encountered in Spanish and ABS can be further grouped under four main categories, which are schematically represented in the following structures:

(141)

(A) D + [e] + Adjectival Phrase: (142a,143a);

(B) D + [e] + Prepositional Phrase: (142b,143b);

(C) D + [e]: (143c,144c);

(D) D + [e] + Relative Clause (142d,143d);

(142) stSp

a. El chico italiano y el [e] boliviano.

The guy italian and the [e] bolivian

‘The Italian guy and the Bolivian one’

b. Muchos chicos de Italia y muchos [e] de Bolivia.

Many guys of Italy and many [e] of Bolivia

‘Many Italian guy and the many Bolivian ones’

c. Estos chicos de Italia y aquellos [e].

These guys of Italy and those [e]

‘These Italian guys and those ones’

d. El chico de Italia y el [e] que viene de Bolivia

the guy of Italy and the [e] who come of Bolivia

‘ The Italian guy and the one who comes from Bolivia’.

234 (143) ABS

a. El chico italiano y el [e] boliviano.

The guy italian and the [e] bolivian

‘The Italian guy and the Bolivian one’

b. Mucho chico di Italia y mucho [e] di Bolivia.

Many guys of Italy and many [e] of Bolivia

‘Many Italian guy and the many Bolivian ones’

c. Ejes chico di Italia y aquellos [e].

These guys of Italy and those [e]

‘These Italian guys and those ones’

d. El chico di Italia y el [e] que viene di Bolivia

the guy of Italy and the [e] who come of Bolivia

‘The Italian guy and the one who comes from Bolivia.

It is important to notice that certain restrictions apply to these constructions; in fact, when the DP is headed by a definite article, constructions like (141C) are ungrammatical. The definite article is incapable of staying by itself in elliptical constructions (144):

(144) stSp

a. El chico italiano y el [e] *(boliviano).

The guy Italian and the [e] Bolivian

‘The Italian guy and the Bolivian one’

235 b. Los chicos de Italia y los [e] *(de Bolivia).

The guys of Italy and the [e] of Bolivia

‘The Italian guys and the Bolivian ones’

(145) ABS

a. El chico italiano y el [e] *(boliviano).

The guy Italian and the [e] Bolivian

‘The Italian guy and the Bolivian one’

b. Lu chico di Italia y lu [e] *(di Bolivia).

The guys of Italy and the [e] of Bolivia

‘The Italian guys and the Bolivian ones’

Note that both de -phrases and APs are grammatical even when an adverb precedes the remnant construction (146, 147).

(146) stSp

a. El chico italiano y el [e] supuestamente italiano.

The guy Italian and the [e] supposedly Italian

‘The Italian guy and the supposedly Italian one’

b. Los chicos de Italia y los [e] supuestamente de Italia

The guys of Italy and the [e] supposedly of Italy

‘The Italian guys and the supposedly Italian ones’

236 (147) ABS

a. El chico italiano y el [e] supuestamente italiano.

The guy Italian and the [e] supposedly Italian

‘The Italian guy and the supposedly Italian one’

b. Lu chico di Italia y lu [e] supuestamente di Italia

The guys of Italy and the [e] supposedly of Italy

‘The Italian guys and the supposedly Italian ones’

Another key characteristic of elliptical constructions is the fact that in presence of a definite article the only prepositional phrases (PPs) allowed are those headed by de

(148) in stSp and by di and cun in ABS (149).

(148) stSp

a. *La plata para mí y la [e] para ti.

The money for me and the [e] for you

‘The money for me and the one for you’

b. *La bandera con rayas y la [e] con manchas.

The flag with stripes and the [e] with spots

‘The flag with stripes and the one with spots’

c. El libro de color verde y el [e] de color rojo.

The book of color green and the [e] of color red

‘The green book and red one’

237 (149) ABS

a. *La plata para mí y la [e] para ti.

The money for me and the [e] for you

‘The money for me and the one for you’

b. La bandera cun raya y la [e] cun mancha.

The flag with stripes and the [e] with spots

‘The flag with stripes and the one with spots’

c. El libro di color verde y el [e] di color rojo

The book of color green and the [e] of color red

‘The green book and red one’

De/di is a multifunctional preposition in stSp and ABS. It can introduce a great array of phrases referring to different nominal theta-roles: possessors (150a, 151a), agents

(150b, 151b), objects (150c, 151c), as well as a variety of non-thematic PPs which provides a description of the N they modify, e.g. color adjectives (150c, 151c). A restriction which applies to stSp and ABS NP ellipsis is the need for thematic role correspondence between the coordinated clauses (compare: 150a,b,c vs. 150d,e and

151a,b,c vs. 151d,e).

(150) stSp

a. El libro suyo poss y el [e] tuyo poss

The book his and the [e] your

‘His book and your book’

238 b. El libro de Cervantesagent y el [e] de Juan agent

The book of Cervantes and the [e] of Juan

‘Cervantes’ book and Juan’s one’ c. El libro de física object y el [e] de sintaxis object

The book of physics and the [e] of syntax

‘The physics book and syntax one’ d. *El libro de María agent/poss y el [e] de física theme

The book of María and the [e] of physics

‘María’ book and the physics one’ e. *El libro de fisica object y el [e] de María agent/poss

The book of physics and the [e] of María

‘The physics book and María’s one’

(151) ABS a. El libro suyo poss y el [e] tuyo poss

The book his and the [e] your

‘His book and your book’ b. El libro de Cervantes agent y el [e] di Juan agent

The book of Cervantes and the [e] of Juan

‘Cervantes’ book and Juan’s one’ c. El libro de física object y el [e] di sintaxis object

The book of physics and the [e] of syntax

‘The physics book and syntax one’

239 d. *El libro de María agent/poss y el [e] di física theme

The book of María and the [e] of physics

‘María’ book and the physics one’

e. *El libro de fisica object y el [e] di María agent/poss

The book of physics and the [e] of María

‘The physics book and María’s one’

The coordination between thematic roles is robust. It is further supported by examples like (150b) and (151b), in which Juan takes an agentive reading rather than a

possessive one. As shown, de/di has the capability of creating different kinds of thematic

PPs. Moreover, it can head a variety of descriptive PPs (152a,b,c; 153a,b,c) free from

theta-assignment.

(152) stSp

a. La manzana muy sabrosa y la [e] de poco sabor

The apple much tasty and the [e] of little taste

‘The very tasty apple and the no-so-tasty one’

b. El vestido rayado y el [e] de cuadritos

The dress with-stripes and the [e] of squares

‘The dress with stripes and the one with squares’

c. La fresa rosada y la [e] de color rojo

The stawberry pink and the [e] of color red

‘The pink strawberry and the red one’

240 (153) ABS

a. La manzana muy sabrosa y la [e] di poco sabor

The apple much tasty and the [e] of little taste

‘The very tasty apple and the no-so-tasty one’

b. El vestido rayado y el [e] di cuadritos

The dress with-stripes and the [e] of squares

‘The dress with stripes and the one with squares’

c. La fresa rosada y la [e] di color rojo

The stawberry pink and the [e] of color red

‘The pink strawberry and the red one’

Note, however, that while thematic PPs must obey the thematic coordination mentioned above, the rest of the PPs, to which a thematic role is not assigned, can occur more freely (154, 155)37 .

(154) stSp

a. Dame el libro de Juan agent/poss y el [e] de color rojo/rojo

give-me the book of Juan and the [e] of color red/red

‘Give me Juan’s book and the red one’

37 As it is well known in the literature, different arrays of theta-roles have been proposed (Belletti & Rizzi 1988; Grimshaw 1990; Speas 1990; Jackendorff 1990; etc.). Dowty (1991) suggests that probably we should better talk about prototypicalality of theta-roles, rather than enumerate potentially endless lists of them. Such an idea may well also apply in this case. However, for the sake of simplicity and conformity, I will here provide an account that considers the nominal theta-roles which have usually been treated in the literature on nominalization, namely: possessor, agent and object (Giorgi & Longobardi 1990; Valois 1991, etc). Grammaticality judgments may vary when certain APs and de-PPs constructions (potentially thematic) are introduced in the picture, e.g. source-role de Italia, italiano . However, the point to be made here is that there is a clear structural distinction between thematic-PPs/APs and non-thematic-ones. Thematic coordination between nominal arguments has also been pointed out by Brucart and Gràcia (1986:9 fn. 9); however, the authors did not develop a theory to account for it.

241 b. Dame el libro de Petrarca agent y el [e] de color rojo/rojo

give-me the book of Petrarca and the [e] of color red/red

‘Give me Petrarca’s book and the red one’

c. Dame el libro de física theme y el[e] de color rojo/rojo

give-me the book of physics and the [e] of color red/red

‘Give me the physics book and the red one’

(155) ABS

a. Dame el libro de Juan agent/poss y el [e] di color rojo/rojo

give-me the book of Juan and the [e] of color red/red

‘Give me Juan’s book and the red one’

b. Dame el libro de Petrarca agent y el [e] di color rojo/rojo

give-me the book of Petrarca and the [e] of color red/red

‘Give me Petrarca’s book and the red one’

c. Dame el libro de física theme y el[e] di color rojo/rojo

give-me the book of physics and the [e] of color red/red

‘Give me the physics book and the red one’

It must be said that not all adjectives behave in the same way; while post-nominal adjectives allow nominal ellipsis, pre-nominal ones block them (156, 157).

242 (156) stSp

a. *El verdadero asesino alto y el supuesto [e] bajo

The true killer tall and the supposed [e] short

‘The true tall killer and the supposed short one’

b. El asesino alto y el [e] bajo

The killer tall and the [e] short

‘The tall killer and the short one’

(157) ABS

a. *El verdadero asesino alto y el supuesto [e] bajo

The true killer tall and the supposed [e] short

‘The true tall killer and the supposed short one’

b. El asesino alto y el [e] bajo

The killer tall and the [e] short

‘The tall killer and the short one’

Finally, only relative clauses introduced by que ‘that’ are licit relative clause remnants, other relativizers result in ungrammatical constructions (158, 159).

(158) stSp

a. La fiesta que me gusta y la [e] que no me gusta

The party that me please and the [e] that no me please

‘The party that I like and the one that I dislike’

243 b. *La fiesta donde hay cerveza y la [e] donde no hay cerveza

The party where have beer and the [e] where no have beer

‘The party where there is beer and the one where there is no beer’

(159) ABS

a. La fiesta que me gusta y la [e] que no me gusta

The party that me please and the [e] that no me like

‘The party that I like and the one that I dislike’

b.*La fiesta donde tiene cerveza y la [e] donde no tiene cerveza

The party where have beer and the [e] where no have beer

‘The party where there is beer and the one where there is no beer’

To summarize, this section has provided an account of the main contexts in which nominal ellipsis may and may not appear in stSp and ABS. We have noticed that the occurrence of this phenomenon follows the same patterns in both languages; the only difference encountered concerns the grammaticality of ellipsis followed by a PP headed by cun in ABS and the ungrammaticality of con-P constructions in stSp. For this reason, before proceeding further, I will briefly provide a description of the uses of prepositions in these two varieties, which, I believe may help us shed light on this issue.

244 7.2. Preposition uses in ABS and stSp

Prepositions in ABS are employed in ways which significantly diverge from their use in stSp. In fact, as Lipski (2008b) points out, they are often omitted in the traditional dialect (160).

(160)

ABS: Nació Mururata

born-past Mururata

stSp: Nació en Mururata

born-past in Mururata

‘She was born in Mururata’

ABS: Aprendió tomá

learn-past drink-inf

stSp: Aprendió a tomar

learn-past to drink-inf

‘She learned how to drink’

ABS: Yo lleva Coroico també

I bring-pres Coroico too

stSp: Yo lo llevo de Coroico también

I it bring-pres of Coroico too

‘I brought this from Coroico too’

245 ABS: Eyu salía mí avisá

They exit-past me tell-inf

stSp: Ellos salían para avisarme

They exit-past for tell-inf-me

‘They went out to let me know’

Moreover, my findings, based on grammaticality judgments and sociolinguistic interviews, indicate that ABS a is often employed as stSp en , with a locative function

(161).

(161)

ABS: Mi hijo vive a Mururata

my-Sg son-M-Sg live to Mururata

stSp: Mi hijo vive en Mururata

my-Sg son-M-Sg live in Mururata

‘My son lives in Mururata’

ABS: Juan nació a La Paz

Juan born-past to La Paz

stSp: Juan nació en La Paz

Juan born-past in La Paz

‘Juan was born in La Paz’

246 ABS: Mis hijo vive a la Argentina

My-Pl son-M-Sg live-pres to the-F-Sg Argentina-F-Sg

stSp: Mis hijos viven en Argentina

My-Pl son-M-Pl live-pres in Argentina

‘My sons live in Argentina’

Also cun is used in different ways. It can replace stSp y ‘and’ (162).

(162)

ABS: Naranja cun cajué

Orange-F-Sg with coffee-M-Sg stSp: Naranjas y cajué

orange-F-Pl and coffee-M-Sg

‘Oranges and coffee’

ABS: Mi tatá cun mi mamá nació Mururata

MySg dad-M-Sg with my-F-Sg mom-F-Sg born-past Mururata stSp: Mi papá y mi mamá nacieron en Mururata

My-Sg dad-M-Sg and my-F-Sg mom-F-Sg born-past in Mururata

‘My father and my mother were born in Mururata’

247 ABS: Mururata cun Chijchipa, nojotro siempre fue uno nomá.

Mururata with Chijchipa we always be-past one no more

stSp: Mururata y Chijchipa, nosotros siempre fuimos uno no más.

Mururata and Chijchipa we always be-past one no more

‘Mururata and Chijchipa, we have always been one single thing’

Most interestingly, ABS cun is employed in many cases in which de would be the preferred preposition in stSp (163).

(163)

ABS: Wawa cun eje edad pesa 20 kilo

kid-F-Sg with this-M-Sg age-F-Sg weight-pres 20 kilo-M-Sg

stSp: Una wawa de esta edad pesa 20 kilo

a-F-Sg kid-F-Sg of this-M-Sg age-F-Sg weight-pres 20 kilo-M-Sg

‘A kid of this age weights 20 kilos’

ABS: El hombre con cabeza blanco

The-M-Sg man-M-Sg with head-F-Sg white-M-Sg

stSp: El hombre de pelo blanco

The-M-Sg man-M-Sg of hair-M-Sg white-M-Sg

‘The white-hair man’

248 ABS: Yo ta cun pelo blanco

I be-pres with hair-M-Sg white-M-Sg

stSp: Yo soy de pelo blanco / Yo tengo pelo blanco

I be-pres of hair-M-Sg white-M-Sg I have-pres hair-M-Sg white-M-Sg

‘I have white hair’

ABS: Eyu ta cun treinta año

they be-pres with thirdy year-M-Sg

stSp: Ellos son de treinta años / Ellos tienen treinta años

they be-pres of thirdy year-M-Pl they have-pres thirdy year-M-Pl

‘They are thirty years old’

ABS: Eyu ta cun la cabeza bien bañadita

they be-pres with the-F-Sg head-F-Sg well wet-F-Sg

stSp: Ellos tiene la cabeza bien bañadita

They have-pres the-F-Sg head-F-Sg well wet-F-Sg

‘They have a very wet head’

In sum, there are several differences in the use of prepositions in ABS and stSp. A

curious peculiarity, which might explain why cun -PP can licitly survive nominal ellipsis in this vernacular, is that ABS cun is often used where de would be preferred in stSp. As all elliptical patterns are the same in ABS and stSp but with exception of cun/con constructions, I will assume that this distinction must have to do with these prepositions.

249 This statement is justified as con/cun presents different grammatical behaviors in the two languages. In the next section, I will describe some of the main analyses which have been proposed to account for the data in stSp.

7.3. Previous analyses

This section provides an overview of the main proposal concerning Spanish NP- ellipsis. These models and their predictions will be contrasted with the data presented in the previous section.

7.3.1. Brucart and Gràcia (1986)

Brucart and Gràcia (1986) suggest that in both Italian and Spanish, the N position is occupied by an empty elided category. They support this claim by providing examples like (164) and (165) (cf. Vanelli 1979:189), where the presence of [e] in (164) blocks the contraction between the determiner ‘quello’ (that) and the adjective ‘alto’ (tall); while such a phonological process is allowed in (165).

(164) Italian

a. Dammi lo sgabello, quello [e] alto

give-me the stool, that [e] tall

‘Give me the stool, the tall one’

b. *Dammi lo sgabello, quell’alto

give-me the stool, that tall

‘Give me the stool, the tall one’

250 (165) Italian

a. Quell’alto monte

that tall mountain

‘That tall mountain’

b. *Quello alto monte

that [e] tall mountain

‘That one tall mountain”

The authors analyze the behavior of Spanish definite articles by claiming that these determiners must cliticize to [+N] elements (adjectives and nouns). The structures proposed are illustrated in (166):

(166) stSp

a. NP

Det N el N NP [+N] [e] (de) Juan

b. * NP

Det N el N PP [-N] [e] P’

P NP con Juan

251 To account for the above-mentioned de -constructions, Brucart and Gràcia claim that de ‘of’ is a false P in Spanish and the category that it introduces is [+N]. In their view, also que ‘that’ would carry [+N] features so that (167) is a licit structure.

(167) stSp

NP

the [e] that saw yesterday

‘The one that you saw yesterday’

The fact that intervening adverbials do not affect the grammaticality of these

constructions would be explained in two possible ways: first, adverbials could be seen as

‘logic operators’ (Guèron 1981) and for this reason they would be ‘transparent’ to

cliticization processes (Brucart and Gràcia 1986:19). Second, the article cliticizes on any

element (including adverbials) if the first maximal category dominating such element is a

category containing a [+N] feature in its head. An example would be (168) where il

cliticizes on the adverbial più , in the Spec of a AP preceding the noun.

252 (168)

NP

The more big problem ‘The bigger problem’

There are at least two problems which emerge from this analysis: firstly, the authors do not explain why number can survive ellipsis while gender cannot (169, 170):

(169) stSp

a. Los gatos negros y el [e] blanco

the-M-Pl cat-M-Pl black-M-Pl and the-M-Sg [e] white-M-Sg

‘The black cats and the white one’

b. * El gato negro y la [e] blanca

the-M-Sg cat-M-Sg black-M-Sg and the-F-Sg [e] white-F-Pl

‘The black cat and the white one’

(170) ABS

a. Los gatos negros y el [e] blanco

the-M-Pl cat-M-Pl black-M-Pl and the-M-Sg [e] white-M-Sg

‘The black cats and the white one’

b. * El gato negro y la [e] blanca

the-M-Sg cat-M-Sg black-M-Sg and the-F-Sg [e] white-F-Pl

‘The black cat and the white one’ 253

Secondly, no distinction is made between pre-nominal and post-nominal adjectives. As they both are [+N] elements, we would expect examples like (171, 172) to be grammatical, but it is not the case.

(171) stSp

*El verdadero asesino alto y el supuesto [e] bajo

The true killer tall and the supposed [e] short

‘The true tall killer and the supposed short one’

(172) ABS

*El verdadero asesino alto y el supuesto [e] bajo

The true killer tall and the supposed [e] short

‘The true tall killer and the supposed short one’

7.3.2. Brucart (1987)

Brucart (1987) begins the section of his book concerned with the elision of

Spanish nominal nuclei (1987:221) by presenting the cases in which empty nouns are grammatical and the instances in which they are not. He provides grammatical examples

(173) in which the definite article is followed by an AP (173a), a de -PP (173b) or a CP

(173c), and ungrammatical constructions (174) lacking an overt determiner (174a), a remnant complement (174b) or presenting a PP complement headed by a preposition other than de (174c).

254 (173)

a. Mi cuñado utiliza e coche antiguo para ir a trabajar y el [e] nuevo para

trasladarse los fines de semana a su casa de campo.

‘My brother in law uses the old car to go to work and the new one to go to

his countryside house during the weekend’.

b. El hijo de Luis y el [e] de Antonio se han hecho muy amigos

‘Luís’s son and Antonio’s one became good friends’

c. La casa que visitaste ayer y la [e] que has visto esta mañana pertenecieron a

un mismo dueño.

‘The house that you saw yesterday and the one that you saw this morning

belonged to one owner’

(174)

a. *El hijo de Luís y [e] se han hecho muy amigos

‘Luís’s son and Antonio’s became good friends’

b. *El hijo de Luís y el [e] se han hecho muy amigos

‘Luís’s son and Antonio’s became good friends’

c. * El tren a Barcelona y el aMadrid ha salido con retraso

‘The train for Barcelona and the one for Madrid left late’

Brucart argues that these cases should be classified as constructions containing empty nouns, rather than structures in which the definite articles behave as pronouns

(contra Bello 1847). In fact, these articles are atonic, as the rest of the definite articles in

255 Spanish. On the other hand, all pronouns bearing nominative case are tonic; for this reason, considering these elements as pronouns implies doubling the systems of personal pronouns able to act as subjects, so that él and el would be variables; this would be a non- economical solution (see also cf. Bucart 1999:2856).

Brucart also disagrees with approaches which fail to identify an empty N as the nucleus of the aforementioned nominal constructions. For example Alarcos-Llorach

(1973) indicates that in cases like el de filosofia ‘the one of philosophy’ the nucleus

would be the element following the definite article. The nature of Brucart’s disagreement

is based on the fact that what follows the article is neither a noun nor a nominal

projection. This fact implies a violation of X’-Theory (175).

(175) Xn ....Xn-1

Rule (175) predicts that projctions and nuclei must belong to the same category, a requirement which is violated by Alarcos’s model. Brucart accounts for cases of Spanish nominal ellipsis as instances of empty nouns acting as nominal nuclei. Such elements are not bound and should be interpreted as pronominal anaphoras PRO. In this way, it is possible to explain why a pronominal element can be understood as a previously mentioned NP. Such nouns would also be able to take on an arbitrary reading, as exemplified in example (176).

256 (176)

El PRO que quiera asistir al concierto deberá pagar la entrada con diez días de antelación.

‘The one who is interested in attending the concert will have to pay ten days in

advance’

The empty category in (176) will receive an arbitrary reading, in line with Control

Theory (Chomsky 1981). Obviously, the entities which will be able to be represented by

such category are restricted by thematic principles. In the case of (176), PRO is an agent

which can only be interpreted as +human (Brucart 1987:229). In Brucart’s view, one of

the crucial components of these constructions is the presence of an overt determiner

preceding the empty N; this follows from the principle of recoverability. In fact, the

adoption of an overt D forces the interpretation of the remnants to the one of

complements of the elided N; on the other hand, the lack of such a restriction in natural

languages could allow the interpretation of similar APs, PPs and CPs as modifiers of

other maximal projections, thus not leading to a clear and univocal reading (1987:231).

According to X’-theory, the nucleus must be obligatorily represented in the

structure, while its complements and determiners are only optional. However, when the

nucleus is realized by an empty category, the nuclear information can only be recovered

by the pronounced non-nuclear elements. For this reason the presence of an overt D

element would be required. Also, the need for an overt D in case of elliptical Ns is related

to its referencial feature and therefore, to its capability of turning nouns into arguments.

Moreover, the structure proposed in (177), where PRO is preceded by an overt D

257 element, can account for the fact that PRO does not bare any nominative case. This is because the maximal projection would act as a barrier (Chomsky 1981) to case assignment, while the overt realization of the definite article would be enough to meet the requirements imposed by the Case Filter Principle (178).

(177) NP

Det N’

N PP

El PRO de Maria

(178) Case Filter Principle (Chomsky 1981:49)

*NP, where NP has a phonetic matrix but not Case

On the other hand, if we had to postulate the presence of a phrase without determiner, it would be impossible to obey the Case Filter Principle. If D is missing, case checking should occur with the nominal nucleus being co-indexed with the constituent bearing case, as in (179). Nevertheless, (179) generates ungrammatical constructions as

PRO is an empty category unable to bear case (Brucart 1987:233).

(179)

NPi

N’

N PP

PROi de Maria

258 The author indicates that the presence of an obligatory complement in phrases containing definite articles seems to be due to superficial factors. In fact, definite articles are clitic elements which need a host to avoid the generation of atonic phrases. For the same reason the remnants of elliptical nuclei should be interpreted as having a restrictive reading. Brucart attributes the grammaticality of de -PP remnants and the ungrammaticality of other PPs to the fact that de is not a ‘true’ preposition. As suggested for English of (Chomsky 1981), de would be inserted at PF. Differently for ‘true’ prepositions, it would simply be a case marker and would not play a role in case assignment.

Brucart also points out that, due to their atonic nature, definite articles differ from other determiners in that they are the only category requiring a following complement in elliptical constructions (180).

(180)

a. *Buscaba el

seek-past the-M-Sg

‘I was looking for the’

b. Buscaba uno

seek-past a-M-Sg

‘I was looking for one’

c. Buscaba el tuyo

seek-past the-M-Sg your-M-Sg

‘I was looking for yours’.

259 d. Buscaba este

seek-past this-M-Sg

‘I was looking for this’

e. Buscaba alguno

seek-past some-M-Sg

‘I was looking for some’

f. Buscaba tres

seek-past three-M-Sg

‘I was looking for three’

He also indicates that definite articles do not accept PPs different from de -PPs, while other determiners do (181).

(181)

a. Compré el/este/alguno/uno/tres de matemáticas

Buy-past the/ this/ some/one/three of mathematics

‘I bought the/ this/ some/one/three of mathematics’

b. Compré *el/este/alguno/uno/tres con lazo

Buy-past the/ this/ some/one/three with lace

‘I bought the/ this/ some/one/three with lace’

The grammaticality of constructions like (181a) vs. the ungrammaticality of structures like (181b) would be due to the fact that these demonstratives allow the recovery of the antecedent so that they enable PRO to participate in the compositional

260 assignment of thematic role across ‘true’ ( con ) and ‘false’ ( de ) prepositions. On the other hand, the definite article does not trigger such operation and matemáticas would be able

to receive thematic assignment only if preceded by a ‘false’ preposition like de (Brucart

1987:245).

Brucart’s account is valuable in that it recognizes the existence of empty Ns as nuclei of nominal phrases and provides a framework to account for differences between definite articles and other determiners; nevertheless, it presents some drawbacks: as for his 1986 study with Gràcia, this new work does not acknowledge the distinction in grammaticality due to variation in gender and number features and it does not provide an account for pre-nominal and post-nominal adjectives. Moreover, it fails to provide an account for thematic correspondences between pronounced and elided Ns.

7.3.3. Torrego (1988)

In Torrego’s (1988) view, the possibility of having nominal ellipsis is related to the null subject parameter, so that ellipsis would be the result of the presence of pro in the language. According to her model, the contrast between the definite articles and the rest of the determiners in Spanish is due to the fact that demonstratives and quantifiers are rich enough to license the null noun (they carry person, gender and number features), while definite articles only have gender and number features and need to be supplied with person features by a [+N] category: adjectives, que or de .

This would account for the following cases (182):

261 (182) stSp

a. Esa carta y ésta pro

this letter and that pro

‘This letter and that one’

b. Nos gustaban tres/algunas/varias pro.

to-us pleased three/some/several pro

We liked three/some/several (of them).’

c. * La carta roja y la pro

the letter red and the pro

‘The red letter and the one’

d. La carta roja y la pro verde/ de color verde

the letter red and the pro green/of color green

‘The red letter and the green one’

This approach indicates that the grammaticality of the ellipsis depends on the nature of the features carried by the determiner and the remnant elements. Therefore, if the D and the remnant express the person, number and gender features of the referent of pro , the structure should be grammatical.

This model accounts for the ungrammaticality of constructions in which the gender has been changed (183a, 184a), but does not explain why the number can vary without affecting the grammaticality of the enunciate (183b, 184b).

262 (183) stSp

a. * El gato negro y la [e] blanca

the-M-Sg cat-M-Sg black-M-Sg and the-F-Sg [e] white-F-Pl

‘The black cat and the white one’

b. Los gatos negros y el [e] blanco

the-M-Pl cat-M-Pl black-M-Pl and the-M-Sg [e] white-M-Sg

‘The black cats and the white one’

(184) ABS

a. * El gato negro y la [e] blanca

the-M-Sg cat-M-Sg black-M-Sg and the-F-Sg [e] white-F-Pl

‘The black cat and the white one’

b. Los gatos negros y el [e] blanco

the-M-Pl cat-M-Pl black-M-Pl and the-M-Sg [e] white-M-Sg

‘The black cats and the white one’

Note that also in this case de and que have been treated as special elements, more precisely they have been assigned [+N] properties to rescue the model, even though they are obviously neither nouns nor adjectives.

In addition, Torrego’s account cannot explain the distinction between pre-nominal and post-nominal adjectives (185, 186).

263 (185) stSp

*El verdadero asesino alto y el supuesto [e] bajo

The true killer tall and the supposed [e] short

‘The true tall killer and the supposed short one’

(186) ABS

*El verdadero asesino alto y el supuesto [e] bajo

The true killer tall and the supposed [e] short

‘The true tall killer and the supposed short one’

Furthermore, as pointed out by Ticio (2003; 2005) the relation between nominal ellipsis and the null subject parameter is problematic for other languages, such as French and German, which have NP-ellipsis but do not have null subjects. This observation is also particularly relevant to the ABS case. This dialect, in fact, is not a pro-drop language. Due to the lack of number and person marking on verbs, the default third person singular form is the only one used. In the most traditional variety the subject must always be pronounced. Compare the ABS data with the stSp ones (187).

(187)

ABS: *(yo) come

stSp: (yo) como

‘I eat’

264 ABS: *(oté) come

stSp: (tú) comes

‘you eat’

ABS: *(ele) come

stSp: (él/ella) come

‘he/she eats’

ABS: *(nojotro) come

stSp: (nosotros) comemos

‘we eat’

ABS: *(otene) come

stSp: (ustedes) comen

‘you eat’

ABS: *(eyu) come

stSp: (ellos) comen

‘they eat’

7.3.4. Ticio (2005)

Ticio (2005) suggests a structure for Spanish DPs following the hypothesis of a strict parallelism between clauses and nominal expressions. This structure assumes

265 Grohmann’s (2000) division of clause structure into three prolific domains and its extension to the DP proposed in Grohmann and Haegeman (2002) which argue in favor of the existence of a Thematic domain containing the predicate and its arguments; an

Agreement domain, in which functional categories such as NumP, GenP and PossP are contained; and a Discourse domain, where discourse information is encoded.

Furthermore, Ticio proposes different positions for Adjectives according to their semantic properties. More concretely, she assumes that prenominal As are generated as the Specifier of NP, while postnominal As are adjoined to NP and Relational As to nP.

(188)

Ticio accounts for the phenomena observed by postulating that only the elements that are base-generated outside of the lower NP node can be left stranded after NP-ellipsis has applied. Therefore, we will expect to observe R(elational)As, Adjuncts,

PostN(onimal) As, Agents and Possessors as possible remnants, while elements such as

PreN(onimal) As and Objects should not be observed as remnants (189).

266 (189)

Such a stipulation seems to account for Adjuncts (190a), PostN As (190b),

Possessors (190c), Agents (190d), R-As (190e), and PreN As (190f); however it does not explain why Objects can be left standed even though they should not (190g).

(190) stSp

a. Compramos bastantes libros para regalo y uno [e] [para consulta] adjunct

(we)bought several books to gift and a [e] to consult

b. Compramos varios libros azules y uno [e] rojo

(we)bought several books blue and a [e] red

c. Compramos varios libros de Luis y uno [e] [de María] poss

(we)bought several books of Luis and a [e] of María

d. Compramos varios libros de Cervantes y uno [e] [de Borges] ag

(we)bought several books of Cervantes and a [e] of Borges

267 e. Compramos varias novelas policíacas y una [e] romántica

(we)bought several books police and a [e] romantic

f. *Ayer vi a la verdadera terrorista y a la supuesta [e]

yesterday (I)saw to the true terrorist and to the alleged [e]

‘Yesterday, I saw the true terrorist and the alleged one’

g. Compramos varios libros de Matemáticas y alguno [e] [de Física] obj

(we)bought several books of Math and some [e] of Physics

To resolve this problem, she proposes that the movement that removes the object from the NP-ellipsis site is not done in the overt syntactic component of the derivation, rather it applies after syntax but before NP-ellipsis. In this way, the stylistic rule can place objects outside of the ellipsis site in the constructions under consideration, probably as adjuncts of some maximal category contained in the DP. The analysis is given in

(191):

(191) stSp

a. varios [e] de Física obj de Juan poss

several [e] of Physics of Juan

268 Finally, she deals with the fact that, in presence of definite articles, no other PPs

can be found as remnants but those headed by de . In line with Brucart & Gràcia (1986)

and Raposo (1999), Ticio highlights the clitic-like character of definite articles, which

seem able to cliticize only on [+N] elements (adjectives and nouns).

(192) stSp

a. Las tres [e] con gafas [e] = chicas

the three [e] with glasses [e] = girls

b. Las tres [e] de gafas [e] = chica

the three [e] of glasses [e] = girl

c. Las tres [e] italianas[e] = chica

the three [e] Italian[e] = girl

d. *La [e] con gafas [e] = chica

the [e] with glasses [e] = girl

e. La [e] de gafas [e] = chica

the [e] of glasses [e] = girl

f. La [e] italiana[e] = chica

the [e] Italian[e] = girl

In fact, as shown in (182), only definite articles need to cliticize on a host. As a consequence examples like (192a,b,c) are always grammatical. On the other hand, NP- ellipsis in definite DPs is ungrammatical whenever the only remnant is not a PP headed by de ‘of’ or a post-nominal adjective.

269 Along the line of Brucart (1987) and Raposo (1999), she also assumes a division of PPs, namely, full PPs headed by Ps such as con ‘with’, which constitute phases by themselves, and false PPs, headed by the P de ‘of’, which do not constitute phases. She proposes that de ‘of’ is inserted at PF.

She departs from Brucart (1987) and Raposo (1999) in the reason why the cliticization of the definite article cannot be satisfied with full PPs: she assumes that cliticization processes cannot operate through the boundaries of phases because phases are spell out units. The derivation of the data ‘*La [e] con gafas’ (‘The [e] with glasses’, where [e] = girl) and ‘La [e] de gafas’ (‘The [e] of glasses’, where [e] = girl) is provided below (where the symbol ‘||’ stands for phase boundary):

(193) stSp

a. [DP La [[NP chica] || [ PP con gafas]]] Spell Out

b. [ DP La [[NP e]] || [ PP con gafas]]] NP-Ellipsis

c. *[ DP La [[NP e]] || [ PP con gafas]]] Cliticization

(194) stSp

a. [ DP La [[NP chica] [NP gafas]]] Spell Out

b. [ DP La [[NP e] [NP gafas]]] NP-ellipsis

c. [ DP La [[NP e] DE-[NP gafas]]] DE(of)-insertion

d. [ DP La[[NP e] + DE-[NP gafas]]] Cliticization

The ungrammaticality of *La [e] con gafas (the e with glasses) is derived from the fact that the cliticization process, which applied from (194b) to (194c), cannot attach

270 the definite article to any element in its phase and cannot operate through the boundaries of the PP phase to cliticize the definite article to the P con ‘with’. In contrast, the

grammaticality of la de gafas ‘the e of glasses’ is explained since the cliticization

requirement of the definite article can be satisfied within its phase.

Also this account presents some drawbacks: first, the postulation of a pre-nominal

and a post-nominal adjective base-generated positions is somehow problematic in that it

goes against hypotheses assuming that all adjectives are originally pre-nominal and the

overt distribution is just a result of N movement (Cinque 1993; 2005; 2007) 38 . Second, the postulation of an Agreement Projection in which all phi-features are checked cannot capture the grammaticality distinctions pointed out in (195-196), where number but not gender can survive to NP-ellipsis. Moreover, it goes also against recent hypotheses which argue in favor of the elimination of an Agreement projection (Chomsky 2002).

(195) stSp

a. Los gatos negros y el [e] blanco

the-M-Pl cat-M-Pl black-M-Pl and the-M-Sg [e] white-M-Sg

‘The black cats and the white one’

b. * El gato negro y la [e] blanca

the-M-Sg cat-M-Sg black-M-Sg and the-F-Sg [e] white-F-Pl

‘The black cat and the white one’

38 For a different account of adjective position and movement within the DP see also Gutiérrez-Rexach & Mallén (2001) 271 (196) ABS

a. Los gatos negros y el [e] blanco

the-M-Pl cat-M-Pl black-M-Pl and the-M-Sg [e] white-M-Sg

‘The black cats and the white one’

b. * El gato negro y la [e] blanca

the-M-Sg cat-M-Sg black-M-Sg and the-F-Sg [e] white-F-Pl

‘The black cat and the white one’

Third, postulating that articles can only cliticize to [+N] elements does not explain why examples like (197, 198) are grammatical.

(197) stSp

a. El chico italiano y el [e] supuestamente italiano.

The guy Italian and the [e] supposedly Italian

‘The Italian guy and the supposedly Italian one’

b. Los chicos de Italia y los [e] supuestamente de Italia

The guys of Italy and the [e] supposedly of Italy

‘The Italian guys and the supposedly Italian ones’

(198) ABS

a. El chico italiano y el [e] supuestamente italiano.

The guy Italian and the [e] supposedly Italian

‘The Italian guy and the supposedly Italian one’

272 b. Los chicos de Italia y los [e] supuestamente di Italia

The guys of Italy and the [e] supposedly of Italy

‘The Italian guys and the supposedly Italian ones’

Examples like (197, 198) clearly show that APs and PPs modified by adverbs [-

N] are actually licit remnants. In order to account for these constructions we should

postulate that adverbs are inserted later in the derivation, after ellipsis applied, like de .

This hypothesis does not seem feasible.

Moreover, the movement of objects before ellipsis and de insertion at PF seems to

be an ad-hoc solution to rescue the framework rather that theoretically motivated

operations. In fact movement and insertion at PF is problematic. By definition, PF cannot

operate on narrow syntactic features, only on phonological matrices. There is no obvious

way in which PF can tell that a PP is a PP. There is no reason why PF movement should

only move PPs to a position in the post-nominal field.

In addition, Ticio (2005) does not provide an account for the differences between relative clauses headed by que and those headed by other relativizers. Finally, this analysis assumes a right-branching structure with rightsided specifiers and rightward movement operations. Such syntactic machinery goes against standard assumptions on left branching and dislocation; it also violates the Liner Correspondence Axiom (Kayne

1994).

273 7.3.5. Eguren (2010; in press)

Eguren (2010; in press) provides an account of Spanish nominal ellipsis which breaks with former analyses supporting proper government condition for empty nominals

(see Brucart 1987; Torrego 1988; Contreras 1989; Lobeck 1995 & Kornfeld and Saab

2004). In fact several approaches share, implicitly or explicitly, the idea that like the empty pro , empty nominals must satisfy the ECP (Chomsky 1981) and must also be identified through strong agreement features.

Such a generalization is formalized by Lobeck (1995) in (199):

(199) Licensing and Identification of pro :

An empty, non-arbitrary pronominal must be properly head-governed, and

governed by an Xº specified for strong agreement (Lobeck 1995:20).

This author takes strong agreement to be a productive, morphological realization

of features, and considers, in particular, that [+plural], [+possessive] and [+partitive] are

strong agreement features in English DPs. Lobeck finally comes to the conclusion that a

determiner is a proper head-governor, and so licenses an empty nominal, only when

specified for strong agreement.

On the other hand, Eguren shows that nominal ellipsis are allowed also with bare

plurals and bare mass nouns, which, in his view, are not headed by a D element.

274 (200)

a. Antes bebía cerveza alemana y ahora sólo bebo [e] española.

before drink-past beer German and now only drink-pres [e] Spanish

‘I used to drink German beer before and I now only drink Spanish beer’.

b. No había leído cuentos de Cortázar, pero sí había leído [e] de Borges.

No have-past read stories of Cortázar but yes have-past read [e] of Borges

‘She had not read stories by Cortázar, but she HAD read stories by Borges’.

Eguren suggests that examples in (200) are to be taken as clear-cut evidence against a proper government condition for empty nominals, since they show that nominal ellipsis can occur in the absence of a head-governing functional category. Of course, this idea only works if bare mass nouns and bare plurals in Spanish really lack a determiner.

A problem to be faced, in this respect, is that bare nouns of this sort have been claimed to contain an empty existential quantifier (see Contreras 1986; Longobardi 1994). Eguren provides a series of examples showing that bare mass nouns and bare plurals are not indefinite quantified expressions in Spanish (Laca 1996; 1999; McNally 2004) and that therefore these empty nominals need not be formally head-governed.

Moreover, Eguren indicates that the presence of agreement markers in the remnant is neither a sufficient nor a necessary condition for nominal ellipsis in Spanish.

On the one hand, in a good number of cases the determiner introducing a deleted nominal bares number and/or gender inflection, but ellipsis is ruled out:

275 (201)

a. Todo estudiante de literatura escribe poemas y a tod-o *(estudiante) de lingüística le gustan las lenguas.

‘Every student of literature writes poems and every student of linguistics loves languages’.

b. Algunos políticos de la derecha dicen que ciert-o-s *(políticos) de la izquierda están arruinando el país.

‘Some right-wing politicians claim that certain left-wing politicians are ruining the country’.

He also states that some uninflected determiners, like cardinals or the comparatives más ‘more’ and menos ‘less’, can combine with an empty nominal.

(202)

a. Juan ha leído dos artículos y María ha leído tres [e].

‘John has read two papers and Mary has read three [e]’.

b. Bebe demasiada cerveza y debería beber menos [e]

‘He drinks too much beer and he should drink less [e]’.

Euger prefers to look for a different explanation to nominal ellipses. Nominal ellipses with bare nouns in Spanish (200) are grammatical constructions because the restrictive modifier is focused and, therefore, it identifies an alternative in a set of alternatives. Such an operation consists of the creation of a set–subset relation between

276 elements of the same class, thus allowing the recovering of the elided element (Eguren

2010:439).

(203)

a. No compró manzanas [verdes]F, pero llenó el cesto de [e] [rojas]F.

‘She did not buy green apples, but filled the basket with red ones’.

By reformulating Sleeman’s (1996) partitiviy condition on nominal ellipsis and

Lopez’s (2000) D-linking condition, Euger provides the following semantic condition on

elliptical phenomena:

(204) The partitivity/D-linking condition on nominal ellipsis:

Elliptical DPs always select a subset from a contextually given set.

Among many phenomena, Eguren (2010) provides a semantic account for the

(un)grammarticality of several adjectival modifications of nominal ellipsis. Eguren

(2010) notes that nominal ellipsis obey a contrastive focus requirement. The creation of a

set–subset relation through the focus operation consists of an intersectional relation

between elements of the same class. Prenominal adjectives are non-intersective elements,

they do not entail the intersection of the adjective and the noun and therefore they cannot

be focused. In fact a false prophet is not a prophet; an alleged murderer is not necessarily

a murderer; a mere general isn’t a general that is mere (see also Braver 2009).

277 (205)

a. *El falso profeta y el verdadero [e]

The fake prophet and the actual [e]

‘The fake prophet and the actual one’

b. *El verdadero asesino alto, y el supuesto [e] bajo

The actual assassin tall and the allegad [e] short

‘The tall actual assassin and the short (*alleged) one’

c. *El gran general y el mero[e]

The grand general and the mere [e]

‘The grand general and the mere one’

Eguren’s proposal is valuable, especially because it can logically account for the differences between pronominal and post-nominal adjectives; moreover, examples like

(200) show that the Case Filter Principle does not make the correct predictions.

Nevertheless, I do not agree with his analysis of bare and mass nouns as elements lacking a preceding D category. As I will show in chapter 8, in my view, mass nouns and bare plurals should not be considered as nominals lacking a D, but rather as nominals lacking a Num projection. This does not imply that they will necessary behave as nouns containing an existential quantifier. On the other hand, I suggest that cover determiners encode a variety of type-shifting functions ( à la Partee). This said, I think that Eguren is correct in suggesting that strong agreement features are not to be held responsible for elliptical constructions, while his semantic condition offers a better explanation of such structures.

278 In addition, Eguren’s model does not seem to account for the fact that, if the pronounced N assignes a theta role (206a), the elided one must do the same to its remnant, or lack a theta-marking altogether (206b); in fact, a remnant with a different theta-assignment would result in an ungrammatical construction (206c); on the other hand, if the first N is not subject to such a constraint, the following elided N may or may not assign a theta-role (206d,e).

(206)

a. El libro de física y el [e] de matemáticas

the book of Physics and the [e]of Math

‘The Phisics book and the Math one’

b. El libro de física y el [e] de color rojo

the book of Physics and the [e] of color red

‘The Phisics book and the red one’

c. *El libro de física y el [e] de Ana

the book of Physics and the [e] of Ana

‘The Phisics book and Ana’s one’

d. El libro de color rojo y el [e] de Ana

the book of color red and the [e] of Ana

‘The red book and Ana’s one’

e. El libro de color rojo y el [e] de color verde

the book of color red and the [e] of color green

‘The red book and green one’

279 7.3.6. Braver (2009)

In order to solve the puzzle and provide an explanation to the clitic-like behavior of definite articles, Braver (2009) proposes what he calls the Clitic Host Selection

Algorithm, here presented in (207):

(207) The Clitic Host Selection Algorithm (CHSA)

a. The Spanish definite articles cliticize to an XP iff:

i. XP is headed by an overt [+N] head X°

ii. Any overt material that intervenes between the clitic and X° must be contained

within XP

b. The Spanish definite articles cliticize to a head X° iff:

i. X° is an overt complementizer head

ii. No overt material intervenes between the clitic and X°

Braver considers, following Kayne (1976; 1994), that de and que are different

from the rest of the prepositions and complementizers, in that they are heads in C°. On

the other hand, other relativizers are XPs, based generated in SpecC, while the rest of the

prepositions are P°, which, according to (207 b.i.), cannot host a clitic article.

Other determiners and demonstratives do not have the restrictions on nominal

ellipsis encountered for articles; this is because they are not clitics, and so are not affected

by the CHSA. In line with Eguren (in press), Braver notes that nominal ellipses obey a

contrastive focus requirement; in other words, the remnant constituent must be

contrastively focused and therefore pronominal adjectives cannot be licit remnants.

280 In Braver’s model, as well as in Brucart’s and Gracia’s (1986) study, adverbials contained between the AP (208) can be accounted for, as the definite article will cliticize to the A. This follows from (207a).

(208)

... y el [e] verdaderamente amarillo

and the [e] truly yellow

‘and the truly yellow one’

Moreover, in elliptical DPs with que relative clause modifiers, the definite article clitic cliticizes to the Cº que (209), while in non-que relative clause modifiers, the definite article has nowhere to cliticize - the Cº is covert, and overt material (namely, the relative pronoun in SpecCP) intervenes before any other potential host (210).

281 (209)

... y el [e] que ...

and the [e] that

‘and the one that...’

(210)

*... y el [e] cuando......

and the [e] when

‘and the one when...’

282 As with relative clauses, Braver (2009) follows Kayne (1994) in determining the categorial status of phrase heads. In particular, he assumes that de is a complementizer

head (marked here as deº/Cº) heading a de-P/CP. In this way the patterns represented in

(211) and (212) find their logical explanation.

(211)

... y el [e] de Australia

and the [e] of Australia

‘and the one from Australia’

(212)

*... y el [e] para el acuario

and the [e] for the aquarium

‘and the one for the aquarium’

283

In addition, due to the non-intersectual properties of prenominal adjectives,

Braver’s (2009) framework can also account for the ungrammaticality of ellipsis involving these elements.

Summarizing, Braver’s (2009) analysis provides a straightforward account for nominal ellipsis appearing with de and que constructions, with adverbials contained between the remnant APs and for the ungrammaticality of ellipsis co-occurring with pre- nominal adjectives. Moreover, contrary to Ticio’s model (2005), and in line with Kayne

(1994), this new proposal does not force a right branching analysis with movement to the right. Nevertheless, Braver’s framework does not account for thematic ordering within the DP; no reference is made to alternation in grammaticality depending on nominal gender and number modifications; PP arguments and adjuncts are all treated in the same way, being all base-generated in a post-nominal position.

Moreover, this model does not provide an account for why the theta-marked complements must match in theta-marking, while non-thematic constructions are not subject to such requirement (213, 214).

284 (213) stSp a. El libro mío poss y el [e] de él poss

the book mine and the [e] of Juan

‘My book and Juan’s one’ b. *El libro de física theme y el [e] de él poss

the book of Physics and the [e] of he

‘The Physics book and his’ c. Dame el libro de Petrarca agent y el [e] de color rojo/rojo

give-me the book of Petrarca and the[e] of color red/red

‘Give me Petrarca’s book and the red one’

d. Dame el libro de física theme y el[e] de color rojo/rojo

give-me the book of Physics and the [e] of color red/red

‘Give me the Physics book and the red one’

(214) ABS a. El libro mío poss y el [e] di él poss

the book mine and the [e] of Juan

‘My book and Juan’s one’ b. *El libro di física theme y el [e] di él poss

the book of Physics and the [e] of he

‘The Physics book and his’

285 c. Dame el libro di Petrarca agent y el [e] di color rojo/rojo

give-me the book of Petrarca and the[e] of color red/red

‘Give me Petrarca’s book and the red one’

d. Dame el libro di física theme y el[e] di color rojo/rojo

give-me the book of Physics and the [e] of color red/red

‘Give me the Physics book and the red one’

7.4. Toward a new proposal

This section provides a new model to explain the data so far presented. The proposal is based on several assumptions concerning the structure of DP, which account for the grammatical and ungrammatical cases of NP-ellipsis. It does not recur to ad-hoc insertion operations or constituent movements. It rather bases the observed differences on pure structural distinctions.

To begin with, the difference between definite articles and the rest of the determiners is based on semantic principles. My proposal assumes that the rest of the determiners possess a [+referential] feature. I hypothesize that the definite article is a restricted element in a semantic sense: although it can precede an empty category, contrary to the English definite article, it is not interpretable in isolation in its Dº position and it therefore requires the presence of a predicate. Under standard assumptions

(Higginbotham 1985), the definite article binds the open position of the predicate, yielding a semantically interpretable construction. Syntactically, when demonstratives, cardinals, etc. are base-generated under their respective projections or possibly move to

SpecDP to check their referential feature, they essentially freeze the DP as it is (see

286 Cardinaletti & Giusti (1991); Zamparelli (2000); Ishane (2008); Gutiérrez-Rodríguez

(2009) for an account of base-generation and movements of these elements within the

DP). The DPs, then freezes into pronouns-like elements regardless of the existence of an overt N 39 . This accounts for cases like (215, 216).

(215) stSp

a. Estas chicas de Italia y aquellas [e].

These girls of Italy and those [e]

‘These Italian girls and those’

b. *Las chicas de Italia y las [e].

The girls of Italy and the [e]

The Italian girls and the

(216) ABS

a. Estas chicas di Italia y aquellas [e].

These girls of Italy and those [e]

These Italian girls and those

b. *Las chicas di Italia y las [e].

The girls of Italy and the [e]

The Italian girls and the

My analysis, as Ticio’s, takes into consideration the well-known existence of thematic layers in the DP, which have shown to hold across several languages by certain

39 See Leonetti (1999) for a more detailed account on this issue. 287 phenomena such as blocking effects in extraction 40 . It is beyond the goal of this chapter

to repeat such tests here, and therefore I will not do it. I will just present the nominal

theta-role hierarchy assumed (217) (Giorgi & Longobardi 1990; Valois 1991, etc.). I also

acknowledge that much research has been done on theta-roles, many hypotheses have

been made, but the precise nature of these thematic relations is still not completely

clear 41 .

(217)

a. Possessor>>Agent>>Object

Moreover, in line with (Cinque 2005; 2007), I argue that all adjectives are based- generated pre-nominally. Therefore, adjectives universally merge to the left of NP and are always phrasal (Bosque & Picallo 1996; Picallo 2010; Cinque to appear)42 . In line with Braver (2009) and Eguren (2010; to appear), I assume that pre-nominal adjectives are generally non-intersective and therefore do not meet the focalizing requirements needed for a nominal ellipsis to obtain. In this way the distinction in grammaticality between pre-nominal and post-nominal adjectives finds its logical explanation.

Moreover, I agree with Braver’s (2009) CHSA and Kayne’s (1994) proposal in assuming que and de to be complementizer heads, differently from the rest of prepositions and relativizer elements. My proposal also addresses the fact that a change in number does not affect the grammaticality of the constructions while a change in gender does. Within my model, gender is an interpretable feature of N, so that a change in

40 For a detailed account see Giorgi and Longobardi (1990) for Italian, Valois (1991) for French, Ticio (2003) for Spanish. 41 See Dowty (1991) for more detailed account. 42 For a different account on adjective merging also read Demonte (1999); Gutiérrez-Rexach & Mallén (2001) and references therein. 288 gender reflects a change in the entities we are referring to. On the other hand, number interpretation is located in Num rather than in N, it has a proper phrasal projection so that it can survive NP-ellipsis.

NP ellipsis targets the lowest DP layer thus erasing only Ns and their intepretable gender features, while leaving NumP, nP and the rest of the higher DP projections.

(218)

DP

D D’

D’ NumP

El AP Num’

Pre-nomA Num’ FP i AP FP’

Post-nom (…)’ nP

(…) nP’ Ellipsis site (…) NP

(…) t i

In line with Kayne (1994), I postulate that all PPs and APs are adjoined to the left and that the surface order is the result of leftward N movement and constituent dislocation. Moreover, I consider stSp de a multifunctional preposition; not only it has the capability of introducing several Spanish argumental PPs, de structures are the

289 syntactic result of commonly used strategies to create adjective-like modificators for nouns 43 .

In line with Carstens (2000), I assume that thematic roles are assigned under nP.

Adjectives and de -phrases carrying Θ-roles are base generated within the nP shells and can move to higher projections in the course of the derivation 44 . On the other hand, the

rest of the post-nominal adjectives (e.i. el libro rojo ), as well as descriptive de -phrases

free from Θ-assignation (e.i. el libro de color rojo ) are generated higher in the structure.

The fact that a strict correspondence between thematic roles is required between

the pronounced and the elided constituents suggests that the information generated under

nP gets frozen. The speaker seems to expect it to match with the thematic configuration

of the elided N. Put it this way, the alternation between APs or de -PPs carrying the same

thematic role represents only a stylistic variation which does not affect the meaning of the

sentence. This piece of evidence further undermines the approaches that do not consider

thematic relations between DPs structures and only focus on the ‘special’ properties of

de . In fact, it is evident that even in presence of de , if there is no thematic

correspondence, the ellipsis is ungrammatical (219, 220).

(219) stSp

a. El libro mío poss y el [e] de él poss

the book mine and the [e] of Juan

‘My book and Juan’s one’

43 Note that multifunctional particles of this kind have been reported for several well-known and no-so- well-known languages. A nice account of such an element in a no-so-well-known language (Rotuman) is the one by Den Dikken (2003). In Rotuman the particle ne seems to resemble Spanish de in many aspects. See Den Dikken (2003: 23-40) for details. 44 The order i obtained by phrasal rising operations (cf. Cinque to appear and Picallo 2010). 290 b. *El libro de física theme y el [e] de él poss

the book of physics and the [e] of he

‘The physics book and his one’ c. La teoría chomskyana agent y la [e] de Marx agent

the theory Chomskyan and the [e] of Marx

‘The Chomskyan theory and the Marxist one’

d. *La teoría Marxista agent y la [e] de física theme

the theory Marxist and the [e] of physics

‘The Marxist theory and the physics one’

(220) ABS a. El libro mío poss y el [e] de él poss

the book mine and the [e] of Juan

‘My book and Juan’s one’ b. *El libro de física theme y el [e] de él poss

the book of physics and the[e] of he

‘The physics book and his one’ c. La teoría chomskyana agent y la [e] de Marx agent

the theory Chomskyan and the[e] of Marx

‘The Chomskyan theory and the Marxist one”

d.*La teoría Marxista agent y la [e] de física theme

the theory Marxist and the [e] of physics

‘The Marxist theory and the physics one’

291

For this reason, I postulate that As and PPs –when free from any potential theta- role assignment- are generated higher in the structure, in a projection different from the one stipulated for the rest of post-nominal non-thematic adjectives.

Note that the present proposal does not need to stipulate ad hoc object movements or procrastinated de insertion at PF (Ticio 2003; 2005); nor it assigns special [+N] features to de and que (Torrego 1988).

The syntactic process involved in NP-ellipsis seems to be able to recover only the

N information, and to freeze the structure projected from the lower thematic nP (object) up to the higher nP (possessor) (see 213). Such a process allows variability for the information contained in higher projections (NumP, non-thematic-APs/de-PPs).

To account for the differences in grammaticality between stSp con and ABS cun ,

I consider cun in ABS to share many of its structural features with stSp de and ABS di .

This statement is based on cross-linguistic data; in fact, as we noticed in section 7.2, prepositions are used differently in the two varieties, and moreover, cun appears in many contexts in which in stSp de would be employed rather than con . As a consequence, constructions like (222) in stSp are ungrammatical while cases like (221) in ABS allow

NP-ellipsis.

(221) ABS

… y la [e] cun gafa

… and the [e] with glasses

‘ …the one with glasses’

292 DP

D D’

D’ NumP

El AP Num’

Pre-nomA Num° cunP/CP i cunP/CP’

cun° DP

cun gafa Ellipsis site

N

t i

(222) stSp

*… y la [e] con gafas

… and the [e] with glasses

‘ …the one with glasses’

293 * DP

D D’

D’ NumP

El AP Num’

Pre-nomA Num° PP i PP’

con DP

con gafas Ellipsis site

N t i

As we can see in examples (221; 222), ABS cun is analyzed as a complementizer head, which, in line with Braver’s (2009) CHSA and Kayne’s (1994) model, allows elliptical constructions. On the other hand, stSp con is a preposition and therefore NP ellipses are ungrammatical. Note that this is true also for non-coordinated constructions; in which the elided noun is not preceded by a pronounced one (223a; 224a). This model is further supported by the ungrammaticality of the examples containing other prepositions (e.g. para , por ‘for’) in both stSp and ABS (223b,c; 224b,c).

(223) stSp

a. *La [e] con gafas es alta.

The [e] with glasses is tall

‘The one with glasses is tall’.

294 b. *La [e] para Bolivia.

The [e] for Bolivia

‘The one for Bolivia’.

c. *La [e] por Bolivia.

The [e] for Bolivia

‘The one for Bolivia’.

(224) ABS a. La [e] cun gafas es alto.

The [e] with glasses is tall

‘The one with glasses is tall’.

b. *La [e] para Bolivia.

The [e] for Bolivia

‘The one for Bolivia’.

c. *La [e] por Bolivia.

The [e] for Bolivia

‘The one for Bolivia’.

295 7.5. Conclusions

This chapter has provided a cross-linguistic comparison between nominal elliptical phenomena in stSp and ABS. Data indicated that models developed for stSp, which ascribed the possibility of NP-ellipsis to its richness of agreement features and the presence of pro, cannot account for the ABS data, where pro is not found and the phi- feature specifications encountered in the DP elements are comparatively reduced.

In contrast with some analyses (Ticio 2003; 2005), the present proposal also attempts to avoid ad hoc post-syntactic movements and insertion operations. It agrees with other models (Kayne 1994; Braver 2009) in ascribing to stSp de and que a special status, namely the status of complenetizers heads, and based on the cross linguistic data, it extends such generalization to ABS di , que and cun . This proposal also accounts for the role played by gender and number features and thematic relationships in elliptical constructions.

296

CHAPTER 8: BARE NOUNS: SEMANTIC AND SYNTACTIC

PROPERTIES

“A ‘nominal expression’ is an argument

only if it is introduced by a category D”

(Longobardi 1994:625).

8.0. Introduction

In this chapter, I focus on two theoretical proposals concerning the semantics and the syntax of bare nouns in natural languages and their implications for the ABS data. In fact, the analysis of such hypotheses will provide a starting point to better understand how bare nouns behave in this specific Spanish dialect. More precisely, I will present

Chierchia’s (1998) Nominal Mapping Parameter and Longobardi’s (1994) proposal that bare nouns are not structurally bare but project a full DP, with an empty D head.

The explanation of these models will be followed by a presentation of the ABS data. Finally, a syntactic and semantic account for the specific patterns encountered in this dialect will be provided.

297 8.1. The Nominal Mapping Parameter

Chierchia’s (1998) Nominal Mapping Parameter attempts to provide an account for the distribution of bare nouns and full DPs cross-linguistically. His classification distinguishes between three language types with the following properties:

(225) The Nominal Mapping Parameter

a. [+arg, -pred] (e.g. Chinese/Japanese)

• generalized bare arguments

• every (lexical) noun is mass

• lack of plural morphology

• generalized classifier system

b. [-arg,+pred] (e.g. Romance Languages)

• count/mass distinction

• lack of bare NPs in argument position

• plural morphology

c. [+arg,+pred] (e.g.Germanic/Slavic Languages)

• count/mass distinction

• bare mass nouns and plurals found in argument position

• lack of bare singular count nouns

• plural morphology

298

d. [-arg,-pred] (non-existent)

According to this system, in languages like Chinese and Japanese, which possess a generalized classifier system and lack plural morphology, NPs are basically argumental.

Bare NPs in argument position are analyzed as names of kinds. These elements are not predicates; as a consequence, they cannot be quantified without the use of the operator ∪, which assigns to them a mass denotation. As a result, all Ns in [+arg –pred] languages will be mass nouns. For this reason, the singular/plural distinction is not found in these systems and an additional classifier is required to distinguish between count and mass.

On the other hand, in Romance languages, NPs are basically predicates; they are turned into arguments by an overtly pronounced D element. The semantic shift from

Predicate to Kind can take place only if this pre-requisite is satisfied in the grammar. A language of this type will also instantiate the count-mass distinction and therefore will present plural marking for count nouns.

Finally, the last possible group includes languages like those belonging to the

Germanic or Slavic families. They show NPs in both argument and predicate position.

These languages parallel Romance in that singular count nouns appear in predicate

position; at the same time, they resemble Chinese and Japanese in that mass and plural

count nouns can appear in argument position. In these systems, argument nouns can be

shifted to predicate position by recurring to ∪, which turns them into mass. On the other

hand, a predicate noun will be shifted to argument position by the “cap” operator ( ∩),

which turns it into a kind.

299 Mass nouns can be directly mapped into argument positions, in this case they are

[+arg]. Count nouns are directly mapped into [+pred]. Given that these languages distinguish between count and mass, they will also be endowed with plural/singular morphology. The ∩ type-shifter is not defined for singulars, so it follows that only plurals

(derived using ∩) and mass nouns (mapped directly) can act as bare arguments.

In Chierchia’s (1998) view, a non-overt type-shifter acting as a definite article

(iota ) is available for languages lacking a pronounced functional item of this kind. On the other hand, if a language has an overt determiner capable of performing such operation, iota will be blocked since it is a last-resort operation.

Chierchia’s proposal has been empirically challenged by Schmitt & Munn (2003) for Brazilian Portuguese, and by several other authors for different languages (e.g

Baptista (2000) for Cape Verdian Creole; Kester & Schmitt (2006) for Papiamentu;

Déprez (2001) for Haitian French). In fact, in these varieties bare singular nouns occur in argument position even though these languages are equipped with plural morphology and lack a generalized classifier system. Moreover, Chierchia’s proposal that the non-overt type shifting iota operator is available only for languages without a definite article does not account for the data in Cape Verdian Creole (Baptista 2000) and Brazilian Portuguese

(Shmitt and Munn 1999).

ABS is another language which does not fit the Nominal Mapping Parameter. As I will show (see section 8.3.), in this Afro-Hispanic vernacular bare singular nouns can appear both in subject and object positions, while definite articles, plural morphology and the count/mass distinction are present.

300 8.2. Longobardi’s proposal

As indicated in Chapter 5, Longobardi makes a distinction between bare common nouns, generic nouns and proper names. He observes that distributionally, at least in

Romance, generics and proper names do not display the same restricted lexically- governed positions as other types of bare nouns. Longobardi accounts for the distributional difference by arguing that generics and proper names can rise from N-to-D overtly or covertly. They can fill the D position. This operation does not take place with bare common nouns. This last type, in fact, is only allowed in a lexically governed position (object position). He suggests that these elements are not actually ‘bare’; rather, they are headed by an empty D, responsible for their existential, indefinite/generic reading. For this reason, in Longobardi’s view, “a ‘nominal expression’ is an argument only if it is introduced by a category D” (1994:620) (cf. Ishane 2008, for a different perspective on the issue).

As far as Italian is concerned, Longobardi (1994:623) shows that proper names are base generated in N and move to D (226, 227).

(226)

a. Il mio Gianni

‘The my Gianni’ (non-contrastive)

b. Il Gianni mio (contrastive)

‘The Gianni mine’

301 (227)

a. Gianni mio

‘Gianni mine’ (non-contrastive)

b. *Mio Gianni

‘My Gianni’

Prenominal and postnominal possessives play a crucial role in Longobardi’s analysis. Postnominal possessives have a contrastive interpretation, while prenominal possessives do not. In (228) and (229) these two different constructions are given.

(228)

DP

D NP

Il Spec N’

mio N

Gianni

(229)

DP

D NP

Il Spec N’

N AP

Gianni mio

302 As (227a) is not contrastive the possessive is supposed to be prenominal, like in

(228). The ungrammaticality of (227b) is due to the fact that in this construction N moves to D and should therefore appear before the possessive. When the possessive is not contrastive it must appear in SpecNP, in this way the co-appearance in Italian of article and possessive can be accounted for (230).

(230)

il mio libro

the my book

‘my book’

According to this view, arguments consist of a noun plus a determiner bearing a

[+Referential] feature. This referential D in turn has an uninterpretable N feature that can be either weak or strong. In languages like Italian, the referential D feature is strong so that, by the point of spell-out, D must be a sister to a constituent with an N feature. This feature is non-selectional; for this reason, it cannot be satisfied simply by merging the D with an NP. On the other hand, this feature can potentially get valued in two other ways

(Snyder 2007). One way for the strong N feature to be satisfied is through movement and adjunction of N to D; this is illustrated in (231). Another way to check off this feature is through the use of an expletive D. The expletive D in Italian takes the form of a definite article, and can bear an N feature (231).

303 (231)

(232)

In Italian empty determiners may occur at S-structure, but only under the following conditions:

(233)

a) They are restricted to morphologically plural (which are neural between the singular/plural distinction) or mass head nouns.

b) They are restricted to lexical government requirement (cannot occur in subject position).

c) They receive the indefinite/generic interpretation of an existential

quantifier (unspecified for number and taking the narrowest possible scope, default

existential).

304 The constraints on grammaticality reported in (233) can be easily verified by looking at the sentences reported in (234) and (235).

(234)

a. * Acqua viene giù dalle colline

water come down from-the hills

‘Water comes down from the hills’

b. Viene giù acqua dalle colline

come down water from-the hills

‘Water comes down from the hills’

d. Ho preso acqua dalla sorgente

have taken water from-the spring

‘I took water from the spring’

(235)

a. * In questo ufficio marocchini telefonano sempre

In this office Moroccans call always

‘In this office Moroccans are always calling’

b. In questo ufficio telefonano sempre marocchini

In this office call always Moroccans

‘In this office Moroccans are always calling’

c. In questo ufficio incontro sempre marocchini

In this office I meet always Moroccans

‘In this office I always meet Moroccans’

305 In sum, the main function of D is to pick out an instance of what is described by

NP. The role of D is to mediate between the description arising from the NP and whatever entity the description refers to. This would suggest that a noun can be argumental only if accompanied by a D, in all languages. Cross-linguistically, D can either be pronounced or silent, in Italian the silent determiner is allowed only if certain constraints are satisfied (cf. 233). In this language silent D is responsible for the existential, generic interpretation of bare-nouns in lexically-governed positions.

Chierchia’s and Longobardi’s proposals provided us with a theoretical background to understand how bare nouns have been treated in the literature. In the following sections I will present data from ABS which will act as a ‘testing ground’

(Kayne 1996) for the previous hypothesis.

8.3. Where to place ABS in the Nominal Mapping Parameter typology

ABS does not seem to fit in any of the four typological categorizations derived from the Nominal Mapping Parameter, (225a,b,c,d). In fact, definite articles, plural morphology and the count/mass distinction are present, while bare singular nouns can be encountered in both subject and object positions.

Let us start with an analysis of plural morphology. Number (plural/singular) is normally conveyed in ABS by the determiner heading a DP. This element can be inherently plural, namely lu (236a), or consist of a standard determiner marked with the inflectional plural marker –s (237).

306 (236) a. Lu buen amigo mayó

the-Pl good-Sg friend-Sg old-Sg

‘The old good friends’

b. El buen amigo mayó

the-Sg good-Sg friend-Sg old-Sg

‘The old good friend’

(237) a. Mis buen amigo mayó

My-Pl good-Sg friend-Sg old-Sg

‘My old good friends’

b. Mi buen amigo mayó

My-Sg good-Sg friend-Sg old-Sg

‘My old good friend’

c. Ejes buen amigo mayó

This-Pl good-Sg friend-Sg old-Sg

‘These old good friends’

d. Eje buen amigo mayó

This-Sg good-Sg friend-Sg old-Sg

‘This old good friend’

As can be observed, the nominal and the adjectival stems remain bare, so that plural marking appears to be lacking or to be not overt on these elements. Plural morphology exists, it clearly appears on possessives and demonstratives, and it is crucial

307 to determine whether we are referring to plural or singular entities (237a,b,c,d), (cf.

Delicado-Cantero & Sessarego 2009; Sessarego & Delicado-Cantero 2009). From this fact, it follows that the count/mass distinction is also present in ABS. To further confirm this point, we can take a look at (238a,b). In these cases the quantifier mucho does not

carry overt plural morphology.

(238) a. Mucho amigo viene a la fiesta

Much friend come to the party

‘Many friends come to the party’

b. Mucho agua hay que beber

Much water have to drink

‘Much water should be drunk’

While amigo in (238a) has a plural reading, in (238b) agua is understood as mass.

This is because the mass/count distinction is present. Note that this fact clearly shows that

ABS differs from languages like Chinese or Japanese, where a special classifier is

required to count nouns, which would be otherwise interpreted as mass. In example

(238a,b) the quantifier mucho remains unchanged; thus, the two different readings (mass

or count) of N have to be specified directly in the lexicon.

What makes ABS even more intriguing is the fact that bare singular nouns can

occur in subject and object position (239). As we will see, they can have a variety of

readings, as a function of their syntactic environment and pragmatic context.

308 (239) Perro come galleta

Dog eat cookie

‘Dogs eat cookies’

However, in contrast to other languages in which argumental plural bare nouns are found (e.g. Brazilian Portuguese (BP)), in ABS, as we have seen, nouns never inflect for number in the traditional dialect and therefore, they cannot carry plural morphology.

These data indicate that, again, contrary to Chierchia’s predictions, there are languages in which singular count bare nouns can be arguments while plural ones are not allowed.

Lastly, ABS is endowed with definite articles: a masculine singular ( el ), a feminine singular ( la ) and a plural article which does not inflect for gender ( lu ) (240).

(240)

a. El gato blanco

The male-cat white

‘The white cat’

b. La gata blanco

The female-cat white

‘The white cat’

c. Lu gato blanco

The male-cat white

‘The white cats’

309 e. Lu gata blanco

The female-cat white

‘The white cats’

In sum, ABS nominal domain possesses the following characteristics (241):

(241)

• singular bare nouns in argument position

• count/mass distinction

• morphological plural

• lack of a generalized classifier system

• presence of definite articles

The aforementioned features make ABS a language which does not fit into

Chierchia’s Nominal Parameter typology. At the same time, ABS also differs from BP, in that this Afro-Hispanic dialect does not allow bare plural nouns to be arguments.

In the next section, we will take a closer look to ABS bare singular nouns in order to understand which semantic and syntactic principles regulate their properties.

8.4. Bare singulars in ABS

Two main hypotheses have been suggested in the literature to account for bare nouns in argument position. Some scholars treated them as names of kinds (Carlson

1977; Chierchia 1998; Dayal 2002 and others), while others have indicated that they should be analyzed as systematically ambiguous between a kind and an indefinite reading

310 (Kratzer 1995; Diesing 1992 and others). In the present section, I will provide a list of examples suggesting that the first hypothesis is the most adequate to account for the data.

I will use the tests presented by Carlson (1977) to treat bare plurals in English as names of kinds and, more recently, re-adopted by Kester & Schmitt (2006) to make a similar point for Brazilian Portuguese and Papiamentu.

Carlson (1977) and Kester & Schmitt (2006) provided three main cues suggesting that the names-of-kinds analysis is on the right track: (i) bare plurals can be subjects of predicates applying to kinds only; (ii) their interpretation varies according to the context; and (iii) they do not take wide scope over negation, intensional verbs and aspectual adverbs 45 .

As we will see, bare singulars in ABS display the same behavior as bare plurals in

English and Portuguese as mentioned above. In (242) I show that bare singulars in ABS appear as subjects of predicates that can only apply to kinds.

(242)

a. Chancho es muy común a Tocaña

Wild-pig is very common in Tocaña

‘Wild-pigs are very common in Tocaña’

Moreover, bare singulars’ generic and existential readings are conditioned by the predicates appearing with the nouns. In fact (242) and (243) show generic readings in subject and object position, while (243) and (245) illustrate existential readings.

45 Many of the examples provided in Muller (2003) and Kester & Schmitt (2006) have been readapted and presented to our Afro-Bolivian informants. 311 (243)

a. A mí me gusta gato

To me to me like cat

‘I like cats’

(244) a. Tiene gallina en la casa

Have chicken in the house

‘There is a chicken/chickens in the house’

(245) a. Oté compró libro

You bought book

‘you bought a book/books’

In addition, when co-occurring with negation (246), intensional verbs (247) and durative adverbials (248), bare singulars allow only narrow scope readings. The distinction between bare singulars and the indefinite singular is clear when examples like

(248) and (249) are compared. While the bare singular allows a plural reading of iguana , the indefinite one does not. This contrast clearly indicates that bare nouns should not be treated as indefinites, contra Diesing (1992) and Kratzer (1995), among others.

312 (246) (neg < object and *object < neg) 46

a. Oté no vió mancha en la ventana

You no saw spot in the windows

‘You did not see the spot on the window’

(247) (want < object and *object < want)

a. Juana quiere ti casar con italiano

Juana want you marry with Italian

‘Juana wants you to marry an Italian’ (any Italian)

(248) (adv < object and *object < adv)

a. Yo mató iguana por dos hora

I killed iguana for two hour

‘I killed iguanas for to hours’

(249) (object < adv and adv < object)

a. Yo mató un iguana #por dos hora

I killed an iguana for two hour

‘#I killed an iguana for to hours’

Summarizing, this section showed that bare singulars in ABS are not specified for number (singular/plural) and can be interpreted as kinds. Indeed, we suggest that this is

46 Read Y

8.5. The semantics of indefinite and definite articles

This section provides a description of ABS article system and shows how overt and null determiners interact in marking the definiteness and indefiniteness of NPs as well as their specificity and non-specificity.

The article system of contact languages has generated much interest in creole linguistics, especially after Bickerton’s (1981) stipulations on how the Language

Bioprogram 47 would be responsible for shaping it in prototypical creoles.

According to Bickerton, creole languages have an article system with “a definite

article for presupposed-specific NP; an indefinite article for asserted-specific NP; and

zero for nonspecific NP” (1981:56). As we will see in more details, this description does

not capture the features of ABS article system. In this vernacular, in fact, there are three

overt definite articles ( el , la , lu ) and two indefinite ones ( un, unos ). Moreover, I claim

that null articles are present, but they are not restricted to non-specific NPs. The

distribution of articles in ABS resembles that of standard Spanish with the exception that

47 The language bioprogram hypothesis argues that the structural similarities between different creole languages cannot be solely attributed to their superstrate and substrate languages. Creolization occurs when the linguistic exposure of children in a community consists solely of a highly unstructured pidgin; these children use their innate language capacity to transform the pidgin, which characteristically has high syntactic variability into a language with a highly structured grammar. As this capacity is universal, the grammars of these new languages have many similarities. 314 bare nouns can take on plural/singular, definite/indefinite, specific/non-specific readings, given the appropriate syntactic and pragmatic environment.

8.5.1. Indefinite articles

I begin this section analyzing the distributional properties and uses of the indefinite un/unos . Indefinite articles in ABS can refer to both specific and non-specific

entities where, from a pragmatic point of view, an entity is specific if it is known by the

speaker by the hearer or by both (cf. Gutiérrez-Rexach 2004). Examples (250a,b) clarify

this distinction between specific and non-specific uses. In (250a) the speaker knows

which chocolate cake(s) he acquired yesterday; in (250b) he does not know exactly which

cake(s) he will buy tomorrow, he only knows that he will be looking for a/some chocolate

one(s).

(250)

a. Ayer yo compró un/unos tarta de chocolate

Yesterday I bought a/some cake of chocolate

‘Yesterday I bought a/some chocolate cake(s)’

b. Mañana yo va comprar un/unos tarta de chocolate

Tomorrow I go buy a/some cake of chocolate

‘Tomorrow I will buy a/some chocolate cake(s)’

The same specific/non-specific reading of indefinites can be given to the noun

when preceded by a null determiner (251a,b).

315 (251)

a. Ayer yo compró tarta de chocolate

Yesterday I bought cake of chocolate

‘Yesterday I bought a/some chocolate cake(s)’

b. Mañana yo va comprar tarta de chocolate

Tomorrow I go buy a/some cake of chocolate

‘Tomorrow I will buy a/some chocolate cake(s)’

The only distinction between the sentences in (250) and those in (251) is that in the first case the number (singular/plural) is conveyed by the overt determiner, while in the second set of sentences it is not. In fact, a strategy commonly used in ABS to distinguish between single and plural readings consists of recurring to indefinite articles, as illustrated below (252).

(252)

a. Oté tiene hijo? Sí, yo tiene hijo/un hijo

You have child? Yes, I have child/a child

‘Do you have children? Yes, I have children-a child/a child.’

In ABS the indefinite singular can also have wide and narrow scope and can be used in generic constructions. As (253) and (254) indicate, ABS indefinites can take both wide and narrow scope interpretations when co-occurring with negation and intensional verbs.

316 (253)

(un > ¬ and ¬ > un)

a. Yo no vió un mancha en el suelo.

I not see a spot on the floor

‘I didn’t see a spot on the floor.’

(want > un and un > want)

b. Juana quiere ti casar con un italiano

Juana want you marry with un Italian

‘Juana wants you to marry an Italian’

Moreover, indefinites in ABS can have a generic reading which, however, should not be confused with the kind-denoting reading. In fact, indefinites can appear with quantificational generics (254), but cannot occur with predicates that can only be true of a kind (255) or with one-event-only episodic predicates (256).

(254)

a. Un inglés habla inglés

an Englishman speaks English

‘An Englishman speaks English.’

317 (255)

a.* Un gallinazo/gallinazo ta en peligro de extinción

A Black-buzzard/ black buzzard is on-the-verge of extinction

Black buzzards are on the verge of extinction

(256)

a. * Un negro/negro votó por primera vez después Reforma Agraria

A black person/ black person voted for first time after Reform Land

‘Black people voted for the first time after the Land Reform’

Therefore, singular bare nouns can be kind-referring expressions while nouns preceded by indefinite articles cannot. As suggested by Müller (2003:78-81) for Brazilian

Portuguese, more differences and similarities regulate the interpretation of indefinite nouns and bare singular nouns. A similar scenario also seems to apply to ABS. In fact, when they are subjects of an episodic predicate, both indefinites and bare singulars, take on a generic reading. Nevertheless, in sentences like (257), the indefinite subject can also have a specific interpretation – “A certain Bolivian eats lechón today”; on the other hand,

such a reading is not available for the bare noun, which is always understood as generic –

“Every Bolivian eats lechón today”.

(257)

a. Un boliviano come lechón hoy.

A Bolivian eats lechon today

‘A Bolivian eats lechon today’

318 b. Boliviano come lechón hoy.

Bolivian eats lechon today

‘Bolivians eats lechon today’

Moreover, in (258) a generic reading obtains for both sentences; this effect is highly dependent on the essential properties of the referred nouns. In fact, a defining, essential property of ‘being a sonnet’ consists of ‘having four strophes’ (cf. Gutiérrez-

Rexach 2006 for a characterization of essential properties). On the other hand, in (259a), it is not possible to ascribe to ‘a romantic song’ the analytical or essential property of

‘being popular’, so that the sentence does not have a salient generic reading. The anomaly of the indefinite noun in (259a) is due to the fact that they are used to describe an analytic or characterizing situation, whereas the bare noun is grammatical in (259b) because it has a descriptive or inductive flavor.

(258)

a. Un soneto tiene cuatro estrofa

a sonnet has four strophes

‘A sonnet has four strophes’

b. Soneto tiene cuatro estrofa

sonnet has four strophes

‘Sonnets have four strophes’

319 (259)

a. #Un canción romántico es popular.

a song romantic is popular

‘A romantic song is popular’

b. Canción romántico es popular.

song romantic is popular

‘Love songs are popular’

In addition, in (260) and (261) bare nouns can only have a generic interpretation.

In fact, the generic reading does not only apply to bare nouns denoting usual classes

(260a); it is also available for bare nouns referring to very unusual ones (261a). In (260a) the class denoted by ‘famous actor’ is quite common and interpreted as generic. In (261a) the unusual class denoted by “Bolivian musician born on the fourth of August in

Cochabamba” is very uncommon but not for this reason is interpreted differently.

On the other hand, both the generic and the specific interpretations are available for the indefinite noun in (260b), while a specific reading is very salient, and strongly preferred for (261b).

(260)

a. Actor famoso gana mucho dinero.

actor famous earns much money

‘Famous actors make a lot of money’

320 b. Un actor famoso gana mucho dinero.

an actor famous earns much money

‘A famous actor makes a lot of money’

(261)

a. Músico boliviano nacido el 4 de agosto en Cochabamba toma harto cerveza

musician Bolivian born the 4 of August in Cochabamba drinks much beer

‘Bolivian musicians born on the fourth of August in Cochabamba drinks a lot of beer’

b. Un músico boliviano nacido el 4 de agosto en Cochabamba toma harto cerveza

a musician Bolivian born the 4 of August in Cochabamba drinks much beer

‘A Bolivian musician born on the fourth of August in Cochabamba drinks much beer’

In sum, both null and indefinite articles can express indefiniteness and +/- specificity; however, if uttered out of the blue, the former can only have a general reading, while the latter can take on either a general or a specific interpretation depending on the presence or absence of triggering presuppositions. In addition, bare nouns never indicate number information (singular/plural), while un/unos inherently convey this feature as a result of their morphological specification.

321 8.5.2. Definite articles

The following data present the properties of pronounced and null definite articles in ABS. Null articles in ABS cannot take on a definite reading if uttered out of the blue.

As I indicated in section 8.5.1, their default interpretation tends to be an indefinite non- specific one (253-259). Nevertheless, we observed that given the right context, bare singular nouns could take on also indefinite specific readings (250).

A similar effect is also encountered for definiteness. In fact, both definite nouns and bare nouns can be associated to an antecedent in the discourse via identity or a part- whole relation. Bare nouns therefore can indicate a [+definite] N. Examples of identity relation are presented in (262-265)

(262)

a. Yo compró un tarta

I bought a cake

‘I bought a cake.’

b. La tarta sabe a fresa

The cake taste like strawberry

‘The cake tastes like strawberry.’

(263)

a. Yo compró dos tarta

I bought two cake

‘I bought two cakes’

322

b. Lu tarta sabe a fresa

The cake taste like strawberry

‘The cakes taste like strawberry.’

(264)

a. Yo compró un tarta

I bought a cake

‘I bought a cake.’

b. Tarta sabe a fresa

cake taste like strawberry

‘The cake tastes like strawberry.’

(265)

a. Yo compró dos tarta.

I bought two cake

‘I bought two cakes’.

b. Tarta sabe a fresa.

cake taste like strawberry

‘The cakes taste like strawberry’.

323 As it is evident from looking at these examples, bare singulars can also refer back to previously mentioned entities; however, as they are not specified for number, they can have either a singular or a plural reading.

Examples (264) and (265) indicate that singular bare nouns can have a [+definite

+specific] reading when occurring in the right context. Such a reading is available for both definite articles and singular bare nouns also in the so-called bridging or associative contexts (Clark 1979; Asher & Lascardies 1998), where the association to an antecedent in the discourse obtains via a part-whole relation.

(266)

a. Yo compró un tarta

I bought a cake

‘I bought a cake’.

b. La crema sabe a fresa

The cream taste like strawberry

‘The cream tastes like strawberry’.

(267)

a. Yo compró un tarta

I bought a cake

‘I bought a cake’.

b. Crema sabe a fresa

cream taste like strawberry

‘ The cream tastes like strawberry’.

324 Definite determiners and bare nouns allow attributive and referential interpretations of the nominal description (Donnellan 1966). They differ, in that attributive readings of bare nouns are only possible when such an attribute is present in the common ground. In (268a,b) both a reading in which we know who the dean is, as well as a reading in which we are looking for the dean whoever he is, are possible. We can utter (268b) only in a scenario with accommodated attributes, for example, the

College’s main office at a particular university, not addressing a server in a bar unless the dean were a regular customer known by everybody working there.

(268)

a. Yo ta buscando el decano

I is looking for the dean

‘I am looking for the dean’.

b. Yo ta buscando decano

I is looking for dean

‘I am looking for the dean’.

The capability of inferring definiteness and specificity from the context in ABS

can also be observed in (269), where the uniqueness of the entities mentioned makes

them automatically [+definite +specific].

(269)

a. Va acabá mundo

Go finish world

‘The World is going to end’.

325 b. Sol ta muy juerte hoy

sun is very strong today

‘The sun is very strong today’.

Definiteness and specificity can also be derived by visual cues. In (270) if the

entities mentioned are pointed at by the speaker to the listener no definite article is

required, even though its presence would not generate an ungrammatical sentence.

(270) (uttered by a speaker pointing to the leaves/clouds)

a. Hoja ta seco

Cloud is dry

‘Leaves are dry’.

b. Nube ta bien rojo

Cloud be good red

‘Clouds are very red’.

In ABS, definite articles, differently from indefinite ones and similarly to bare singular nouns, can combine with kind predication (271) and with one-event only episodic predicates (272).

(271)

El gallinazo ta en peligro de extinción

the black buzzard is on-the-verge of extinction

Black buzzards are on the verge of extinction.

326 (272)

El negro votó por primera vez después Reforma Agraria

the black person voted for first time alter Reform Land

‘Black people voted for the first time after the Land Reform’.

Nevertheless, definite articles are different from bare nouns in that they cannot have wide scope over adverbs (273).

(273)

(adv < object and *object < adv)

Yo mató la iguana #por dos hora

I killed the iguana for two hour

‘I killed iguanas for to hours’.

In sum, this section shows that bare singular nouns can alternate with definite articles. Such an alternation is obviously not random. Bare nouns are allowed in the position in which a definite article would be found in standard Spanish only when their definiteness and/or specificity is clearly inferred from the common ground. Said interpretation can be made explicit through an anaphoric identity relation (265-266), part- whole relation (267), environmental cues (268), uniqueness presupposition (269), or visual deixis (270).

These examples also show that bare nouns are not subject to number constraints, differently from nouns headed by (in)definite articles. This is because the overt article inherently carries number features. However, it is important to highlight that such a lack

327 of specification for number does not imply a mass denotation for bare singular nouns. In the next section we will take a closer look at this distinction.

8.6. Number and mass interpretations

Below I present data indicating that bare singulars in ABS are unmarked for number but they are not necessarily denoting mass entities. When occurring in the scope of a time-span aspectual adverbial, a bare singular behaves as a plural, and allows an activity interpretation (Vendler 1967) of the event phrase:

(274)

Yo mató iguana/#un iguana/#la iguana por dos hora

I killed iguana/ an iguana / the iguana for two hour

‘I killed iguanas/# an iguana for to hours’.

Bare singulars have a plural reading when found as objects of verbs imposing a non-atomicity entailment on their internal argument (275). However, the singular interpretation is available if the verb does not impose such a constraint (276).

(275)

Pedro coleciona sello

Pedro collects stamp

‘Pedro collects stamps’.

328 On the other hand, when encountered with verbs imposing an atomicity entailment on its internal argument the singular reading is also possible. It is the case of the verb to marry (at least in monogamous societies) (276).

(276)

María quere ti casar con italiano

Maria wants you marry with Italian

‘Maria wants you to marry an Italian.’ (any Italian)

Discourse anaphora provides further evidence for the unmarkedness of bare singulars for number. The bare singular in ABS can be referred back to with either a plural or a singular pronoun, as illustrated in (277).

(277)

Yo tiene hijo. El/Eyu vive a Mururata

I have child. He/They live to Mururata

‘I have a child/children. He/They live in Mururata’.

The data indicate that bare singulars allow atomic and non-atomic readings and can be linked to singular or plural pronoun. In ABS mass terms can surface as bare arguments, with an existential reading as in (278a) or a generic reading, as in (278b):

(278)

a. Agua ta friu .

water is cold

‘The water is cold’

329 b . Oro ta caro .

Gold is expensive

‘Gold is expensive’.

Another piece of evidence indicating that bare nouns can act as mass but are also able to take on a count reading is provided by the occurrence of bare nouns with predicates that requires individuation. In fact, mass nouns are generally incapable of acting as subjects of predicates requiring individuation (279).

(279)

* Gold weighs two grams.

Such a constraint does not apply to bare singular count nouns in ABS:

(280)

Wawa pesa 20 kilos cun eje edad

child weighs 20 kilos in-this age

‘Children weigh 20 kilos at this age.’

Similar phenomena have also been reported for Papiamentu and Brazilian

Portuguese (cf. Kester & Schmitt (2006)).

To summarize, the data above show that bare singulars allow mass and count readings. With respect to the latter, they can be interpreted as atomic or non-atomic individuals, and they can be linked to singular or plural pronouns. The next section will

330 provide a syntactic account of the distribution of bare nouns in ABS along the lines proposed by Longobardi (1994).

8.7. The syntax of bare nouns

When compared to the examples provided by Longobardi (1994) (cf. 271-280), the distributional patterns found in ABS are different, but parallel to the Italian data in many ways. In fact, also in this language proper names can co-occur with a definite article (281a) or appear bare (281b), thus suggesting that they can value strong N features on D.

(281)

a. El Juan come pasta

the Juan eat pasta

‘Juan eats pasta’

b. Juan come pasta

Juan eat pasta

‘Juan eats pasta’

However, what distinguish ABS from Italian (and standard Spanish) is that not

only bare proper nouns can be encountered in argument position, and common ones can

be found when subjects to the (233a,b,c) constraints. In this language constraint (233b)

seems not to apply, so that bare common nouns can appear without being preceded by a

(pronounced) D in subject position.

331 (282)

a. Lu perro come pasta

the dog eat pasta

‘Dogs eat pasta’

b. Perro come pasta

dog eat pasta

‘Dogs eat pasta’

Nevertheless, it is important to notice that their interpretation will always be subject to conditions (233a,c), namely: a default indefinite/generic reading either as mass or non-specified for number. Moreover, arguing N-to-D raising is more difficult for ABS, at least for cases where pre-nominal adjectives can co-occur with bare nouns, indicating that N could not move to D (283-284). For such cases, it seems more appropriate to postulate an empty D category.

(283)

a. El viejo Juan come pasta

‘the old Juan eat pasta’ (non-contrastive)

b. Viejo Juan come pasta

‘old Juan eat pasta’ (non-contrastive)

c. El Juan viejo come pasta

‘the Juan old eat pasta’ (contrastive)

d. Juan viejo come pasta

‘Juan old eat pasta’(contrastive)

332 (284)

a. El viejo perro come pasta

‘the old dog eat pasta’ (non-contrastive)

b. Viejo perro come pasta

‘old dog/dogs eat pasta’ (non-contrastive)

c. El perro viejo come pasta

‘the dog old eat pasta’ (contrastive)

d. Perro viejo come pasta

‘dog/dogs old eat pasta’(contrastive)

I would like to argue that three elements in ABS appear to be able to check the N feature on D: proper names (285), pronounced articles (286-287) and silent Ds (288).

(285)

(286)

333 (287)

(288)

As far as common nouns in argument position are concerned, by comparing

examples (283) and (284), at a first look, we may intuitively assume that the underlying

structures of such constructions are identical. Nevertheless, I analyze common nouns as

non-referential, and therefore unable to check off the N feature in D, (differently from

proper names).

Two DP variants for common nouns seem to be available in ABS. On one hand,

we find DPs endowed with Number Projection (NumP), which can be counted and

manifest their number pre-nominally (289, 290); on the other hand we have DPs lacking

such a projection, bare singular nouns (291)48 . They tend to have an out-of-the-blue default interpretation which is unspecified for number. Nevertheless, these DPs can take on a plural or singular number reading according to the context in which they are uttered.

48 Here, for a matter of visual clarity, I am not graphically representing D and NP branching, as well as N features. Such elements are however supposed to form part of the DP structure. 334 (289)

(290)

(291)

8.8. Bare Nouns in ABS: A Unified Account

In this section, I provide a theoretical account of the attested patterns. First, it seems clear that the Nominal Mapping Parameter (Chierchia 1998) does not explain adequately the ABS data. As observed, ABS does not fit into Chierchia’s framework; in this language, in fact, we can encounter single bare nouns in argument position as well as plural morphology and the count/mass distinction.

335 Following Longobardi (1994), we can assume that bare nouns are not structurally bare but do, in fact, project a full DP, with an empty D head (cf. also Contreras 1986).

Only DPs can serve as arguments, NPs are predicates turned into arguments by D.

Given this theoretical background, I postulate that the null determiner plays a semantic role in the structure. It is the locus of semantic type-shift operations. Given that common nouns are of type >, i.e., denote properties, the nominalization operation

(nom ∩) shifts the nominal into an individual (Chierchia 1998).

I propose that covert determiners in ABS encode a variety of functions, not only nom. Covert type-shifting operations are last resort operations. If DPs are not canonically mapped into predicates, their interpretation as kinds or quantificational elements will be the result of last resort operations to fix a type mismatch (cf. also Kester & Schmitt

(2006)).

Moreover, contrary to what has been claimed for Papiamentu and Brazilian

Portuguese (Kester & Schmitt 2006), the Derived Kind Predication (DKP) does not apply as the only means to obtain object readings of bare nouns in ABS. The prediction of DKP is that bare existentials always take scope under other operators (Dayal 2002).

Nevertheless, (i) bare singulars in ABS not only allow narrow scope readings; (ii) they can also be interpreted as definites or indefinites with wide scope.

I analyze ABS bare singular nouns as lacking number specification. Number and quantificational force are encoded at the D level. In the case of bare nouns, a covert determiner encodes a variety of type-shifting functions ( à la Partee):

1. Nom: For the shift of predicates to kind readings in the default case.

336 2. Existential or definite ( iota ) operators: For existential and definite-like interpretations. In the case of the former (existential), pure existential readings are favored; for the latter (definite), familiarity presuppositions have to be satisfied in the common ground.

The present analysis provides support to hypotheses suggesting the presence of empty Ds in Romance (Contreras 1986; Longobardi 1994); in doing this, this work rejects positions arguing in favor of the lack of empty categories in DPs presenting non- overt determiners (Eguren 2010). However, while Contreras (1986) and Longobardi

(1994) indicated that empty Ds must be lexically governed in Spanish and Italian respectively, my data suggest that in ABS such a requirement does not have to be satisfied to generate grammatical constructions.

8.9. Conclusions

In this chapter, I have provided an analysis of the nature of bare nouns in ABS. I have also compared pronounced definite and indefinite articles with the properties of their respective null realizations.

Similarly to English (Carlson 1977) and Brazilian Portuguese (Kester & Schmitt

2006), in ABS bare nouns are better analyzed as names of kinds, contra approaches which analyzed them as systematically ambiguous between kinds and indefinites (Kratzer

1995; Diesing 1992).

Also, it has been shown that Chierchia’s (1998) Nominal Mapping Parameter fails to account for the nominal phenomena encountered in ABS. A different approach has been suggested here to analyze the realizations of pronounced and silent determiners in

337 this Afro-Hispanic dialect. My argument is in line with Longobardi’s (1994) proposal that NPs are predicates while DPs are arguments. From this it follows that a non- pronounced D element must be postulated to account for ABS bare nouns in [+arg] position. Bare nouns in ABS seem to obey certain specific syntactic and pragmatic constraints, which ultimately determine the semantic interpretation of their unpronounced

D-categories.

338

CHAPTER 9: CONCLUSIONS

Stairs are climbed step by step. -Turkish proverb-

Microparametric syntax is a growing field of research, which can be employed to test formal theories (Barbiers & Cornips 2001). The field of syntactic microvariation has grown significantly during the last decade and particular attention has been paid to the study of microparametric variation across Italian, Dutch and English dialects (see for example Poletto

2000; Benincà 1989; Adger and Smith 2005; Barbiers and Cornips 2001). Unfortunately, so far, not many works of this kind have been conducted on Spanish; in particular, Afro-

Hispanic contact varieties have never been studied by adopting this approach. This opens up a new field of investigation, which, if addressed in a methodological and systematic way, could lead to interesting discoveries. The main purpose of this dissertation is to place the first stone in building such a research program.

From a theoretical point of view, Afro-Hispanic contact varieties represent an important group of Spanish dialects. In fact, these languages are rich in constructions which would be considered ungrammatical in standard Spanish (stSp), but which form the core grammar of these less-prestigious, but equally efficient, syntactic systems. Such differences can provide researchers with a great linguistic laboratory (Kayne 2000) where generative

339 hypotheses, usually developed on the intuitions of standard language speakers, can be tested and evaluated against a different, but closely related, set of linguistic data.

The majority of the dialects which developed in Latin America from the contact of

African languages and Spanish at the time of slavery, are not “radical creoles”, languages highly influenced by substrate patterns which would be unintelligible for a stSp speaker; rather, they often consist of comprehensible vernaculars with a comparatively reduced inflectional morphology. Afro-Bolivian Spanish (ABS) is one of these languages. By virtue of its close relatedness to stSp, ABS appeared to me as a variety offering a great opportunity to embark on a microparametric study, the first one on Afro-Hispanic contact languages. In doing this, the present dissertation adopted a novel approach to account for language variability. In fact, it combined sociolinguistic methodology with formal analyses in line with recent micro-parametric studies (see for example Cornips and Poletto 2005).

In order to build more robust, fine-grained generalizations, data were analyzed by comparing grammaticality judgments to the syntactic structures extracted from sociolinguistic interviewes. Nevertheless, my data were collected with consideration to the fact that not all linguistic phenomena can be studied by using the same methodology. In fact, while syntactic phenomena occurring with high frequency in natural speech can efficiently be recorded employing sociolinguistic interviews; less frequent constructions are better collected by asking informants for grammaticality judgments. Therefore, while a comparison between natural data and inducted elicitations has been a prerogative in the study of Afro-

Bolivian Spanish constructions, this dissertation also considered necessary methodological adjustments imposed by the nature of the different phenomena under analysis.

340 In this work, my attention was primarily directed at two different objectives. The first goal was to shed light on the unclear origin of ABS by analyzing the available socio- historical data as well as the linguistic evidence found in this language. The second goal was to explore certain aspects of the Afro-Bolivian Spanish (ABS) Determiner Phrase (DP) to provide a testing ground for verifying the feasibility of current linguistic hypotheses and, when appropriate, to propose new solutions in light of the empirical data collected.

As far as the origin of ABS is concerned, my findings indicate that ABS is a language which approximated Spanish from its inception and which probably did not develop from the nativization of an earlier pidgin, as suggested by Lipski (2009). In fact, several factors have affected the dimension of African slavery in Bolivia and consequently the presence of black population in the territory from the 16 th century through the middle of the 20 th century. We observed that the Spanish Crown’s monopoly of slave trading, the geographic location of

Bolivia and the availability of native workforce affected the cost of Africans, raising their price and, as a result, reducing the number and the dimension of slave transactions. The non- massive introduction of a black workforce into the territory favored the acquisition of a closer approximation to Spanish by the slaves. Moreover, the Yungan hacienda was not a plantation society of the kind found elsewhere in the Americas; it was characterized by low

Black/White ratio and relatively high social mobility. Linguistic findings suggest that the grammatical elements of ABS, which had been invoked in the literature as potential creole indicators, can also be ascribed to a conventionalized advanced/intermediate second language, which probably crystallized in these rural valleys and did not undergo processes of standardization imposed elsewhere by urban society and the linguistic norm.

341 As far as my second goal is concerned, the close proximity of ABS to stSp provided us with a great “testing lab” for syntactic theories. In particular, the last chapters of this dissertation focused on phi-features, on NP ellipsis and on bare nouns.

The DP structure adopted is the one put forward by Carstens (2000), where the loci of

interpretation for person, number and gender are D °, Num ° and N °, respectively. However, differently from Carstens (2000) and contra Chomsky (2000; 2001), I adopted a feature- sharing view of Agree (Frampton and Gutmann 2000) and I argued in favor of the elimination of the Valuation/Interpretability Biconditional principle (Pesetsky and Torrego

2007). Therefore, in line with Pesetsky and Torrego (2007), I assumed features to come from the lexicon in four sorts: uF [ val ] uninterpretable, valued; iF [ val ] interpretable, valued; uF [ ]

uninterpretable, unvalued; iF [ ] interpretable, unvalued. Moreover, to account for the ABS

“DP impoverished agreement” (Lipski 2006), I postulated the existence in this language of elements coming from the lexicon without a specification for a certain feature, No-F [].

In order to account for the agreement patterns encountered in stSp, where gender and

number are marked redundantly across all the DP elements, I proposed that stSp nouns come

from the lexicon with a set of valued phi-features, of which only gender is interpretable. Num

carries an unvalued interpretable instance of the number feature, while D elements are

endowed with an unvalued interpretable instance of the person one. Such a configuration

explains the stSp data; nevertheless, this model, given the c-command restriction on Agree

and the valued number feature on N, cannot account for the ABS agreement patterns. In fact,

all stSp demonstratives, quantifiers, nouns and articles come from the lexicon with a

specification for number and gender features. Such specification is what will result in overt

number and gender morphological marking after all the Agree operations have applied. On

342 the other hand, traditional ABS does not posses the richness in feature specification characteristic of stSp and other Romance languages. In traditional ABS, nouns are specified for gender, but this feature is not morphologically marked on the majority of the DP elements

(it only appears on definite articles). Also, the morphological distribution of number marking is much more restricted, it is limited to determiners, and it never applies to adjectives, nouns and quantifiers. For this reason, I suggested that in traditional ABS nouns do not carry number features at all, while Num enters the derivation with a valued and interpretable number specification. These parametric differences between ABS and stSp shed some light on currents debates concerning the role of Agree in languages. Namely, that ABS’s poverty of feature specifications does not prevent this language from presenting the same adjective+noun and noun+adjective order combinations encountered in Romance. In fact, as

Carstens (2001:154) and Alexiadou (2001:223), among others, have demonstrated, raising of

N to Num is not prompted by number feature checking, but rather by other mechanisms such as EPP or categorial features. This indicates, in line with Atkinson’s (2002) observation, that agreement at least in these clear examples, cannot feed movement.

After having provided a layout of the phi-feature distribution across the DP in traditional ABS and in stSp, we noticed that the data collected through means of grammaticality judgments and sociolinguistic interviews indicated several different agreement patterns. The study of language variability seems at first to be at odds with formal linguistic analyses that strive to hypothesize principles and generalizations based on

Universal Grammar allowing a very limited range of inter-speaker variation (Chomsky 1957,

1965). Nevertheless, recent minimalist research entailed a derivational approach that is inconsistent with parameter-based accounts. Thus, an extension of the parameter/micro-

343 parameter idea to intra-speaker variation seems to not be a straightforward matter (Adger &

Smith 2005). In the most recent formulations of the Minimalist Program (Chomsky 2001;

2002; 2006), syntactic derivations are viewed as strictly dependent on feature valuation and checking. The distinction between interpretable and non-interpretable features, though controversial, has proven very useful. Several features have an interpretation at LF, thus they are semantically-interpretable features. Other features, however, lack such semantic import and are present to trigger the necessary merger or agreement operations during the derivation.

Said uninterpretable features have to be matched via Agree and are finally deleted before

Spell-Out. Given these theoretical assumptions, Adger and Smith (2005) argue for characterizing syntactic variation in terms of (un)interpretable features. Certain uninterpretable features may be present in one category but absent in another. Since they are uninterpretable, they would have no semantic repercussion, thus being equally legitimate for a convergent derivation.

In this way, variation is reduced to the specification of the uninterpretable features in a derivation (Adger & Smith 2005: 161). As expected, syntax per se remains invariable or

“perfect” (Brody 1997), given that variation is located only in the lexical component.

Variation will occur when one item or other enters the numeration and takes part in a syntactic derivation. Therefore, the variable gender and number agreement patterns encountered in my fieldwork can be formally accounted for by postulating variable feature specification across the ABS DP.

As far as NP-ellipses are concerned, some important parametric differences and similarities could be encountered between stSp and ABS. In fact, comparison of the ABS and the stSp data indicated that even if ABS is not inflectionally rich and does not have pro , it

344 allows all the elliptical configurations encountered in stSp. This suggests that, as pointed out by Ticio (2003; 2005), the relation between nominal ellipsis and the null subject parameter is problematic (contra Torrego 1988); not only ABS data point to this conclusion, other languages, such as French and German, present NP-ellipsis but do not have null subjects. The examples provided also indicated that, in ABS as well as in stSp, gender, differently from number, is a feature of the noun, which gets deleted in the process of elision. This fact strongly undermines approaches which postulate a unique projection for number and gender

(e.g. Ritter 1991), as well as those that argue in favor of two separate projections NumP and a

GenP (e.g. Picallo 1991; Bernestein 1993). In fact, masculine and feminine nouns do not seem to be derivable from the same lexical entry. Data, on the other hand, back a framework in which lexical entries are clearly different in gender specifications before entering the syntactic numeration. Therefore, only NumP is a licit projection while ‘gender’ represents an intrepetable feature of N. Moreover, in this Afro-Hispanic dialect, nouns can often be elided also in contexts for which an ungrammatical construction would obtain in the standard language, namely when the elided noun is followed by cun ‘with’. A curious peculiarity, which might explain why cun -PP can licitly survive nominal ellipsis in this vernacular, is that

ABS cun is often used where de would be preferred in stSp. As all elliptical patterns are the same in ABS and stSp with the exception of cun/con constructions, I assumed that this

distinction must have to do with such prepositions. This statement is justified as con/cun

presents different grammatical behaviors in the two languages.

In contrast with some previous analyses (Ticio 2003; 2005), the model I proposed

attempted to avoid ad hoc post-syntactic movements and insertion operations. It agrees with

other proposals (Kayne 1994; Braver 2009) in ascribing a special status to stSp de and que ,

345 namely the status of complenetizer heads, and based on the cross linguistic data, it extends such a generalization to ABS di , que and cun . This proposal also accounts for the role played

by gender and number features and thematic relationships in elliptical constructions, issues

which had not been completely analyzed in previous works (Brucart and Gràcia 1986;

Brucart 1987; Torrego 1988; Eguren 2010; etc.).

Finally, the study of ABS bare nouns revealed that Chierchia’s (1988) Nominal

Mapping Parameter does not hold. In fact, contrary to what is predicted by Chierchia, in this

Afro-Hispanic vernacular bare singular nouns can appear both in subject and object

positions, while definite articles, plural morphology and the count/mass distinction are

present. Following Longobardi (1994) (contra Eguren 2010), I suggested that bare nouns are

not structurally bare but do, in fact, project a full DP, with an empty D head (cf. also

Contreras 1986). Only DPs can serve as arguments, NPs are predicates turned into arguments

by D. From this it follows that a non-pronounced D element must be postulated to account

for ABS bare nouns in [+arg] position. In my opinion, this determiner plays a semantic role

in the structure. It is the locus of semantic type-shift operations. Given that common nouns

are of the type >, i.e., denote properties, the nominalization function shifts the

nominal into an individual via a nominalization operation (nom ∩) (Chierchia 1998). I

argued that covert determiners in ABS encode a variety of functions, à la Partee. Covert type-shifting operations are last resort operations. If DPs are not canonically mapped into predicates, their interpretation as kinds or quantificational elements will be the result of last resort operations to fix a type mismatch (cf. also Kester & Schmitt (2006)). Similarly to

English (Carlson 1977) and Brazilian Portuguese (Kester & Schmitt 2006), in ABS bare nouns are better analyzed as names of kinds, contra approaches which analyzed them as

346 systematically ambiguous between kinds and indefinites (Kratzer 1995; Diesing 1992).

Moreover, ABS bare nouns seem to obey certain specific syntactic and pragmatic constraints, which ultimately determine the semantic interpretation of their unpronounced D-categories.

This dissertation explored some aspects of the syntax of the ABS DP, and, in doing so it adopted a novel approach to obtain more fine-grained generalizations on the structure of this dialect. The present study represents the first microparametric work on an Afro-Hispanic contact variety. Afro-Hispanic dialects offer a great opportunity for microparametric studies, there is plenty to do for those who are willing to take on this challenge.

347

REFERENCES

Abney, S. (1987). The Noun Phrase in Its Sentential Aspect. MIT Ph. D. Dissertation.

Cambridge.

Aboh, E.O. (1998). On the syntax of Gungbe noun phrase. Paper presented at the annual

meeting of the African Language and Linguistics Colloquium (27 th , Leiden, Netherlands).

2-57.

Aboh, E.O., N. Corver, M. Dyakonova and M. van Koppen (2009). DP-internal information

structure: Some introductory remarks. Lingua : http://www.sciencedirect.com/science .

Adger, D. & Hrabour D. (2008). Why Phi? In Harbour, D. Adger, D. and Béjar, S (eds.). Phi

Theory , 1-34.

Adger, D. (2006). Combinatorial Variability. In: Journal of Linguistics 42 : 503–530.

Adger, D. and G. Trousdale (2007). Variation in English syntax: theoretical implications. In

English Language and Linguistics , 11:2:261-278. Cambridge University Press.

Adger, D. and J. Smith (2005). Variation and the minimalist program. In Syntax and Variation.

Reconciling the Biological and the Social, Leonie Cornips and Karen P. Corrigan (eds),

149–178. Amsterdam: Benjamins.

Alacros-Llorach, E. (1973). Estudios de gramática functional del español. Madrid, Gredos

(BRH, Estudios y ensayos, 147).

348 Alexiadou, A. (2001). Adjective syntax and noun raising: Word order asymmetries in the DP

as the result of adjective distribution. Studia Linguistica 55.3: 217-248.

Alexiadou, A., L. Haegeman and M. Stavrou (2007). Noun Phrases in Generative Grammar.

Berlin: Mouton de Gruyter.

Alleyne, M. (1980). Comparative Afro-American: An Historical-Comparative Study of

English-based Afro-American Dialects of the New World . Ann Arbor: Karoma.

Álvarez, A., and E. Obediente (1998). El español caribeño: Antecedentes sociohistóricos y

lingüísticos. In Matthias Perl &Armin Schwegler (eds.), América negra: Panorámica

actual de los estudios lingüísticos sobre variedades hispanas, portuguesas y criollas , 40–

61. Madrid: Iberoamericana.

Andersen, R. (1980). Creolization as the acquisition of a second language as a first language.

In A.Valdmann, & A. Highfield (Eds.), Theoretical orientations in creole studies. 273–

295. San Diego: Academic Press.

Andersen, R. (ed.). (1983). Pidginization and creolization as language acquisition . Rowley:

Newbury House.

Anderson, S. R. (1986). Disjunctive Ordering in Inflectional Morphology. Natural Language

and Linguistic Theory, 4:1–31.

Angola Maconde, J. (MS): Unpublished manuscripts. La Paz, Bolivia.

Aoun, J. (1985). A grammar of anaphora , MIT Press, Cambridge MA.

Aronoff, M. (1976). Word Formation in Generative Grammar. Linguistic Inquiry Monograph

No. 1. Cambridge: MIT Press.

Asher, N. and A. Lascarides (1998) Bridging, Journal of Semantics , 15(1), 83-113, Oxford

University Press.

349 Atkinson, M. (1992). Children’s Syntax: An introduction to principles and parameters.

Oxford: Blackwell.

Auger, J. (1998). Le redoublement des sujets en francais informel quebecois: Une approche

variationiste. Canadian Journal of Linguistics , 43 (1), 37-63.

Baker, P. (1990) Off target?, Journal of Pidgin and Creole Languages 5, 107-119.

Baker, P. (1994) Creativity in creole genesis. In Dany Adone, & Ingo Plag (Eds.), Creolization

and language change. 65–84. Tübingen: Niemeyer.

Baptista, M. (2000). Variabilité référentielle des syntagmes nominaux nus en créole cap-

verdien. Paper presented at the University of Paris III Linguistic Symposium . April 7th,

2000.

Barbiers, S. (2005). Word order variation in three verb clusters and the division of

laborbetween generative linguistics and sociolinguistics. In Syntax and variation:

reconciling the biological and the social . Amsterdam/Philadelphia: John Benjamin 233-

264.

Barbiers, S. (2009). Locus and limits of syntactic microvariation. Lingua

119, (11): 1607-1623

Barbiers, S. and Cornips, L. (2001). Introduction to syntactic microvariation. In: Barbiers, S.,

Cornips, L., van der Kleij, S. (Eds.), Syntactic Microvariation. Electronic publication of

the Meertens Instituut. http:// www.meertens.knaw.nl/projecten/sand/synmic/.

Béjar, S. (2008). Conditions on Phi-Agree. In Harbour, D., Adger, D and Béjar, S. (eds), Phi-

Theory. Oxford: Oxford University Press.

Belletti, A. (1978). Strutture coordinate e possessive . Rivista di Grammatica Generativa , 3, 1.

350 Belletti, A. (ed.) (2004) Structures and Beyond: The Cartography of Syntactic Structures,

Vol.3, Oxford University Press, New York.

Belletti, A. and L. Rizzi (1988). Psych-verbs and Θ-theory. NLLT 6:291-352.

Bello, A. (1847). Gramatica de la lengua castellana destinad a al uso de los americanos.

Santiago de Chile: Imprenta de Progreso

Benincà, P. (1994). La variazione sintattica. Studi di dialettologia romanza . Bologna: Il

Mulino.

Beninclà, P. (1989). Dialect Variation and the Theory of Grammar . Dordrecht: Foris.

Bernstein, J. (1993). The Syntactic Role of Word Markers in Null Nominal Constructions,

Probus 5, 5-38.

Bickerton, D. (1981). Roots of Language . Ann Arbor: Karoma.

Bickerton, D. and A. Escalante (1970). Palenquero: A Spanish-based creole of Northern

Colombia. Lingua 13:63-92.

Blackburn, R. (1997). The making of new world slavery. London: Verso

Boskovic, Z. (1997). The Syntax of Nonfinite Complementation. An Economy Approach .

Linguistic Inquiry Monograph 32. Cambridge, Mass.: MIT Press.

Bosque, I. & M. C. Picallo. 1996. Postnominal Adjectives in Spanish DPs. Journal of

Linguistics 32: 349-385.

Braver, A. (2009). Restrictions on DP-Internal Ellipsis in Spanish. In Proceedings of the

Hispanic linguistic Symposium 2009 (HLS2009). San Juan: Universidad de Puerto Rico.

Bridikhina, E. (1995a). El tráfico de esclavos negros a La Paz a fines del siglo XVIII. Estudios

Bolivianos 1:183-191.

351 Bridikhina, E. (1995b). La Mujer Negra en Bolivia . La Paz, Bolivia: Ministerio de Desarrollo

Humano.

Brockington, L. (2006). Blacks, Indians, and Spaniards in the Eastern Andes. Lincoln &

London : University of Nebraska Press.

Brody, M. (1997). Perfect chains. In L. Haegeman (ed.): Elements of grammar , 139-167.

Dordrecht: Kluwer Academic Publishers.

Brody, M. (2003). Lexico-logical form . Cambridge, Mass: MIT Press.

Browser, F. (1974). The African slave in colonial Peru 1524-1650 . Stanford: Stanford

University Press.

Brucart, J. M. (1999). La elipsis. In Bosque, I. & Demonte, V. Gramática descriptiva de la

lengua española . Vol. II. Espasa, Madrid. 2787-2866 .

Brucart, J. M. and M. L. Gràcia (1986). ‘I Sintagmi Nominali Senza Testa’, Rivista di

Grammatica Generativa 11: 3-32.

Brucart, J.M. (1987). La Elisión Sintática en Español. Dissertation. Universitat Autònoma de

Barcelona.

Bruhn de Garavito, J. and L. White (2000). L2 acquisition of Spanish DPs: the status of

grammatical features. In Howell, C., Fish, S. and Keith-Lucas, T., editors, Proceedings of

the 24th Annual Boston University Conference on Language Development . Somerville,

MA: Cascadilla Press. 164-75.

Bruyn, A. (1996). On identifying instances of grammaticalization in Creole languages. In P.

Baker &A. Syea (Eds.), Changing meanings, changing functions. Papers relating to

grammaticalization in contact languages (29-46). London: University of Westminster

Press.

352 Busdiecker, S. (2006). We are Bolivians too: The experience and meaning of blackness in

Bolivia. PhD Dissertation. University of Michigan.

Bybee, J. (1985). Morphology: A Study of the Relation between Meaning and Form .

Amsterdam/Philadelphia: John Benjamin.

Cabrera, L. (1983). El monte . 3 rd ed. Miami:Editorial C.R.

Cardinaletti, A. and G. Giusti (1991), “Partitive ne and the QP-Hypothesis. A Case Study”,

University of Venice Working Papers in Linguistics , Venice.

Carlson, G. (1977). Reference to kinds in English . Doctoral dissertation, University of

Massachusetts, Amherst.

Carstens, V. (1997). Empty nouns in Bantu locatives. The Linguistic Review 14: 361-410.

Carstens, V. (2000). Concord in Minimalist Theory. Linguistic Inquiry 31.2: 319-355.

Carstens, V. (2001). Multiple Agreement and Case-Deletion: Against Φ-(In)Completeness.

Syntax 4: 147-163

Cedergren, H. J. and D. Sankoff (1974). Variable rules: performance as a statistical reflection

of competence. Language . 50(2): 333-55.

Chambers, J.K. (2003). Sociolinguistic Theory: Linguistic Variation and its social

Implications. Oxford, UK/Malden, US: Blackwell.

Chambers, J.K. (2004). Dinamic typology and vernacular universals. In Kortmann B. (ed.)

Dialectology meets typology: dialect grammar from a cross linguistic perspective . 127-

145

Chaudenson, R. (1979). Les creoles francais. Evreux: Nathan.

Chaudenson, R. (1992). Des iles, des homes, des langues. Paris: L’Harmattan.

Chaudenson, R. (2001). Creolization of language and culture . London: Routledge.

353 Chierchia, G. (1998). Reference to kinds across languages. Natural Language Semantics 6:

339-405.

Chomsky, N. & M. Halle (1968) The Sound Pattern of English . New York: Harper and Row.

Chomsky, N. (1957). Syntactic Structures . The Hague: Mouton. Berlin and New York.

Chomsky, N. (1962). Discussion. In Archibald Hill (Ed.), Third Texas Conference on

Problems of Linguistic Analysis in English (pp. 124–169). Austin: University of Texas.

Chomsky, N. (1965). Aspects of the Theory of Syntax . Cambridge: The MIT Press.

Chomsky, N. (1981). Lectures on Government and Binding . Dordrecht: Foris.

Chomsky, N. (1982). Some Concepts and Consequencesof the Theory of Government and

Binding . Cambridge, MA: MIT Press.

Chomsky, N. (1986a) Knowledge of Language , Praeger.

Chomsky, N. (1986b) Barriers . Cambridge, MA: MIT Press.

Chomsky, N. (1990). Some notes on economy and derivation . MIT Working Papers.

Chomsky, N. (1992) The minimalist program for linguistic theory. MIT Occasional Papers in

Linguistic 1, 1-62.

Chomsky, N. (1995) The Minimalist Program . Cambridge, Mass.: MIT Press.

Chomsky, N. (1998). Minimalist inquiries: The framework. MIT Occasional Papers in

Linguistics , 15.

Chomsky, N. (2000). Minimalist inquiries: The framework. In Step by Step . Ed. Roger Martin,

David Michaels, and Juan Uriaguereka. 89-155. Cambridge, Mass.:MIT Press.

Chomsky, N. (2001). Derivation by phase. Ken Hale: A life in linguistics, M. Kenstowicz (ed.),

1-52. Cambridge, Mass.: MIT Press.

354 Chomsky, N. (2002). Beyond explanatory Adequacy. MIT Occasional Papers in Linguistics ,

No.20 Department of Linguistics and Philiosophy, MIT.

Chomsky, N. (2006). Language and Mind. (3 rd edition) Cambridge: Mass.: MIT Press.

Chomsky, N. and H. Lasnik. (1993). The theory of principles and parameters. In J. Jacobs et al.

(eds.) Syntax: An International Handbook of Contemporary Research , 1. 506-569. Berlin:

Walter de Gruyter.

Cinque, G. (1980). On Extraction from NP in Italian. Journal of Italian Linguistics , 5, 47-99.

Cinque, G. (1981) Sulla nozione di soggetto di NP in italiano. Cultura Neolatina 41

Cinque, G. (1993). On the Evidence for Partial N-Movement in the Romance DP. Venice

Working Papers in Linguistics 3(2): 21-40.

Cinque, G. (2002) Issues in Adverbial Syntax, ms., University of Venice.

Cinque, G. (2005). Deriving Greenberg’s Universal 20 and Its Exceptions. Linguistic Inquiry

36.315-332.

Cinque, G. (2007). The fundamental left-right asymmetry of natural languages University of

Venice Working Papers in Linguistics. 17, 77-107.

Cinque, G. (to appear). The syntax of adjectives. A comparative study . Oxford. Oxford

University Press.

Clark, H.H. (1979). Bridging. In: P.N. Johnson and P.C. Wasow, eds., Thinking , 411-420.

Cambridge: Cambridge University Press.

Coene, M. and Y. D’hulst (2003a). From NP to DP. Volume I: The expression of possession in

noun phrases. Amsterdam: John Benjamins.

Coene, M. and Y. D’hulst (2003b). From NP to DP. Volume II: The syntax and semantics of

noun phrases. Amsterdam: John Benjamins.

355 Comrie, B. (1981). Language universals and linguistic typology: Syntax and morphology.

Chicago: University of Chicago Press.

Contreras, H. (1986). Spanish bare NPs and the ECP. In Generative studies in Spanish syntax ,

eds. Ivonne Bordelois, Heles Contreras and Karen Zagona, 25–49. Dordrecht: Foris.

Contreras, H., (1989). On Spanish empty N’ and N. In: Kirschner, C., DeCesaris, J. (Eds.),

Studies in Romance Linguistics. John Benjamins, Amsterdam, pp. 83–95.

Corbett, G. (1991). Gender. Cambridge: Cambridge University Press.

Corbett, G. (2000). Number. Cambridge: Cambridge University Press.

Corbett, G. (2006). Agreement. Cambridge: Cambridge University Press.

Cornilescu, A. (1993). Notes on the structure of the Romanian DP and the assignment of the

genitive case. Venice Working Papers in Linguistics 3(2): 107-33.

Cornips, L. (2006). Variationist Sociolinguistics – Syntax. Meertens Institute. 2-12.

Cornips, L. and C. Poletto (2005) On standardising syntactic elicitation techniques (part 1). In:

Lingua 115: 939–957.

Cornips, L. and K. P. Corrigan (2005). “Toward an integrated approach to syntactic variation:

A retrospective and prospective synopsis”. In Syntax and Variation , Cornips, Leonie and

Karen P. Corrigan (eds.), 1–27.

Crespo, A. (1977). Esclavos Negros en Bolivia . La Paz, Bolivia: Librería Editorial Juventud.

Crisma, P. (1990). Functional categories inside the NP: a study on the distribution of nominal

modifiers. Ph.D. dissertation. Università di Venezia

Crisma, P. (1993). On adjective placement in Romance and Germanic event nouns. Rivista di

grammatical Generativa 18: 61-100.

356 Curtin, P. (1969). The Atlantic Slave Trade: A Census. Madison: University of Wisconsin

Press.

Dayal, V. (2002). Number marking and (In)defi niteness in kind terms . Linguistics and

Philosophy 27: 393–450.

DeGraff, M. (2005). Linguists’ most dangerous myth: the fallacy of Creolist Exceptionalism.

Language in Society 34, 533-591.

Delance, J. M. (1975). Bosquejo estadístico de Bolivia. Bolivia en su historia. La Paz: Editorial

Universitaria.

Delicado-Cantero, M. and S. Sessarego. (2010). Variation and Syntax in Number Expression in

Afro-Bolivian Spanish. In Proceedings of the Hispanic linguistic Symposium 2009

(HLS2009). San Juan: Universidad de Puerto Rico.

Demonte, V. (1999). El adjetivo, clases y usos. La posición del adjetivo en el sintagma

nominal. In Demonte, V. & I. Bosque (eds). Gramática descriptiva de la lengua

española . Madrid. Espasa Calpe. 129-215

Demonte, V. (2008). Meaning-form correlations and adjective position in Spanish. Ms.

Universidad Autónoma de Madrid.

www.uam.es/personal_pdi/filoyletras/vdemonte/adjective.pdf den Dikken, M. (2003). The Structure of the Noun Phrase in Rotuman, Lincom Studies in

Austronesian Linguistics 05, LINCOM Europa den Dikken, M., J. Bernstein & C. Tortora (2006). Data and grammar: means and individuals.

Ms. http://web.gc.cuny.edu/dept/lingu/dendikken/docs/featherston_reply_final.pdf

357 Depiante M. and P. Masullo (2001). Género y número en la elipsis nominal: consecuencia para

la hipótesis lexicalista. Paper presented at I Encuentro de gramática generativa , Gral

Roca, 22-24 November 2001.

Déprez, V. (2001). On the syntactic and semantic nature of Haitian Bare NPs. In: Cresti, D.,

Tortora, C., Saterfield, T. (Eds.), Current Issues in Romance Linguistics: Selected Papers

from the 29th Linguistics Symposium on Romance Languages . John Benjamins,

Amsterdam.

Di Domenico, E. and M. De Vincenzi. (1999). Introduction to special issue on ‘Gender and

number in normal and impaired language processing’. Rivista di Linguistica, 11 (1), 3-9.

Díaz-Campos, M. and C. Clements (2008). A Creole origin for Barlovento Spanish? A

linguistic and sociohistorical inquiry. Language in Society 37, 351–383.

Diesing, M. (1992). Indefinites . Cambridge MMA: The MIT Press.

Dixon, R. M. W. (1994). Ergativity . Cambridge: Cambridge University Press.

Donnellan, K. (1966). Reference and Definite Descriptions The Philosophical Review , Vol. 75,

3. 281-304

Dowty, D. (1991). Thematic proto-roles and argument selection. Language. 67, 3 547-619.

Eguren, L. (2010). Contrastive Focus and Nominal Ellipsis in Spanish . Lingua (in press).

Eguren, L. (to appear). Adjectives and Deleted Nominals in Spanish . In Enoch Aboh, Elisabeth

van der Linden, Josep Quer, and Petra Sleeman (eds.) Romance Languages and

Linguistic Theory 2007 . John Benjamins, Amsterdam.

Emonds, J. (1978). The verbal complex V’-V in French. Linguistic Inquiry 9.

Faingold, E. D. 2003. The development of grammar in Spanish and the Romance languages.

Basingstoke, UK: Palgrave Macmillan.

358 Frampton, J. and S. Gutmann. (2000). Agreement is Feature Sharing.

http://www.math.neu.edu/ling /pdffiles/agrisfs.pdf

Franceschina, F. (2002). Case and φ-feature agreement in advanced L2 Spanish grammars.

EUROSLA Yearbook , Foster-Cohen, Susan H., Tanja Ruthenberg and Marie Louise

Poschen (eds.), 71–86.

Franceschina, F. (2005). Fossilized second language grammars . Amsterdam: Benjamins.

Giorgi, A and G. Longobardi (1990), The Syntax of Noun Phrases, Cambridge, Cambridge

University Press.

Giusti, G. (1993), La sintassi dei determinanti, Padova, Unipress.

Givón, T. (1976). Topic, pronoun and grammatical agreement. In C. N. Li (ed.), Subject and

Topic. New York, NT: Academic Press.

Goodman, M. (1987). The Portuguese element in the American creoles. In Gilbert 1987, 361-

405.

Granda, G. de. (1978). Estudios linguisticos hispanicos, afrohispanicos y criollos. Madrid:

Editorial Gredos.

Greenberg, J. H. (1966). Language Universals, with Special Reference to Feature Hierarchies.

The Hague: Mouton.

Grimshaw, J. (1997). Projection, heads, and optimality. Linguistic Inquiry 28: 373-422.

Grimshaw, J. (1990) Argument Structure . Cambridge, Mass.: MIT Press.

Grimshaw, J. (1991). Extended projection. Manuscript, Brandeis University.

Grohmann, K.K. (2000). Prolific Peripheries: A radical view from the left . Ph. D. dissertation,

University of Maryland, College Park.

359 Grohmann, K.K. and L. Haegeman. (2002). Resuming reflexives. Proceedings of the 19th

Scandinavian Conference in Linguistics, Tromsø, Norway.

Guèron, J. (1981). Logical Operators, Complete Constituents, and Extraction Transformations,

in May, R. & Koster, J. (eds.) Levels of Syntactic Representation. Dordrecht, Foris. 65-

142.

Gutiérrez, J. and E. Mallén (2001). NP movement and Adjective position in the DP Phases.

Features and interfaces in Romance: essays in honor of Heles Contreras , J.

Herschenshon, E. Mallén and K. Zagona (eds.), 107-132. Amsterdam: John Benjamins.

Gutiérrez-Rexach, J. (2004). La semántica de los indefinidos . Madrid: Arco Libros.

Gutiérrez-Rexach, J. (2006). Beyond the indefiniteness restriction. Sinn & Bedeutung 11.

Gutiérrez-Rexach, J. and Sessarego S. (2010). On the nature of bare nouns in Afro-Bolivian

Spanish. Peper presented at the 40 th Linguistic Symposium on Romance Languages

(LSRL40). University of Washington, March 2010.

Gutiérrez-Rodríguez, E. (2009). Rasgos gramaticales de los cuantificadores débiles.

Universidad Computense de Madrid. Ph. D. dissertation. Madrid.

Guy, G. (1981). Linguistic variation in Brazilian Portuguese: Aspects of the phonology, syntax,

and language history . Ph. D. dissertation, University of Pennsylvania.

Hagoort, P. & Brown, C.M. (1999). Gender electrified: ERP evidence on the syntactic nature

of gender processing. Journal of Psycholinguistic Research, 28 (6), 715-728.

Harley, H. & E. Ritter (2002). “Person and Number in Pronouns: A Feature Geometric

Analysis”. Language, 78 , 482–526.

Harris, J. W. (1991). The exponence of gender in Spanish. Linguistic Inquiry 22(1):27-62.

360 Hawkins, R. (1998). The inaccessibility of formal features of functional categories in second

language acquisition. Paper presented at the Pacific Second Language Research Forum ,

Tokyo, March 1998.

Heap, D. (2001). Split Subject Pronoun Paradigms: Feature Geometry and Underspecification.

In T. Satterfield, C. Tortora & D. Crest (Eds.) current Issues in Linguistic theory:

Selected papers from the XXIXth Linguistic Symposium on romance Languages (pp124-

139) Amsterdam & Philadelphia: John Bemjamins.

Henry, A. (2005). Idiolectal variation and syntactic theory. In: Cornips Leonie / Corrigan,

Karen P. (eds.): Syntax and Variation. Reconciling the Biological and the Social,

Amsterdam / Philadelphia: Benjamins. pp. 109–122.

Higginbotham, J. (1985). On semantics. Linguistic Inquiry 16: 547-593.

Holm, J. (1992). Popular Brazilian Portuguese: A semi-creole. In E.d’Andrade and A. Kihm,

eds., 37-66.

Holm, J. (2004). Languages in Contact: The Partial Restructuring of Vernaculars . Cambridge:

Cambridge University Press.

Hornstein, N. (1995), Logical Form: From GB to Minimalism, Oxford, Basil Blackwell.

Ishane, T. (2008) The layered DP: form and meaning of French indefinites. Linguistik Aktuell

/ Linguistics Today, Volume 124 John Benjamin: Amsterdam.

Jackendoff, R. (1990). Semantic structures. Cambridge, Mass.: MIT Press.

Jakobson, R., F., G. & Halle, Morris (1952) Preliminaries to Speech Analysis . Cambridge,

MA: MIT.

Kayne, R. (1976). French Relative ‘que’ . In Marta Lujan and Fritz Hensey (eds.) Current

Studies in Romance Linguistics , pp. 63–75. Georgetown University Press, Washington.

361 Kayne, R. (1994). The antisymmetry of syntax . Cambridge, MA: MIT Press.

Kayne, R. (1996). Microparametric syntax: Some introductory remarks. In: Black, James R. /

Motapanyane, Virginia (eds.): Microparametric Syntax and Dialect Variation . ix-xxviii.

Amsterdam: John Benjamins.

Kayne, R. (2000). Parameters and Universals. Oxford: Oxford University Press.

Keenan, E. L. (1974). The functional principle: Generalising the notion of ‘subject of’.

Chicago Linguistic Society (CLS), 10, 298-309.

Kempen, G., & E. Hoenkamp (1987) An incremental procedural grammar for sentence

formulation. Cognitive Science 11, 201–258.

Kester E-P. & Schmitt C. (2006). Papiamentu and Brazilian Portuguese: A comparative study

of bare nouns. In Baptista, M. Guéron, J. (Eds.), Noun phrases in Creole languages .107-

144. Amsterdam: John Benjamins.

King, R. & T. Nadasdi (1997). Left dislocation number marking and Non-Standard French.

Probus , 9, 267-284.

Kinyalolo, K.K.W. (1995). Licensing in DP in Fon. Linguistique Africaine 14: 61-92.

Kiparsky, P. (1972). Explanation in phonology. In Stanley Peters (ed.): Goals of linguistic

theory . Englewood Cliffs, NJ: Prentice-Hall. 189-227.

Kiparsky, P. (1982). Word Formation and the Lexicon. Mid-America Linguistics Conference

Proceedings. Lawrence, Kansas: University of Kansas.

362 Klein, H. (1999). The Atlantic Slave Trade. Cambridge University Press.

Klein, W. and C. Perdue (1997). The Basic Variety. Or: Couldn't natural language be much

simpler? Second Language Research 13: 301-347.

Koopman, H. (1997). The Spec Head Configuration. In E. Garrett and F. Lee (eds.): Syntax at

Sunset, UCLA Working Papers in Syntax and Semantics 1. 37-65.

Kornfeld, L. and A. Saab (2004). Nominal ellipsis and morphological structure in Spanish. In:

Bok-Bennema, R., Hollebrandse, R.B., Kampers-Manhe, B., Sleeman, P. (Eds.),

Romance Languages and Linguistic Theory 2002 . John Benjamins, Amsterdam, 183–

198.

Kouwenberg, S. (2006). L1 transfer and the cut-off point for L2 acquisition processes in Creole

formation. In Claire Lefevbre, Lydia White, & Christine Jourdan (Eds.), L2 Acquisition

and Creole Genesis: Dialogues. 205–219. Amsterdam: John Benjamins.

Kouwenberg, S. and Patrick (2003). Reconsidering the role of second language acquisition in

pidginization and creolization. Special issue of Studies in Second Language Acquisition

25.2, 175–306.

Kratzer, A. (1995). Stage-level and individual-level predicates.. In The Generic Book , ed. G.N.

Carlson and F.J. Pelletier. Chicago: University of Chicago Press, 125-175.

Kroch, A. (1994). Morphosyntactic Variation. In: K. Beals et al., eds. 1994. Papers from the

30th Regional Meeting of the Chicago Linguistics Society: Parasession on Variation and

Linguistic Theory

Kroch, A. (1989). Function and Grammar in the History of English: Periphrastic ‘do’. In Ralph

Fasold and Deborah Schiffrin, eds. Language Change and Variation. Amsterdam:

Benjamins. pp. 133–172.

363 Labov, W. (1966). The Social Stratification of English in New York City . Washington, DC:

Center for Applied Linguistics.

Labov, W. (1969). Contraction, deletion and inherent variability of the English copula.

Language , 45, 715-762.

Labov, W. (1972). Language in the Inner City. Philadelphia: Univ. of Pennsylvania.

Labov, W. (1984). Field Methods of the Project on Linguistic Change and Variation . In Baugh

and Sherzer (eds.), Language in Use , 84-112. Englewood Cliffs: Prentice Hall.

Laca, B. (1996). Acerca de la sema´ntica de los plurales escuetos del espan˜ol. In: Bosque, I.

(Ed.), El sustantivo sin determinacio´n. La ausencia de determinante en la lengua

espan˜ola. Visor, Madrid, pp. 241–268.

Laca, B. (1999). Presencia y ausencia de determinante. In: Bosque, I., Demonte, V. (dirs.),

Grama´tica descriptiva de la lengua espan˜ola, vol. 1. Espasa, Madrid, 891–928.

Larsen-Freeman, D. (1975a). The acquisition of grammatical morphemes by adult learners of

English as a second language. Ph.D. Thesis. University of Michigan, Ann Arbor.

Larsen-Freeman, D. (1975b). The acquisition of grammatical morphemes by adult ESL

students. TESOL Quarterly 9, 409–430.

Larson, B. (1998). Cochabamba, 1550-1900 . Duke University Press.

Larson, R. (1988). On the double object construction. Linguistic Inquiry, 19(3), pp. 335-391.

Lattewitz, K. (1995). Eine Analyse des deutschen Genitives. Linguistiche Berichte 150, 118-

146.

Laurence, K. (1974). Is Caribbean Spanish a case of decreolization? Orbis 23:484–99.

364 Lefebvre, C. (1998). Creole Genesis and the Acquisition of Grammar . Cambridge University

Press.

Lefevbre, C., L. White and C. Jourdan (eds.) (2006) L2 Acquisition and Creole Genesis:

Dialogues. Amsterdam: John Benjamins.

Lehmann, C. (1982). Universal and typological aspects of agreement. In Teil II, H. Seiler & C.

Lehmann (eds.). Apprehemsion: Das Sprschlich Erfassen von Gegenständen. Tubingen:

Gunter Narr.

Leonini, C. (2006). The acquisition of object clitics and definite articles: Evidence from Italian

as L2 and L1 . Ph. D. dissertation. University of Florence.

Leons, W. (1984). Some notes on the demographic history of the Negro in the Bolivian

Yungas. Antropological Investigation in Bolivia 58, 13-27. Greeley, CO: University of

Northern Colorado, Museum of Anthropology, Miscellaneous series.

Levelt, W. J. M. (1989) Speaking. From intention to articulation. Cambridge, Mass.: MIT

Press.

Lipski, J. M. (1982). El Valle del Chota: enclave lingüístico afroecuatoriano. Boletín de la

academia Puertorriqueña de la Lengua Española 10: 21-36.

Lipski, J. M. (1985). The Spanish of Equatorial Guinea . Tübingen: Max Niemeyer.

Lipski, J. M. (1987). The Chota Valley : Afro-Hispanic Language in Highland. Ecuador, Latin

American Research Review, 22:1 155-170.

Lipski, J. M. (1990). The language of the ISLENOS: Vestigial Spanish in Louisisna. Baton

Rouge: Louisiana State University Press.

Lipski, J. M. (1993). On the non-creole basis for Afro-Caribbean Spanish. Albuquerque, NM:

University of New Mexico Press.

365 Lipski, J. M. (1994). Latin American Spanish . New York: Longman.

Lipski, J. M. (1996). Los dialectos vestigiales del español en los Estados Unidos: estado de la

cuestión. Signo y Seña 6, 459-489.

Lipski, J. M. (1998). El español bozal. In Matthias Perl & Armin Schewgler (eds.), América

negra: Panorámica actual de los estudios lingüísticos sobre variedades criollas y

afrohispanas , 293–327. Frankfurt: Vervuert.

Lipski, J. M. (2000). Bozal Spanish: Restructuring or creolization? In Edgar W. Schneider &

Ingrid Neuman-Holzschuh (eds.), Degrees of restructuring in creole languages , 55–83.

Amsterdam & Philadelphia: John Benjamins.

Lipski, J. M. (2005). A history of Afro-Hispanic language. Five centuries, five continents.

Cambridge: Cambridge University Press.

Lipski, J. M. (2006a). Morphosyntactic implications in Afro-Hispanic language: new data on

creole pathways. Paper presented at The 35 th New ways of Analyzing Variation

Conference (NWAV35). Columbus: The Ohio State University. October 2006.

Lipski, J. M. (2006b). Afro-Bolivian Spanish and Helvetia Portuguese: Semi-creole parallels.

Papia, 16. 96-116 .

Lipski, J. M. (2007). Afro-Bolivian Spanish: The survival of a true creole prototype. In

Synchronic and Diachronic Perspectives on Contact Languages Eduted by Magnus

Huber, Viveka Velupillai. Creole Language Library 32 . 175-98. . Amsterdam: John

Benjamins.

Lipski, J. M. (2008a). Morphosyntactic implications in Afro-Hispanic language: new data on

creole pathways. Available at: http://www.personal.psu.edu/ jml34/afmorph.pdf

Lipski, J. M. (2008b). Afro-Bolivian Spanish . Madrid & Frankfurt: Iberoamericana – Vervuert.

366 Lipski, J. M. (unpublished) Spanish-based creoles in the Caribbean.

http://www.personal.psu.edu/jml34/spcreole.pdf

Lobeck, A., (1995). Ellipsis: Functional Heads, Licensing and Identification. Oxford

University Press, New York.

Lockhart, J. (1994). Spanish Peru, 1532-1560: A Social History . Madison, Wisconsin:

University of Wisconsin Press.

Longobardi, G. (1994). Reference and proper names: A theory of N-movement in syntax and

logical form. Linguistic Inquiry 25: 609-665.

López, L. (2000). Ellipsis and discourse-linking. Lingua 110, 183–213.

Mallén, E. (1997). A Minimalist Approach to Concord in Noun Phrases. Theoretical

Linguistics . 23. 1/2. 49-77. Berlin: Walter de Gruyter.

Martinus, F. (1989). West African connection: The influence of the Afro-Portuguese on the

Papiamentu of Curazao. In Estudios sobre espanol de America y linguistica

afroamericana (no ed). Bogota: Publicaciones del Instituto Caro y Cuervo.

McNally, L. (2004). Bare plurals in Spanish are interpreted as properties. Catalan Journal of

Linguistics 3, 115-133.

McWhorter, J. (1997). Towards a new model of creole genesis. New York: Peter Lang.

McWhorter, J. (1998). Identifying the creole prototype: vindicating a typological class.

Language 74, 788-818.

McWhorter, J. (2000). The Missing Spanish Creole .University of California Press. Berkley.

Megenney, W. (1985). África en Venezuela: su herencia lingüística y cultura literaria.

Montalbán 15.3-56.

367 Mellafe, R. (1984). La introducción de la esclavitud negra en Chile: tráfico y rutas. Santiago:

Editorial Universitaria.

Migge, B. (2003). Creole Formation as Language Contact: The Case of the Suriname Creoles.

Amsterdam: John Benjamins .

Milner, J. C. (1982). Ordres et raison de langue , Seuil, Paris.

Milsark, G. (1974). Existential sentences in English, MIT: Ph.D. dissertation.

Mintz, S. (1971). The socio-historical background to pidginization and creolization. In Hymes

1971, 481-98.

Mufwene, S. (1996). The Founder Principle in creole genesis. Diachronica 13:83-134.

Mufwene, S. (1997). Jargons, pidgins creoles and koines: what are they?. In Spears A. &

Winford D. (eds.): The structure and status of pidgins and creoles . Amsterdam: John

Benjamins.

Mufwene, S. (2001). The ecology of language evolution . Cambridge: Cambridge University

Press.

Mufwene, S. (2002) .Competition and selection in language evolution. Selection 3, 45–56.

Muhlhausler, P. (1997). Pidgin and creole linguistics. London: University of Westminster

Press.

Müller, A. (2003). Generic sentences with indefinite and bare subjects in Brazilian Portuguese.

In Anderssen, J., Menéndez-Benito, P. Werle, A., eds. The Proceedings of SULA2 .

Vancover, BC: 71-86.

Munn, A. & Schmitt, C. (2001). Bare nominals and the morphosyntax of number. Current

issues in Romance Linguistics , T. Satterfield, c. Tortora & D. Cresti (eds), 217-213.

Amsterdam: John Benjamins.

368 Murphy, S., (1997). Knowledge and production of English articles by advanced second

language learners. Ph.D. Thesis. University of Texas at Austin.

Olsen, S. (1989). AGR(eement) in the German Noun Phrase. In Bhatt, Chr., Lobrl, E. &

Schmidt, C. (eds.). Syntactic Phrase Structure Phenomena in Noun Phrases & Sentences

(39-29). Amsterdam: John Benjamins.

Ormazabal, J. (1991). Asymmetries on Wh-movement and some theoretical consequences. Ms.

University of Connecticut, Storrs.

Parkvall, M. (2000). Reassessing the role of demographics in language restructuring. In

Degrees of Restructuring in Creole Languages. Edited by I. Neumann-Holzschuh and E.

W. Schneider. Creole Language Library 22. 185-214. Amsterdam – Philadelphia : John

Benjamins.

Parodi T., B.D. Schwartz and H. Clahsen (1997). On the acquisition of the morphosyntax of

German nominals. Essex Reports Research in Linguistics , 15: 1- 43.

Parodi, C. (1994). On the case of agreement in Spanish and English DPs. In Issues and Theory

in Romance Linguistics. M.L. Mazzola (ed.), Selected papers from the Linguistic

Symposium on Romance Languages 1993 , 403-16. Washington DC: Georgetown UP.

Parrot, J. (2007). Distributed Morphological Mechanisms of Labovian Variation in

Morphosyntax . Ph. D. Dissertation Georgetown University. Washington, DC

Partee, B. (1987). Noun Phrase Interpretation and Type-Shifting Principles in J. Groenendijk,

D. de Jongh y M. Stokhof (eds.) 115-143

Pesetsky, D. and E. Torrego. (2007). The syntax of valuation and the interpretability of

features. In Simin Karimi , Vida Samiian, and Wendy K. Wilkins (eds.): Phrasal and

369 Clausal Architecture: Syntactic derivation and interpretation, 262-

294.Amsterdam/Philadelphia : Benjamins.

Picallo, C. (1991). Nominals and Nominalizations in Catalan, Probus 3, 279-316.

Picallo, C. (2008). Gender and Number in Romance. Lingue e Linguaggio 1: 47-66.

Picallo, C. (2010). Structure of the Noun Phrase. In Handbook of Hispanic Linguistics. Oxford:

Blackwell.

Pienemann, M. (1998) Language processing and second language development: Processability

Theory . Amsterdam: John Benjamins.

Pienemann, M. (ed.) (2005) Cross-lingistic aspects of Processability Theory . Amsterdam: John

Benjamins.

Pienemann, M., M. Johnston and G. Brindley. (1988). Constructing an acquisition-based

procedure for second language assessment. Studies in Second Language Acquisition , 10

(2) 217-243.

Pizarroso-Cuenca, A. (1977a). La Esclavitud Negra en las Epocas Colonial y Nacional de

Bolivia. La Paz, Bolivia: Universo.

Pizarroso-Cuenca, A.(1977b). La Cultura Negra en Bolivia. La Paz, Bolivia: ISLA.

Plag, I. (2008a) Creoles as interlanguages: inflectional morphology. Journal of Pidgin and

Creole Languages 23(1):109–130.

Plag, I. (2008b) Creoles as interlanguages: syntactic structures. Journal of Pidgin and Creole

Languages 23(2):307–328.

Plag, I. (2009a) Creoles as interlanguages: word-fromation. Journal of Pidgin and Creole

Languages 24(2):339–362

370 Plag, I. (2009b) Creoles as interlanguages: phonology. Journal of Pidgin and Creole

Languages 24 (1):119–138.

Poletto, C. (2000). The higher functional field. Evidence from Northern Italian dialects. Oxford

University Press, Oxford & New York .

Pollard, C. and I. A. Sag. (1994). Head-driven phrase structure grammar. Chicago: University

of Chicago Press.

Pollock, J.-Y. (1989) Verb movement, Universal Grammar and the structure of IP. Linguistic

Inquiry 20 : 365-424.

Poma de Ayala F. G. (1615). El primer nueva corónica y buen gobierno (København, Det

Kongelige Bibliotek, GKS 2232 4°) Available at

http://www.kb.dk/permalink/2006/poma/info/en/frontpage.htm

Pomino N. and E. Stark. (2008). Adnominal Adjectives in Romance. Where Morphology

seemingly meets Semantics, In M. T. Espinal, M. Leonetti and L. McNally (eds.):

Proceedings of the IV Nereus International Workshop “Definiteness and DP Structure in

Romance Languages”, Konstanz: Arbeitspapiere des Fachbereichs Sprachwissenschaft.

Poplack, S. (1980). Deletion and disambiguation in Puerto Rican Spanish. Language 56.2: 371-

385.

Portugal-Ortíz, M.(1977). La esclavitud negra en las epocas colonial y nacional de Bolivia.

Instituto Boliviano de Cultura, La Paz.

Radford, A. (2000). NP Shells. Essex Research Reports in Linguistics 33.2-20.

Raposo, E. (1999). Towards a Minimalist Account of Nominal Anaphora in Spanish and

English. In http://www.ling.umd.edu/courses/Ling819/Papers/edu.html.

371 Ritter, E. (1991). Two functional categories in noun phrases: Evidence from Modern Hebrew.

In S. Rothstein (ed.), Syntax and Semantics 25: Perspectives on Phrase Structure: Heads

and Licensing . 37-62. San Diego: Academic Press.

Ritter, E. (1993). Where is gender? Linguistic Inquiry, 24 , 795-803.

Rizzi, L. (1997). The fine structure of the left periphery. In L. Haegeman (ed.), Elements of

grammar: A handbook of generative syntax . Dordrecht: Kluwer.

Rizzi, L. (ed.) (2004). The Structure of CP and IP. The Cartography of Syntactic Structures,

vol.2 , New York: Oxford University Press.

Roberts, I. and A. Roussou. (2003). Syntactic Change. Minimalist Approaches to

Grammaticalization. Cambridge: Cambridge University Press.

Romaine, S. (1988). Pidgin and creole languages . London and New York: Longman.

Rosemblat, A. (1954). La Población indígena y el mestizaje en America . Buenos Aires:

Editorial NOVA.

Saab, A. (2004). El dominio de la elipsis nominal en español: identidad estricta e inserción

tardía. Master thesis. Universidad Nacional del Comahue.

Saito, M. and K. Murasugi. (1999). Subject Predication within IP and DP. Beyond Principles

and Parameters , eds. K. Johnson and I. Roberts, 167-188. Dordrecht: Kluwer.

Sànchez L., M.J. Giménez (1998). The L2 Acquisition of Definite Determiners: from Null to

Overt. In BUCLD 22 Proceedings . 640-650.

Sankoff G, S. Tagliamonte, and E. Smith (2005). Goldvarb X: A variable

rule application for Macintosh and Windows. Department of Linguistics, University of

Toronto

372 Sankoff, G. (1990). The grammaticalization of tense and aspect in Tok Pisin and Sranan.

Language Variation and Change (2), 295-312.

Schmitt, C. & Munn, A. (1999). Against the Nominal Mapping Parameter: Bare nouns in

Brazilian Portuguese. Paper presented at NELS 29.

Schmitt, C. & Munn, A. (2003). The syntax and semantics of bare arguments in Brazilian

Portuguese. Linguistic Variation Yearbook 2: 185-216.

Schneider, E. W. (1990). English World-Wide: The cline of creoleness in English-oriented

creoles and semi-creoles of the Caribbean. In English World-Wide . 11, 79-113.

Schwegler, A. (1991). Negation in Palenquero: synchrony. Journal of Pidgin and Creole

Languages, 6.165-214.

Schwegler, A. (1993). Rasgos (afro-)portugueses en el criollo del Palenque de San Basilio

(Colombia). In homenaje a José Pérez Vidal, ed. Cameron Díaz D. Alayon, 667-96. La

Laguna, Tenerife: Litografia A. Romero S. A.

Schwegler, A. (1996). La doble negación dominicana y la génesis del español caribeño.

Hispanic Linguistics 8:247–315.

Schwenter, S. (2005). The Pragmatics of negation in Brazilian Portuguese. Lingua 115, 1427-

1456.

Sessarego, S. (2009a). Gender Agreement Variation in the Afro-Bolivian Determiner Phrase:

The Interplay of Social and Linguistic Factors. In Revista Iberoamericana de Lingüística .

4. 135-146.

Sessarego, S. (2009b). On the Evolution of Afro-Bolivian Spanish Subject-Verb Agreement:

Variation and Change. In Sintagma . 21. 107-119.

373 Sessarego, S. and J. Gutiérrez-Rexach (2009b). Gender Agreement in Afro-Bolivian

Determiner Phrase: The Interplay of Social and Structural Factors. Paper presented at

The Joint Summer Meeting of the Society for Pidgin and Creole Linguistics (SPCL) and

the Associação de Crioulos de Base Lexical Portuguesa e Espanhola (ACBLPE).

University of Cologne. August 2009

Sessarego, S. and J. Gutiérrez-Rexach. (2009a). A Minimalist Account for Gender Agreement

in the Afro-Bolivian Determiner Phrase. Paper presented at The XIXth International

Conference on Historical Linguistics (ICHL19). Radboud University Nijmegen . August

2009.

Sessarego, S. and M. Delicado-Cantero (2009). A syntactic approach to variable number

marking in Afro-Bolivian Spanish. New Ways of Analyzing Variation (NWAV) 38.

University of Ottawa, Ottawa, Ottawa (Canada). October 2009.

Seuren, P. A. M. and H. Wekker (1986). Semantic transparency as a factor in Creole genesis.

In Muysken P. and Smith N. (eds.), Substrata versus Universals in Creole Genesis , 57-

70. Amsterdam: Benjamins.

Shakespeare, W. (1597). Romeo and Juliet . Available at

http://shakespeare.mit.edu/romeo_juliet/

Sholonsky, U. (1991), “Quantifiers as Functional Heads: A Study of Quantifier Float in

Hebrew”, Lingua 84, págs. 159-180.

Sigurðsson, H. A. (1993). The structure of the Icelandic NP. Studia Linguistica 47(2), 177-197.

Silverstein, M. (1976[1986]). Hierarchy of features and ergativity, in P. Muysken and H. van

Riemsdijk (eds.), Features and Projections. Dordrecht: Foris, 163-232. Originally

374 published in R. M. W. Dixon (ed.), (1976) Grammatical categories in Australian

Languages. Canberra: Australian Institutes of Aboriginal Studies.

Simioni, L. (2007). A concordância de número no DP: propostas minimalistas. Estudos

Lingüísticos XXXVI.1: 117-125.

Sleeman, P. (1996). Licensing Empty Nouns in French. Ph.D. dissertation, University of

Amsterdam.

Snyder, W. (2007). Child language : the parametric approach . Oxford ; New York : Oxford

University Press.

Soux, M. L. (1993). Esclavos, peones y mingas: apuntes sobre la fuerza de trabajo en las

haciendas Yungueñas a principios de la república. Historia y Cultura 21:51-58.

Speas, M. (1990). Phrase Structure in Natural Language. Dordrechr: Kluwer, 1 990.

Sportiche, D. (1988). A Theory of Floating Quantifiers and its Corollaries for Constituent

Structure, Linguistic Inquiry 19: 2, 425-451.

Sportiche, D. (2002). Movement, agreement and case.

http://www.linguistics.ucla.edu/people/sportich/papers/mac.pdf

Szabolcsi, A. (1983), The Possessor that Ran Away from Home, The Linguistic Review 3, 89-

102.

Szabolcsi, A. (1987). Functional Categories in the Noun Phrase in I. Kenesei (ed.), 167-189.

Szabolcsi, A. (1994). The noun phrase. In: Kiefer, F., Kiss, K.E. (Eds.), The Syntactic

Structure of Hungarian. Syntax and Semantics , 27: 179–274. Academic Press, San Diego.

Tagliamonte, S. (2006). Analysing Sociolinguistic Variation . Cambridge: Cambridge

University Press.

375 Thomas, M. (1989). The acquisition of English articles by first- and second-language learners.

Applied Psycholinguistics 10, 335–355.

Thomson, R. W. and T. Kaufman (1988). Language contact, creolization and genetic

linguistics. Berkeley: University of California Press.

Ticio, M. E. (2003). On the Structure of DPs Doctoral dissertation, UConn, Storrs.

Ticio, M. E. (2005). NP-Ellipsis in Spanish . Proceedings of the 7th Hispanic Linguistics

Symposium 7:128-141

Torrego, E. (1984), Determinerless NPs , ms., University of Massachusetts, Boston MA.

Torrego, E. (1988). On empty categories in nominals , Ms., UMass. Boston.

Travis, L. (1984) Parameters and effects of word Order Variation . Ph.D. dissertation,

University of Washington. Seattle

Troconis de Veracoecha, E. (1969). Documentos para el estudio de los esclavos negros en

Venezuela . Caracas: Academia Nacional de la Historia.

Trubetzkoy, N.S. (1939). Grundzüge der Phonologie . Christiane Baltaxe, trans. Principles of

Phonology, Berkeley: University of California Press, 1969.

Valois, D. (1991). The internal structure of DP. UCLA. Ph. D. Dissertation. van Gelderen, E. (2004). Grammaticalization as Economy. Amsterdam: John Benjamins. van Gendeleren, E. (2005). Principles and parameters in change. In L. E., A. Cornips, K. P.

Corrigan (eds.) Syntax and variation: reconciling the biological and the social .

Amsterdam/Philadelphia: John Bengiamins. 179-198.

Vanelli, L. (1979). Una forma suppletiva dell’articolo e la sua fonosintassi. Rivista di

Grammatica Generativa . 4, 183-206.

376 Vendler, Z. (1967). Verbs and times. In: Z. Vendler, Linguistics in Philosophy, pp. 97-121.

Ithaca: Cornell University Press.

Weinrich U, W. Labov, M. Herzog. (1968). Empirical foundations for a theory of language

change. In: Lehmann W P, Malkiel Y (eds.) Directions for Historical Linguistics.

University of Texas Press, Austin, TX

Weiß, H. (2001). On Two Types of Natural Languages. Some Consequences for Linguistics.

In: Theoretical Linguistics 27 : 87–103.

Winford, D. (2000). “Intermediate” creoles and degrees of change in creole formation: The

case of Bajan. In Degrees of Restructuring in Creole Languages. Edited by I. Neumann-

Holzschuh and E. W. Schneider. Creole Language Library 22. 215-46. Amsterdam –

Philadelphia : John Benjamins.

Wilde, O. (1890). The Picture of Dorian Gray . In The Complete Works of Oscar Wilde ;

Halifax: Magpie Books (1993).

Wolff, I. (1981) . Esclavitud y tráfico de negros en el Alto Perú, 1545-1640 . Historia y Cultura

4:37-63.

Wosow, T. (2002). Postverbal Behavvior. Stanford: CSLI.

Zagona, K. (2006). Sintaxis generative del español. Madrid: Visor Libros.

Zamparelli, R. (2000). Layers in the determiner phrase . New York NY:Garland.

Zribi-Hertz, A. and J-F Hanne. (1995). Pronoms, determinants et relatives Bambara de

Bamako. Linguistique Africaine 15: 91-135.

Zubizarreta, M.L. (1986). Levels of representation in the Lexicon and in the Syntax, ms.,

University of Tilburg and University of Maryland.

377