The L2 Acquisition of Spanish Subjects: Multiple Perspectives 161451593X, 9781614515937

This volume has two principle objectives: the first is to understand what it is that L2 learners are capable of acquirin

588 97 1MB

English Pages 306 Year 2015

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

The L2 Acquisition of Spanish Subjects: Multiple Perspectives
 161451593X, 9781614515937

Table of contents :
Table of contents
Abbreviations used in glosses
Abbreviations used in text
List of tables
List of figures and graphs
Acknowledgements
1 Introduction: History and theories of second language acquisition
1.1 Formal/generative approaches
1.1.1 Universal grammar and language acquisition
1.1.2 The parametric approach in second language acquisition
1.1.3 Interface approaches
1.2 Processing models
1.3 Discourse-pragmatic approaches
1.4 The variationist approach
1.5 Concluding remarks regarding theoretical approaches to L2 acquisition
1.6 Organization of the volume
2 Subjects in Spanish
2.1 Overview of subject forms in Spanish
2.2 Syntactic properties of Spanish subjects
2.2.1 The Null Subject Parameter
2.2.2 Interpretative constraints
2.2.3 Strong agreement and Spanish subjects
2.3 Discourse-pragmatic features and Spanish subject distribution
2.3.1 Focus, topic-continuity and topic-shift
2.3.2 Cohesion, saliency and cognition
2.4 Referent features and verbal semantics in Spanish subject distribution
2.4.1 Person, number and verb semantics
2.4.2 Specificity and animacy of referents
2.5 Regional variations of Spanish subjects
2.6 Chapter summary
3 Parametric studies in the acquisition of subject pronouns
3.1 Early research into the L2 acquisition of Spanish subject pronouns
3.2 Research on the acquisition of the Overt Pronoun Constraint (OPC) and Topic/Focus
3.3 Recent studies from a parametric approach: The OPC and naturalistic input
3.4 Conflicting evidence and common findings
4 The syntax-pragmatics interface in the acquisition of subject pronouns
4.1 Cross-linguistic influence in bilinguals and the Unidirectional Hypothesis
4.2 The Unaccusative Hypothesis and word order
4.3 The OPC (again)
4.4 Person and number
4.5 Contributions of findings of the syntax-pragmatic interface model
5 Processing studies and L2 subject pronoun acquisition in Spanish
5.1 Anaphoric resolution in pro-drop languages and the PAH
5.1.1 The PAH in Italian
5.1.2 The PAH in Spanish
5.1.3 The PAH in Spanish and Italian
5.1.4 Summary of effects of the PAH in pro-drop languages
5.2 Anaphoric resolution strategies in bilingual and L2 acquisition (pro-drop and non-pro-drop languages)
5.2.1 Strategies of anaphoric resolution in Italian L2
5.2.2 Strategies of anaphoric resolution in Italian-English and Italian-Spanish bilinguals
5.3 Strategies of anaphoric resolution in Spanish heritage and L2 speakers
5.3.1 PAH biases in early (heritage) and late (L2) bilingual speakers
5.3.2 PAH biases and discourse structure in L2 learners
5.3.3 Conclusions of anaphoric resolution in bilingual and Spanish L2 speakers
5.4 Chapter summary
6 Discourse-pragmatics approaches in the acquisition of subject expression
6.1 Discourse structure and the acquisition of subject expression
6.2 Cohesion theory and the anaphoric hierarchy
6.2.1 Testing the anaphoric hierarchy and cohesion theory
6.2.2 Native speakers’ use of subject forms according to the anaphoric hierarchy
6.2.3 L2 Learners’ use of subject forms according to the anaphoric hierarchy
6.2.4 Predictions and universality of the anaphoric hierarchy
6.3 Emphasis, contrast and disambiguation
6.3.1 A longitudinal case study
6.3.2 Quantitative findings
6.3.3 Qualitative findings
6.3.4 Comparing the quantitative and qualitative findings
6.4 Pragmatic constraints of saliency, focus, and pragmatic weight
6.4.1 The rule of salient referent
6.4.2 The rule of switch focus of attention
6.4.3 The rule of contrastive focus
6.4.4 The rule of pragmatic weight
6.4.5 The rule of epistemic parentheticals
6.4.6 Saliency, focus, and pragmatic weight in L2 learners’ narrative structure
6.4.7 Results of analysis of pragmatic constraints
6.5 Cognitive status and the Givenness Hierarchy
6.5.1 Cognitive status of type identifiable and referential
6.5.2 Cognitive status of uniquely identifiable
6.5.3 Cognitive status of familiar
6.5.4 Cognitive status of activated
6.5.5 Cognitive status of in focus
6.5.6 A revised hierarchy for Spanish
6.5.7 The revised hierarchy and predicting subject use in L2 learners’ narratives
6.5.8 Results of testing the revised hierarchy
6.6 Contributions of findings of the discourse-pragmatics approach
7 A variationist approach to L2 subject pronoun acquisition
7.1 Constraining factors in variationist studies
7.1.1 Linguistic variables
7.1.2 Discourse variables
7.2 Conditioning factors in L2 variationist studies
7.2.1 Linguistic variables in L2 studies
7.2.2 Discourse variables in L2 variation studies
7.3 Chapter summary
8 Summary of findings
8.1 The parametric model
8.1.1 The initial state
8.1.2 Automatic learning of a cluster of properties
8.1.3 Acquisition along a hierarchical/implicational path
8.1.4 The role of input in resetting parameters
8.1.5 Conclusions of findings of the parametric approach
8.2 The syntax-pragmatics interface model
8.2.1 Subject pronoun use constrained by discourse context
8.2.2 Unidirectional transfer
8.2.3 Word order constrained by lexical verb class and focus
8.2.4 Subsequent vs. simultaneous acquisition of syntax and discourse-pragmatic constraints
8.2.5 Why do discourse-pragmatic deficits linger?
8.2.6 Non-uniformity of L2 acquisition
8.2.7 Conclusions of findings of the syntax-pragmatics interface model
8.3 Processing models
8.3.1 Strategies for resolving conflicting cues
8.3.2 Processing at the interface
8.3.3 Mediation between input and universal linguistic properties
8.3.4 Innate or L1 principles
8.3.5 Conclusions of findings of the processing model
8.4 Discourse-pragmatic approach
8.4.1 Discourse constraints/pragmatic rules
8.4.2 Impact of information structure in discourse
8.4.3 Predictions of implicational hierarchies
8.4.4 Universal cognitive abilities vs. language specific principles
8.4.5 Conclusions of findings of the discourse-pragmatics model
8.5 Variationist model
8.5.1 Linguistics contexts/factors and variability affecting subject expression in L2
8.5.2 Discourse contexts/factors and variability affecting subject expression in L2
8.5.3 Effects of verbal lexical semantics and variability of subject expression in L2
8.6 Common findings and concluding remarks
References
Index

Citation preview

Margaret Lubbers Quesada The L2 Acquisition of Spanish Subjects

Studies on Language Aquisition

| Edited by Peter Jordens

Volume 50

Margaret Lubbers Quesada

The L2 Acquisition of Spanish Subjects | Multiple Perspectives

ISBN 978-1-61451-593-7 e-ISBN (PDF) 978-1-61451-436-7 e-ISBN (EPUB) 978-1-5015-0041-1 Library of Congress Cataloging-in-Publication Data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at http://dnb.dnb.de. © 2015 Walter de Gruyter Inc., Boston/Berlin Typesetting: PTP-Berlin, Protago-TEX-Production GmbH, Berlin Printing and binding: CPI books GmbH, Leck ♾ Printed on acid-free paper Printed in Germany www.degruyter.com

Table of contents Abbreviations used in glosses | x Abbreviations used in text | xi List of tables | xii List of figures and graphs | xiv Acknowledgements | xv 1 1.1 1.1.1 1.1.2 1.1.3 1.2 1.3 1.4 1.5 1.6 2 2.1 2.2 2.2.1 2.2.2 2.2.3 2.3 2.3.1 2.3.2 2.4 2.4.1 2.4.2

Introduction: History and theories of second language acquisition | 1 Formal/generative approaches | 3 Universal grammar and language acquisition | 3 The parametric approach in second language acquisition | 6 Interface approaches | 9 Processing models | 10 Discourse-pragmatic approaches | 12 The variationist approach | 14 Concluding remarks regarding theoretical approaches to L2 acquisition | 17 Organization of the volume | 18 Subjects in Spanish | 20 Overview of subject forms in Spanish | 20 Syntactic properties of Spanish subjects | 23 The Null Subject Parameter | 23 Interpretative constraints | 27 Strong agreement and Spanish subjects | 30 Discourse-pragmatic features and Spanish subject distribution | 32 Focus, topic-continuity and topic-shift | 32 Cohesion, saliency and cognition | 36 Referent features and verbal semantics in Spanish subject distribution | 38 Person, number and verb semantics | 38 Specificity and animacy of referents | 40

vi | Table of contents 2.5 2.6 3 3.1 3.2 3.3 3.4 4 4.1 4.2 4.3 4.4 4.5

5 5.1 5.1.1 5.1.2 5.1.3 5.1.4 5.2 5.2.1 5.2.2 5.3 5.3.1

Regional variations of Spanish subjects | 42 Chapter summary | 44 Parametric studies in the acquisition of subject pronouns | 47 Early research into the L2 acquisition of Spanish subject pronouns | 49 Research on the acquisition of the Overt Pronoun Constraint (OPC) and Topic/Focus | 61 Recent studies from a parametric approach: The OPC and naturalistic input | 69 Conflicting evidence and common findings | 74 The syntax-pragmatics interface in the acquisition of subject pronouns | 77 Cross-linguistic influence in bilinguals and the Unidirectional Hypothesis | 79 The Unaccusative Hypothesis and word order | 87 The OPC (again) | 107 Person and number | 114 Contributions of findings of the syntax-pragmatic interface model | 117 Processing studies and L2 subject pronoun acquisition in Spanish | 120 Anaphoric resolution in pro-drop languages and the PAH | 122 The PAH in Italian | 122 The PAH in Spanish | 124 The PAH in Spanish and Italian | 131 Summary of effects of the PAH in pro-drop languages | 133 Anaphoric resolution strategies in bilingual and L2 acquisition (pro-drop and non-pro-drop languages) | 135 Strategies of anaphoric resolution in Italian L2 | 135 Strategies of anaphoric resolution in Italian-English and Italian-Spanish bilinguals | 138 Strategies of anaphoric resolution in Spanish heritage and L2 speakers | 139 PAH biases in early (heritage) and late (L2) bilingual speakers | 139

Table of contents

5.3.2 5.3.3 5.4 6 6.1 6.2 6.2.1 6.2.2 6.2.3 6.2.4 6.3 6.3.1 6.3.2 6.3.3 6.3.4 6.4 6.4.1 6.4.2 6.4.3 6.4.4 6.4.5 6.4.6 6.4.7 6.5 6.5.1 6.5.2 6.5.3 6.5.4 6.5.5 6.5.6

| vii

PAH biases and discourse structure in L2 learners | 144 Conclusions of anaphoric resolution in bilingual and Spanish L2 speakers | 148 Chapter summary | 150 Discourse-pragmatics approaches in the acquisition of subject expression | 152 Discourse structure and the acquisition of subject expression | 153 Cohesion theory and the anaphoric hierarchy | 154 Testing the anaphoric hierarchy and cohesion theory | 157 Native speakers’ use of subject forms according to the anaphoric hierarchy | 158 L2 Learners’ use of subject forms according to the anaphoric hierarchy | 162 Predictions and universality of the anaphoric hierarchy | 166 Emphasis, contrast and disambiguation | 168 A longitudinal case study | 169 Quantitative findings | 171 Qualitative findings | 172 Comparing the quantitative and qualitative findings | 177 Pragmatic constraints of saliency, focus, and pragmatic weight | 178 The rule of salient referent | 178 The rule of switch focus of attention | 180 The rule of contrastive focus | 182 The rule of pragmatic weight | 184 The rule of epistemic parentheticals | 185 Saliency, focus, and pragmatic weight in L2 learners’ narrative structure | 187 Results of analysis of pragmatic constraints | 189 Cognitive status and the Givenness Hierarchy | 192 Cognitive status of type identifiable and referential | 193 Cognitive status of uniquely identifiable | 194 Cognitive status of familiar | 194 Cognitive status of activated | 194 Cognitive status of in focus | 195 A revised hierarchy for Spanish | 196

viii | Table of contents 6.5.7 6.5.8 6.6

The revised hierarchy and predicting subject use in L2 learners’ narratives | 196 Results of testing the revised hierarchy | 197 Contributions of findings of the discourse-pragmatics approach | 203

7 7.1 7.1.1 7.1.2 7.2 7.2.1 7.2.2 7.3

A variationist approach to L2 subject pronoun acquisition | 206 Constraining factors in variationist studies | 209 Linguistic variables | 209 Discourse variables | 216 Conditioning factors in L2 variationist studies | 222 Linguistic variables in L2 studies | 222 Discourse variables in L2 variation studies | 234 Chapter summary | 242

8 8.1 8.1.1 8.1.2 8.1.3 8.1.4 8.1.5 8.2 8.2.1 8.2.2 8.2.3 8.2.4

Summary of findings | 246 The parametric model | 246 The initial state | 247 Automatic learning of a cluster of properties | 248 Acquisition along a hierarchical/implicational path | 249 The role of input in resetting parameters | 249 Conclusions of findings of the parametric approach | 250 The syntax-pragmatics interface model | 251 Subject pronoun use constrained by discourse context | 252 Unidirectional transfer | 253 Word order constrained by lexical verb class and focus | 253 Subsequent vs. simultaneous acquisition of syntax and discourse-pragmatic constraints | 254 Why do discourse-pragmatic deficits linger? | 254 Non-uniformity of L2 acquisition | 255 Conclusions of findings of the syntax-pragmatics interface model | 256 Processing models | 256 Strategies for resolving conflicting cues | 258 Processing at the interface | 259 Mediation between input and universal linguistic properties | 259 Innate or L1 principles | 260 Conclusions of findings of the processing model | 260

8.2.5 8.2.6 8.2.7 8.3 8.3.1 8.3.2 8.3.3 8.3.4 8.3.5

Table of contents | ix

8.4 8.4.1 8.4.2 8.4.3 8.4.4 8.4.5 8.5 8.5.1 8.5.2 8.5.3 8.6

Discourse-pragmatic approach | 261 Discourse constraints/pragmatic rules | 262 Impact of information structure in discourse | 263 Predictions of implicational hierarchies | 264 Universal cognitive abilities vs. language specific principles | 264 Conclusions of findings of the discourse-pragmatics model | 265 Variationist model | 265 Linguistics contexts/factors and variability affecting subject expression in L2 | 267 Discourse contexts/factors and variability affecting subject expression in L2 | 267 Effects of verbal lexical semantics and variability of subject expression in L2 | 268 Common findings and concluding remarks | 269

References | 273 Index | 283

Abbreviations used in glosses 1 2 3 ACC COND DAT EMPH FUT IMP INF OBL PL PRET pro REFL SG SUBJ

first person second person third person accusative case (direct object) conditional dative case (indirect object) emphatic future imperfect infinitive oblique object plural preterit null pronoun reflexive marker singular subjunctive

Abbreviations used in text AGR AgrSP CF CFC Def NP DP EPP ExpS FCH Foc FocP Gen NP GH Ind NP INFL IP L1 L2 N NP NSP OPC PAH ProS Prp QDP SLA Spec T TAM TP UG V VP

agreement subject agreement phrase contrastive focus contrastive focus constraint definite noun phrase determiner phrase extended projection principle null expletive pronoun functional compensation hypothesis focus node focus phrase semantically general noun phrase givenness hierarchy indefinite noun phrase inflection inflectional phrase first language second language noun noun phrase null subject parameter overt pronoun constraint position of antecedent hypothesis referential pronominal subject proper noun quantifier determiner phrase second language acquisition specifier tense node tense, aspect and mood tense phrase universal grammar verb node verb phrase

List of tables Table 1 Table 2 Table 3 Table 4 Table 5 Table 6 Table 7 Table 8 Table 9 Table 10 Table 11 Table 12 Table 13 Table 14 Table 15

Table 16

Table 17

The generative approach. | 6 Specific research agenda of the parametric approach in L2 acquisition. | 8 Specific research questions of the syntax-pragmatics interface model in L2 acquisition. | 10 Specific research questions of processing models in L2 acquisition. | 12 Research questions of discourse-pragmatic approaches in L2 acquisition. | 14 Research questions of the variationist approach in L2 acquisition. | 16 Specific research questions in the L2 acquisition of subject expression from a variationist approach. | 16 Verbal paradigm in Spanish for present and preterit morphology (–ar verbs). | 22 Syntactic properties of subjects in pro-drop and non-pro-drop languages. | 32 Distribution of null and overt subjects and discourse-pragmatic notions in Romance languages. | 36 Summary of cross-linguistic influence from bilinguals’ non-pro-drop to pro-drop languages. | 88 Summary of results for lexical and pragmatic principles in L2 subject acquisition. | 106 Monolingual studies testing the Position of Antecedent Hypothesis (PAH). | 134 Percentage selection of subject [Spec, IP] as antecedent of null and overt pronouns. | 142 Percentage of overt subject pronouns interpreted as subject antecedent in English according to discourse structure – English speakers. | 146 Percentage of null and overt subject pronouns interpreted as subject antecedent in Spanish according to discourse structure – Spanish speakers. | 146 Percentage of null and overt subject pronouns interpreted as subject antecedent in Spanish according to discourse structure – L2 learners. | 147

List of tables |

Table 18

Table 19

Table 20

Table 21

Table 22 Table 23 Table 24

Table 25 Table 26

Table 27 Table 28 Table 29 Table 30 Table 31

xiii

Summary of native speaker use of forms occurring in subject position according to character role and anaphoric hierarchy. | 161 Summary of native speaker and learner use of forms occurring in subject position for main character according to anaphoric hierarchy. | 164 Summary of native speaker and learner use of forms occurring in subject position for secondary character according to anaphoric hierarchy. | 165 Summary of native speaker and learner use of forms occurring in subject position for tertiary character according to anaphoric hierarchy. | 167 Overall pronoun distribution for first person singular subject pronoun referential contexts. | 189 Distribution (%) of subject pronouns by person and specificity. | 226 Summary of Gudmestad, House, and Geeslin’s (2013) rank for number and specificity (order of estimated posterior probabilities for subject expression). | 228 Summary of Abreu’s (2009) results for the effect of priming on overt subject pronouns (three participant groups). | 236 Summary of Geeslin and Gudmestad’s (2011) results for the effect of perseveration (priming) on overt subject pronouns (three participant groups). | 238 Summary of findings of parametric approach. | 250 Summary of findings of syntax-pragmatics interface model. | 257 Summary of findings of the processing model. | 261 Summary of findings of the discourse-pragmatics model. | 266 Summary of findings of the variationist model. | 269

List of figures and graphs Figure 1 Figure 2

Pronominal feature geometry analysis. | 115 Forward and backward anaphora picture interpretation task. | 137

Graph 1

Summary of SpecIP antecedents by discourse structure and pronoun type. | 148 Distribution of overt first person subject pronoun by pragmatic rule – all groups. | 190 Distribution of null first person subject pronoun by pragmatic rule – all groups. | 190 In focus status (expect null pronouns). | 198 Type identifiable status (expect indefinite NP). | 199 Activated status (expect overt pronouns). | 199 Familiar status (expect definite NP). | 200 Uniquely identifiable status (expect definite NP or proper noun). | 200 Activated and recoverable status (expect null subjects). | 201 Activated non-recoverable status (expect semantically general NPs). | 202

Graph 2 Graph 3 Graph 4 Graph 5 Graph 6 Graph 7 Graph 8 Graph 9 Graph 10

Acknowledgements I am indebted to the many individuals and institutions that have contributed either directly or indirectly to this book seeing the light of day. First, a special thanks goes to Peter Jordens of the Max Planck Institute for Psycholinguistics and SOLA editor, for his enthusiasm and encouragement for this project from the beginning; to Lara Wysong, the project editor at Walter de Gruyter; and to anonymous reviewers. A UCMEXUS Collaborative Research Grant (No. CN0255) from the University of California and CONACYT (Mexican National Council of Science and Technology) allowed me to compile, with my colleague and co-principal investigator, Robert Blake (University of California Davis), the rich database that served as the catalyst for two of the analyses reported on in chapter six of this volume, as well as many of the examples that appear throughout the chapters. My deep gratitude goes to the participants of this study from the Universidad Autónoma de Querétaro, the University of California Davis, and the University of Georgia, and especially to my graduate research assistants, Matthew Kanwit, Adam Lassila, Abnériz Meléndez Cuadra, Wendy Orozco Lecona, Aránzazu Pascual Ortiz, Benjamin Schmeiser, and Eve Zyzik, who tirelessly collected and transcribed hours of recorded narratives and organized or reorganized reams of data. My friend, colleague, and co-author, Sarah Blackwell, deserves an enormous amount of credit and gratitude. It was she who first challenged me to consider the complexity of pragmatic rules that guide and determine the use and acquisition of Spanish subjects. Together we explored the seemingly irrational and unpredictable uses of overt and null subjects in the oral narrative structure of both native and second language speakers and discovered that order and predictability could be found in the chaos of real data. My sincere appreciation goes to my students and colleagues in the Department of Romance Languages at the University of Georgia where the intellectual environment is both inspiring and supportive and the numerous casual and formal academic exchanges with fellow linguists have infinitely enriched my research. Thanks are also due to my friend, colleague and head of the department, Stacey Casado, who encouraged me every step of the way in the long and arduous process of completing this book. Immense gratitude is also owed to the many colleagues whose scholarship has directly stimulated mine and whose research is discussed in this volume; I hope I have done their body of work justice. I would like to name specifically those who have provided me with invaluable feedback or additional materials, both published and unpublished; they are: Sarah Blackwell, Kimberly Geeslin, Aarnes Gudmestad, Nuria López-Ortega, Cristóbal Lozano, Silvina Montrul, Diana Ranson, Jason Rothman,

xvi | Acknowledgements and Antonella Sorace. My heartfelt thanks also go to my son-in-law, Kyle Williams, for technical support with the graphics. I am grateful for and to my children, Sarah Quesada Williams, Sergio J. Quesada, and Amanda Quesada. They did not always make academic life easy when they were growing up, but they have filled my personal life with endless joy. Most of all, I am eternally indebted to my best friend, life-long companion and husband, Sergio Quesada. During the preparation of this book, he endured many long evenings, weekends, and vacations without his wife mentally present and yet despite it all, continued to inspire and cheer me on to its completion. He has my everlasting devotion and gratitude. Finally, as I write these words, a new family member is on the way but I do not know her name yet. By the time this book is published, however, she will have arrived. I dedicate this book to her – mi nieta.

1 Introduction: History and theories of second language acquisition Introduction Spanish is a pro-drop or a null subject language where an overt subject is not syntactically obligatory and in most cases the verbal morphology makes clear who or what the subject is. Nonetheless, research has shown that null and overt subjects are not in free variation but rather are subject to a complex interaction of linguistic, discursive and perceptual properties for their use and distribution. This causes difficulties for late (adolescent and adult) learners of Spanish and especially for, but not necessarily restricted to, those whose first language (L1) is nonpro-drop such as English. Pronominal and lexical subjects are among the most widely studied expressions in the acquisition of Spanish as a second language and furthermore, due to their multi-pronged complexity, have been studied from a wide variety of theoretical perspectives. In Spanish the form of a subject (null, pronominal, lexical, or clausal) and its placement (null, preverbal or postverbal) depend on complex semantic and discourse-pragmatic constraints. This leads to multiple questions regarding what guides the acquisition process. Just a few of the questions addressed in research include: 1) Do the difficulties lie in how the parametric properties of the structure are set (or re-set)? 2) Do acquisition problems lie at the interface between syntactic properties and discourse-pragmatic constraints of use? 3) Does the semantics of the verb affect subject use and placement? 4) What role do person, number, tense, mood, aspect, etc. play in the use of one particular form over another? 5) How do speakers and learners interpret pronominal reference when there are two or more possible antecedents in complex sentences? 6) What effect does referent saliency have on the subject form used? A plethora of questions regarding the structure will necessarily lead to a variety of theoretical perspectives and methodologies employed for the study of its acquisition. Thus, it is the purpose of this book to examine research over the past two decades on the second language (L2) acquisition of Spanish subjects, both pronominal and lexical, from several theoretical perspectives, including generative approaches (specifically parametric and syntax-pragmatics interface models), processing theory, discourse-pragmatics, and the sociolinguistic-variationist model. This volume has two overriding goals; the first is to understand what it is that L2 learners are capable of acquiring in terms of subject expression in Spanish. In other words, what specific linguistic properties and extra-linguistic factors promote and constrain acquisition and use, how does this acquisition compare to that of native

2 | 1 Introduction: History and theories of second language acquisition speakers, and why do learners, even at the advanced level, display lingering nonnative behavior in the use of this structure? The second goal is to identify the distinct contributions of each theoretical perspective and the comparable findings among them in an attempt to find common ground among researchers of multiple perspectives in L2 acquisition research. One might wonder why and if it is a worthwhile endeavor in the 21st century to seek common ground among diverse theories in formal linguistics and language acquisition studies. No student or practitioner in contemporary theoretical linguistics and/or the field of L2 acquisition is unaware of the sharp philosophical and methodological divide that has developed over the past generation between formal-generative and functional-discourse perspectives of language. Researchers working from within one framework often do not examine or discuss the work of colleagues working from opposing models, unless it is to point out their methodological or explanatory shortcomings. For those of us in graduate school in the 1960’s, 1970’s and even the 1980’s, who can forget the linguistic wars¹ that seethed in theoretical circles or the second language theory war² that raged in the 1990’s? Hubris aside, much of the divide is simply because different approaches ask different questions related to the properties of language and its acquisition. Fortunately, for the field of Spanish linguistics and Spanish second language acquisition in particular, opposing theoretical points of view are less divisive; indeed I would go as far as to say that Spanish linguistics and applied linguistics are marked by a more comprehensive attitude towards diverse theoretical perspectives, as witnessed by the inclusion of multiple perspectives in several current volumes on both theoretical and applied issues in Spanish.³ In the area of language acquisition, researchers from several distinct perspectives seek to understand what lies at the heart of learners’ acquisition of certain properties of a subsequent language and whether or not this process is constrained by internal linguistic properties of the first or second language or external, non-linguistic factors, such as the nature and context of the linguistic input, the discourse context in which the structure is used, and speakers’ intentions and cognitive state of mind. Although the questions and goals guiding different research paradigms

1 See the volume, The Linguistics Wars (Harris 1993) for a detailed description of the famous disputes in theoretical linguistics in the 1960’s and 1970’s. 2 See the volume, Theory Construction in Second Language Acquisition (Jordon 2004) for a brief description of disputes in the 1990’s between generativists and variationists regarding second language acquisition theories, as well as volume 11 of the journal, Applied Linguistics. 3 The reader is directed to two recent volumes that include studies of Spanish from multiple theoretical perspectives: The Handbook of Hispanic Linguistics (Hualde, Olarrea & O’Rourke 2014) and The Handbook of Spanish Second Language Acquisition (Geeslin 2014).

1.1 Formal/generative approaches |

3

may vary, the subject matter of this volume, the L2 acquisition of Spanish subject expression, brings many of the same issues to light among the different perspectives. It is indeed rare for one particular structure to receive such detailed analyses from such a wide variety of perspectives. An examination and analysis of this diverse body of research can thus bring us closer to a unified account of the acquisition and use of Spanish subject expression and of L2 acquisition processes. It will be argued throughout this volume that central to native speaker use and L2 acquisition of subject expression in Spanish are universal properties that range from highly specific syntactic principles to more general characteristics of human cognition and that the range of properties is responsible for language acquisition and use. In the following sections, I examine the objectives and issues important to each particular theoretical model, discuss the underlying research questions and examine the contributions of each to our understanding of language and L2 acquisition. Although not intended to be an exhaustive review of all the theories in L2 acquisition, the chapter is necessary for the reader to understand how a diverse range of theoretical perspectives informs research in the acquisition of Spanish subjects. Furthermore, and I believe this is a credit to the field of second language acquisition of late, many of the approaches share multiple features - so much so that it is oftentimes difficult to categorize the theoretical approach of many of the studies reviewed here. Therefore, rather than depending on a variety of very distinct theories which are mutually exclusive, our field appears to be coming to a consensus that multiple perspectives, multiple questions and issues need to be examined in order to understand the complex and interrelated processes related to second language acquisition in the 21st century.

1.1 Formal/generative approaches 1.1.1 Universal grammar and language acquisition The generative enterprise views language structure as a specific cognitive domain of the human mind, independent of other types of cognition, although it interfaces with other domains in important ways. What this means is that humans have, and are born with, a biological endowment that determines very specific abstract properties of language, that these properties are universal in nature, and their development or acquisition do not depend on the development of other aspects or types of cognition. The central concept of generative grammar is that language consists of a system of principles, conditions and rules, which are elements or properties of all human languages, and this is not due to a mere accident but is because of necessity (Chomsky 1975). This system constitutes a Universal Gram-

4 | 1 Introduction: History and theories of second language acquisition mar, which according to Chomsky forms the basis for a theory of innate principles that comprises the initial state of language acquisition in children. Therefore, the goals of explaining this Universal Grammar (UG) are, in very general terms, to describe language as a quality of the human mind, and to explain its acquisition. In the principles and parameters model, it is set forth that UG consists of a set of principles that exist in all languages and another set of parameters that vary from language to language. Languages differ in how they set those parameters. In the more recent Minimalist model, UG contains an abstract set of principles and constraints, where the properties are “linked to the specification of features of individual lexical items, more specifically, functional categories (inflection, tense, complementizers, determiners)” (Rothman and Pascual y Cabo 2014: 49). In other words in the Minimalist model, the feature specification of lexical items determines syntactic operations and differences in languages can be explained by variations in these feature specifications. Thus, Chomsky claims that children are born with very specific innate knowledge of what is structurally possible in a human language and what is not possible. It is this knowledge that constitutes UG in all human beings. Language acquisition is a process whereby children learn how particular parameters or feature specifications apply to their language. In the principles and parameters approach, this innate knowledge of what is possible in a language determines which way the parameters are set, whether they are set for a positive value or a negative one. In the minimalist model, this knowledge determines how rules are interpreted. The study of child language acquisition is especially relevant to the generative model in theoretical linguistics because it is claimed “that children exhibit knowledge of language which cannot have been derived by observation of, or by induction from, their linguistic environment” (Thomas 2004: 3). Thus it holds that an examination of what the child knows and what she brings to the undertaking of learning her language can deliver important insights into what constitutes the core of universal grammar (Thomas 2004: 3). In other words, there is an innate linguistic faculty responsible for the acquisition of the native language. This accounts for, in part, the logical problem (also referred to as the learnability problem or the projection problem) of language acquisition, which attempts to justify the observed discrepancy between the complexity of a human language and the apparent ease with which a child acquires any language to which she is exposed in such a short time. This observation has also lead to the proposition of the poverty of the stimulus hypothesis, which explains how the child is able to learn such a complex system in a relatively uniform way when exposed to such a wide variety of inadequate (some more than others) input situations. Thus, the UG approach does not eliminate the need for environmental input; indeed linguistic input is a necessary trigger for the development of primary language acquisition. Rather,

1.1 Formal/generative approaches |

5

UG knowledge is viewed as the mediator “between the input (or primary linguistic data), the intermediate grammatical systems the child constructs until he reaches the final steady state (adult grammar), and the output (what the child produces)” (Montrul 2004a: 4). In other words, upon exposure to language-specific data, the child selects the options of UG that are in accord with the language of her linguistic environment. In terms of second language acquisition, if UG grammar imposes restrictions on what is possible in a human language, then it necessarily must impose these same restrictions on the second language system (i.e. the interlanguage). Generative approaches in L2 acquisition ask whether or not the principles of UG are available to adult learners and to what extent they guide the L2 learning process even in the face of underdetermined input, and whether or not the first language (L1) works in tandem with or against UG principles. Traditionally, two positions have been proposed regarding the relationship between UG and L2 acquisition. There are those who have proposed that if UG is a set of principles of languages derived from the human mind, then it must necessarily be available throughout life; therefore adult learners have direct access to UG in the same way that children do and the initial state of language acquisition is UG. In this view, language learners make no assumptions about the structure of the L2. Others have claimed that although UG is the initial state for L1 in early childhood (both monolingual and bilingual acquisition), for adult L2 acquisition the initial state is the L1 and UG is accessible to the learner and constrains interlanguage development only in cases where the L1 knowledge can no longer accommodate the input or provide a suitable representation. This has been termed the Full Transfer/Full Access Model (White 1985, 1989; Schwarz and Sprouse 1996; Montrul 2004a). Therefore, learners have access to UG but only as constrained by the rules of their L1. For example, UG stipulates that languages can be either pro-drop or non-pro-drop. Children acquire the setting for their language upon exposure to linguistic input and when they attempt to learn an L2 as an adult, they assume that the L2 has the same setting as their L1 for pronouns. In this case, adults have to “unlearn” the L1 setting in order to “reset” the L2 parameter. A third position has also been set forth and this is attributed to those who deny that UG plays any role at all in adult L2 acquisition. In this position, it is claimed that the innate capacity is available only to children up until a certain age and after that, the capacity is no longer available; thus, the adult language learner only has the principles of UG that are instantiated in the L1. Proponents of this stance have called it The Fundamental Difference Hypothesis or the Partial Access Model (Schachter 1990; Bley-Vroman 1990a, 1990b). The claim is that in adulthood there is no access to UG and adult language learners must resort to other cognitive capacities for learning an L2, such as hypothesis testing, analytical problem solving, analogy, etc. Consequently, research in the generative model focuses on addressing these issues and looking for evidence to

6 | 1 Introduction: History and theories of second language acquisition support one claim or another. In other words, if UG represents the initial state of knowledge of a child learning her first language, what we want to know in second language acquisition is ultimately, what constitutes the initial state of the adult learner and what can account for the fact that the end state is wholly unlike that of the child. In other words, children all become native speakers (barring unusual circumstances) of a language, but adults rarely (if ever) become native speakers of a second language. A summary of the tenets, goals and issues of the generative approach as they relate to second language acquisition appears in Table 1. Table 1. The generative approach. Major tenets: – A formal system of syntactic rules independent of other aspects of cognition. – A set of principles shared by all languages. – Cross-linguistic differences due to how principles or parameters are applied to languages or differences in feature specifications of lexical items. – The capacity for the development of language is innate; syntactic information is specific and hard-wired (i.e. predetermined) in the brain. Goals of generative approach: – To describe language as a quality of the human mind – To explain its acquisition Role of UG to L2 acquisition: – Learners have direct access to UG; i.e. the initial state is UG. – Learners have partial access to UG, which is accessible only through the L1; i.e. the initial state is the L1 but the interlanguage is constrained by the principles of UG. – In adulthood there is no access to UG; the innate capacity is available only to children up until a certain age.

1.1.2 The parametric approach in second language acquisition The principles and parameters model of Chomsky (1980, 1981, 1982, 1986) was set forth in order to “account for both the diversity of languages and for the rapid and uniform development of language among children” (Flynn 1990: 91). In other words, the model attempts to explain why there are cross-linguistic differences among languages despite the fact that all languages have a universal grammar (basic abstract principles) in common, and to provide a theoretical explanation for how children are able to acquire their language in such a short amount of time and in such a seemingly effortlessly manner given the indeterminateness

1.1 Formal/generative approaches |

7

of the input (i.e. given the poverty of the stimulus). It is this particular model of the generative theory of language that was first put to the test in the study of L2 language acquisition. What are parameters? Parameters are a subset of apparently unrelated properties or structures that are governed by an abstract principle of the grammar (Ayoun 2003: 9). For example, the null subject parameter or pro-drop parameter (the topic of Chapter 3 of the present volume), one of the most widely studied of the parameters in Spanish and other Romance languages, has been hypothesized to subsume several properties such as allowing both overt and null referential subjects in tensed clauses (due to a rich verbal morphology), requiring null expletive subjects (pro Llueve. ‘pro Is raining.’), allowing subject-verb inversion (María llega temprano ‘Mary arrives early’ or Llega temprano María ‘Arrives early Mary’), and violating the that-trace filter in sentences such as ¿Quién crees que verá a Juan? (‘Who do you think will see John?’)⁴. Thus, whereas Spanish is set for the positive value of the parameter, and possesses all of these properties, English is set for the negative value. Therefore, English does not allow for null referential subjects, does not have a rich verbal morphology or allow for subject-verb inversion, and adheres to the that-trace filter. The claim of the principles and parameters model is that children learning Spanish as their L1 set the parameter for [+pro-drop] and by doing so learn all of the properties of the parameter as a cluster; they do not learn each property separately or one by one. In the case of English speakers learning Spanish as adults, the task at hand is to reset the parameter from a negative to a positive value based on the evidence in the input. It goes without saying that the opposite is true of native Spanish speakers learning English as a second language, where learners have to “unlearn” the value of a positive parameter and reset the value to a negative one. The argument in the principles and parameter approach is that L2 learners use the same principles of syntactic organization that are available in L1 acquisition; however, when there is a mismatch in the values between the L1 and L2 parameter, the task of the learner is to reassign a new value to that parameter so as to coincide with the evidence in the input. When the values of the L1 and L2 match, there is no need to reset the value. The principles and parameters approach also subsumes markedness theory (Hyams 1986), which claims “core grammar is the grammar that is determined by fixing the parameters of UG in one of the permitted ways. Peripheral grammar is the set of marked elements and constructions that

4 The that-trace filter blocks the extraction of subject wh-words with overt complementizers in English but not in Spanish. See Chapter 2 for a more complete description of all these properties as they relate to the pro-drop parameter.

8 | 1 Introduction: History and theories of second language acquisition are outside of core grammar, but still constitute what is actually represented in the mind of an individual” (Liceras 1989: 109). In terms of L2 acquisition, it has been proposed that learners begin the acquisition process with the unmarked setting of a parameter regardless of its value in the L1. For example, Liceras (1989) and Hyams (1986) have proposed that for the pro-drop parameter, the unmarked value is the positive one (i.e. [+pro-drop]. Therefore, markedness theory would predict that English speakers learning Spanish would start with the L2 setting rather than the L1 setting for learning the properties of the pro-drop parameter. Tsimpli and Roussou (1991), however, maintain that the pro-drop parameter does not derive from UG but rather is associated with the functional model, which “is inaccessible to the adult L2 learner” (Tsimpli and Roussou 1991: 151). Finally, Liceras has suggested that the acquisition of the properties of certain parameters may not be acquired as a cluster but rather in an implicational hierarchical manner. This is to say that one principle may imply the acquisition of another principle but not necessarily the other way around (1990: 110). It has been claimed that this model allows us to explain both the similarities of the L1 and L2 acquisition processes, as well as account for the differences in ultimate attainment between native and non-native speakers of languages, and for the individual differences and varying success rates among adult L2 learners (Flynn 1989: 93). As the first model of the generative theory of language to be put to the test in the field of L2 acquisition, the basic tenets and issues are, of course, those that were previously presented in Table 1. However, the parametric approach within the generative model has, in addition, its own very specific research questions and these are presented in Table 2 below. Table 2. Specific research agenda of the parametric approach in L2 acquisition. –

– – –

Do adult L2 learners start the learning process with the same initial state as the child and set the parameters of their L2 as a child does in L1 acquisition or do they start with their L1 setting and gradually reset the parameter to the L2 option? Does the learning of one property lead to “automatic” learning of a cluster of structural properties associated with any given parameter? If the learning of a cluster of structural properties is “automatic”, is it learned all at once or in a hierarchical manner along an implicational path? What is the role of input in the triggering or “re-setting” of parameters?

1.1 Formal/generative approaches |

9

1.1.3 Interface approaches Also included among the generative perspectives and a more recent model is the syntax-pragmatics interface approach. This model is considered generative in that it maintains that syntactic domains are modular and that the grammatical properties of these domains are determined by functional features. However, the model differs from previous generative models in that it claims that grammatical properties that are completely syntactic are in theory learnable by non-native speakers, but that those “that require integration of modular linguistic domains with domains of general cognition are subject to residual optionality” (Rothman and Pascual y Cabo 2014: 57). In other words, syntactic structures that lie at the interface of non-syntactic properties for their interpretation and use do not ever become native-like. It is proposed that the acquisition of new syntactic features places higher demands on speakers in terms of managing the different cognitive resources. This model has been proposed as a way to explain why the end result of L2 acquisition is non-uniform, unlike the uniform attainment of children’s L1 acquisition. The interface hypothesis was first proposed in the study of pro-drop languages by Sorace and Filiaci (2006) who observed that even highly advanced or near-native adult learners of Italian as an L2 showed patterns of non-convergence and residual optionality (i.e. non-native-like behavior) in their interpretations of overt and null subject pronouns, a syntactic structure whose appropriate use is highly dependent upon pragmatic and contextual variables. Sorace points out The term ‘interface’ refers to syntactic structures that are sensitive to conditions of varying nature [. . . and] the meaning of the term therefore denotes the fact that these conditions have to be satisfied in order for the structure to be grammatical and/or felicitous. Thus, the interface between the structure and the domain that defines the conditions on its grammaticality and/or felicity is critical for its appropriate use. (Sorace 2011: 6)

Studies in L2 acquisition based on the interface model seek to examine to what extent learners acquire syntactic and pragmatic properties at the same time or in a subsequent manner, and why pragmatic deficits linger even in very advanced learners. Research from this model recognizes that although L2 learners are able to acquire, produce and comprehend native-like morphosyntactic structures, there is also observable variability in their development such that the end result of L2 acquisition is non-uniform, unlike the uniform attainment of L1 acquisition. In the study of the null subject or pro-drop parameter, research has turned to the complex (and therefore, late-learned) pragmatic-discourse properties of language distribution. Some investigators have claimed that transfer from a pragmatically less complex non-pro-drop language to a pragmatically more complex pro−drop

10 | 1 Introduction: History and theories of second language acquisition one is unidirectional (Sorace 2004; Serratrice, Sorace, and Paoli 2004; Sorace, Serratrice, Filiaci, and Baldo 2009) and that this accounts for non-native-like production and comprehension. The claim is that acquisition at the syntax-pragmatic interface is particularly vulnerable among both children and adults and that formal morphosyntactic properties are acquired before the discourse-pragmatic constraints. The interface studies explored in this volume all take into account, to some degree (some more than others), the processing difficulties that emerge at the interface of syntax and pragmatics. Nonetheless, whereas some researchers focus on the interface itself, others center more on the psycholinguistic effects of processing at the interface. Therefore, in this volume I believe that processing studies deserve examination apart from the others of the generative model. The specific research questions that the interface hypothesis seeks to answer are presented in Table 3. Table 3. Specific research questions of the syntax-pragmatics interface model in L2 acquisition. – – –

Do adult L2 learners acquire the formal morphosyntactic properties before the discoursepragmatic properties at the same time or in a subsequent manner? Why do discourse-pragmatic deficits linger even in very advanced learners who apparently have learned the formal morphosyntactic properties of a given structure? Why is the end result of L2 acquisition non-uniform among adult learners, unlike the uniform attainment of children’s L1 acquisition?

1.2 Processing models Processing models have a long tradition in L2 acquisition studies and share with UG studies an attempt to answer the questions regarding the extent to which innate restrictions and/or L1 principles influence how learners process or interpret L2 structures that are similar to or different from their L1. Indeed, as Dowens and Carreiras (2014) point out, “the first psycholinguistic studies of L2 Spanish were carried out to validate predictions from a generativist framework, with researchers examining issues of the nature and extent of L2 access to Universal Grammar [. . . ] and the role of the first language [. . . ] in the initial state and development of the L2” (2014: 64). Furthermore, recent studies have examined the syntactic-pragmatics interface and its impact on L2 acquisition. Nonetheless, psycholinguistic studies in L2 acquisition have mushroomed over the past two decades with diverse theoretical agendas (i.e. generativist, emergentist, connectionist) and with a myriad of experimental and technological designs (i.e. offline

1.2 Processing models |

11

or online production or reading tasks, eye-tracking, neuroimaging). On the other hand, what they have in common is the focus on exploring the relationship between language structure and cognitive processing. More specifically studies from this perspective seek to explain L2 deficits as the result of the difficulties in the processing of different types of information found at the interface between syntax and other cognitive domains. Processing can be examined from two distinct perspectives. The first views processing as part of acquisition and mediates between input data and universal linguistic properties (i.e. Universal Grammar). Thus, the object of processing is the acquisition of the formal features of language that leads to alterations in a learner’s underlying linguistic system. Another perspective accepts that certain underlying knowledge is already present in the learners’ grammar and seeks to understand how speakers/learners use this knowledge in interpreting certain features of the language. It is this latter perspective that is of interest in the study of the acquisition of subject expression in L2 Spanish. Processing studies in anaphoric interpretation have revealed that monolingual English speakers prefer to link a subject pronoun (the most phonetically minimal form available in the language) to the most discourse-prominent antecedent, the sentential subject (Ariel 1994). In one study, when asked to read two-sentence discourses that included a previous sentence that introduced two referents (Jack apologized profusely to Josh) and either a second sentence that started with a subject pronoun that referred to the subject of the previous clause (He had been rude to Josh) or one that referred to the object (He had been offended by Jack’s comment), it was found that participants processed the second sentence faster when it referred to the subject rather than the object (Hudson-D’Zmura and Tanenhaus 1998). Monolingual speakers of pro-drop languages tend to link the null subject to the most discourse-prominent antecedent and the overt subject pronoun to an antecedent other than the sentential subject. Because languages differ in the resolution of overt subject pronouns, “these competing strategies pose a potential conflict to bilinguals who speak a null-subject language alongside a non-null-subject language” (Keating, VanPatten, and Jegerski 2011: 199) and perhaps present an even greater conflict for L2 learners of pro-drop languages whose native language is non-pro-drop. Recent research in pro-drop Romance languages (Spanish and Italian) has focused on the difficulties in interpreting discourse-pragmatic conditions that constrain subject expression. In examining the interpretation strategies of subject pronouns, researchers assume that both overt and null subjects exist in the underlying grammar of bilingual speakers and language learners. It is also assumed that the related syntactic properties of at least rich verbal morphology and subjectverb inversion (and perhaps other properties of the null subject parameter) are

12 | 1 Introduction: History and theories of second language acquisition already present in the L2 Spanish grammar. What is not assumed is that the existence of the null subject parameter will automatically result in native-like processing strategies for interpreting antecedent resolution with different types of pronominal subjects (VanPatten and Jegerski 2010). Thus, it is held that the studies reviewed in this volume are clearly processing due to the fact that the central objective is to establish non-native (and native) speakers’ interpretation strategies when resolving conflicting (L1 and L2) syntactic cues. Table 4. Specific research questions of processing models in L2 acquisition. – – – –

Do innate restrictions or L1 principles influence how L2 learners process or interpret L2 structures that are similar to or different from their L1? How does processing mediate between input data and universal linguistic properties (i.e. Universal Grammar) in adult L2 acquisition? How do adult L2 learners (and bilingual speakers) process different types of information found at the interface between syntax and other cognitive domains? What strategies do adult L2 learners (and bilingual speakers) use for resolving conflicting (L1 and L2) morphosyntactic cues?

1.3 Discourse-pragmatic approaches The L2 acquisition of pragmatic notions has long been relegated to the study of what has been termed pragmatic competence, understood by some “as knowledge of the linguistic resources available in a given language for realizing particular illocutions, knowledge of the sequential aspects of speech acts and finally, knowledge of the appropriate contextual use of the particular languages’ linguistic resources” (Barron 2003: 10). Pragmatics is an area of language competence much disregarded by generativists, who up until recently were concerned solely with the knowledge of the rules of grammar which were claimed to be independent of contextual appropriateness. Recent research in L2 acquisition is now more in line with Hymes’ (1972) claim that competence includes not only knowledge of syntactic structures but also the capacity to use this knowledge in discourse appropriate ways. However, much L2 research from a pragmatic perspective has been grounded in speech act theory and discourse analysis and is concerned with learners’ comprehension and production of speech acts such as requests, offers and refusals, all based on the pragmatic notion of politeness (Barron 2003), or discourse markers that function as cohesive devices (Kasper and Rose 2002). And indeed, this wide body of research has shown that the use and acquisition of

1.3 Discourse-pragmatic approaches | 13

appropriate communicative units (i.e. speech acts and discourse markers) are subject to discourse-level constraints as they are in native-speaker speech. However, it has been shown that not just speech acts and discourse markers but also very specific syntactic structures are governed by discourse-level rules. López Ortega points out that grammatical structures that appear to be acquired at the sentence level continue to be used in non-native ways when examined at the discourse level (López Ortega 2002: 3). Subject expression is particularly sensitive to discourse context, both conversational (López Ortega 2002) and narrative (Lubbers Quesada and Blackwell 2009). Thus, because linguistic competence includes the knowledge of discourse constraints or pragmatic rules that guide appropriate language use and acquisition, specifically “. . . how coherence and sequential organization in discourse is produced and understood” (Levinson 1983: 286), research from a discourse-pragmatic approach necessarily considers language beyond the level of the sentence and examines larger portions of discourse or text. Research on the acquisition and use of subjects from this framework in both native speaker and L2 learner language is based on a wide variety of discourse notions that affect language use. These include, but are not limited to narrative cohesion and coherence, referent saliency, cognitive focus, contrastive focus, pragmatic weight, and abstract properties such as cognitive status. This framework takes into account the role, familiarity, relative importance and function of referents in narrative structure. For example, research has found that the use of pronouns and noun phrases in narrative discourse adheres to a universal anaphoric hierarchy (indefinite noun phrase > definite noun phrase > proper noun > overt pronoun > null pronoun), based on what the speaker expects the addressee to already know. The higher up in the hierarchy a referring expression is, the less information the speaker assumes the listener has in regards to the expression’s potential referent. Thus, the use of an indefinite noun phrase presupposes the listener has no information regarding the referent, whereas the use of a null pronoun presupposes the maximum amount of information (Saunders 1999). The discourse-pragmatic model seeks to understand how speakers mark contrast, emphasis or disambiguation, in an effort to create cohesive discourse; however, researchers also examine discourse conditions that lie beyond these traditional pragmatic notions (of contrast, emphasis and disambiguation) and look at discursive aspects as they relate to speaker identity, intention and attitude, topic involvement, and narrative structure. In the study of subject expression, two notions are particularly prominent: focus and presupposition. These are pragmatic notions that affect subject pronoun use among all persons (first, second and third person) and number (singular and plural), but notions such as speaker commitment, intention and evaluation are pragmatic notions only available to speakers themselves and thus can determine only first person (and mostly first

14 | 1 Introduction: History and theories of second language acquisition person singular) subject pronoun use. Work that centers on the abstract notions of cognitive status is based on the underlying assumptions cooperative speakers make about the informativeness and saliency of referents in discourse and assume that greater referent saliency corresponds to more minimal referring expressions (Levinson 2000). Like the anaphoric hierarchy, the Givenness Hierarchy proposes a set of implicationally related statuses also based on assumptions speakers make about the informativeness and saliency of referents in discourse (Gundel, Hedberg and Zacharski 1993). Greater referent saliency corresponds to the use of more phonetically minimal referring expressions and lesser referent saliency is consistent with more phonetically elaborate expressions (Ariel 1990; Givon 1983; Levinson 1987, 2000). Like the formal-generative and processing models, the discourse-pragmatic model asks to what extent universal properties are responsible for language acquisition and use. However in this case, they are general cognitive abilities related to the notions of textual cohesion and coherence, saliency, focus, memory, attention state, and speaker involvement that all appear to guide use and L2 acquisition. The specific questions that are of concern to researchers working within a discoursepragmatic approach are presented in Table 5. Table 5. Research questions of discourse-pragmatic approaches in L2 acquisition. – – – –

What are the discourse constraints or pragmatic rules that guide language use and adults’ L2 acquisition in discourse? How does knowledge of information structure in discourse contribute to appropriate use and acquisition in adult L2 learners? To what extent do forms acquired and used appropriately in discourse adhere to implicational hierarchies proposed to account for speaker presupposition and cognitive state? To what extent is this knowledge related to universal cognitive abilities (and which ones), and to what extent are the constraints that guide use and acquisition language specific?

1.4 The variationist approach The variationist model has a long tradition in L2 acquisition studies in English but is new to the methodological inventory of studies in Spanish L2 acquisition; studies from this perspective document the frequency of a variant and its conditioning linguistic and social factors in order to understand better the L2 acquisition process of variation. Like the discourse-pragmatic approach outlined above, the variationist approach also examines language beyond the sentence level. Dif-

1.4 The variationist approach |

15

ferent from the former approach however, variationist research often takes into account extra-linguistic social factors, such as social or economic status, social distance, and social context. Early research from this approach attempted to understand how learners develop the ability to use competence in different kinds of language situations. It was claimed that by analyzing what learners do systematically, we can infer a great deal about their competence, both grammatical and communicative. In this sense, output that displays variability, both systematic and unsystematic, can give us clues as to the competence and developing competence of language learners (Ellis 1985, 1989). Variationist studies in second language acquisition have two characteristics in common. The first is that they use as their models variationist studies in adult native language or sociolinguistic studies. Many of these studies advocate for an analysis of the variation of learner interlanguage at different levels (e.g. phonological, syntactic, lexical) while considering a variety of sociolinguistic variables (e.g. social context, task, topic, interlocutors, attention to task or form, ethnolinguistic background) in order to examine how these factors interact with purely linguistic factors, such as linguistic environment (phonological or syntactic) and function-form relationships (Preston 1989, 1993). External factors influence the degree to which speakers focus on the way that they speak, producing a more or less formal or careful style of speech. Some claim that L2 learners, especially in the beginning stages, do not have sufficient linguistic competence in the second language to be sensitive to many relevant external social contextual factors; specifically, they do not have the competence to manipulate a variety of styles ranging from informal to formal (Hulstijn 1989; Preston 1989). Variation is thus believed to be due to other factors. Therefore, the second characteristic of variationist models in second language acquisition, especially as they apply to subject expression, is a focus on internal linguistic features and external discourse-pragmatic features. Other researchers distinguish two kinds of variability manifested in L2 learners’ interlanguage system. The first is synchronic, which is the alternation in the use of forms at one point in time due to social and contextual features of the situation. Many current studies on variation have taken this approach using sociolinguistic models such as Labov’s to explain variation in interlanguage (Tarone 1983, 1985a, 1985b, 1989; Ellis 1985). The second kind of variation is diachronic, which is the gradual change over time in the learner’s use of certain forms and which shows that the learner’s system is evolving (Bialystok 1990). Current research in Spanish L2 acquisition from a variationist perspective examines learner language taking into account a wide range of factors, both linguistic and contextual and takes advantages of advances in statistics to make use of multivariate regression analyses that can handle large amounts of data. This helps researchers establish the effects of a broad range of linguistic and social factors

16 | 1 Introduction: History and theories of second language acquisition (Bayley and Tarone 2012). The assumption among sociolinguistic studies is that because variability of non-native speaker language is most likely seen in contexts where native speaker language does not vary, analyses of learner language must include all the contexts for subject expression and not just a subset of contexts. The important research questions that many L2 variationists have examined are presented in Table 6. Table 6. Research questions of the variationist approach in L2 acquisition. –

– – –

How does the variation of L2 learners’ language vary at the phonological, syntactic, and lexical levels according to the sociolinguistic variables of social context, task, topic, interlocutors, attention to task or form, and ethnolinguistic background? How do sociolinguistic factors interact with function-form relationships? Does L2 learner language exhibit variability in the same contexts where native speaker language varies? Do L2 learners have sufficient linguistic competence in the L2 to manipulate a variety of styles ranging from informal to formal?

These are important questions that have traditionally occupied L2 variationists; nonetheless, in the acquisition of subject expression in L2 Spanish, the issue that has received the most attention is the third one from the list above. Thus, in Chapter 6, where I examine the research in recent years, I focus almost exclusively on the linguistic and discourse contexts that have been shown to promote variability in both native speaker and L2 learner language in the acquisition of Spanish subjects. This research question is further divided into the following more specific questions presented in Table 7. Table 7. Specific research questions in the L2 acquisition of subject expression from a variationist approach. – – –

Does L2 learners’ language exhibit variability in the same linguistic contexts where native speaker language varies? Does L2 learners’ language exhibit variability in the same discourse contexts where native speaker language varies? Does the lexical semantics of the verb affect variability of subject expression?

1.5 Concluding remarks regarding theoretical approaches to L2 acquisition

|

17

1.5 Concluding remarks regarding theoretical approaches to L2 acquisition Each of the perspectives outlined in this chapter has made major contributions to the field of L2 acquisition and has brought us to a greater understanding of the structural and functional underpinnings of subject pronouns in Spanish and how they influence L2 acquisition. I am well aware that the theoretical models discussed here are not the only ones on which research in second language acquisition is based. For example, there are no chapters or discussion of several well-known models in L2 acquisition such as input processing (VanPatten and Cadierno 1993; VanPatten 2007), processability theory (Pienemann 1998, 2007), interaction hypothesis (Long 1981; Gass and Mackey 2007) or sociocultural theory (Lantolf and Thorne 2007). This is not an oversight on my part, but simply is due to the fact that as far as I know, no research on the L2 acquisition of Spanish subjects has emerged based on these models. I am also cognizant of the fact that the models discussed in this volume are not necessarily mutually exclusive and/or may appear to some readers to be arbitrarily labeled. For example, in the chapter on processing models, much of the research is, indeed, generative in terms of the researchers’ views of the basic tenets of the underlying structure of language, but in terms of the questions asked (i.e. ‘how do L2 learners process or interpret L2 structures?), the approach is labeled as ‘processing’ because of the researchers’ interest in understanding the mental processes of learners. In addition, the research that I describe as ‘pragmatic’ can also be considered variationist in terms of data collection and its focus on multiple linguistic and extra-linguistic factors that contribute to appropriate use of subject forms. However, the focus is explicitly on the pragmatic notions of topic involvement, speaker identity, intention and attitude, and presupposition of knowledge states and cognitive statuses. Oftentimes approaches to the study of L2 acquisition are divided between either theoretical or applied, linguistic or psycholinguistic in nature according to the researcher’s point of view regarding the nature of language, the type of research questions asked and the kind of data collected and analyzed. In the present volume, the theoretical models used to examine the L2 acquisition of Spanish subjects can broadly be divided into generative (Chapters 3, 4, and 5), which tend to be theory driven; and functional (Chapters 6 and 7), which tend to be data driven. Additionally, one could consider much of the research discussed in Chapters 4, 5, and 6 as psycholinguistic because they focus on cognitive processes and interpretation of subject forms in different syntactic and discursive structures. In reading up on theories of L2 acquisition and the vast literature related to the L2 acquisition of subjects in Spanish, I have come to the conclusion that theoretical approaches,

18 | 1 Introduction: History and theories of second language acquisition like language itself, are never categorical in nature but can and do share characteristics with each other.

1.6 Organization of the volume In the following chapters I will examine more closely how research from the diverse perspectives has helped to shed light on the complex process of the L2 acquisition of subject expression in Spanish. Chapter two presents an overview of subjects in Spanish and discusses the syntactic, pragmatic-discursive properties of the structure, as well as some regional variations that are important to take into account and can impact methodological considerations in much of the work done in L2 acquisition of subjects. The third chapter outlines early research into the L2 acquisition of Spanish subject pronouns which adopted a parametric approach in order to examine learners’ access to a clustering of syntactic properties of the Null Subject Parameter (NSP) (Chomsky 1981; Jaeggli 1982; Jaeggli and Safir 1989). The examination of parametric studies is important because they establish the role of UG in parameter resetting and have led to more recent studies that look to the importance of pragmatic information for explaining the overt/null distinction. The research examined in chapter four attempts to pinpoint the source of lingering interlanguage errors by examining the interface between morphosyntactic and discourse-pragmatic properties. These studies recognize the importance of information structure in understanding the overt/null distribution; nonetheless, the role of learners’ L1 is an unresolved issue among these studies. The question of how learners interpret pronominal reference when both L1 and L2 processing strategies are available is the topic of chapter five. Although research based on this paradigm for now also assumes a UG perspective, it differs from that discussed in the previous two chapters in that of interest are the interpretation strategies speakers and adult L2 learners of Spanish use to interpret null and overt pronoun antecedents in intra and intersentential contexts. Chapter six presents a model from a non-generative, discourse-pragmatic perspective that examines discourse constraints or pragmatic rules that guide subject use in L2 narrative discourse. These studies are important because they address general cognitive abilities related to the notions of saliency, focus, memory, attention state, and speaker involvement that all appear to guide subject pronoun usage for both native speakers and learners alike. Chapter seven discusses research from a sociolinguistic/variationist model which examines a wide range of factors, both linguistic and contextual, that largely have been ignored in generative studies and only touched on in processing and pragmatic studies. They are essential, nonetheless, for a more complete understanding of how learners use and acquire Spanish sub-

1.6 Organization of the volume

|

19

ject expression. The final chapter summarizes the findings of the research on L2 learners’ use and interpretation of subject expression in Spanish from the multiple paradigms discussed. The purpose of this final chapter is to point out both the contradictory evidence and the common findings among the different studies and to suggest avenues for future collaborative, “inter-theoretical” research and possible pedagogical implications of the findings.

2 Subjects in Spanish Introduction Subject expression has been studied extensively in Spanish in native speaker and bilingual-speaking communities, and in second language (L2) learners from a wide variety of theoretical perspectives. This is due to the fact that there are welldefined syntactic propositions that seek to explain the structural properties for the presence or absence of subject forms and that this phenomenon appears to be explained by highly specific universal syntactic principles. Subject expression is also constrained by semantic, functional, and cognitive and discourse conditions, which it will be argued in this volume, are also universal in nature. Although the syntactic principles and semantic, functional, and cognitive-discursive conditions responsible for the acquisition, representation and use of subject expression are universal in nature, they are also language specific for their interpretation and distribution in distinct linguistic and extra-linguistic environments. Finally, there are regional differences in the use of overt and null subjects in Spanish. These facts have generated the extensive body of research that exists today and which merit examination in order to make sense of how these proposed universal properties impact L2 acquisition. This chapter is intended to give the reader a general overview of the syntactic, lexico-semantic, functional cognitive and discursive properties of Spanish subjects, (as well as a brief mention of some dialectal differences), and will serve as a general background for many of the issues and investigations discussed in subsequent chapters.

2.1 Overview of subject forms in Spanish Spanish is a null subject or pro-drop language, which means that the language allows referential subjects to have overt or null forms. Why languages allow both overt and null subjects and what related properties result from this has been the source of numerous propositions and research. Referential subjects are those that refer to concrete or abstract entities in the real world and, therefore, can be assigned a thematic role (sometimes referred to as thematic subjects). Overt referential subjects can be full lexical noun phrases, as in (1), pronominal as in (2), or a null subject as in (3) below.

2.1 Overview of subject forms in Spanish

(1)

El niño camina, The boy walk-3sg⁵ ‘The boy walks.’

(2)

Él camina, he walk-3sg ‘He walks.’

(3)

pro Camina. pro walk-3sg

|

21

‘[He/she] walks.’ Lexical subjects can also be proper nouns (e.g. Susana) and pronominal subjects moreover can be demonstrative (e.g. éste ‘this’), indefinite (e.g. nadie ‘no one’), quantifier (e.g. todos ‘all’), and interrogative (e.g. quién ‘who’). Null subjects are allowed because Spanish has a rich and uniform verbal morphology and is a strong subject-verbal agreement language. All finite verbs agree in person and number with their syntactic subject. It is possible to omit the pronominal and lexical subjects “because the information is recoverable from the rich person and number morphology on the verb, which licenses and identifies the null subjects” (Montrul 2004a: 175). Table 8 illustrates the verbal paradigm for the Spanish verb, caminar ‘walk’ in the present and preterit (perfective) indicative tenses. In all cases, the verbal ending indicates (in addition to tense, aspect and mood) the person and number. Spanish verbal endings, like other Romance languages, do not indicate gender. The verb forms in Table 8 demonstrate the distinctiveness of the verbal morphology for each person and number⁶ for each tense. It is evident from the En-

5 Here, and throughout this volume, verbs in the present indicative are not marked in the glosses. Verbs in the preterit (pret), imperfect (imp), future (fut), conditional (cond), present subjunctive (subj) and imperfect subjunctive (imp-subj) are all marked as such. Verb morphology for perfect or progressive tenses is indicated on the auxiliary verbs and together with the participle and gerund makes clear that they are complex verb forms. Thus, the absence of a tense marker indicates present indicative tense. 6 The second person singular form, usted ‘you-sg’, is the formal form of address and contrasts with the familiar form, tú ‘you-sg” in most dialects. The second person plural form, ustedes ‘youpl’, is the formal form of address in peninsular dialects but used as both the formal and familiar forms in most Latin American and Caribbean dialects. The verbal endings for these forms are indistinguishable from the third person singular and third person plural forms and are potentially ambiguous, although most contexts will make clear who the intended referent is. There is also the

22 | 2 Subjects in Spanish Table 8. Verbal paradigm in Spanish for present and preterit morphology (–ar⁷ verbs). Present indicative (yo) (tú) (él/ella) (nosotros) (vosotros) (ellos/ellas)

camin- o camin- as camin- a camin- amos camin- áis camin- an

(‘I’) (‘you’) (‘he’/’she’) (‘we’) (‘you’) (‘they’)

‘walk’ ‘walk’ ‘walks’ ‘walk’ ‘walk’ ‘walk’

(‘I’) (‘you’) (‘he’/’she’) (‘we’) (‘you’) (‘they’)

‘walked’ ‘walked’ ‘walked’ ‘walked’ ‘walked’ ‘walked’

Preterit indicative (yo) (tú) (él/ella) (nosotros) (vosotros) (ellos/ellas)

camin- é camin- aste camin- ó camin- amos camin- asteis camin- aron

glish glosses that English does not share this property with Spanish. English is a weak agreement language with a morphologically poor and non-uniform verbal agreement paradigm (Montrul 2004a: 179). A consequence of this poor verbal morphology for languages like English is that the syntactic subject must always be explicit, except in coordinating parallel clauses, whereas for Spanish (as well as Italian and other null subject languages), it can be explicit or implicit (i.e. null)⁸. This observation has led to the proposal of the null subject parameter within the

vos ‘you-sg’ informal form used in parts of Latin America (predominantly in Argentina, Paraguay, Uruguay, but also in other dialectal areas). Verbal forms of vos are variable among the voseante dialects. 7 Spanish has three classes or conjugations of verbs. The verb caminar belongs to the first class whose infinitives end in –ar; verbs of the second class end in –er and those of the third class end in –ir. Verbal endings for each class are different, although it is estimated that approximately 90 % of verbs belong to the first class. 8 These facts, of course, cannot explain why a language like Chinese is also a null subject language but lacks a rich verbal morphology. Some have suggested that it is not the rich verbal morphology but rather a uniform verbal morphology (Jaeggli and Safir’s [1989] Morphology Uniformity Principle) that allows for null subjects. Thus Chinese, which is uniform in its lack of verbal morphology, is also a null subject language. But Lu (1995, cited in López Ortega 2002: 19) has challenged this explanation. According to Camacho (2013), the apparent contradiction is dealt with more satisfactorily in the Minimalist Program (Chomsky 1995). See Camacho (2013) for more detail.

2.2 Syntactic properties of Spanish subjects |

23

Principles and Parameters approach (Chomsky 1981) in order to account for several properties that were thought to cluster together and to explain in a unified manner why some languages permit only overt subjects and others accept both overt and null subjects.

2.2 Syntactic properties of Spanish subjects 2.2.1 The Null Subject Parameter The Null Subject Parameter (NSP) is derived from the Extended Projection Principle (EPP), which states that all clauses must have a subject in the Spec, IP position (Chomsky 1986). Languages that appear to not have subjects in some clauses actually do have them, but in the null or implicit form. Ayoun points out, as proposed by Jaeggli and Safir (1989), that it was “believed [. . . ] the null subject parameter involved AGR (an abstract agreement element which is part of the inflectional element INFL). In languages with a rich AGR, i.e., with rich inflection systems, the subject pronoun may be omitted because it can be easily identified” (Ayoun 2003: 80). Furthermore, several authors (Hyams 1986, 1994; Rizzi 1982) have proposed “that AGR was a type of pronominal which could properly govern the null subject to fulfill the requirement that all empty categories be governed by lexical heads or INFL” (Suñer 1982, cited in Ayoun 2003: 81). Initially six properties were associated with the NSP to account for the properties of all languages. As the parameter relates to null subject Romance languages, we will illustrate four of the properties identified by Rizzi (1982, 1986) for Italian and which have been hypothesized to cluster together for both Italian and Spanish. One of the first properties of the NSP identified by Rizzi is that in null subject languages, referential subjects can be overt or null. (4)

Overt vs. null referential subjects in tensed clauses a. Yo estudio español. I study-1sg Spanish ‘I study Spanish.’ b. pro Estudio español. pro study-1sg Spanish ‘[I] study Spanish.’

Although either an overt (4a) or null (4b) subject can be used, as will be discussed shortly, certain semantic and discourse constraints apply and thus overt and null subjects are not considered to be in free variation. As previously mentioned, due

24 | 2 Subjects in Spanish to the fact that verbal morphology in Spanish is marked for person and number (among other features), a subject can be null if it is clear from the inflectional morphology or discourse conditions to what or to whom the subject refers. Whereas referential subjects (distinguished for person, number and gender features) can be overt or null, expletive subjects are obligatorily null in Spanish, the second property of the NSP identified by Rizzi. Expletive subjects have no referents in the real world and are thus, third person, singular and gender-neutral (Lozano 2002: 38). They appear in sentences that express meteorological (5a) and existential events (5b), and with pleonastic constructions (5c). (5)

Obligatory null expletive subjects a. pro Llueve. pro rain-3sg ‘[It]’s raining.’ b. pro Hay mucha gente en la calle. pro exist-3sg much people in the street ‘[There] are a lot of people in the street.’ c. pro Es importante que estudies. pro is-3sg important that study-2sg-subj ‘[It]’s important that you study.’

Again, the English glosses illustrate the fact the English does not allow null subjects, even in these types of sentences, where an expletive pronoun, it or there, is required in subject position. When first proposed, the NSP was claimed to explain the ease with which a child learned her first language. It was assumed that the properties of the NSP clustered together and were learned at the same time. If, for example, the child were learning Spanish, upon noticing that null referential subjects are possible in the input, she would automatically know that subjects of expletive sentences were obligatorily null, or vice versa. The English-speaking child, on the other hand, would notice that referential subjects are always overt and would automatically learn the related property that subjects of expletive sentences were also obligatorily overt. Thus, part of the implication of the NSP was that languages are either one way or another; i.e. they are either like Spanish and Quechua and allow null referential subjects and require null expletive subjects, or they are like English and French and require overt referential and expletive subjects. Recent developments in the analysis of null subject languages have revealed, however, that languages can have some combination of null and overt subjects. For example, Cape York Creole requires overt referential subjects and null expletive subjects and Finnish allows null referential subjects and requires

2.2 Syntactic properties of Spanish subjects | 25

overt expletive subjects (Camacho 2013: 14–16). The extent to which these differences reveal a deeper property of grammars is an avenue for future research. A third property of Romance pro-drop languages is the possibility to freely invert the subject and verb. Although the default unmarked word order for Spanish in unfocused (neutral) declarative sentences is S–V–(O), other word orders (V–S, V–S–O, V–O–S) are possible as well. This means that the subject may appear either preverbally as in (6a) and (6c), or postverbally, as in (6b), (6d) and (6e). (6)

Subject-verb inversion a. María llegó. (S–V) Mary arrive-3sg-pret ‘Mary arrived.’ b. Llegó María. (V–S) arrive-3sg-pret Mary ‘Mary arrived.’ c. Juan leyó un libro interesante. (S–V–O) Juan read-3sg-pret a book interesting ‘Juan read an interesting book.’ d. Leyó Juan un libro interesante. (V–S–O) read-3sg-pret Juan a book interesting ‘Juan read an interesting book.’ e. Leyó un libro interesante Juan. (V–O–S) read-3sg-pret a book interesting Juan ‘Juan read an interesting book.”

Despite the fact that the syntax (i.e. the NSP) allows for “free” inversion, this property is subject to syntactic, semantic and pragmatic/interpretative constraints (Isabelli 2003). For example, in unfocused declarative sentences, preverbal subjects are most common with unergative verbs as in (7a), and postverbal subjects are more common with unaccusative verbs as in (6b) above (repeated as (7b) below for contrastive purposes). (7)

a. Unergative verb: S–O order preferred María llamó. María call-3sg-pret ‘María called.’

26 | 2 Subjects in Spanish b. Unaccusative verb: V–S order preferred Llegó María. arrive-3sg-pret Mary ‘Mary arrived.’ This observation in languages with free subject-verb inversion has lead to what is known as the Unaccusative Hypothesis (UH) (Perlmutter 1978). The UH and how it affects word order is discussed at length in Chapter 4, but basically its ramifications for word order of declarative sentences in Romance languages such as Spanish, as well as other languages, is dependent upon principles operating at the lexico-syntactic interface and the syntactic-pragmatic interface. In other words, the lexical subcategory of the verb (whether transitive or intransitive and unergative or unaccusative) determines in part whether an overt subject is preverbal or postverbal. Other factors, such as the discursive information status (e.g. focus or non-focus) of the subject, interact with this lexico-syntactic property to determine subject use and word order in Spanish. The fourth property of the NSP of Romance languages identified by Rizzi is the apparent violation of the that-trace filter in Spanish. The that-trace filter was proposed by Chomsky and Lasnik (1977) to account for the asymmetries in English for allowing the extraction of wh-question words from subject and object positions across clausal boundaries. In English, object wh-words can be extracted from object position with and without overt complementizers (i.e. that), as evidenced in (9a), but the that-trace filter blocks extraction of subject wh-words with overt complementizers, as seen in (9b). Spanish appears to violate the filter and allows both object and subject extraction of wh- words with overt complementizers⁹, as seen in (10a) and (10b).

9 Unlike English, Spanish requires an overt complementizer. In the sentences below, the lack of a complementizer results in ungrammatical sentences in Spanish. (8)

a. b. c.

d.

Who do you think he will see? Who do you think will see John? *¿A quién crees él verá? to whom think-2sg he see-3sg-fut ‘Who do you think he will see?’ *¿Quién crees verá a Juan? Who think-2sg see-3sg-fut to John ‘Who do you think will see John?’

2.2 Syntactic properties of Spanish subjects | 27

English (9)

Object extraction a. Who do you think that he will see who? Subject extraction b. *Who do you think that who will see John?

Spanish (10)

Object extraction a. ¿A quién crees que él verá a quién? to whom think-2sg that he see-3sg-fut to whom ‘Who do you think that he will see?’ Subject extraction b. ¿Quién crees que quién verá a Juan? who think-2sg that who see-3sg-fut to John ‘Who do you think is going to see John?’

Isabelli explains that the reason the that-trace filter is in effect for English but not for pro-drop languages like Spanish is because pro-drop languages allow “free verb-subject inversion in finite clauses, [where] the embedded subject is theoretically extracted post-verbally so as to avoid leaving an immediate trace behind the complementizer” (Isabelli 2003: 152). The prediction of the NSP for child language acquisition and adult L2 acquisition is that the properties are not learned one by one, but rather as a clustering of properties, or in some cases, in an implicational manner. Whether or not these predictions are borne out has been the subject of much empirical research in the past three decades. Additional structural properties not originally subsumed in the NSP are outlined in the following sections.

2.2.2 Interpretative constraints 2.2.2.1 The Overt Pronoun Constraint (Montalbetti 1986) An additional principle of null subject languages considered in a number of studies in Spanish L2 acquisition is the Overt Pronoun Constraint (OPC) as set forth by Montalbetti (1986), which states that overt pronouns are avoided when the pronoun is bound to a formal variable, such as the trace from wh-movement as in (11a) and (11b) or quantifier raising as in (12a) and (12b).

28 | 2 Subjects in Spanish (11)

a. ¿Quiéni cree que él∗i/j ganará el premio? who think-3sg that he win-3sg-fut the prize ‘Whoi thinks that he will win the prize?’ b. ¿Quiéni cree que proi/j ganará el premio? who think-3sg that pro win-3sg-fut the prize ‘Who thinks that [he] will win the prize?’

(12)

a. Nadiei dice que él∗i/j ganará el premio. nobody say-3sg that he win-3sg-fut the prize ‘Nobody says that he will win the prize.’ b. Nadiei dice que proi/j ganará el premio. nobody say-3sg that pro win-3sg-fut the prize ‘Nobody says that [he] will win the prize.’

In (11a) and (12a), the overt subject of the subordinate clause, él ‘he’, cannot be co-referenced with the subject of the main clause, quién ‘who’ in 10a and nadie ‘nobody’ in (12b). It can only be interpreted as co-referential with another identifiable referent in the discourse. In (11b) and (12b), however, the null subject can be interpreted as either co-referential with the subject of the main clause or with another known referent in the discourse. The OPC is in effect for pro-drop languages such as Spanish (as well as [+topic-drop] languages such as Chinese), but not for non-pro-drop languages like English. On the other hand, English and Spanish parallel each other for pronoun distribution based on certain pragmatic factors, such as contrastive focus. Whereas English uses stress to signal contrastive focus, Spanish uses overt pronouns. In response to focus questions, for example, the use of a stressed pronoun would be used in English if, in (13a) and (13b), John believes that he himself will win the prize; an unstressed subject pronoun would not be appropriate. (13)

a. Who does Johni think will win the prize? b. Johni thinks that HEi /*hei will win the prize.

For the same situation in Spanish, the use of an overt pronoun is appropriate for contrastive focus and a null subject pronoun would not be appropriate as illustrated in (14a) and (14b). (14)

a. ¿Quién cree Juani que ganará el premio? who think-3sg John that win-3sg-fut the prize ‘Who does John think that will win the prize?’

2.2 Syntactic properties of Spanish subjects | 29

que éli /*proi ganará el premio. b. Juani cree John think-3sg that he/*pro win-3sg-fut the prize (Adapted from Strozer 1992, cited in Pérez-Leroux and Glass 1997, p. 152) The effects of the OPC on the L2 acquisition of Spanish subject pronouns is examined further in Chapters 3 and 4 of the present volume.

2.2.2.2 The Position of Antecedent Hypothesis (Carminati 2002) Carminati noticed for Italian, a pro-drop language that shares many of the NSP properties with Spanish, that subjects differ in their degree of prominence in sentences. Specifically, she states “that pronouns retrieve highly accessible/prominent referents, but crucially that the null pronoun retrieves a more prominent referent than the overt pronoun” (Carminati 2002: 25). Stemming from these observations, Carminati proposes the Position of Antecedent Hypothesis, (PAH) for the interpretation of Italian null and overt subject pronouns. The hypothesis is stated as follows: (15)

The Position of Antecedent Hypothesis for the Italian null and overt pronouns in intra-sentential anaphora: the null pronoun prefers an antecedent which is in the Spec IP position, while the overt pronoun prefers an antecedent which is not in the Spec IP position. (Carminati 2002: 57)

Thus, the PAH assumes there is a ‘division of labor’ with null and overt subjects: the null pronoun is linked more often than the overt pronoun with a more prominent antecedent, usually the one in preverbal subject position of the antecedent clause and conversely, the overt pronoun (in null subject languages) is linked more often than the null subject with a less prominent antecedent usually in a complement position. No such ‘division of labor’ exists in non-pro-drop languages such as English, which depends on a subject assignment strategy, i.e. overt pronouns tend to be linked to the subject antecedent. In intra-sentential contexts such as (16), the PAH predicts that the structural configuration, i.e. the syntactic position, determines antecedent prominence and thus, anaphoric resolution.

30 | 2 Subjects in Spanish (16)

a. Quando Marioi ha telefonato a Giovannij , proi aveva when Mario have-3sg telephoned to John pro have-3sg-past appena finite di mangiare. just finished of eat-inf ‘When Mario telephoned John, [he] had just finished eating.’ b. Quando Marioi ha telefonato a Giovannij , luij aveva when Mario have-3sg telephoned to John he have-3sg-past appena finite di mangiare. just finished of eat-inf ‘When Mario telephoned John, he had just finished eating.’ (Adapted from Carminati 2002: 58)

The PAH predicts that for (16a), the null subject (pro) of the second clause refers to Mario, the subject of the first clause, and in (16b), the overt subject pronoun (lui) links to Giovanni, the direct object of the first clause, because a pre-verbal subject in the Spec IP position is more prominent than antecedents found in positions lower in the hierarchical syntactic structure, e.g. in an object position. The hypothesis, as stated, implies that non-structural factors (e.g. the semantics of the referent or the discourse context), should not override the preferences of antecedents for pronouns. Furthermore, the hypothesis states that the preferred antecedent for a null subject is not just a subject but also an antecedent in the Spec IP position. That means that only pre-verbal antecedents (not post-verbal ones) are preferred to link to null subject. This has implications for dative subjects because the PAH predicts that they are the preferred antecedents for null pronouns, as well. Finally, Carminati claims that the PAH is an even better predictor of anaphoric interpretation for quantified subjects than the OPC. The implications of the PAH in the study of L2 acquisition of Spanish subjects is detailed in Chapter 5.

2.2.3 Strong agreement and Spanish subjects To what extent the identified properties of the NSP and the OPC influence the L2 acquisition of Spanish subjects is explored in Chapters 3 and 4 and the predictive power of the PAH is examined in Chapter 5. However, later developments in in generative theory questioned the clustering nature of the properties of the NSP, and indeed their very parametric variation. Tsimpli and Roussou (1991) assumed “that parameters are not associated with UG principles but with lexical items and in particular, functional categories. Parametric variation is exclusively determined by the different values associated with functional categories” (Tsimpli

2.2 Syntactic properties of Spanish subjects | 31

and Roussou 1991: 151). Montrul (2004a) states that “parameters are now in the lexicon, which comprises language specific lexical and functional categories (Montrul 2004a: 7), and Camacho (2013) claims that in the Minimalist Program “parameters no longer have an independent place in accounting for variation across grammars” (Camacho 2013: 5). Specifically, Montrul has suggested that the properties subsumed under the NSP are associated with other structures in the grammar (Montrul 2004a: 179). In more current accounts of null subject languages within the Minimalist Program, the Extended Projection Principle (EPP) is revised so as to check nominal (D) features (Case and agreement-related features) in INFL. A language such as Spanish does not project a spec AgrSP position because the verbal agreement is nominal and has semantic content. English, on the other hand, does not have verbal agreement morphemes with semantic content (Montrul 2004a). Thus, within this model, English personal pronouns (I drink, you drink, etc.) have the same status as person and number verbal morphology in Spanish (beb-o, beb-e-mos). In Spanish, the subject DP can check EPP features by attaching to the head of AGR whereas in English the subject doesn’t move but rather merges with the specifier of AGR to check EPP features. In terms of the L2 acquisition of Spanish subjects by learners of non-pro-drop languages such as English, learning the language involves going from the [-D, -interpretable] value in English to the [+D, +interpretable Φ -feature] value in Spanish. Rothman explains that “(i)n Spanish, head-to-head movement of the verb to AgrSP is able to check the EPP-feature since the agreement morpheme of the inflected verb is able to check the D-feature of AgrSP and to meet the identification requirement” (Rothman 2007: 302) and thus, null pronouns are possible. Whereas in English, with no overt pronoun to merge to the Spec of AgrSP, it is not possible to check the strong EPP-feature and thus, overt pronouns are required. Therefore, speakers of languages like English must learn the appropriate nominal features of agreement in Spanish in order to license null subjects in their L2 Spanish. Montrul points out the Minimalist Program “analysis captures the null subject phenomenon, the word order facts, and the lack of overt expletives in Spanish” (Montrul 2004: 180). Some of these features and how they impact the L2 acquisition of Spanish subjects are explored in Chapter 4. A summary of the syntactic and interpretative properties of the NSP discussed so far is presented in Table 9. These properties are compared with those of English, which is not a null subject language.

32 | 2 Subjects in Spanish Table 9. Syntactic properties of subjects in pro-drop and non-pro-drop languages. Pro-drop language (Spanish) Properties

– – – – – – – –

Non-pro-drop language (English)

morphologically rich and uniform verbal agreement paradigm overt and null referential subjects obligatory null expletive subjects free subject-verb inversion violation of that-trace filter obeys OPC (no bound variable interpretation with overt) obeys PAH (division of labor for null and overt subject pronouns) subject attaches to head of AGR (can check strong EPP-feature)



– – – – –



morphologically poor and nonuniform verbal agreement paradigm overt referential subjects overt expletive subjects strict SVO word order that-trace filter in effect OPC has no effect depends on subject assignment strategy (overt pronouns link to subject antecedents subject merges with specifier of AGR (can not check strong EPPfeature)

(Based on and expanded from Montrul 2004a: 179)

2.3 Discourse-pragmatic features and Spanish subject distribution 2.3.1 Focus, topic-continuity and topic-shift I have discussed how the distribution of null and overt subjects in Spanish, like other Romance languages, is constrained by a formal set of syntactic principles, although which properties of the NSP affect L2 language acquisition is far from a settled issue. Pro-drop languages such as Spanish, however, possess an additional level of complexity where discourse-pragmatic principles govern their apparent free alternation. Native Spanish speakers tend to encode non-focus and same reference (i.e. topic continuity) with null pronouns and to express focus and switch reference (i.e. topic shift) with overt pronouns or full noun phrases. This is not to say that non-pro-drop languages such as English have no means by which to express these universal pragmatic principles; languages use different mechanisms to transmit the notions of topic and focus, including prosody, morphology, syntax or any combination of these (Lozano 2009). Although researchers use slightly different terminology in referring to these pragmatic notions, in this book I follow that used by Lozano to discuss how overt and null subjects encode focus, topic-continuity and topic-shift in Romance languages and how these notions impact acquisition.

2.3 Discourse-pragmatic features and Spanish subject distribution

|

33

New information that is introduced into the discourse is focus and is introduced normally with a full noun phrase; if this information continues to be the topic of discourse, it then represents known or shared information and is referred to as topic information or topic-continuity (also referred to in the literature as ‘same referent’) and in Spanish is normally expressed with a null subject. (17)

(Context: the native speaker is relating a series of events that took place in a Charlie Chaplin silent film) Pues, Charlesi llegó, proi vio a la muchacha que well Charles arrive-3sg-pret pro see-3sg-pret to the girl that estaba sola; proi se sienta a platicar con ella pero, bueno, pues be-3sg-imp alone pro refl-sit-3sg to chat-inf with her but well then casi, casi proi tropezó con ella almost almost pro trip-3sg-pret with her ‘Well, Charles came, saw the girl that was alone, sits down to talk with her but, well, then almost, almost trips on her’ (Lubbers Quesada and Blake 2003)

In (17), an informant introduces Charles (new information or focus), the main character of the film, into the discourse using a full noun phrase (in this case, a proper noun) and in subsequent references to Charlie Chaplin, which represent a continuity of the immediately previous topic (‘Charles’), the informant uses a null subject, pro, the preferred form in Spanish (also for English in coordinating parallel clauses). Although syntactically an overt pronoun (‘él’) or a full noun phrase (or proper noun) can be used in these contexts, pragmatically they are infelicitous. When there is a change of referent in the discourse, this signals a topic-shift or a ‘switch referent’; in Spanish, normally an overt subject (i.e. a full noun phrase or an overt pronoun) is required, as seen in example (18). (18)

(Context: Charlie Chaplin has blindfolded a man and is playing with him, pulling him around with his cane. The man thinks that Chaplin is a girl he has been flirting with) Entonces proi jala al señorj de los ojos vendados y cuando then pro pull-3sg to-the man of the eyes blindfolded and when élj se da cuenta de que proi no es la muchachak , he refl-give account-3sg of that pro no be-3sg the girl ‘Then [he] pulls the man with the blindfolded eyes and when he realizes that [he] is not the girl,’ (Lubbers Quesada and Blake 2003)

34 | 2 Subjects in Spanish In the first clause Chaplin, who had been the topic of the previous discourse, continues to be the topic and therefore is encoded with a null subject, pro; then the blindfolded man is brought into the discourse via a direct object, al señor ‘to the man’. Now there are two referents to contrast in the discourse. Thus, in the second clause the overt pronoun is used to refer to the man (not Chaplin); there is a switch in topic or referent from Chaplin to the man. A null pronoun would have indicated a continuation of the topic or referent (or non-focus) and would have referred again to Chaplin. But clearly this is not the case as it is the blindfolded man that realizes that the man leading him around (i.e. Chaplin) is not the girl. A full noun phrase or overt pronoun also encodes the subject (focus) in response to a question, which serves to focus in on a new topic or introduce new information, as observed in (19). (19)

¿Quién vino? Él/Mario/*pro vino. who come-3sg-pret he/Mario/pro come-3sg-pret ‘Who came? He/Mario/pro came.’ (Adapted from Montrul and Rodríguez Louro 2006: 404).

In response to the question ¿Quién vino? (‘Who came?’) illustrated in (19), a null pronoun would be infelicitous; only an overt pronoun or full noun phrase (proper noun) is pragmatically felicitous. In parallel clauses there is another type of contrastive focus which speakers use to specifically contrast two referents already introduced into the discourse; this type of contrastive focus, as seen in (20), is particularly marked and usually accompanied by prosodic stress. (20)

(Context: the informant is describing her best friend and how they aren’t able to spend much time together anymore) Me gusta pasar mucho tiempo con ella aunque a veces me-dat please-3sg spend-inf much time with her although at times pro no tenemos tiempo ya que . . . YO estoy acá y ELLA pro no have-1pl time now that I-emph be-1sg here and she-emph está allá. be-3sg there ‘I like to spend a lot of time with her although sometimes [we] don’t have time now that . . . I am here and SHE is there.’ (Lubbers Quesada and Blake 2003)

Prosodic stress can also signal the difference between a referent, which is a topic if unstressed and a contrastive focus if stressed. This is exemplified in (21).

2.3 Discourse-pragmatic features and Spanish subject distribution

(21)

|

35

que éli /ÉLi no había escrito El periodistai dijo the journalist say-3sg-pret that he/he-emph no have-3sg-imp written ese reporte. that report ‘The journalisti said that hei /HEi (himself) had not written that report.’ (Adapted from Montrul 2004b: 128)

Without prosodic stress, the pronoun él (‘he’) makes no particular contrast with another journalist, but with prosodic stress ÉL (‘HE’), the implication is that it is the journalist who is speaking, and not some other journalist, who had not written the report. Of course, depending on the context, both pronouns (stressed and unstressed) could refer to another antecedent in the discourse. In contrast, a null subject could only be a topic and could only refer to the journalist speaking, as in (22). (22)

El periodistai dijo que proi no había escrito ese the journalist say-3sg-pret that pro no have-3sg-imp written that reporte. report ‘The journalist said that [he] had not written that report.’ (Adapted from Montrul 2004b: 128)

Thus, contexts where a null subject is expected include those in which information is known or shared among discourse participants, and contexts where overt subjects (i.e. pronouns or full noun phrases) are normally used include those in which new information is introduced into the discourse or when a special contrast is marked. Table 10 summarizes the pragmatic notions of focus, topic-continuity and topic-shift and their functions and provides alternate terminology for these notions used by distinct authors. The information summarized in Table 10 highlights the tendencies of native speakers for null/overt distribution of subjects. An understanding of these discourse-pragmatic notions is essential in detecting and analyzing deficits in the interpretation and production of pronoun distribution in L2 Spanish. These pragmatic notions and how they interface with syntactic properties are examined in research of L2 acquisition of subject expression in Chapter 4. These notions and how they respond to cognition, memory and presupposition are explored in Chapter 6 and their interaction with a range of subject forms and how they vary in L2 learner language is the subject of Chapter 7.

36 | 2 Subjects in Spanish Table 10. Distribution of null and overt subjects and discourse-pragmatic notions in Romance languages. Type of subject

Pragmatic notions

Related terminology

Function

null

topic (topic-continuity)

‘same referent’ ‘non-focus’ ‘non-contrastive focus’ ‘non-topic shift’

refers to old/shared information in discourse

overt

focus (topic-shift)

‘change of/switch referent’ ‘contrastive focus’

introduces new information or serves to mark a specific contrast

2.3.2 Cohesion, saliency and cognition The use and interpretation of subject forms depend not only on syntactic properties and the discourse-pragmatic notions of topic and focus, they also depend on a wide range of discourse notions related to narrative cohesion, referent saliency, and the presuppositions speakers have regarding their addressee’s memory state and cognitive status (Gundel, Hedberg, and Zacharski 1993). In the following examples, which describe the same scene in a Charlie Chaplin film, the role of null subjects in (23) contributes to the narrative cohesion whereas the use of repeated overt subjects, in this case full lexical noun phrases, in the learner example (24) breaks down the cohesion. Although the text is coherent in (24), the lack of cohesion causes processing difficulties for a listener. (23)

Pasa una mujer y coquetea con el papá, y ella sigue pass-3sg a woman and flirt-3sg with the father and she continue-3sg su camino pero el papá se queda viendo y pro ve que her way but the father refl-keep-3sg looking and pro see-3sg that la mamá y la hija están dormidas; pro se levanta y the mom and the daughter be-3pl asleep pro refl-rise-3sg and la pro sigue her-acc pro follow-3sg ‘A woman goes by and flirts with the dad and she continues on her way but the dad watches her and [he] sees that the mom and daughter are asleep; [he] gets up and [he] follows her.’ (Lubbers Quesada and Blake 2003)

2.3 Discourse-pragmatic features and Spanish subject distribution

(24)

|

37

Una mujer coquetea con el padre de la familia y el padre a woman flirt-3sg with the father of the family and the father camina con la mujer y el padre va a obtener una bebida walk-3sg with the woman and the father go-3sg to get-inf a drink para la mujer for the woman ‘A woman flirts with the father of the family, and the father walks with the woman and the father is going to get a drink for the woman.’ (Lubbers Quesada and Blake 2003)

The use of a particular subject form is conditioned also by the role, familiarity and relative importance of a given referent in a narrative, the function of that referent at different times throughout the narrative, as well as that referent’s syntactic position in a clause. Propositions of informativeness hierarchies have been proposed to account for anaphoric forms that reflect the importance of a referent’s role, familiarity and function within discourse. For example, Saunders (1999) proposes the anaphoric hierarchy in (25) to predict the use of pronouns and lexical noun phrases. Null pronouns are used to refer to referents that are highly salient and their use presupposes that the listener has the maximum amount of information regarding that referent. On the other hand, indefinite noun phrases are used when introducing new referents into the discourse and assume no previous knowledge on the part of the listener. (25)

indefinite NP > definite NP > proper noun > overt pronoun > null pronoun (Saunders 1999: 51)

Gundel, Hedberg and Zacharski (1993) propose a similar hierarchy known as the Givenness Hierarchy, but in this case it is the cognitive status that conditions the form and distribution of referring expressions in discourse. The cognitive statuses are a set of implicationally related statuses based on assumptions speakers make about the informativeness and saliency of referents in discourse and about their addressee’s knowledge and attention state. The implications of these statuses is that greater referent saliency corresponds to the use of more phonetically minimal referring expressions and lesser referent saliency is consistent with more phonetically elaborate expressions (Givon 1983; Ariel 1990; Blackwell 1998, 2003; Levinson 1987, 2000). The impact of these notions of cognitive status on the L2 acquisition of Spanish subjects is explored in Chapter 6 of the present volume.

38 | 2 Subjects in Spanish

2.4 Referent features and verbal semantics in Spanish subject distribution 2.4.1 Person, number and verb semantics It has been questioned why first and second person (mostly singular) pronouns would ever need to be overt in Spanish since they are deictic in nature and the morphological ending, except in rare cases, and the discourse context (usually) make clear who the referent is (i.e. the speaker is first person and the interlocutor(s) is the second person). In fact, Rosengren (1974) suggests that first and second person expressions are not true pronouns at all since they do not substitute for a noun (1974: 28). This is not the case for third person (both singular and plural) pronouns, which are anaphoric in nature and refer to lexical or proper noun phrase antecedents introduced previously in the discourse. However, several studies have shown that the use of overt first and second person subject pronouns among native Spanish speakers is higher than that of third person (Enriquez 1984; Ranson 1991; Davidson 1996), suggesting a functional distinction among the different persons and numbers. Thus, researchers have claimed that in the examination of the use and acquisition of subject pronouns in Spanish, different persons and numbers should be analyzed separately (López Ortega 2002; Travis 2005; Lozano 2009; Lubbers Quesada and Blackwell 2009). Gili y Gaya (1961) suggests that the use of first and second person overt subject pronouns is for emphasis and/or contrast or disambiguation when the verbal morphology does not make clear who the referent is, e.g. in the case of first person singular and third person singular forms in the imperfect (hablaba ‘[I] was speaking/would speak’ vs. hablaba ‘[he/she] was speaking/would speak’). Davidson (1996), on the other hand, claims that the use of the overt first person subject pronoun is poly-functional – that is, in addition to the emphatic or contrastive functions, first person overt pronouns can add a certain amount of ‘pragmatic weight’ to a speaker’s utterance. He goes on to explain that these pronouns “are used to increase the speaker’s stake in what is being said, and as such will be interpreted as either signaling an increased speaker commitment to the information in the utterance, or as adding semantic ‘weight’ to the verb to which they may be associated” (Davidson 1996: 544). Example (26) reveals the poly-functional nature of the first person overt subject.

2.4 Referent features and verbal semantics in Spanish subject distribution

(26)

| 39

O sea, yo por la noche, a las once de la noche, estoy que or be-3sg-subj I by the night at the eleven of the night be-1sg that me caigo. refl-fall-1sg ‘That is, (as for) me, by night time, at eleven at night, I’m (like) falling down.’ (Adapted from Davidson 1996: 547)

The preverbal overt subject pronoun in (26) functions as the subject of both verbs (estoy ‘I am’ and me caigo ‘I’m falling down’) and as the topic of the sentence. The verbal morphology of both verbs clearly indicates the person and number and thus, the overt subject pronoun is syntactically redundant. However, pragmatically it tells the listener what or whom the sentence is about and in doing so, adds pragmatic weight to the speaker’s utterance. Davidson hypothesizes that first person singular overt subject pronouns are more common in conversational discourse conceivably due to the fact that people like to talk mostly about themselves. Thus, first person overt pronouns are used for a variety of functions in addition to the emphatic and contrastive ones, including: to introduce a topic (the speaker), to take a stand, to negotiate turns, and to signal a speaker’s wish to take the floor (Davidson 1996: 553). As shown in Section 2.2.1, the transitivity of a verb can affect word order in sentences with overt subjects (i.e. preverbal subjects are expected with unergative verbs and postverbal subjects are expected with unaccusative verbs in unfocused declarative sentences), but research in native speaker use has shown that the semantics of the verb, as well as the person and number, can determine whether a subject pronoun is overt or null. Certain semantic classes of verbs trigger more overt subject use, especially first person subject pronouns. Enríquez (1984) points out that verbs that belong to the semantic field of knowledge or perception or those that express an opinion are more likely to be used with overt subject pronouns. It seems logical, she explains, that it would be these kinds of verbs that imply a greater subjectivity that would trigger more frequent uses of overt subjects. These types of verbs allow speakers to become more involved in and to individualize their opinions or their emotions in a more subjective way (Enríquez 1984: 118). Travis (2007) and Travis and Torres Cacoullos (2014) refer to them as psychological verbs; García-Miguel (2005) refers to them as verbs of sight, within the class of verbs of perception; and Lubbers Quesada and Blackwell (2009) use a range of terms to identify them, including: verbs of opinion, belief, claiming, emotion or knowledge. These verbs include: creer ‘to believe’, saber ‘to know’,

40 | 2 Subjects in Spanish pensar ‘to think’, ver ‘to see’, sentir ‘to feel’, querer ‘to want’, decir ‘to say’, among others. The deictic and anaphoric functions of subject pronouns as they relate to person and number in the L2 acquisition of Spanish are explored in Chapters 4, 6 and 7 of the present volume; the extent to which L2 learners acquire the function of pragmatic weight is examined in Chapter 6; and the influence of semantic class of verb in the use and acquisition of first person subject pronouns is examined in Chapters 6 and 7.

2.4.2 Specificity and animacy of referents Other features that have been examined for their impact on the use of overt or null subjects in Spanish are specificity and animacy (especially humanness) of the referent. Specificity refers to whether a given referent can be differentiated by name in the discourse context or the immediate non-linguistic environment. Human referents are necessarily animate but not all animate referents are necessarily human (i.e. animal referents). Whereas specific animate (human or humanized) referents are indicated by full lexical noun phrases (including proper nouns), overt, null, and demonstrative pronouns, non-specific animate (human or humanized) referents are generally referred to by indefinite pronouns and null pronouns (usually third person plural). Non-human (specific and non-specific) referents (unless humanized as in the case of household pets) are not generally used with subject pronouns, only full lexical noun phrases or null subjects. Plural definite noun phrases can also be used to refer to non-specific human, nonhuman and inanimate subjects. First and second person subjects are by nature [+specific] and [+human/animate]. Only third person subjects can be inanimate; therefore, all of the examples below illustrate only third person referents. Examples (27–29) are [+specific] and [+human/animate]; examples (30–31) are [+specific] and [−human/animate]¹⁰; and examples (32–34) are [−specific] and [+human/animate]. Example (35) is an illustration of a subject that is [−specific] and [−human/animate]¹¹.

10 Note that both English and Spanish allow full lexical noun phrases to refer to specific and inanimate subjects, but whereas English requires a pronominal subject (‘it’), Spanish favors a null subject. 11 Note that [−specific] generic entities in English, both [+animate/human] and [−animate/human] cannot be expressed with definite noun phrases as they are in Spanish. In English, the generic bare plural (no determiner) is used, as illustrated by the English glosses for examples (33) and (34).

2.4 Referent features and verbal semantics in Spanish subject distribution

Specific referents (27)

El chico comió un plátano. the boy eat-3sg-pret a banana ‘The boy ate a banana.’

(28)

María trabaja en la universidad. Mary work-3sg in the university ‘Mary works at the university.’

(29)

Ella/pro quiere más café. she/pro want-3sg more coffee ‘She wants more coffee.

(30)

La cama no cabe en la recámara chica. the bed no fit-3sg in the bedroom small ‘The bed doesn’t fit in the small bedroom.’

(31)

?Ella/pro no cabe en la recámara chica she/pro no fit-3sg in the bedroom small ‘She/[it] doesn’t fit in the small bedroom.’

Non-specific referents (32)

Alguien tiró la leche. someone spill-3sg-pret the milk ‘Someone spilled the milk.’

(33)

Por fin pro arreglaron la calle. At last pro fix-3pl-pret the street ‘Finally [they] fixed the street.’

(34)

Los jóvenes ya no respetan a los mayores. the youngs any more no respect-3pl to the elders ‘Young people don’t respect their elders anymore.’

(35)

En la costa los libros se cubren de moho. on the coast the books refl-cover-3pl of mold ‘On the coast books get covered with mold.

| 41

42 | 2 Subjects in Spanish Enriquez (1984), in her study of Madrid Spanish, found that the features [+specific] and [+human/animate] for referents highly favor the use of overt subject pronouns and that the opposite, that is [−specific] and [−inanimate] referents disfavor the use of overt subjects. Overt subject pronouns that are [+specific] and [−human] are generally used to humanize to some degree a non-human referent, such as a family pet. Chapter 7 of the present volume discusses at greater length the impact of these features on subject expression in Spanish and explores to what degree L2 learners are also constrained by these features.

2.5 Regional variations of Spanish subjects Throughout this chapter we have been discussing the syntactic, lexico-semantic, discursive-pragmatic and cognitive properties that characterize subject expression in Spanish as if all varieties of Spanish were the same. Although all varieties of Spanish retain “a fundamental cohesiveness throughout the world, social and geographical variation is considerable” (Lipski 2014: 2) and Spanish subject expression is not an exception. Variation exists in the use of overt and null subjects and in pre-verbal or post-verbal subject placement. Even introductory textbooks in Spanish linguistics mention the differences between most major dialects and those spoken in the Caribbean where the tendency of the latter dialects is to un-invert the subject pronoun in wh- interrogatives such as in (36b). The non-Caribbean variety exhibited in (36a) would either leave out the overt subject pronoun or place it post-verbally for contrastive focus. (36)

Non-Caribbean variety a. ¿Qué dices pro/tú? ¿Cómo estás pro/tú? What say-2sg pro/you how be-2sg pro/you ‘What do you say?’ ‘How are you?’ Caribbean variety b. ¿Qué tú dices? ¿Cómo tú estás? what you say-2sg how you be-2sg ‘What you say?’ ‘How you are?’ (Adapted from Escobar 2010: 409)

Montrul (2004a) points out that a distinctive feature of Dominican Spanish is the less flexible word order in general which tends to be subject-verb regardless of pragmatic conditions. Another characteristic mentioned in Caribbean, Venezuelan, Panamanian (Morales 1997; Escobar 2010), and Dominican Spanish (Montrul

2.5 Regional variations of Spanish subjects |

43

2004a) is the use of overt pronouns as subjects of infinitival verbs as evidenced in (37). (37)

para yo poder venir for I able-inf come-inf ‘for me to be able to come’

Probably one of the most commonly cited differences between Caribbean and nonCaribbean dialects is the greater use of overt subjects in Caribbean varieties. Many authors have examined whether or not this is due to a phonological effect of /s/weakening or elision which, in turn, contributes to ambiguity in the verbal agreement paradigm. For example, if word-final /s/ is weakened or eliminated in the second person singular verb, hablas → hablas ‘you-SG speak’, then the distinction with third person singular, habla ‘he/she speaks’, is lost. The use of the overt subject pronoun would be required then, it has been suggested, in order to compensate for possible referent ambiguity. This has been set forth as the Functional Compensation Hypothesis (FCH) (Hochberg 1986). The FCH would also account for overt subject use in Andalusian dialects, although these are not particularly known for greater overt pronoun use. In fact, however, research has shown that overt pronouns tend to be used to eliminate contextual rather than verbal ambiguity in both Caribbean and Andalusian dialects where /s/ weakening is common (Bentivoglio 1987; Ranson 1991; Silva-Corvalán 1994; Hurtado 2005; Torres Cacoullos and Travis 2011; Travis and Torres Cacoullos 2014) and thus, the accumulated evidence over several studies for several dialectal regions does not appear to support the FCH. Another commonly cited difference in subject use of Dominican Spanish is the custom of placing a non-referential neutral subject in expletive clauses as in (38), where non-Caribbean varieties require an obligatory null subject or an overt subject pronoun for inanimate subjects. This use (also stipulated to be obligatorily null in NSP) is illustrated in (39) below. (38)

Ello es difícil entender. it be-3sg difficult understand-inf ‘It is difficult to understand.’

(39)

Me gusta el café dominicano; él es muy sabroso. me-dat please-3sg the coffee Dominican he be-3sg very delicious ‘I like Dominican coffee; he is very delicious.’ (Lipski posted on http://www.personal.psu.edu/jml34/pronouns.pdf)

44 | 2 Subjects in Spanish Although regional variation in the use of subject expression, and particularly of Caribbean varieties, has been extensively studied in monolingual and bilingual communities, the extent to which regional variation of Spanish subjects impacts L2 acquisition, to my knowledge, has not been investigated. In fact, it is generally the norm in L2 acquisition studies to eliminate participants from Caribbean dialects for comparison between learner groups and native-speaker control groups because pronoun use among these speakers is so diverse¹². Nonetheless, it is important for the reader to keep in mind that when considering L2 learner data, not all uses of subjects would be considered anomalous in all varieties of Spanish. Abreu (2009) points out that studies of the L2 acquisition of Spanish subject pronouns often assume that learners’ overuse of subject pronouns in non-obligatory contexts are ungrammatical when they are, in fact, quite common in the speech of native speakers of several regional dialects. The extent to which ambiguity of verbal forms leads to greater overt subject use (the effect of the FCH) and the complicating factor of L2 learners’ lack of control over verbal morphology is examined in Chapter 7.

2.6 Chapter summary The purpose of this chapter has been to present a general overview of the syntactic, lexico-semantic, pragmatic-discursive, cognitive, and subject referential properties related to Spanish subjects, as well as to provide a brief description of some of the regional variations. Spanish is a pro-drop language and, because of its rich morphological system, syntactically allows both phonetically overt and null (pro) subject pronouns. Textbooks for Spanish learners will often assume or even explicitly state that the use of null vs. overt subject pronouns is optional or used only for emphasis or contrast. This is not the case in many instances and is a simplified explanation of a complex phenomenon. Furthermore, although overt referential subjects may be ‘optional’, expletive subjects that have no referent in the real world are obligatorily null in Spanish. When there is the option to choose between overt and null referential subjects, the use of one or the other produces a

12 Two prominent studies that go against this trend are Geeslin and Gudmestad (2008) and Abreu (2009), which included Puerto Rican (PR) speakers of Spanish in their control and comparative groups respectively. Geeslin and Gudmestad included PR speakers among a variety of regional dialects and Abreu compared L2 learners’ use of subject pronouns to both monolingual and heritage PR Spanish speakers. Nonetheless, neither of these studies specifically targeted the possible influence of the PR dialect on the use and acquisition of subjects by L2 Spanish learners.

2.6 Chapter summary

| 45

different interpretation. For example, in the sentence, Nadiei dice que él∗i/j ganará el premio (‘Nobodyi says that he∗i/j will win the prize’), the Overt Pronoun Constraint, a principle not available in English, stipulates that the overt pronoun él (‘he’) cannot be co-referenced with nadie (‘nobody’). The Position of Antecedent Hypothesis assumes a ‘division of labor’ with null and overt subjects where the null pronoun is linked more often with a more prominent antecedent and the overt pronoun is linked with a less prominent antecedent. No such ‘division of labor’ exists in languages like English. Discourse features of emphasis, contrastive focus, switch vs. same reference, among others, also dictate subject pronoun use as well as a wide range of discourse notions such as cohesion, saliency, presupposition, memory and cognition, speaker attitude, and status of the referent (i.e. person, number, specificity and animacy). One might argue that there are too many aspects of subject expression to be able to cover them all in the foreign language classroom, or that research may tell us that many of these features are general principles related to conversational and narrative discourse that guide subject expression cross-linguistically. Finally, regional dialectal variations dictate the use of null and overt subject pronouns; Caribbean dialects tend to be more ‘pronoun heavy’ and often employ overt subject pronouns for both expletive and inanimate subjects. Dialectal variation affects word order as well; whereas non-Caribbean dialects display variable word order depending on several semantic and pragmatic conditions, Caribbean dialects have been shown to adhere to a stricter subject-verb order regardless of these conditions. Furthermore, Caribbean dialects will use overt pronouns as subjects of infinitival verbs whereas non-Caribbean varieties do not. To date researchers have not examined the effect of these dialectal variations on the L2 acquisition of subject expression; in fact, L2 research is careful to avoid including Caribbean speakers in their experimental control groups for comparison. In sum, the ‘rules’ that determine subject expression in Spanish are far more complex than learners’ textbooks indicate and the acquisition and use of subjects is indeed a challenging task for even advanced learners of Spanish. The challenges for researchers examining the L2 acquisition of Spanish subjects are equally complex. Researchers have spent their academic careers examining just the syntactic properties related to the NSP or analyzing only the discourse features constraining subject expression in discourse, or merely documenting the wide variety of dialectal differences. Where to begin and how to consider all of the intricacies of subject expression and its L2 acquisition would be an overwhelming task. It is the job of the researcher, however, to limit the scope of inquiry and that leads us back to the discussion in the first chapter. What we investigate will largely be determined by what our research questions are, what our view of language is, and what part of language we deem worthy of study. Each investigation and

46 | 2 Subjects in Spanish the culmination of findings will ultimately help us to solve the larger puzzle of what is innately known (i.e. what are general or universal principles), what is learned (i.e. what is language specific), and how it is learned by second language learners.

3 Parametric studies in the acquisition of subject pronouns Introduction As discussed in chapter one of the present volume, proponents of generative approaches in L2 acquisition ask whether or not the principles of Universal Grammar (UG) are available to adult learners and to what extent they guide the L2 learning process even in the face of underdetermined input, otherwise known as the learnability problem. In the parameterized model of the generative approach, it is assumed that L1 learners set a parameter either one way or another according to the options allowed by UG. In this model, researchers ask whether L2 learners start the learning process with the same initial state as the child and set the parameters of their L2 much like the child does in L1 acquisition, or whether the adult learner starts with their L1 setting and gradually resets the parameter to the L2 option. Some proponents of the parametric model have suggested that the learning of one property would lead to an “automatic” learning of related structural properties associated with any given parameter. The approach also allowed researchers to investigate the role of input in the triggering of parameters. Since some of the properties of the Null Subject Parameter (NSP) (Chomsky 1981; Jaeggli 1982) are not necessarily derivable from the input, the investigation of the acquisition of null subjects was considered ideal for testing the poverty of stimulus (POS) hypothesis in adult L2 acquisition. In sum, the parametric study of the acquisition of subject pronouns in L2 Spanish has been concerned with three basic issues: 1) the question of access to UG either directly or via the L1, 2) the acquisition of a cluster of properties associated with the NSP (and to some extent the question of which properties truly form part of the parameter), and 3) the role of input, specifically the quality and quantity (i.e. classroom vs. naturalistic) of the input. Although not all research from this perspective has taken an interest in the role of input in triggering acquisition, there seems to be a consensus as to the importance of establishing the role of UG and of identifying the properties of the NSP (the latter of which has not been uncontroversial). The four properties of the NSP identified by Rizzi (1982, 1986) that have been examined most often in Spanish L2 acquisition are: 1) overt and null referential subjects in tensed clauses, 2) obligatory null expletive subjects, 3) subject-verb inversion, and 4) that-trace effects. Recall from Chapter 2 that because verbal morphology in Spanish is marked for person and number (among other features), a referential subject can be null if it is clear who the referent is given the inflectional verbal morphology or discourse

48 | 3 Parametric studies in the acquisition of subject pronouns context. Expletive subjects, which have no referents in the real world, are obligatorily null in Spanish with meteorological (e.g. pro Llueve mucho. ‘It rains a lot.’) and existential sentences (e.g. pro Hay mucha gente en la calle. ‘There are a lot of people in the street.’), whereas in English an expletive ‘it’ or ‘there’ is required in the subject position. Free subject-verb inversion allows overt subjects to appear either pre- or post-verbally however this property is also subject to syntactic and pragmatic-interpretative constraints (Isabelli 2003). The apparent violation of the that-trace filter in Spanish has been associated with the NSP and investigated in several studies of L2 Spanish, but researchers have questioned whether or not it forms part of the NSP. Nonetheless that-trace is in effect for English but not for prodrop languages like Spanish. An additional principle of null subject languages is the Overt Pronoun Constraint (OPC) (Montalbetti 1986), which states that overt pronouns cannot be bound to a formal variable. There is not a consensus as to whether all of these principles cluster together. Whereas some researchers, at least in the early years, supported and tested Rizzi’s original proposal (Liceras 1989; Al-Kasey and Pérez-Leroux 1998, among others), there are those who have claimed more recently that only referential and null expletives cluster together with the OPC to form the NSP (Rothman and Iverson 2007a; Judy 2011, among others). Furthermore, while some attest to a prominent role for the L1 (Liceras and Díaz 1998, 1999), others make no mention of the issue. All, however, strongly support adult learners’ access to some of the properties of UG. Research examined in this chapter which has studied either one or more of Rizzi’s (1982, 1986) proposed properties includes those who examine the possibility that L2 acquisition is one of learning a clustering of properties (Liceras 1989; Al-Kasey and Pérez-Leroux 1998; Isabelli 2003), those who reject the notion of parameter resetting for subject pronoun acquisition and explore the possibility that the type of morphology of learners’ L1 affects how null subjects are licensed in the L2 (Liceras and Díaz 1998, 1999), and those who claim that referential and expletive subjects may have different underlying representations (Lozano 2002a). A discussion of work that has investigated L2 learners’ knowledge of Montalbetti’s (1986) proposed OPC includes those that have explored the effects of both the OPC and contrastive focus for the L2 Spanish acquisition of null subjects among English speakers (Pérez-Leroux and Glass 1997, 1999) and English and Greek speakers (Lozano 2002b), and a series of studies that investigated whether learners exposed to more abundant and naturalistic input in the study abroad context were at an advantage for resetting the NSP, specifically the OPC (Rothman and Iverson 2007a, 2007b, 2007c). The examination of parametric studies is important because although they offer conflicting evidence of the acquisition of several properties of the NSP and question the extent to which apparently related princi-

3.1 Early research into the L2 acquisition of Spanish subject pronouns |

49

ples are part of the parameter, they also reveal what it is that learners are capable of acquiring. Furthermore, they have brought to the forefront of L2 acquisition studies a renewed interest in the role of language specific L1 representations and the interaction with the more general principles of UG. The organization of the chapter is as follows: earlier studies that examined the possible L2 acquisition of a clustering of properties of the pro-drop parameter and the role of L1 morphological typology are the subject of the next section; research that focused specifically on the L2 acquisition of the OPC and topic/focus contexts are discussed in Section 3.2. In Section 3.3, more recent research that has investigated the acquisition of all of the controversial properties of the NSP and the role of linguistic input is discussed. In the final section, the contributions of these studies and possible commonalities among them are examined.

3.1 Early research into the L2 acquisition of Spanish subject pronouns One of the earliest investigations in L2 Spanish to study the use and acquisition of null subject pronouns was that of Liceras (1989) who examined English and French speakers learning Spanish in the classroom. Assuming that L2 learners do not necessarily begin the acquisition process with the L1 setting of the parameter if the L2 option represents the unmarked option, Liceras set out to establish whether the pro-drop setting or the non-pro-drop setting was the unmarked status of the NSP. Furthermore, she sought to understand the relationship between parameterized (i.e. null subjects) and nonparameterized aspects of the grammar (i.e. Spanish personal a-marking of direct objects). The author suggested that the acquisition of null subjects was a prerequisite for acquiring other related properties of the pro-drop parameter, such as subject-verb inversion and that-trace violations; in other words, there is an implicational hierarchy linking the three properties of the pro-drop parameter such that if that-trace has been acquired, then inversion and pro-drop have also been acquired, but acquisition of inversion or pro-drop does not necessarily mean that that-trace has been acquired. This is a hierarchy of difficulty where of the three properties, pro-drop is the first learned and that-trace is the most difficult and last to be acquired. Liceras asked four different groups of French and English speakers of L2 classroom Spanish (from beginning to advanced levels) and native Spanish speakers to consider a variety of written sentences, to judge their grammaticality, and to correct the ones they considered ungrammatical. The task included sentences with overt and null subject pronouns (both referential and expletive), verb-subject in-

50 | 3 Parametric studies in the acquisition of subject pronouns version, and that-trace effects, including sentences with either empty or lexicalized complementizers in the subordinate clause. Following are examples of the types of sentences participants were asked to judge: (40)

Ungrammatical overt expletive *Ello hace mucho frío en Canadá. It make-3sg much cold in Canada ‘It’s very cold in Canada.’

(41)

Grammatical null expletive Sí, pro hay dos libros de español. yes pro exist-3sg two books of Spanish ‘Yes, [there] are two books of Spanish.’

(42)

Grammatical verb-subject inversion Ya han salido los profesores. Already have-3pl left the professors ‘Already the professors have left.’

(43)

Ungrammatical that-trace filter *¿Quién pro dices t estudia español contigo? who pro say-2sg t study-3sg Spanish with-you ‘Who do [you] say studies Spanish with you?’ (Adapted from Liceras 1989: 117–118)¹³

The author found that all learners accepted null subjects with expletives and rejected expletives with overt pronouns; they also uniformly accepted null subjects with third person plural arbitrary reference. In the case of subject-verb inversion, she found that although the native speakers accepted inversion with all sentences, the learners did so more often with ergative verbs (e.g. llegar ‘arrive’) than transitive verbs (e.g. telefonear ‘to telephone [someone]’), especially the beginning learners. Learners who had not yet acquired the obligatory Spanish a- ‘to’ marking of direct objects that refer to people may have been interpreting the post-verbal subject of transitive verbs as the object. Liceras suggests that this is an indication

13 Examples taken from studies are sometimes shortened or combined and the formatting may have changed from the original so as to conform to the same format used throughout the chapter. Specifically, all null pronouns in clauses are indicated with pro whether they appear as such in the original examples or not, and all Spanish examples are translated in single quotation marks.

3.1 Early research into the L2 acquisition of Spanish subject pronouns |

51

of the role of a non-parameterized aspect of grammar (a-marking) in triggering a property of the parameterized aspect (inversion). Finally, the results for thattrace effects were mixed. In general all groups, including the native speakers, had difficulty interpreting interrogative sentences with the missing complementizer que ‘that’ and that-trace effects. On the other hand, all groups showed gradual improvement in interpreting and correcting declarative sentences with a missing complementizer, although the French speakers did slightly better than the English speakers. Liceras’ results revealed that resetting the pro-drop parameter from English and French to Spanish is not difficult as evidenced by all levels of learners’ acceptance of null subjects. The author interprets this as an indication that the pro-drop option is the unmarked setting of the parameter and that Spanish L2 learners, at least English and French speakers, do not start the learning process with the L1 setting, but rather rely on UG. Nonetheless, the data from this study could not provide evidence that suggests a clustering of these properties for the acquisition process. Learners were successful in resetting the parameter for the null subject and subject-verb inversion for Spanish but not for that-trace effects. Liceras could not establish whether the acquisition of null subjects was a prerequisite for acquiring inversion or that-trace. She did, however, maintain that the three properties are related but may not have the same status and suggested that both verbal morphology and the nonparameterized aspect of the Spanish personal a-marking of direct objects may trigger verb-subject inversion. Following Liceras, Isabelli (2003) examined the acquisition of the same three properties of the null subject parameter in a group of native English speakers studying Spanish in a nine-month study abroad immersion program in Barcelona, Spain. Specifically, Isabelli wanted to see if exposure to an abundance of positive evidence (i.e. linguistic input) in a naturalistic setting would aid learners in resetting the [−null] parameter of their L1 to the [+null] setting of the L2, and if this greater amount of input would help learners to acquire the cluster of properties associated with the parameter, in this case, null subjects, verb-subject inversion, and apparent violations of that-trace sequences. Subjects in the study included 31 intermediate-level students from U.S. institutions participating in the study abroad program and who did not speak Spanish at home or had had prior experience living in a Spanish-speaking country, and a control group of 18 monolingual native speakers from Spain. Data were collected from the learner group one month after arrival in Spain and again after nine months; data from the native speaker group were collected only once. Participants were asked to complete two tasks: a grammaticality judgment task (GJT) with both grammatical and ungrammatical (when possible) representations of the three properties, and oral interviews based on different pictures and questions. Examples of stimuli for the GJT are as follows:

52 | 3 Parametric studies in the acquisition of subject pronouns (44)

Null subject pronoun pro Salieron de la casa a las ocho. pro leave-3pl-pret of the house at the eight ‘[They] left the house at eight.’

(45)

Null expletive pro Hay un perro grande en el jardín. pro exist-3sg a dog big in the garden ‘[There] is a big dog in the yard.’

(46)

Ungrammatical null expletive *Ahora lo está lloviendo afuera. now it be-3sg raining outside ‘Now it’s raining outside.’

(47)

Verb-subject inversion Durmió el bebé 3 horas. sleep-3sg-pret the baby 3 hours ‘The baby slept 3 hours.’

(48)

that-trace effects ¿Quién cree la policía que mató al joyero? who believe-3sg the police that kill-3sg-pret to-the jeweler ‘Who do the police believe that killed the jeweler?’

(49)

Ungrammatical that-trace effects *¿Quién creen Uds. t pegó al hombre? Who believe-3pl you-pl t hit-3sg-pret to-the man ‘Who do you believe hit the man?’ (Adapted from Isabelli 2003: 155–156)

In general, the results of Isabelli’s analysis confirmed Liceras’ (1989) findings that learners were successful in resetting the parametric values for null subjects and verb-subject inversion, but not for that-trace effects. The results of the GJT revealed no significant differences between month one and month nine for judgments of null subjects and both grammatical and ungrammatical expletives indicating that the learners were already performing at ceiling levels (perfect or almost perfect scores) upon their arrival in Spain. There was a significant increase in their accurate judgments of verb-subject inversion and grammatical that-trace repre-

3.1 Early research into the L2 acquisition of Spanish subject pronouns |

53

sentations indicating improvement over time. However, for ungrammatical thattrace items, although there was a slight increase among the learners in rejecting these, the difference from month one and month nine was not statistically significant. In fact, learners did quite poorly on these sentences and were unable to identify ungrammatical instances of that-trace effects during both data-collecting sessions. In general, learners’ nine-month scores for the GJT revealed that they performed like the native speakers for both null subjects and verb-subject inversion but not for ungrammatical that-trace effects. The results of the oral interviews supported those of the GJT for nulls subjects and verb-subject inversion (there were no instances either among the learners or the native-speaker group of that-trace sentences). Learners produced more null subjects and fewer overt subjects (both referential and expletive) at month nine than at month one and also increased their production of verb-subject inversion sentences; in addition, there were no ungrammatical expletive sentences produced. Again the behavior of the learners at month nine was not significantly different from that of the native speakers for two of the three parametric properties. Isabelli suggests that there are several explanations for learners’ persistent problems in acquiring the Spanish setting for that-trace effects. It may be that because sentences in which that-trace effects are evident are rare in the input. She asks if learners might well acquire the property with even more exposure to the L2 language (perhaps after another year of study in Spain?). Or it might possibly be, as Liceras postulates, that that-trace is the last and most difficult property to acquire clearly suggesting, as does Liceras, that a hierarchy of difficulty exists for the different properties of the parameter. A subsequent analysis of individual learners revealed that some of the more advanced L2 learners were successful in consistently rejecting the ungrammatical that-trace sentences indicating that acquisition is, indeed, possible, although not as a cluster, as previously suggested. There are, of course, alternative explanations ranging from the possibility that that-trace is not part of the null subject parameter and therefore not learned in the same way as referential and expletive null subjects and verb-subject inversion, to the claim of others that adult learners no longer have access to the principles of UG and only have access to those principles as they are activated in the L1. Isabelli’s final claim is that the properties of the NSP are acquirable, although perhaps not as a cluster, and that abundant, naturalistic L2 input is the necessary trigger for the activation of these properties. These claims have been challenged more recently by others (Rothman and Iverson 2007a, 2007b, 2007c). The possibility that the morphological typology of learners’ native languages affects how null subjects are licensed within their interlanguages was investigated by Liceras and Díaz (1998). Based on more recent findings and contrary to the conclusions of Liceras (1989), in this study the authors assume a full transfer view,

54 | 3 Parametric studies in the acquisition of subject pronouns which predicts that for adult L2 acquisition the initial state is L1 knowledge, but not necessarily parametric options. Current developments in linguistic theory at the time, which sought to offer a formal account of the link between semantic and pragmatic values of morphological features and null and overt pronouns, allowed for a typology of languages consistent with the f -features they project. According to this proposal, languages vary according to “the realization of the morphosyntactic features that instantiate infl as person, number, tense, and so forth” (Liceras and Díaz 1998: 308). Thus, this gave morphology a prominent role in the triggering of UG in L2 acquisition because the operation of the principles of UG in the grammar of the target language would depend largely on learners’ grasp of morphology. Liceras and Díaz therefore set out to establish whether learners would realize the abstract features of their L1 in the construction of their L2 interlanguage grammar. In other words, would the inflection type of learners’ L1 affect their L2 in beginning and even more advanced stages? The authors analyzed the production of both beginning and advanced learners of three different typological languages, those whose functional heads are specified for person (French, Spanish, Italian) or type-a languages; for number (English, Danish, Swedish), or type-b languages; or for tense (Japanese), a type-c language. The full-transfer view provided the authors with the theoretical foundation by which to hypothesize that learners of Spanish whose L1 was a type-b language would not allow null subject pronouns because as adults they possibly are not able to grasp the abstraction that allows them to project the functional feature (f -features) of person rather than number. French is considered a special case because although it is typologically a type-a language like Spanish, its f features – clitic subject pronouns – are mostly preverbal and thus it would be expected that person is already projected. However the authors predicted that their French learners would redundantly employ a set of overt tonic pronouns like those in French until they recognize that Spanish has post-verbal f -features and lacks a second set of tonic pronouns. Finally, it was supposed that the Japanese speakers (type-c) would allow null subjects independently of the verbal forms since their L1 does not have the f -features of person or number and pronouns would not substitute for these features. In order to test their predictions, the authors analyzed the use of subject pronouns in the spontaneous oral production of English, Danish, Swedish, bilingual French-English, and Japanese speakers from beginning and advanced levels of Spanish. All speakers were asked to participate in story-telling activities that were based on pictures presented to them. The results of the analysis of speakers’ production of subject pronouns were not conclusive. All beginning learners produced a considerable number of null subjects but, whereas among the advanced type-b (English, Danish, Swedish) and French-English bilingual speakers the percentage of null subjects increased

3.1 Early research into the L2 acquisition of Spanish subject pronouns |

55

substantially, the type-c (Japanese) speakers maintained the same level. Furthermore, whereas the type-b speakers produced null subjects if the referent was identified in the discourse and used lexical or pronominal subjects when there was a topic shift, the type-c speakers produced null subjects even in topic shift contexts. The authors concluded that, in general, L2 learners are not sensitive to the morphosyntactic triggers that were predicted to license null subjects, and since learners do not make use of syntactic clues, it is most likely that they depend on the pragmatic functions of null and overt pronouns to convey and retrieve meaning, at least the speakers of type-a and type-b languages. Another possible explanation for the data is that the strength of the L2 morphological values is not available to adult learners. The authors suggest that early L2 grammars do not clearly reflect the native language syntactic representations, which lends support to the importance of L2 input and UG principles in L2 adult acquisition. Nonetheless, the authors are convinced that although the surface production is the same (e.g., null subjects appearing at early stages), this output originates from distinct underlying grammatical representations that are based on their different L1 typologies. This is in contrast to Liceras’ earlier study (1989) where she claimed that L2 learners of Spanish do not start the acquisition process from the L1 knowledge state, at least in terms of the acquisition of the NSP. In Liceras and Díaz (1999) the authors continued to explore the notion that subject acquisition among adult learners of different native languages is not due to the resetting of the L1 value of the pro-drop parameter but rather is the result of a default licensing procedure where learners use subject pronouns for identification purposes. The authors again turned to the recent modifications to the Principles and Parameters model of the time, which had separated UG principles from parametric options and suggested that all parameters were located within functional categories (Chomsky 1991). This led researchers to propose that adult L2 learners have access to UG but not for the resetting of L2 parameters; that is, the triggering experience responsible for the development of L1 is not available for L2 acquisition (Lightfoot 1991; Pinker 1994). Following Tsimpli and Roussou (1991), Rizzi (1994) and Hyams (1994), Liceras and Diaz suggested that non-native pronominal subjects do not have the same value as native Spanish subjects and that the production of null subjects in L2 Spanish of speakers from different typological languages is the result of distinct grammatical representations. The argument is that if the functional component is subject to maturation, it can only mature once. What we maintain is that L1 learners set parameters by letting an initial representation grow out of exposure to triggering data in the adult L1 system. L2 learners, on the other hand, do not grow initial representations because what they possess are grown representations. This

56 | 3 Parametric studies in the acquisition of subject pronouns implies that they are not sensitive to the specific triggers that lead to parameter setting in L1 acquisition. (Liceras and Diaz 1999: 5)

The authors proposed that L2 acquisition proceeds via secondary level domainspecific procedures that restructure previous linguistic representations according to the learner’s L1. Languages differ according to whether or not they are [+/−prodrop] where null subjects are licensed if the INFL features are [+strong] (French, Spanish), and not licensed if they are [−strong] (English); or [+/−topic-drop] where null subjects are possible if they are both licensed via the [+strong] feature located at Spec-C󸀠 and identified (Chinese, Japanese), but are not licensed if they are [−strong] (Standard German). Because L2 learners are not aware of the abstract [+/−strong] features of the functional categories, they cannot reset the Spanish [+pro-drop] option of the parameter. What L2 learners need to do to acquire Spanish null subjects is restructure their L1 representations based on verbal morphology (i.e. the person marker), the Spanish subject pronouns or null topics. To test these claims, Liceras and Diaz examined the spontaneous oral production data of L2 Spanish learners, both adolescents and adults, of several L1s (Chinese, English, French, German and Japanese) to ascertain whether or not there were differences in the production of null or overt subjects according to L1 background, learner level, and syntactic position (matrix vs. subordinate clause subjects). The data were elicited via oral interviews where subjects were asked to answer questions about their family and school life and told stories based on comic strips. The authors had hypothesized that speakers of [−pro-drop] (English) and [−topic-drop] (German) languages would produce more overt subjects than native speakers of [+topic-drop] languages (Chinese, Japanese). They assume that French is a [+pro-drop] language due to the [+strong] feature of French infl which allows licensing of pro at Spec-I’ in the form of a pronominal subject clitic, and that the L2 Spanish subject pronoun production of French speakers would most closely resemble that of native Spanish speakers. They also proposed that there would be differences in the use of null and overt subjects in matrix and subordinate clauses according to learners’ levels of competency. It was found that the French-English speakers in the early stages of acquisition, both adolescent and adult learners, produced null subjects in both matrix and subordinate clauses although, like the production of the native speakers, there were more null subjects for subordinate clauses. Liceras and Diaz claim that these learners take French as their initial representation and identify subjects based on their L1 knowledge of agreement features and therefore do not identify based on subject pronouns. The more advanced learners from the various

3.1 Early research into the L2 acquisition of Spanish subject pronouns |

57

language groups displayed similar results, although the authors had expected different results based on the particular languages. The advanced intermediate French speakers were expected to mirror the native speaker control group but they found that this group used even fewer overt subjects than native speakers and more null subjects in subordinate clauses. The advanced intermediate English speakers, who were expected to overuse overt subject pronouns, were found to have already abandoned pronouns as identifiers and used more null subjects in the matrix clauses, although the difference between null subject production between matrix and subordinate clauses was not statistically significant in the data. The authors had hypothesized that the advanced intermediate German speakers would transfer the [topic-drop] parameter of their L1 to their L2 Spanish and expected null subjects only in the matrix clauses, but did not find this to be the case; the German speakers produced more null subjects in the subordinate clauses. Finally, it had been predicted that speakers of Chinese and Japanese, both [+topicdrop] with [+strong] features, would license Spanish null subjects because their L1s also license them, albeit via a discourse null operator which is present in the specifier of the root; however, when identified by null topics, they produce ambiguity which Spanish does not permit. This prediction was not borne out, as none of the advanced intermediate Chinese and Japanese speakers produced ambiguous sentences in Spanish. Furthermore, they used null subjects in both matrix and subordinate sentences much like the learners from the other languages. The authors interpret their results as evidence that parameter resetting is not the cause of these learners’ interlanguage production of null and overt subject pronouns but rather is the result of “a default licensing procedure which is responsible for the production of null subjects provided they are identified” (Liceras and Díaz 1999: 34) and that it is, in fact, the identification procedure that undergoes restructuring in the acquisition of L2 Spanish subject pronouns. Thus far, the research discussed here have left us with inconclusive results regarding both the role of the L1 and the question of access to UG for the resetting of the L2 parameter and the status of the acquisition of a cluster of properties associated with the NSP. Studies that examined more closely the distinct properties of the NSP, however, were in the works. Al-Kasey and Pérez-Leroux (1998) looked at the acquisition of a cluster of two distinct domains related to the NSP, specifically null expletives and null thematic subjects. The authors looked to support or reject three hypotheses that would explain learners’ path to the resetting of the NSP. Their first hypothesis stated that there is no parameter underlying both null expletives and optional thematic subject pronouns and the evidence in favor would be a slow increment in the use of null subjects with no correspondence between the two types or no progress at all; the second hypothesis projected that both null subjects form part of the

58 | 3 Parametric studies in the acquisition of subject pronouns same parameter but that optional thematic subjects are more difficult to learn than obligatory null expletives for which the evidence would be an increase in the use of null expletives without a resultant decline in thematic null subjects; and finally, their third hypothesis indicated that a rapid increase in the use of both null subjects at the same time would prove it is possible to acquire a single parameter with two distinct domains of the language. In order to test these hypotheses, the authors established three linguistic contexts where the NSP operates: (50)

Expletives (learners may assume a subject pronoun is possible) pro Es divertido volar. pro be-3sg fun fly-inf ‘[It]’s fun to fly./Flying is fun.’

(51)

Raising verbs (pronoun is possible but is not a subject) a. pro Resulta interesante ir al museo de Boston. pro result-3sg interesting go-inf to-the museum of Boston ‘[It]’s interesting to go to the Boston Museum.’ b. Le pro resulta interesante ir al museo de Boston. him-DAT pro result-3sg interesting go-inf to-the museum of Boston ‘[It] is interesting for him to go to the Boston Museum.’

(52)

Referential sentences (subject pronoun is optional and assigned a thematic role) Esta es divertida de volar. this be-3sg fun of fly-inf ‘This is fun to fly.’ (Adapted from Al-Kasey and Pérez-Leroux 1998: 168)

English-speaking L2 Spanish learners from five different levels ranging from elementary to advanced majors and a native speaker group were asked to participate in both comprehension and production tasks. In the comprehension task, the participants were asked to read sentences and link them to a series of visual stimuli where the obligatory null pronoun should correspond to a picture with a generic interpretation and the use of the overt referential pronoun should correspond to a picture with a referential interpretation. The production task consisted of a cloze test with blanks at the beginning of each sentence of a story. Participants were asked to fill in the blanks with an appropriate pronoun if needed, a null symbol (0) if no pronoun were required, and a question mark (?) if they thought something was required but weren’t sure what it was.

3.1 Early research into the L2 acquisition of Spanish subject pronouns |

59

The results showed that all learners used both null expletives and null thematic subject pronouns at about the same time, although the null expletives were used slightly earlier and more consistently. In addition, there was a strong correspondence between type of error with type of verb, with expletives occurring with more categories of subject errors (impossible noun phrases, demonstrative pronouns, es, and lo). Furthermore, no learner used null thematic subject pronouns before null expletive pronouns, which according to the authors, is evidence that these two properties are clustered and that learning one property implies learning the other. Additionally they state that their findings are evidence that L2 learners have access to UG principles, although they claim that “(t)here is both a UG and a non-UG explanation for such behavior” (Al-Kasey and Pérez-Leroux 1998: 178). The UG explanation is that adult learners have two grammatical systems which temporarily coexist, much like the simultaneous use of goed and went, for a while, in the child grammar. Or adults may cautiously allow the new parameter to apply in an attempt to avoid incorrect generalizations. The non-UG explanation is that less experienced learners have not set the new parameter but are merely producing unanalyzed lexical blocks, which appear native-like at times, yet cause errors at others. The authors do not ascribe to this view but how one is to tease these possibilities apart is not discussed. The study is an advancement, nonetheless, in that the authors recognize the importance not only of establishing the role of UG and parameter setting in the acquisition of subject expression in Spanish L2, but also in the fact that they recognize, and empirically analyze, the different linguistic contexts, both referential and expletive, that license null subject use in the Spanish grammar. Lozano (2002a) expanded on Al-Kasey and Pérez-Leroux’s (1998) examination of learners’ acquisition of the obligatorily null expletive pronouns (which he abbreviates as ExpS) and the so-called optional referential pronominal (thematic) subjects (abbreviated in his study as ProS). He set out to specifically test whether or not learners treat overt and null ProS differently from overt and null ExpS. Arguing that although the two types of null pronouns are identical on the surface, they possess different underlying mental representations as witnessed by their distinct semantic and syntactic behavior. Syntactically null ProS refers to and can alternate with overt subject pronouns (yo ‘I’, tú ‘you’, etc.) while in contrast, for ExpS the null option is the only possibility. Semantically, although referential pro distinguishes for person, number and gender, expletive pro is obligatorily third person singular and always gender neutral. Lozano believed that if he could show that learners are sensitive to these differences from the earliest stages of acquisition, then this would be proof that learners have different mental representations of null ProS and ExpS. Specifically, he hypothesized that learners would show an awareness of the parametric differences between the two types of null pronouns

60 | 3 Parametric studies in the acquisition of subject pronouns early on and that they would treat grammatical null and ungrammatical overt ExpS differently but grammatical null and overt ProS similarly. Furthermore, although he expected learners to accept both null and overt ProS from the earliest stages of acquisition, he hypothesized a gradual shift in learners’ acceptance of null ExpS and rejection of overt ExpS as proficiency increased. In order to test these hypotheses, Lozano examined three levels of native speakers of English studying L2 Spanish and a group of native Spanish speakers for control. All participants were presented with two different tests of paired grammaticality judgment tasks, one focused on ExpS and the second on ProS, and participants had to decide whether both sentences were wrong, both were right or one was wrong and the other one right. For example, both referential subjects in (53a) and (53b) are correct, whereas only the null option for expletive sentences like that in (54b) is correct in Spanish. (53)

a. Yo voy a la Universidad en coche. I go-1sg to the university in car ‘I go to the university in car.’ b. pro Voy a la Universidad en coche. pro go-1sg to the university in car ‘[I] go to the university in car.’

(54)

a. *Lo nieva en Finlandia en invierno. It snow-3sg in Finland in winter ‘It snows in Finland in winter.’ b. pro Nieva en Finlandia en invierno. pro snow-3sg in Finland in winter ‘[It] snows in Finland in winter.’ (Adapted from Lozano 2002a: 47)

The results revealed that although all learner groups rejected overt ExpS more often than they accepted them, the very beginning group treated both null and overt ExpS more similarly than the more advanced and native speaker groups. Also acceptance of overt ExpS tended to decrease and acceptance of null ExpS increased as proficiency improved and a statistically significant linear trend confirmed that resetting the ExpS parameter is a gradual process and is not as instantaneous as the resetting of ProS. The results of the second paired grammaticality judgment task confirmed that learners treated both null and overt ProS the same from the earliest stages of learning and accepted null and overt referential pronouns at the same rate.

3.2 Research on the acquisition of the OPC and Topic/Focus |

61

Lozano’s results showed that his learners were sensitive to the differences between null and overt ExpS from the earliest stages of acquisition, although their judgments improved slightly with time, whereas they accepted both null and overt ProS early on. Lozano concluded that because ungrammatical overt ExpS is different from grammatical overt ProS, learners (and native speakers) have different mental representations for these two types of pronouns. Although his findings are similar to those of Al-Kasey and Pérez-Leroux (1998), Lozano makes no claims regarding learners’ access to Universal Grammar; furthermore, whereas Al-Kasey and Pérez-Leroux claim that null expletive and referential pronouns are part of the same parameter (although null ProS does not occur until null ExpS has been acquired), Lozano claims to have provided evidence that learners have different mental representations for these two types of pronouns. He suggests that further empirical studies are required in order to understand the internalized representations of pro features among different learning levels. Further empirical studies were also needed to understand more fully learners’ acquisition of the different properties hypothesized to form part of the NSP given the conflicting evidence for the status of referential and expletive subjects, subject-verb inversion and thattrace effects among the different investigations.

3.2 Research on the acquisition of the Overt Pronoun Constraint (OPC) and Topic/Focus One property of the NSP that has been investigated extensively apart from others is L2 learners’ knowledge of the OPC (Montalbetti 1986). Recall that the OPC applies to pro-drop languages such as Spanish (as well as topic-drop languages like Chinese), but not to non-pro-drop languages like English. Learners are not formally taught the restrictions of the OPC and complex sentences where it operates are not abundant in the input. Therefore, testifying that learners have knowledge of this property could lead to valuable insights regarding learners’ access to UG and lend support to the “poverty of stimulus” hypothesis for adult L2 acquisition. Although pronoun distribution in English and Spanish differs in regards to OPC behavior (see Chapter 2 for a detailed discussion of OPC in Spanish and English), it is similar for focused environments or contrastive focus (CF). Thus, in the sentence, Juani cree que éli/j ganará el premio (‘Juan believes that hei/j will win the prize’), the overt pronoun in the subordinate clause can either refer to Juan or to another referent depending on the discourse context. This corresponds to the English sentence, John thinks HE will win the prize, where the stressed pronoun can refer to either John or another referent.

62 | 3 Parametric studies in the acquisition of subject pronouns Pérez-Leroux and Glass (1997) explored different semantic effects on both OPC and focused environments for the acquisition of null subject pronouns among highly proficient English-speaking learners of L2 Spanish and native speakers. The authors proposed three possible acquisition scenarios, which if were to be confirmed, would yield different outcomes. The first is that if learners are merely achieving “a surface redescription of their pronominal system” (Pérez-Leroux and Glass 1997: 155), then they would view null and overt subject pronouns as translation equivalents of English pronouns and use them randomly in all contexts. The second scenario predicted that if UG properties were accessible only via the L1, then learners would be able to acquire the properties of pronoun distribution for focused environments, which parallel those of their native language, but not for OPC effects which are not operable in the L1. The third scenario proposed was that if learners have full access to the principles of UG and are not dependent on L1 properties, then they would be able to use subject pronouns in Spanish in accordance with the OPC. The authors proposed that since contrastive focus in Spanish parallels English, learners would learn this use of subject pronouns easily. If learners have access to the principles of UG, they will also learn the effects of OPC. Two groups of participants, learners and native speakers, were asked to complete two contextualized tasks where they were asked to read short stories of two to three sentences; in addition, for the OPC task, they were asked to translate the last sentence of the story into Spanish, and for the topic/focus task, participants were asked to answer a question. The tasks were designed to test whether learners and native speakers would produce null and overt pronouns for bound and referential contexts, following the OPC, for the first task, and topic and contrastive focus for the second. An example of an expected bound variable interpretation for the OPC task is presented in (55). (55)

Bound Variable Story. The Spanish class has decided to have a pot-luck party at the end of the semester. It will be a time to celebrate. To translate: Each student said that he would bring something to eat. Prompt: Cada estudiante . . . each student Target translation: Cada estudiante dijo que pro traería algo de each student say-3sg-pret that pro bring-3sg-cond something of comer. eat-inf ‘Each student said that [he] would bring something to eat.’ (Adapted from Pérez-Leroux and Glass 1997: 157)

3.2 Research on the acquisition of the OPC and Topic/Focus |

63

In this case, the researchers expected both native speakers and learners to be guided by the OPC and to select the null option (pro) in the subordinate clause since the intended meaning is a bound variable, i.e. coreference of the subordinate subject with the matrix subject, cada estudiante (‘each student’). Example (56) demonstrates the topic/focus contexts, where the expected answer for the object question (i.e. a topic context) is a null pronoun (pro) and the expected response for the subject question (i.e. a focus context) is an overt pronoun (él ‘he’). These two contexts correspond to English use where the expected response for a topic context is an unstressed subject pronoun and the expected response for a focus context is a stressed subject pronoun. (56)

Topic/Focus Story: Sólo hay una galleta para comer después de la cena, y todos only exist-3sg one cookie for eat-inf after of the dinner and all tienen mucha hambre. have-3pl much hunger ‘There is only one cookie to eat after dinner and everyone is very hungry.’ Object Question: ¿Qué piensa Cookie Monster que hará después de la cena? what think-3sg Cookie Monster that do-3sg-fut after of the dinner ‘What does Cookie Monster think he will do after dinner?’ Expected Response: Cookie Monster piensa que pro comerá la galleta. Cookie Monster think-3sg that pro eat-3sg-fut the cookie ‘Cookie Monster thinks that [he] will eat the cookie.’ Subject Question: ¿Quién piensa Cookie Monster que comerá la galleta? Who think-3sg Cookie Monster that eat-3sg-fut the cookie ‘Who does Cookie Monster think that will eat the cookie?’ Expected Response: Cookie Monster piensa que él comerá la galleta. Cookie Monster think-3sg that he eat-3sg-fut the cookie ‘Cookie Monster thinks that HE will eat the cookie.’ (Adapted from Pérez-Leroux and Glass 1997: 158)

64 | 3 Parametric studies in the acquisition of subject pronouns The results of the two tasks were surprising. Although the authors had expected that native speakers’ use of pronouns would be constrained by the OPC in the first task and would select null pronouns 100 % of the time for bound variable contexts and overt pronouns 100 % of the time for referential contexts, it was seen that the learners produced even more null subject pronouns than the native speakers for the bound variable contexts (88 % compared to 75 %). However, the learners also showed a clear preference for the use of null pronouns in the referential contexts, although to a lesser extent than for bound contexts, whereas native speakers tended to avoid null pronouns in these contexts. There was a significantly main effect for group and a highly significant main effect for context type, which the authors interpreted as evidence that although there is some overgeneralization on the part of the L2 learners to overuse null subject pronouns across conditions, there is also a significant effect of pronoun use that attests to learners’ knowledge of OPC effects. The results of task two which tested for knowledge of the null/overt alternation in focused contexts were also unexpected and both groups preferred null pronouns for both topic and focus contexts, with the L2 learners somewhat outperforming the native speakers for the topic contexts with 81 % use of nulls compared to 71 % nulls for the native speaker group. For the focus contexts, where the authors expected a higher use of overt pronouns, the native speaker group outperformed the learner group, but with only 36 % use of overt pronouns compared to 20 % use among the learners. Again it is seen that despite the fact that the L2 learners overgeneralize the use of null subject pronouns, there is a significant main effect for pronoun use between topic or focus context, although this is not as strong as the contrast of pronoun use seen in the OPC task. This is evidence that adult L2 learners of Spanish are capable of mastering both the semantic and pragmatic properties of the null/overt contrast and the effects of a principle, the OPC, which is not operable in their L1. What is surprising is that there was less evidence for L1 transfer than that hypothesized by the authors. As discussed in chapter four of this volume, it will be seen that pragmatic constraints such as focus and topic are not the same in English and Spanish as supposed by the authors. Although it is commonly accepted that where English has prosodic stress for contrastive focus, Spanish uses overt pronouns. Nonetheless, L2 learners of Spanish must learn to associate obligatory overt pronouns with topic and focus contexts and null pronouns with non-focus and non-topic contexts. In a subsequent study, Pérez-Leroux and Glass (1999) again tested learners’ development of null and overt pronouns in both OPC and topic/focus contexts, this time with the purpose of discrediting Ellis’ (1996a, 1996b) claims that language acquisition can be explained by the frequency of L2 input. Because OPC contexts are infrequent in the data and topic/focus contexts are very common, the authors believe that an investigation of the use of null subject pronouns in

3.2 Research on the acquisition of the OPC and Topic/Focus |

65

these two contexts can lend support to the generativist explanation of L2 acquisition that “language learners attain knowledge of structural properties which are underdetermined by the available input” (Pérez-Leroux and Glass 1999: 221). If the authors could provide evidence that learners early on have knowledge of and are sensitive to the OPC, a property that is underdetermined in the input, while the topic/focus contrast needs to be learned, this would be support for the poverty of stimulus argument for adult L2 learners. Furthermore, the overt/null contrast in topic/focus contexts is constrained by discourse functions, whereas the same contrast in OPC contexts is a purely formal distinction. Thus the two contexts have different statuses in terms of grammatical knowledge and should display different patterns of acquisition. The authors predicted that 1) “Knowledge of the overt/null pronoun distinction in OPC contexts should be present from early stages of L2 development” . . . and that 2) “Knowledge of the semantic properties of pronouns in topic and focus environment should develop with experience” (Pérez-Leroux and Glass 1999: 239). The second prediction here is, in fact, contra their previous claims that topic and focus contexts would be easier to learn due to positive transfer from L1 English. The research reported on two studies designed to once again test participants’ interpretation of both referential vs. bound variables, as constrained by the OPC, and focused vs. non-focused contexts. The authors made use of the same contextualized translation tasks as in their (1997) work but expanded the two studies to include learner groups from three different proficiency levels, including elementary, intermediate and advanced levels, and native-speaker control groups. The results of the OPC study showed that all learner groups and native speakers produced a higher proportion of null pronoun subjects for the bound variable contexts and a greater percentage of overt pronouns, with the exception of the advanced learner group, for the referential contexts. As in their (1997) study, the authors found that the highest proficiency group tended to overgeneralize and overuse null subject pronouns in referential contexts where the overt pronoun would be expected, although a null pronoun is not ungrammatical in these contexts. Although there were some violations of the OPC (i.e. overt pronouns in bound variable contexts), these were mostly seen in the two lower proficiency groups and native speakers of Caribbean dialects (known for a higher use of overt pronouns). Nonetheless, none of the speakers produced more overt pronouns in the bound variable contexts than in the referential contexts. Furthermore, all groups made a statistically significant distinction between environments that called for null subjects and those that did not, which the authors claim is evidence that the OPC operates even at the beginning levels of language acquisition. For the focus study, results revealed that as proficiency level improved among the learner groups, their production of null pronouns in topic contexts and overt

66 | 3 Parametric studies in the acquisition of subject pronouns pronouns in focus contexts increased. Although the differences among the groups was statistically non-significant due to the fact that groups produced comparable rates of null pronouns, the distribution of the null pronouns across story types showed a significant interaction. Specifically, in comparison with the two lower proficiency groups, the advanced learners and native speakers increased their use of null subject pronouns in topic contexts and decreased their use in focus contexts, as expected. Thus the results of the two studies revealed that language proficiency had a different effect on the use of null pronouns in OPC and focus studies. In short, in the OPC study, as language proficiency improved among the learner groups, null pronoun use increased in both bound variable and referential contexts. On the other hand, in the focus study, the use of null pronouns increased with language proficiency only in the topic contexts but remained stable in the focused contexts. The authors interpreted these results as evidence that OPC effects are available to learners in the early stages whereas knowledge of the semantic and pragmatic properties of pronoun use in topic and focus contexts must be learned and only develops as language proficiency improves. However, a closer look at the OPC contexts that were tested reveal that there are also discourse constraints that appear to be at work. In the contexts where the researchers expected an overt referential pronoun, they claim that there is little improvement and thus this is evidence that the OPC is impervious to learning. However, even 31 % of the native speakers produced a null subject for these contexts. In the story, see (57) below, that the authors present as an example of a referential context in which participants should produce an overt pronoun, él (‘he’), the use of a null subject would be grammatical and would not be ambiguous at all; there is only one antecedent who could possibly be considered guilty and that is the defendant (i.e. O.J. Simpson), not the press and not a journalist. The discourse context makes it clear that the subject of the subordinate clause could only refer to “the defendant”. This is a problem in studies that assume the OPC is a purely syntactic principle and not susceptible to discourse or contextual constraints. (57)

Referential story: In the O.J. Simpson trial, it is clear that the press has a negative bias against the defendant in their reporting. Some journalist said that he was a wife-beater. To translate: ‘But no journalist said that he is guilty.’ Target translation: Ningún periodista dijo que él era culpable. not one journalist say-3sg-pret that he be-3sg-imp guilty ‘No journalist said that HE was guilty.’ (Adapted from Pérez-Leroux and Glass 1999: 232)

3.2 Research on the acquisition of the OPC and Topic/Focus |

67

Nonetheless, the results of this study are revealing because although the authors make no claims as to the role of the L1 in the development of the null/overt pronoun distinction in these two types of context (OPC and focus), apparently the L1 does not have an effect, either for the acquisition of narrow syntactic properties of the OPC or of the semantic-pragmatic constraints of topic and focus. This also calls into question the supposed discourse parallels of English (and other non-pro-drop languages) and Spanish (and other pro-drop languages) for topic/focus contexts. In both studies, the authors claim that access to the constraints of OPC is available to L2 Spanish learners even if evidence in the input is limited. They also claim that learners are sensitive to and capable of mastering the semantic-pragmatic contrasts that guide null pronoun use in Spanish even though not explicitly taught such a distinction; the former acquired early and dependent on UG knowledge and the latter learned more gradually and dependent on discourse knowledge. Future research in the L2 acquisition of Spanish null and overt subjects would, indeed, corroborate Pérez-Leroux and Glass’ findings for pragmatic knowledge. In a similar study, Lozano (2002b) considered learners’ knowledge of both the OPC and focused contexts, or what he terms the contrastive focus constraint (CFC), in L2 Spanish among speakers of English and Greek (a language where the OPC is also operative and with similar feature specifications for the CFC). Like Pérez-Leroux and Glass (1999), Lozano hypothesized that it is not the L1 but rather the principles of UG that constrain pronoun use for the OPC and predicted that both groups of learners would perform similarly on the tasks that tested for this constraint. On the other hand, in the tasks that tested for the CFC, which operates with similar feature specifications of pro in Greek and Spanish but with less specified feature specifications in English, the author hypothesized that Greek learners would perform like the native Spanish speakers but English learners would show persistent difficulties for this parameter. In both Spanish and Greek the feature specification of pro has to be specified for either [+masc] or [−masc], but for English the feature specification of pro is [+/−masc] and thus, unlike the Greek learners, the English learners’ L1 will not aid them in selecting the full representation of the features of pro in Spanish. Participants included three groups: a native speaker control group and two experimental groups consisting of Greek speakers studying Spanish as an L3 in their home country and English speakers of Spanish as an L2, also in their home country. Participants were presented with contextualized sentences and two target sentences that they had to judge as more or less acceptable on a 5-point Likert rating scale where −2 indicated completely unacceptable and +2 indicated completely acceptable. The sentences were designed so that an overt pronoun would be most acceptable for CFC cases and a null for OPC cases. An example of ac-

68 | 3 Parametric studies in the acquisition of subject pronouns ceptable (overt pronoun) and unacceptable (null pronoun) responses to a CFC stimulus is presented in (58). (58)

El señor López y la señora García trabajan en la Universidad y en una famosa editorial. No obstante . . . (a) cada estudiante dice que él tiene poco dinero. −2 −2 0 +1 +2 (b) cada estudiante dice que pro tiene poco dinero. −2 −2 0 +1 +2 ‘Mr. López and Ms. García work at the university and at a famous publishers. However . . . ’ (a) ‘each student says that he has little money’ (b) ‘each student says that pro has little money’ (Lozano 2002b: 59)

The results showed that English and Greek learners of Spanish performed similarly to native speakers for the OPC cases, rejecting the overt pronoun for a bound variable interpretation and accepting the null pronoun in these contexts, a finding consistent with the OPC. However, in the CFC cases, the English speakers performed slightly differently from both the Greek learner and native speaker groups, a finding similar to what Pérez-Leroux and Glass (1997, 1999) found for contrastive focus cases. The English speakers judged the grammatical CFC cases (overt pronouns) favorably, like the Greek and Spanish speakers, but also judged the ungrammatical CFC cases (null pronouns) favorably, unlike the other participants. A one-way ANOVA with post-hoc Tukey HSD comparisons revealed that there were no significant differences between groups for the grammatical constructions, no significant difference between the Greek and Spanish speakers but a significant difference between the English and Spanish groups for these constructions. The author suggests that this may be due to the fact that in English, pronoun coreference is less restrictive, and indeed, adult English grammars allow sentences such as that seen in (59): (59)

The speakeri in the sample seems to make particular mistakes in this part of theiri utterances. (Lozano 2002b: 64)

where the speaker is specified for [+/−masc] since the referent’s sex is unknown. In addition, the pronoun, their, is also specified for [+/−masc], which is the least restrictive option. This is as Lozano had predicted, namely that “English natives’ low rejection of ungrammatical pro suggests that their grammatical knowledge allows

3.3 Recent studies from a parametric approach: The OPC and naturalistic input | 69

an option with the least restrictively specified set of features, namely [+/−masc]” (2002b: 64). In sum, Lozano claims that the results are evidence that both UG and the L1 are sources of knowledge in adult second language learning, a finding that has been more readily accepted in generative approaches to L2 acquisition by the early 2000’s. Also, at that time, it was highly questioned whether subject-verb inversion and that-trace effects were truly part of the NSP. One issue that had not been thoroughly investigated was the role of input in the triggering of aspects of the NSP.

3.3 Recent studies from a parametric approach: The OPC and naturalistic input Building on the findings of previous research, in a series of studies Rothman and Iverson (2007a, 2007b, and 2007c) set out to investigate whether or not learners exposed to more abundant and naturalistic input in the study abroad context were at an advantage for resetting the NSP, specifically the OPC. Developments in generative syntactic theory have assumed that parametric values are located within the lexicon of specific grammars and “(b)ecause language-specific lexicons vary in terms of which functional categories, features and strengths they instantiate, parametric differences arise cross-linguistically; that is, the acquisition of the functional lexicon has a number of syntactic consequences, which explains cross-linguistic surface variation” (2007b: 286). Rothman and Iverson sought to test the Full Access approach, which assumes L1 transfer where the L2 initial state is the L1 grammar, which adults use to make initial hypotheses about the L2 grammar. Nonetheless, they are capable of restructuring the transferred grammar (i.e. resetting the L2 features) but in order to do so they must have direct access to UG. Their major tenet in these studies was that more prolonged input is not necessary for triggering knowledge that is part of UG as long as learners at some point, either in the classroom or in a naturalistic setting, are exposed to the proper data for resetting parameters. They suggest that Isabelli’s (2003) claim that extended exposure to naturalistic input is a necessary ingredient for complete parameter resetting is contrary to the notion of parameterization. They also suggested, as have Jaeggli and Hyams (1988), that the NSP is much narrower in scope than what was previously believed and consists of only two or three properties: 1) the licensing of pro (i.e. null subjects), 2) OPC effects, and perhaps 3) obligatory null expletives (Rothman and Iverson 2007a: 330), and that knowledge of the OPC is “the ultimate criterion for determining whether the NSP has been reset” (Rothman and Iverson 2007b: 293). The authors claimed that because the OPC responds

70 | 3 Parametric studies in the acquisition of subject pronouns to poverty of stimulus effects associated with the NSP, if L2 learners showed knowledge of the OPC but not of subject-verb inversion or that-trace, then that would be evidence that these latter properties do not derive from the NSP. The data reported on in the three studies came from 30 intermediate L2 learners tested at the beginning and end of a five-month study abroad program in Madrid, Spain. Additionally, the (2007a) and (2007c) studies included eight monolingual native Spanish speakers from Madrid for control; the control group in the (2007b) study was different and consisted of 20 native Spanish speakers from Spain, Colombia, Mexico and Peru. Finally, for the (2007c) study, the authors also included a group of 24 classroom-only L2 intermediate learners. This group was included to challenge specifically Isabelli’s (2003) conclusion that abundant naturalistic input via study abroad is a necessary ingredient for parameter resetting. The participants from all three studies completed several tasks, including a grammaticality judgment/correction test, which was designed to test for the cluster of properties traditionally associated with the [+null] value of the NSP; a logical sentence-formation test created to assess participants’ capability for producing null subjects; and two tests that were designed to test specifically for the OPC: one taken from Kanno (1998), which was a coreference-matching test where participants were presented with sentences that included either quantifier determiner phrases or wh-matrix clause subjects with null and overt subjects in the subordinate clause and had to select either a bound or disjointed referential interpretation for the embedded subject pronoun, and the second one which was a context translation task based on a version used in the Pérez-Leroux and Glass (1999) study. The grammaticality judgment/correction task that tested for a cluster of properties of the NSP was only reported on in the (2007a) and (2007c) studies. The properties included in this task are those presented in the examples (21–24); the OPC was tested separately in these studies. Grammaticality judgment/correction task (60)

Null expletive subject pro Creo que pro hace frío hoy pro think-1sg that pro make-3sg cold today ‘[I] think [it]’s cold outside today.’

3.3 Recent studies from a parametric approach: The OPC and naturalistic input | 71

(61)

Null referential subject muy simpático. A menudo pro decimos que proi es Roberti es Robert be-3sg very likeable often pro say-3pl that pro be-3sg demasiado simpático. too likeable ‘Roberti is very friendly. Often we say that [he]i ’s too nice.’

(62)

Inverted subject Nadie sabe que lo dije yo. nobody say-3sg that it say-3sg-pret I ‘Nobody knows that I said it.’

(63)

wh-extraction of subject or that-trace ¿Quiéni dijiste que ti había salido temprano? who say-2sg-pret that t have-3sg-imp left early ‘Whoi did you say that ti left early? (Adapted from Rothman and Iverson 2007a: 332)

Participants were asked to judge the grammaticality of these sentences and if they considered them to be ungrammatical, were asked to correct them. For the logical sentence-formation task, reported on in Rothman and Iverson (2007a) and (2007b), participants were tested for knowledge of obligatory null expletive subjects and referential subjects with both contrastive focus, where an overt subject pronoun is contextually expected, and without contrastive focus, where a null subject is normally used. Participants were presented with a context followed by uninflected lexical items and were asked to form logical sentences. Examples included in this task are presented in (64–66). Logical sentence-formation task (64)

Expletives A lo lejos se ve una cola muy larga a la entrada del bar. ‘From far away a very long line is seen at the entrance of the bar.’ Parecer/ que/ mucha gente/ estar en el bar. ‘To seem/ that/ a lot of people/ to be in the bar.’ Expected: pro Parece que mucha gente está en el bar. pro seem-3sg that much people be-3sg in the bar ‘[It] seems that a lot of people are in the bar.’

72 | 3 Parametric studies in the acquisition of subject pronouns (65)

Referential (null) (without contrastive focus) María y Pedro salieron anoche por primera vez. ‘María and Pedro went out last night for the first time.’ Comer/ en un restaurante/ muy/ caro/ y/ ver/ una película ‘To eat/ in a restaurant/ very/ expensive/ and/ to see/ a movie’ Expected: pro Comieron en un restaurante muy caro y pro vieron pro eat-3pl-pret in a restaurant very expensive and pro see-3pl-pret una película. a movie ‘[They] ate in a very expensive restaurant and [they] saw a movie.’

(66)

Referential (overt) (contrastive focus) Laura, Mónica y Yolanda son profesoras de ingles en Puerto Rico en una escuela de niñas. Todos los padres hablan inglés bastante bien y creen que es importante saber hablarlo. Les agrada saber que esta escuela tiene las mejores profesoras de inglés. ‘Laura, Mónica and Yolanda are English teachers in Puerto Rico in a school of girls. All the parents speak English quite well and believe that it’s important to know how to speak it. It pleases them to know that this school has the best English teachers.’ Por lo tanto/ los padres/ estar/ seguros de/ que/hablar/ inglés/mejor ‘Therefore/ the parents/ to be/ sure of/ that/to speak/ English/best’ Expected: Por lo tanto, los padres están seguros de que ellas hablan inglés therefore the parents be-3pl sure of that they speak-3pl English mejor. better ‘Therefore, the parents are sure that THEY speak English the best.’ (Adapted from Rothman and Iverson 2007a: 332–333)

A third task, which was reported on in all three studies, was a coreferencematching test designed to test specifically for the OPC and was based on Kanno’s (1998) study of L2 Japanese. Examples appear in (67) and (68).

3.3 Recent studies from a parametric approach: The OPC and naturalistic input | 73

(67)

Overt embedded pronoun with wh-matrix clause subjects (OPC forces (b) as the only answer) ¿Quién ha dicho que él nunca se enfada? who have-3sg said that he never refl-get mad-3sg ‘Who do you suppose never gets angry?’ (a) the same person as Quién (b) someone else (Adapted from Rothman and Iverson 2007c: 199)

(68)

Null embedded pronoun with wh-matrix subject (No OPC restriction; (a) and (b) are possible) ¿Quién no sabe que pro tiene derecho a tomar cervezas a who no know-3sg that pro have-3sg right to drink-inf beer at los 21 años? the 21 years ‘Who do you suppose does not know that pro has the right to drink beer at 21? (a) the same as Quién (b) someone else (Adapted from Rothman and Iverson 2007b: 297)

After reading the stimuli sentence, participants were asked to select a boundvariable interpretation (answer a), a disjoint referential interpretation (answer b) or both possibilities (selecting both a. and b.). Because sentences of this type are not often heard in the input nor taught in the L2 classroom, the authors believed that if the participants showed knowledge of the OPC without sufficient command of subject-verb inversion and/or that-trace effects, then this would be strong evidence that these properties are not part of a cluster of the NSP. Finally, a context translation task, a version based on that used in Pérez-Leroux and Glass (1999) and also designed to test for OPC knowledge, was reported on only in the Rothman and Iverson (2007b) study. Participants were provided with stories in English followed by questions and appropriate answers, also in English, and were asked to translate the answers into Spanish. The questions were designed to test for OPC effects. Results showed that for the grammaticality judgment task, learners performed native-like with null referential subjects at the onset of study abroad and thus, were unable to demonstrate improvement. Unlike the native speaker groups, learners had difficulty in judging ungrammatical overt expletive sentences, a finding in accord with the results of the beginning learners of Lozano’s (2002a) study; they accepted both null and overt subject pronouns at both testing intervals and showed L2 target-deviancy for long-distance wh-subject extraction and

74 | 3 Parametric studies in the acquisition of subject pronouns subject-verb inversion. Also, the authors found no differences between the study abroad and classroom groups in the judgment of null referential and expletive subjects and subjects in SVO word order in their (2007c) study. For the logical sentence-formation task, the learners that performed at ceiling levels at the onset of the study abroad program created both referential and expletive null-subject sentences but, compared with the native speakers, used fewer overt subjects for CF contexts, although there was a slight improvement from interval one to two in the (2007a) and (2007b) studies. The authors claim that this shows that naturalistic input helps learners acquire the discourse-pragmatic constraints, but not necessarily the parameter resetting. Twenty of the study abroad L2 learners who showed knowledge of the OPC at the onset of the study abroad experience in the (2007a) study, also demonstrated native-like interpretations of bound variable co-reference contexts when the main clause subject was a quantifier determiner phrase or a wh-element and the embedded subject was a null or overt pronoun. Those who did not demonstrate knowledge of the OPC at the onset showed no improvement over time in the (2007b) study, evidence that naturalistic input is not helpful in the acquisition of this property. Additionally, seventeen of the classroom learners made OPC distinctions much like the native speakers and seven did not. For the study abroad group that already had knowledge of the NSP prior to the onset of their foreign stay, an increase in naturalistic input did not help them acquire other properties. Since both the classroom and study abroad learner groups performed similarly, the authors claimed that this is evidence that increased naturalistic input is not beneficial for NSP resetting, although it may be for acquiring the discoursepragmatic constraints of the parameter. The authors were confident in claiming that relatively early on in the development of L2 Spanish, the OPC is learned notwithstanding its poverty of stimulus status (Rothman and Iverson 2007c: 207) and that the type of learning environment (abroad or classroom) did not significantly affect acquisition. In addition, the fact that the learners in their studies had native-like knowledge of the OPC but incomplete command of thattrace effects and subject-verb inversion in Spanish provided evidence that these properties mostly likely are not part of the NSP.

3.4 Conflicting evidence and common findings The parametric studies offer conflicting evidence of learners’ resetting of a cluster of hypothetically related properties associated with the NSP and of learners’ acquisition of the distinction between obligatory null expletive and “optional” referential sentences, of topic/focus contexts, and of the restrictions of the OPC.

3.4 Conflicting evidence and common findings

|

75

Furthermore, the roles of both the L1 and input, although not the topic of all of the studies discussed here, has been controversial. None of the studies reviewed here could provide evidence that Rizzi’s (1982, 1986) four hypothesized properties of the NSP were learned as a cluster, although Liceras (1989) and Isabelli (2003) both found that learners could reset the parameter for null subjects and subject-verb inversion but not for that-trace effects. Rothman and Iverson (2007a, b, c), on the other hand, found evidence that both subject-verb inversion and that-trace effects are problematic even at later stages although their participants also were nativelike in regards to null referential subjects. Each of the authors who examined this property has offered conflicting explanations for learners’ persistent problems acquiring that-trace effects. Liceras suggests that it is because there is a hierarchy in the learning of the clustered properties and that-trace is the most difficult and perhaps last-learned property. Isabelli proposes that it is because that-trace effects are infrequent in the input and thus, even study abroad students with access to abundant input are not at an advantage to acquire this property. Rothman and Iverson, on the other hand, reject this possibility. If the property is part of the NSP, then some access to the data should trigger this setting of the parameter, as in the case of OPC. Liceras and Díaz, for their part, reject the notion that parameter resetting is responsible for the acquisition of the NSP, but rather suggest that it is due to a default licensing procedure resulting from different grammatical representations. It appears that all of these studies indicate that at least that-trace is not part of the NSP and also most possibly subject-verb inversion. In regards to the acquisition of null expletive and referential pronouns, it was seen that both Lozano (2002a) and Al-Kasey and Pérez-Leroux (1998) had similar results revealing that learners treat these two types of pronouns differently. However, on the one hand, whereas Al-Kasey and Pérez-Leroux uphold that these two types of subject pronouns cluster together in the NSP, Lozano claims to have provided evidence that learners have different mental representations for null expletive and referential pronouns. The data seem to support the latter claim. Although most of the research discussed here has concluded that the OPC is acquirable even in the face of underdetermined input, the data examined offered variable results among the learners studied regarding knowledge and access to the OPC. Lozano (2002b) discovered that Greek learners of Spanish L2 performed better than English speakers on determining OPC effects, suggesting L1 influence on the acquisition process, whereas Liceras (1989) and Pérez-Leroux and Glass (1997, 1999) found no support for L1 effect. Rothman and Iverson (2007a, 2007b, 2007c) noted differences among learners’ knowledge of the OPC but also attested that additional input via study abroad was not beneficial in helping to reset the constraint. This finding at least, is in accord with Pérez-Leroux and Glass who also concluded that access to the OPC is available to learners even in the face of limited

76 | 3 Parametric studies in the acquisition of subject pronouns input. Thus, the role of the L1 and input for acquiring the syntactic properties of subjects is still not clear. That some of the properties of the pro-drop parameter are learned early and are guided by universal principles is evident and yet problems among learners persist even at advanced and near-native levels, including the overproduction of both null and overt subjects and knowledge of that-trace effects. Several authors have hinted that the deficits remaining in subject pronoun use is due to a lack of knowledge of the pragmatic functions of null and overt pronouns and the discourse constraints responsible for expressing and retrieving meaning and that this type of knowledge is learned late (Pérez-Leroux and Glass 1997, 1999) and after exposure to sufficient input (Rothman and Iverson 2007a, b). Nonetheless, the examination of parametric studies is important because although they questioned the extent to which apparently related principles are part of the NSP and offer conflicting evidence of learners’ early acquisition of the distinction between obligatory null expletive and “optional” referential sentences and the restrictions of the OPC, they also revealed that learners are capable of acquiring the formal syntactic properties that license null subjects in L2 Spanish, even at early stages, specifically the OPC. They established the role of UG in L2 acquisition and brought to light a new understanding of the influence of learners’ L1. Finally, these studies have highlighted important syntactic properties of Spanish and set the stage for more recent studies that looked to the importance of syntactic and pragmatic interfaces in attempting to explain the acquisition of the overt/null distinction of Spanish subject pronouns.

4 The syntax-pragmatics interface in the acquisition of subject pronouns Introduction Generative studies from the parametric view have confirmed that learners of L2 Spanish are capable of resetting the null subject parameter as evidenced by their acceptance and production of null and overt referential subjects. They also provided evidence that related properties of the parameter, such as bound and variable interpretations according to the OPC and obligatory null expletives, are acquirable. However, the parametric studies were not successful in explaining the source of lingering interlanguage errors, especially the overuse of null subjects in contexts of topic and focus and the overuse of overt subjects for topic continuity and non-focus. Although not the central issue under investigation, several of the parametric studies noted that the semantic and pragmatic properties of topic and focus were learned late by adult L2 learners of Spanish (Pérez-Leroux and Glass 1997, 1999) and that they were learnable through exposure to naturalistic L2 input (Rothman and Iverson 2007a, b). More recent research from a generative perspective has approached the problem of L2 learners’ and bilingual speakers’ lingering deficits in subject pronoun distribution by examining the interface between morphosyntactic and discourse-pragmatic properties of the NSP (Hertel 2003; Montrul 2004b; Montrul and Rodríguez Louro 2006; Margaza and Bel 2006; Rothman 2007, 2009; Lozano 2006, 2009). These studies recognize that there is observable variability in the development and end result of L2 acquisition, unlike the uniform attainment of L1 acquisition, and that this is best explained by the complex (and therefore, late-learned) pragmatic-discourse properties of language distribution. While acknowledging that child L1, adult L2 and simultaneous bilingual acquisition are all different processes, the studies on adult L2 acquisition of Spanish subject pronouns are informed by but also complement the ever-growing body of work in child L2 and bilingual acquisition in English and several Romance languages (Bel 2003 for Spanish and Catalán; Serratrice, Sorace, and Paoli 2004 for Italian; Sorace, Serratrice, Filiaci, and Baldo 2009 for Italian and Spanish; and Pladevall Ballester 2010 for Spanish). Some have claimed that transfer from the pragmatically less complex non-pro-drop language to the pragmatically more complex pro-drop one is unidirectional (Sorace¹⁴ 2004; Serratrice, Sorace, and

14 The work of Sorace and her colleagues is discussed more fully in Chapter 5 where processing strategies are discussed in relation to the acquisition of subjects. The main thrust of her

78 | 4 The syntax-pragmatics interface in the acquisition of subject pronouns Paoli 2004; Sorace, Serratrice, Filiaci, and Baldo 2009) and that this accounts for the overuse of overt subjects in bilingual speakers. This prediction is only partially borne out, however, in bilingual children and L2 adult learners who appear to overuse both overt and null subjects. It is becoming increasingly clear, however, that the distribution of null and overt subjects at the syntax-pragmatic interface is particularly vulnerable in acquisition among children and adults and it appears that perhaps the formal morphosyntactic properties of the null subject parameter may be acquired before the discourse-pragmatic constraints. Examining why this is so is one of the central thrusts of this chapter. Of particular interest in the L2 and bilingual studies are the still unanswered questions of which particular syntactic and lexical properties interact with discourse-pragmatic features to constrain subject distribution in Spanish and whether discourse-pragmatic knowledge develops concurrently with the development of lexico-syntactic knowledge or subsequently. Also of interest is the role of L1 and/or cross-linguistic interference in the production and interpretation of null and overt subjects and the impact of the quality and quantity of L2 input in acquisition. Finally, there are those who have suggested that the discourse-pragmatic constraints are perhaps never acquired and are therefore, responsible for adult L2 learners’ fossilization (Valenzuela 2006). Although in Chapters 2 and 3 the syntactic features that have been associated with the NSP were discussed at length, in this chapter we reconsider some of the syntactic constraints that have been examined at the interface with lexical, semantic and discourse-pragmatic properties of the NSP in both L1 and L2 acquisition studies, specifically lexical verb class and word order, the OPC, and person and number of subjects. The chapter is organized as follows: the next section includes a discussion of how cross-linguistic interference in bilingual children and adults affects the distribution of null and overt subjects at the semantic and pragmatic interfaces and whether or not this interference is unidirectional; the consequences of the interaction of lexical category (of the verb), discourse status (of the subject) and word order on child L1/L2 and adult L2 language acquisition is the topic of the following section; the subsequent section revisits the overt pronoun constraint (OPC) (Montalbetti 1986) and examines how the OPC interfaces with discourse constraints in adult L2 acquisition of subject pronouns; the follow-

work is that deficits in the production and interpretation of overt and null subjects in bilinguals who speak/acquire both pro-drop and non-pro-drop languages are due to processing difficulties. However, to the extent that her work discusses the properties of the syntax-pragmatic-discourse interface I include discussion of her work in the present chapter (see also footnote 16 in this chapter).

4.1 Cross-linguistic influence in bilinguals and the Unidirectional Hypothesis |

79

ing section deals with how pronominal features of person and number of subjects can account for the lingering deficits L2 learners of Spanish exhibit at the syntaxdiscourse interface; and in the final section, I discuss the contributions of research to date and their diverse and common findings.

4.1 Cross-linguistic influence in bilinguals and the Unidirectional Hypothesis We can look to research in child bilingual development for some guidance as to the difficulties in acquiring properties of null and overt subjects in Romance languages at the syntax-pragmatics interface and to the role of L1 transfer. Research in child L1 acquisition has found that children’s ungrammatical behavior can be traced to an overdependence on pragmatics (Serratrice, Sorace, and Paoli 2004: 184), which leads to optionality of alternating syntactic forms. More specifically, work by Serratrice and Sorace (2003) found that monolingual Italian (a pro-drop language much like Spanish) children and bilingual Italian-English children used null and overt subjects according to the pragmatic constraints of the informative status of subject referents (see Chapter 2 for a complete discussion of the pragmatic notions of topic and focus). Further research has examined the possibility that the syntax-discourse interface is particularly vulnerable to cross-linguistic influence, especially in the direction of non-pro-drop languages to pro-drop languages (Paradis and Navarro 2003, Serratrice, Sorace, and Paoli 2004; Montrul 2004b, Sorace et al. 2009). Whereas in Romance languages the distribution of null/overt and preverbal/postverbal subjects interfaces with lexico-syntactic and discourse-pragmatic properties, in English it does not. In Romance, overt pronominal subjects are [+topic shift] or [+focus] and in English they are underspecified for these features. The pragmatics of information status plays a role in non-prodrop languages such as English, but the distinction is between lexical subjects, which mark high informative status, and pronominal subjects, which mark referents of low informative status. Prosodic stress as well is used to indicate topic shift and/or focus in non-pro-drop languages such as English. Faced with these two competing systems in bilingual acquisition, Serratrice, Sorace, and Paoli (2004) have proposed the unidirectional hypothesis, which predicts that simultaneous and systematic contact with a non-pro-drop language and a pro-drop language “might lead to the bleaching of the interface features that constrain subject realization in the latter” and thus “when faced with a choice that requires the coordination of sophisticated pragmatic knowledge with two syntactic alternatives, the bilingual child might optionally select the pragmatically unconstrained option available” of the non-pro-drop language (Serratrice,

80 | 4 The syntax-pragmatics interface in the acquisition of subject pronouns Sorace, and Paoli 2004: 188). The effect of this would be that overt subject pronouns in the Romance language would lose their association with the features of [+topic shift] and [+focus] and be interpreted as coreferential with non-focus and/or topic continuity; that is, the child would overuse overt pronouns in the prodrop language where a null subject would be expected. Although not syntactically incorrect, this overuse would be considered pragmatically inappropriate. Serratrice, Sorace, and Paoli (2004) tested the unidirectional hypothesis in a case study of a bilingual English-Italian child and in ten monolingual Italian and English children (for control). The data were collected from the bilingual child in a series of spontaneous child-adult conversations from the ages of 1 year, ten months to four years, six months; the monolingual data came from different corpora of the CHILDES database (MacWhinney 2000) with children of similar age ranges. The data were coded for overtness (null vs. overt), morphosyntactic form (full noun phrase, proper name, personal pronoun, demonstrative pronoun, etc.), and informativeness features (person, absence, activation, contrast, differentiation in discourse, query and predicate transitivity). Referents were considered informative if they were third person subjects, physically absent (from the conversational context), had not previously been introduced into the discourse, were being contrasted with another referent, were identified among one or more potential antecedents, were the subject of or the response to a question, or the subject of an intransitive predicate. If referents possessed the opposite of these features, they were considered uninformative. Results indicated that the monolingual Italian children, at all stages, produced more null subjects than the monolingual English children, an indication that the syntactic property licensing null subjects is acquired early on in L1 Italian acquisition and that English-speaking children as well take note of the language-specific occurrence of subjects in their language. In addition, the authors observed that at all stages of development the distribution of null and overt subjects in Italian depended on the informativeness features of referents. In other words, null subjects tended to be associated with uninformative features and overt subjects were more often associated with informative features. Results for the bilingual child revealed that he used a larger number of overt subjects in English than the English monolingual children but also more overt subjects in Italian than the Italianspeaking children. This shows influence (unidirectional) from the non-pro-drop language to the pro-drop language but not the other way around, in accord with the unidirectional hypothesis. However, it was observed that his null subjects also tended to be significantly more often associated with uninformativeness, which provides evidence that even at a young age, this bilingual speaker was sensitive to the discourse-pragmatic status of null referents in Italian. There were also a number of inappropriate overt subjects used (in 9 % of the cases) in contexts that

4.1 Cross-linguistic influence in bilinguals and the Unidirectional Hypothesis |

81

were uninformative - evidence of cross-linguistic influence of the underspecified features in English for [+focus] and [+topic shift]. According to the authors, this is to be expected since cross-linguistic influence is unidirectional from the language with fewer pragmatic constraints in the distribution of null and overt subjects (English) to the language where the overt subjects is constrained by the complex pragmatic features of topic shift and focus. If the integration of syntactic and pragmatic knowledge is a challenging task for any young child, then it is even more so for the bilingual child who must navigate two competing systems for the resolution of the overt/null subject distinction. In order to corroborate (or reject) the unidirectionality of cross-linguistic effects in older bilingual children, Sorace, Serratrice, Filiaci, and Baldo (2009) conducted a large-scale study that included English-Italian (from the UK and Italy) and Spanish-Italian (from Spain) bilingual children (ages 6–7 and 8–10) and, for control, monolingual children (from the UK and Italy) of the same age groups and monolingual adults (from the UK and Italy). The authors had predicted that if cross-linguistic differences were the source of bilingual errors in null and overt subject interpretation, then the performance of the EnglishItalian bilinguals would be significantly different from that of the Spanish-Italian (both pro-drop languages) bilinguals and the Italian monolinguals and that this would be evidence in favor of the unidirectional hypothesis. Data were collected via an acceptability judgment task based on a short animated story involving Mickey Mouse, Minnie Mouse, Donald Duck and Daisy. English-Italian bilinguals completed the task in English and Italian and the Spanish-Italian bilinguals completed the task only in Italian. The task tested for knowledge of the use of null and overt subject pronouns in non-topic-shift and topic-shift contexts. Below are examples from the Italian task. (69)

No topic shift (Minnie and Daisy in the foreground; Mickey and Donald in the background) Minnie: pro Sono caduta! pro be-1sg fallen ‘[I]’ve fallen!’ Donald: Minnie ha detto¹⁵ che pro è caduta3 . Minnie have-3sg said that pro be-3sg fallen ‘Minnie has said that [she] has fallen.’

15 Both verb forms, ha detto (‘has said’) and è caduta (‘has fallen’) are passato prossimo (‘proximate past’ or present perfect) but the verb dire (‘to say’) forms the passato prossimo with the

82 | 4 The syntax-pragmatics interface in the acquisition of subject pronouns Micky:

(70)

Minnie ha detto che lei è caduta. Minnie have-3sg said that she be-3sg fallen ‘Minnie has said that she has fallen.’

Topic shift (Minnie and Daisy in the foreground; Mickey and Donald in the background) Minnie: Daisy è caduta! Daisy be-3sg fallen ‘Daisy has fallen!’ Donald: Minnie ha detto che pro è caduta. Minnie have-3sg said that pro be-3sg fallen Micky:

‘Minnie has said that [she] has fallen.’ Minnie ha detto che lei è caduta. Minnie have-3sg said that she be-3sg fallen ‘Minnie has said that she has fallen.’ (Taken from Sorace et al. 2009: 467)

Participants were told that the cartoon characters were learning Italian (or English) and had to decide which of the characters spoke ‘better.’ In (69), Donald’s response is the most pragmatically appropriate in the context of non-topic-shift where the null subject refers to the subject antecedent of the main clause (Minnie), and in (70), the preferred response is Micky’s where there is a shift in topic and the overt subject (lei ‘she’) of the subordinate clause does not refer to the subject antecedent of the main clause (Minnie), but rather is understood to refer to Daisy in this context. The version in English included both null and overt options as well but because the null options are not grammatical in English (‘Micky said that pro sang.’), the task tested only for syntactic knowledge (acquired more easily or earlier), whereas the Italian task required the integration of syntactic and discourse-pragmatic knowledge for topic-shift and non-topic-shift contexts. Thus, the authors expected that the Italian-English bilinguals would perform better on the English task than the Italian one and reject ungrammatical null subjects. If they were to accept null subjects in English (not expected), this would then be evidence against the unidirectional hypothesis.

auxiliary verb, avere (‘to have’) and the verb cadere (‘to fall down’) forms the passato prossimo with the auxiliary verb essere (‘to be’).

4.1 Cross-linguistic influence in bilinguals and the Unidirectional Hypothesis |

83

The results of pronominal preferences in English revealed that the adults, the monolingual children and the older (8–10 year olds) bilingual children all scored 100 % grammatically correct on the task (accepting the sentences with overt subject pronouns) in both topic-shift and non-topic-shift contexts and rejecting those with null subjects. The younger bilingual children (6–7 year olds) scored only marginally less accurately than the other groups and this difference was only significant in the topic-shift contexts. This shows that the syntactic constraint for English, overt subjects, is easily learned and that there is very little, in any, cross-linguistic influence from Italian. For pronominal preferences in Italian, the results were very different; not even the adult speakers chose the pragmatically appropriate option 100 % of the time, although they scored significantly better than the child monolingual and bilingual groups. Younger children selected overt pronouns in non-topic-shift contexts significantly more than older children who, in turn, chose overt pronouns significantly more than the adults. This finding reveals that the pragmatic constraints governing null and overt options in Italian are more difficult to learn and develop over time. In addition, the younger children living in the UK were more likely than those living in Italy or Spain to select an overt pronoun in non-topic-shift contexts, suggesting a greater cross-linguistic influence from English. On the other hand, the Spanish-Italian bilinguals performed much like the English-Italian bilinguals from both countries, selecting an overt pronoun in non-topic-shift contexts more often than the monolingual children. For the topic-shift contexts, adults and monolingual children selected overt pronouns more often than both the English-Italian and the Spanish-Italian bilinguals, suggesting that more than cross-linguistic influences are at work in the interpretation of null and overt pronouns. Both the monolingual and bilingual children in this study were more likely than the adults to accept overt pronouns in discourse contexts where they are redundant, but the bilingual children take even longer to sort this out.¹⁶ In addition, the bilingual children’s problems were not only with redundancy but also with ambiguity in that they also accepted more null subjects in contexts that required an overt pronoun in Italian. The data in this study show that with age it becomes easier to reject ambiguous null subjects in topic-shift contexts than it is to reject redundant overt pronouns in non-topic-shift contexts. However, because

16 The authors moreover suggest that the differences between the monolinguals and bilinguals may be because bilinguals, when confronted with two distinct systems for linking pronouns to antecedents, are less efficient in integrating the multiple sources of information (syntactic, pragmatic, semantic) necessary for such antecedent assignment. This claim will be discussed in Chapter 5, which examines more fully the implications for processing difficulties.

84 | 4 The syntax-pragmatics interface in the acquisition of subject pronouns there were no significant differences between the English-Italian and the SpanishItalian bilingual children in their acceptance of redundant overt pronouns, the authors were cautious in attributing the cross-linguistic influences only to English. Finally, there was no evidence of any effect of Italian on the English-Italian children’s preferences in English because even the youngest children were performing at ceiling or almost ceiling levels in rejecting null subject pronouns. The syntactic task of rejecting ungrammatical null subjects in English is by far less difficult for the bilingual children than sifting through the pragmatically less optimal options available in Italian and, according to the authors, provides further evidence for the unidirectionality of cross-linguistic influences from the non-prodrop language to the pro-drop one, but still leaves unexplained the overuse of ambiguous null pronouns in Italian. This would suggest a certain amount of directionality (although more limited) from the pro-drop language to the non-pro-drop language as well. The implication of the child bilingual studies is that with age the integration of syntactic and discourse-pragmatic knowledge that is necessary for the optimal interpretation and use of null and overt subjects stabilizes as the two languages become increasingly independent. Research on adult bilinguals, however, does not appear to support this suggestion. Sociolinguistic studies on overt subject distribution have revealed that Spanish-English bilinguals appear to ignore the pragmatic constraints for overt subject distribution and employ redundant subject pronouns in non-topic-shift contexts (Silva-Corvalán 1994; Lipski 1996) in their non-dominant language Spanish, suggesting influence from the dominant language, English. The effect of cross-linguistic influence from English to Spanish was further investigated by Montrul (2004b), who examined pronominal preferences¹⁷ in two groups of adult bilingual (heritage) speakers and a group of monolingual Spanish speakers. Montrul considers the possibility that attrition (language loss) and/or incomplete acquisition in these types of populations are particularly vulnerable at the syntactic-pragmatic interfaces, the same areas that are susceptible in L1 and L2 acquisition. She hypothesized that attrition or incomplete acquisition would influence the semantic-pragmatic interface of pronoun distribution more than the narrow syntactic domain and that Spanish heritage speakers would exhibit robust knowledge of null subjects, despite the fact that their dominant language (English) does not share these syntactic properties, but would also display variable performance with the discourse-related distribution of null and overt subjects in Spanish (Montrul 2004: 131). Twenty-four bilingual

17 Although the Montrul (2004) study examined subject and object pronouns, of interest for the present volume are the results of the subject pronoun tasks.

4.1 Cross-linguistic influence in bilinguals and the Unidirectional Hypothesis |

85

college-aged adults raised in the U.S. but exposed to Spanish from an early age and twenty monolingual Spanish speakers from different Spanish-speaking countries (excluding speakers of Caribbean dialects) participated in the study; the bilinguals were further divided into two groups of intermediate and advanced heritage speakers based on their scores on an independent proficiency measure taken from the DELE (Diploma de Español como lengua extranjera ‘Diploma of Spanish as a Foreign Language’). The elicitation instrument consisted of an oral production task where participants were asked to narrate in Spanish the story of Little Red Riding Hood based on a series of pictures presented to them. Null and overt subjects were coded for pragmatically appropriate and inappropriate contexts. Null subjects were considered appropriate in contexts of same referent (topiccontinuity) and inappropriate in contexts of change of referent (topic-shift) or for emphasis (contrastive focus). Conversely, overt subjects were inappropriate in same referent contexts and appropriate in change of referent or emphatic contexts. The results indicate that all groups produced a substantial number of both overt and null subjects, which signifies that the syntactic option for null subjects in Spanish is in place for these heritage speakers. The monolinguals and the advanced heritage speakers performed much in the same way and overall produced a higher percentage of null subjects than overt subjects, whereas the intermediate heritage speakers produced a greater number of overt subjects than null subjects. Furthermore, both the monolingual and advanced groups used null and overt pronouns in a more pragmatically appropriate manner than the intermediate group. An examination of the results of just the heritage speakers reveals that the intermediate group is less accurate than the advanced group for inappropriate uses of overt and null subjects. Both groups, however, are less accurate pragmatically in the use of null subjects than they are in the use of overt subjects, a finding that partially supports unidirectionality, which predicts the overuse of pragmatically redundant overt subjects but not the overuse of ambiguous null subjects. The question of why these speakers overproduced more null subjects than overt merits further scrutiny. In the examples Montrul provides (all taken from the intermediate heritage speakers’ production), there are cases where it is indeed true that a null pronoun is used in a switch referent context, however, that referent is not ambiguous as the following excerpt from one of the intermediate subjects illustrates:

86 | 4 The syntax-pragmatics interface in the acquisition of subject pronouns (71)

Caperucita Roja salió a ir a la casa de su Little Red Riding Hood leave-3sg-pret to go-inf to the house of her abuelita con una canasta de comida porque pro estaba, pro grandma with a basket of food because pro be-3sg-imp pro iba a visitarla porque *pro estaba enferma. go-3sg-imp to visit-inf-her because *pro be-3sg-imp sick ‘Little Red Riding Hood went out to go to her grandmother’s house with a basket of food because [she] was, [she] was going to visit her because [she] was sick.’ (Montrul 2004b: 133)

First of all, as Sorace (2004) points out, it can be assumed that the investigator and participants are familiar with the story of Little Red Riding Hood and therefore know that the main protagonist goes to visit her grandmother because she (the grandmother) is sick. Knowledge shared by interlocutors can help to identify the referent of a syntactically minimal pronoun (Blackwell 2003), in this case a null pronoun. Furthermore, the grandmother is brought into focus as a prepositional object in the first line (la casa de su abuelita lit. ‘the house of her grandmother’) and then again as a pronominal direct object la (‘her’) of the verb visitar (‘to visit’) in the second line. These two mentions of the grandmother are sufficient to bring the referent into focus and thus, a null pronoun is licensed in the discourse (Gundel, Hedberg, and Zacharski 1993). This specific type of cognitive information in the discourse structure is not explored in the research discussed so far but will be taken up in detail in Chapter 6 of the present volume. Thus, the claim of Montrul that there are more instances of overproduction of null subjects than overt subjects must be considered with precaution and is perhaps not sufficient evidence against unidirectionality in these cases. Finally, Montrul (2004b) examined the distribution of preverbal and postverbal subjects and found little significant differences among the three groups of speakers. Preverbal subjects were more commonly produced than postverbal subjects, which appeared mostly as subjects of unaccusative verbs (see Section 4.1.2.1 for further discussion of verb class and word order). Although the results could not positively identify whether it was language loss or incomplete acquisition that affected the deficits in the use of null and overt subjects in the production data of these heritage speakers, it is evident that the syntactic domain is intact and deficits can be attributed to the vulnerabilities at the semantic-pragmatic interface of pronoun distribution. Montrul suggests, as do Serratrice, Sorace, and Paoli (2004) for very young bilingual children and Sorace et al. (2009) for older bilingual children, that adult heritage speakers who

4.2 The Unaccusative Hypothesis and word order | 87

are faced with two competing systems may tend to rely on the least pragmatically constrained option offered by the two languages. This erosion of the complex semantic and pragmatic features in the grammar of Spanish would lead to a convergence towards the less pragmatically complex morphosyntactic system of English (Montrul 2004b: 138) for overt and null subject distribution. In the studies discussed here, summarized in Table 11, there certainly appears to be an effect of unidirectionality, although there is also evidence of the overuse of null subjects as well and this overuse is seen in children whose L1s are both pro-drop and non-pro-drop and in adult heritage speakers of pro-drop whose dominant language is non-pro-drop. This implies that more is at play than cross-linguistic influence, at least in the case of child development. What each of these studies has in common, nonetheless, is evidence that the overuse of overt subjects in pragmatically anomalous contexts is more difficult to overcome than the overuse of null subjects. The former expresses a redundancy in the discourse and the latter leads to ambiguity. The evidence suggests, however, that the need to resolve ambiguity in discourse is a more pressing task than that of eliminating redundancy, but also that opting between null and overt subjects in same or switch referent contexts is a cognitively more taxing task. The earlier elimination of pragmatically incorrect null subjects in the pro-drop language does appear to suggest that these bilingual speakers have converged towards a pragmatically less complex system such as that found in the non-pro-drop language and speaks to the difficulties in acquiring and maintaining the null/overt contrast at the syntaxpragmatics interface. That the syntactic properties of subjects in pro-drop languages are acquired early in child and adult L2 development is evident and yet deficits linger at the discourse-pragmatics interface. The situation is more complex than that as suggested by the unidirectional hypothesis. Lexical-semantic properties of verbs interact with the interpretable [+focus] Feature and this interface also has syntactic consequences for subject use in Spanish. This adds another layer of complexity to the acquisition process. The following section discusses research that has considered how the lexical, pragmatic and syntactic properties of Spanish subjects affect their acquisition and use among learners whose L1 is either non-pro-drop, such as English, or pro-drop, such as Italian or Greek.

4.2 The Unaccusative Hypothesis and word order Word order for declarative sentences in Romance languages such as Spanish, as well as other languages (e.g. Greek), is dependent upon principles operating at the lexico-syntactic interface and the syntactic-pragmatic interface. In other words,

88 | 4 The syntax-pragmatics interface in the acquisition of subject pronouns Table 11. Summary of cross-linguistic influence from bilinguals’ non-pro-drop to pro-drop languages. Study

Overuse of overt subjects

Overuse of null subjects

Sensitivity to discoursepragmatics in pro-drop language

Serratrice et al. (2004)

Yes

Yes

Yes and No Some errors

Child English–Italian 1;10–4;6 years Sorace et al. (2009) Child English–Italian Spanish–Italian

Yes

Yes

Younger children: No Older children: Yes and No

More overuse with UK bilinguals but improvement with each age group

Easier to reject ambiguous null than redundant overt subjects

6–7 and 8–10 years Montrul (2004b)

Yes

Yes

Adult English– Spanish

More overuse with intermediate speakers

More overuse with intermediate speakers

2 levels of proficiency

Intermediate speakers: Yes and No Advanced speakers: Yes Both groups are pragmatically more accurate in the use of overt subjects

both the lexical subcategory of the verb (i.e. transitive or intransitive) and the discursive information status (i.e. focus or non-focus) of the subject determine whether a subject is preverbal or postverbal in Spanish. For example, simple transitive declarative sentences allow the following word orders: (72)

a. El esposo golpea a Charlie Chaplin. (SVO) the husband hit-3sg to Charlie Chaplin ‘The husband hits Charlie Chaplin.’ b. Golpea a Charlie Chaplin el esposo. (VOS) hit-3sg to Charlie Chaplin the husband ‘Hits Charlie Chaplin the husband.’ c. Golpea el esposo a Charlie Chaplin. (VSO) hit-3sg the husband to Charlie Chaplin ‘Hits the husband Charlie Chaplin.’ (Lubbers Quesada and Blake 2003)

4.2 The Unaccusative Hypothesis and word order | 89

According to Zubizarreta (1998), in Romance languages new information introduced into the discourse (i.e. focus) is normally the last constituent of an utterance (and in many cases is also accompanied by prosodic stress) (Zubizarreta 1998: 78), unless all of the information in an utterance is new, in which case, the sentence usually follows the canonical SVO word order. In (72a), all of the information in the sentence is new and brought into focus and could be uttered in response to a question such as ¿Qué pasó? (‘What happened?’). On the other hand, in (72b), the new information introduced into the discourse is el esposo (‘the husband’) and might be uttered in order to clarify that it was the husband (and not the wife, for example) that hit Charlie Chaplin. If the speaker wants to clarify that it is Charlie Chaplin that the husband hit and not someone else then (72c), which places the object of the verb, Charlie Chaplin, at the end of the utterance, would be an appropriate response. Whereas with transitive verbs, the unfocused (or neutral) word order of declarative sentences is SVO, the word order of intransitive verbs is constrained by its lexical class as set forth in the Unaccusative Hypothesis (UH) (Perlmutter 1978) or the Split-Intransitivity Hypothesis (Burzio 1986; Levin and RappaportHovav 1995). The UH distinguishes two types of intransitive verbs: unaccusatives, of which the subject is a theme (i.e. non-agentive), and unergatives, of which the subject is an agent. Examples of unaccusative verbs in Spanish include llegar (‘to arrive’) or venir (‘to come’) and unergatives include such verbs as gritar (‘to shout’) or saltar (‘to jump’). In neutral contexts in response to questions such as What happened?, or ‘out of the blue’ statements, native speakers of Romance languages like Spanish tend to prepose subjects with unergative verbs (SV order), as exemplified in (73a), and postpose subjects with unaccusative verbs (VS order), as seen in (73b). (73)

a. Unergative verb: preferred word order in neutral contexts SV (subject is agent) La chica gritó. the girl shout-3sg-pret ‘The girl shouted.’ #Gritó la chica. shout-3sg-pret the girl ’Shouted the girl.’

90 | 4 The syntax-pragmatics interface in the acquisition of subject pronouns b. Unaccusative verb: preferred word order in neutral contexts VS (subject is not an agent) #Dos señores llegaron. two men arrive-3sg-pret ’Two gentlemen arrived.’ Llegaron dos señores. arrive-3sg-pret two men ‘Arrived two gentlemen.’ (Lubbers Quesada and Blake 2003) However, in response to a question which focuses the subject, such as ¿Quién gritó? (‘Who shouted?), an interlocutor would most likely respond with Gritó la chica, placing the subject, la chica, the focus element at the end of the utterance in accord more with pragmatic constraints than with syntactic ones. In unaccusative sentences, the subject is most likely to be postposed in both neutral and focus contexts. The details of this distinction can be explained syntactically according to where the subject is base-generated (Zagona 2002; Hertel 2003; Lozano 2006). Whereas the subject of unergative verbs is considered an external argument basegenerated preverbally in [Spec, VP] and must raise to [Spec,TP] for the purposes of feature-checking, the subject of unaccusative verbs is an internal argument base-generated postverbally in object position in [V,Comp] and does not raise to [Spec, TP]. These facts are made evident in the derivations of (73a) and (73b) for neutral contexts, repeated and illustrated as (74a) and (74b) below. (74)

a. Unergative (unfocused subject) La chica gritó. (‘The girl shouted.’) TP T󸀠

DP la chica T gritó

DP (la chica)

VP

V (gritó)

4.2 The Unaccusative Hypothesis and word order |

91

b. Unaccusative (unfocused subject) Llegaron dos señores. (‘Arrived two men.’) TP T󸀠

DP pro T llegaron

VP

V DP (llegaron) (dos señores)

(Adapted from Lozano 2006: 374–375) In (74a) and (74b), the verb raises from V to the T position of T󸀠 ; however, only in (74a) does the subject raise from the spec position of VP to the spec position of TP (raised lexical items are, of course, deleted and appear between parentheses). The subject in (74b) remains in the comp position of VP. Thus subject order for intransitives is determined syntactically according to the lexical verb class. When subject word order is determined by the information structure of discourse however, most notably subject focus, the derivations for unergative and unaccusative sentences are as illustrated in (75a) and (75b). In both cases the subject is focused and thus has an interpretable feature [Foc]; the head of focus merges with VP and because the focus head has the unvalued uninterpretable feature [uFoc] and an [EPP] feature, the subject can be displaced, eventually raising to the spec position of FocP. In subject focus contexts for both unergative and unaccusative sentences this results in postverbal subjects; the only difference is that the subject of unaccusatives is based generated in object position of the VP (not in the Spec position as for unergatives) and the verb merges with this subject to create a VP. (75)

a. Unergative focused subject ¿Quién gritó? (‘Who shouted?’) Gritó la chica. (‘Shouted the girl.’)

92 | 4 The syntax-pragmatics interface in the acquisition of subject pronouns TP T󸀠

DP pro T gritó

FocP

Foc󸀠

DP la chica [Foc]

Foc0 VP ([uFoc]) ([EPP]) DP V (gritó) (la chica) (gritó) [Foc] b. Unaccusative focused subject ¿Quienes llegaron? Llegaron dos señores. (‘Arrived two men.’) TP T󸀠

DP pro T llegaron

FocP

DP Foc󸀠 dos señores [Foc] Foc0 VP ([uFoc]) ([EPP]) V DP (llegaron) (llegaron) (dos señores) ([Foc]) (Adapted from Lozano 2006: 377)

4.2 The Unaccusative Hypothesis and word order | 93

The consequences of this lexico-syntactic and discourse-pragmatic distinction have implications for both L1 and L2 language acquisition. Bel (2003) found that children acquiring L1 Spanish and Catalan set the positive value of the pro-drop parameter early (ages ranged from 1 year, 1 month to 2 years, 6 months) and that there is evidence that overt and null subjects, as well as preverbal and postverbal subjects, are constrained by lexical verb class and discourse-pragmatic properties in the same way as they are in adult speech. Overuse of overt subjects in pragmatically inappropriate contexts was not found, even at the earliest ages. Furthermore, the children in her study preferred preverbal subjects in most of their utterances but not with unaccusative verbs, where postverbal subjects were more frequent. However, when the children produced preverbal subjects with unaccusatives, there was a discourse reason, namely, the subject is the known information and the verb is the new information, justifying a left-dislocated subject (preverbal). The author claims that this is evidence that the distribution of null and overt subjects and the distribution of pre- and postverbal overt subjects do not derive only from the pro-drop parameter but rather are dependent upon discourse principles. In short, children of L1 Spanish and Catalan have both syntactic knowledge of different verb types (transitive, stative, unaccusative and unergative) and pragmatic knowledge of the restrictions that apply to these verbs in conversational discourse contexts. Bel’s study suggests an early convergence towards the adult native grammar at the lexical-syntactic and discourse-pragmatic interfaces in child L1 acquisition of null and overt subjects in their pro-drop languages. The question arises as to whether these same patterns arise in child and adult L2 acquisition of the distribution of null and overt subjects and preverbal and postverbal subjects in pro-drop. The evidence suggests that this is not the case; that is, in cases of L2 acquisition of null and overt subjects in pro-drop, syntactic knowledge, whether available via the L1 or UG, appears to develop before pragmatic knowledge. In a study of child L2 acquisition of Spanish subjects, Pladevall Ballester (2010) found that the syntactic properties of Spanish subjects are acquired, although not completely in younger children, before properties at the syntax-discourse interface, which remained problematic even among the older children. Participants in the study were five-, ten-, and seventeen-year old L1 English speakers learning L2 Spanish in an immersion context in Britain, starting at the age of four, and corresponding native-speakers (monolingual) of the same age from primary and secondary schools in Spain. The author was specifically interested in testing if the children would acquire the strong pronominal [+interpretable] Agreement features of the verb whereby the verb is raised to T by EPP checking, therefore allowing null subjects and postverbal subjects. In addition, she wanted to test their knowledge of topic vs. focus contexts and co-reference in backwards anaphora contexts, and of lexical verb class in preverbal and postver-

94 | 4 The syntax-pragmatics interface in the acquisition of subject pronouns bal subject contexts. Data were collected through a paired grammaticality judgment and preference task, which was oral for the five-year-old L2 and monolingual control groups, and written for the ten- and seventeen-year-old L2 and control groups. Participants were asked to judge grammatical and ungrammatical sentences that displayed a discourse contrast. In order to assure that participants were not responding randomly and were focused on the task, they were also asked to justify their answers. The sentences presented to the five-year-old groups were uttered by finger puppets that, the children were told, were “learning Spanish”. The children had to decide if the puppets’ utterances “sounded right” or “sounded funny”. The specific linguistic properties that all groups were tested on included: null and overt subjects in main and subordinating clauses and expletive sentences and postverbal and preverbal subjects in neutral and presentational focus contexts with transitive, unaccusative and unergative verbs. Several of the test items are illustrated in examples (76–79). (76)

Pronominal subjects main clause a. pro Creo que Ana ha recogido los juguetes. pro believe-1sg that Anna have-3sg gathered the toys ‘[I] believe that Ana has picked up the toys.’ b. Yo creo que Ana ha recogido los juguetes. I believe-1sg that Anna have-3sg gathered the toys ‘I believe that Ana has picked up the toys.’

(77)

Pronominal subjects subordinate clause (backwards anaphora) a. Cuando ellos trabajan, mis padres no vienen a dormir. when they work-3pl my parents no come-3pl to sleep-inf ‘When they work, my parents don’t come to sleep.’ b. Cuando pro trabajan, mis padres no vienen a dormir. when pro work-3pl my parents no come-3pl to sleep-inf ‘When [they] work, my parents don’t come to sleep.’

(78)

Pre- and postverbal subjects with unaccusative verbs (with presentational focus on subject) ¿Quién ha llegado al colegio? who have-3sg arrived to-the school ‘Who has arrived to the school?’ a. Ha llegado la nueva profe al colegio. (VS) have-3sg arrived the new prof to-the school ‘Has arrived the new prof to the school.’

4.2 The Unaccusative Hypothesis and word order | 95

b. La nueva profe ha llegado al colegio. (SV) the new prof have-3sg arrived to-the school ‘The new prof has arrived to the school.’ (79)

Pre- and postverbal subjects with unergative verbs (with presentational focus on subject) ¿Quién llamó desde España? who call-3sg-pret from Spain ‘Who called from Spain?’ a. Llamó mi abuelo desde España. (VS) Call-3sg-pret my grandfather from Spain ‘Called my grandfather from Spain.’ b. Mi abuelo llamó desde España. (SV) my grandfather call-3sg-pret from Spain ‘My grandfather called from Spain.’ (Based on Pladevall Ballester 2010: 194)

Pladevall Ballester (2010) hypothesized that child L2 learners would gradually develop native-like judgments of the syntactic and discursive properties of Spanish subjects but that there would be L1 transfer and that the acquisition of discourse properties would be more difficult and would develop later. A comparison of the results of the three L2 groups revealed that indeed there is a gradual increase in the behavior of the children towards the native speakers’ judgments for each age level. All learner groups had more difficulty rejecting subjects in subordinate clause contexts (where a null subject is expected) than in main clauses and all groups did better rejecting overt subjects in expletive sentences, although the five- and ten-year-old age groups performed slightly less successfully than the older L2 and native speaker groups, revealing a purely syntactic deficit. Younger children were also more likely to accept redundant subjects in subordinate clauses in sentences such as (77a) revealing another syntactic deficit, although the author suggests this may have been due to processing difficulties young children experience since “backwards anaphora interpretation requires more processing effort” (Pladevall Ballester 2010: 209). For transitive sentences, all learners accepted the SVO word order, as expected. Overall the five and ten-year-old groups showed a greater preference for overt subjects and preverbal word order regardless of verb class or presentational focus, although the ten-year-old group was slightly more accurate than the younger group for most properties except in regards to the acceptance/rejection of preverbal/postverbal subjects, where the latter group actually performed less native-like than the five-year olds. The seventeen-year-old

96 | 4 The syntax-pragmatics interface in the acquisition of subject pronouns group showed native-like judgments for all properties except null/overt pronominal subjects in subordinate clauses and with unergative verbs in neutral contexts. However, the seventeen-year-old monolingual control group also showed difficulties with the acceptance/rejection of null/overt subjects in subordinate clauses, which led the author to conclude that these results must be due to methodological issues and not from “lack of syntactic competence” (Pladevall Ballester 2010: 212). The results of all three groups nonetheless showed that the L2 learners are capable of acquiring the syntactic properties of Spanish subjects as early as five years of age as evidenced in this group’s acceptance of null and postverbal subjects. Nonetheless, acquisition at the syntax-discourse interface is more difficult and takes longer to acquire. Even as late as seventeen years of age, L2 learners show lingering deficits related to discourse properties. L1 influence, which is clearly evident in the two younger groups’ interpretations, has largely disappeared among the older groups’ judgments. This confirms that “the strong pronominal [+interpretable] Agreement features on the verb, by which Spanish checks the EPP by raising the verb to T” (Pladevall Ballester 2010: 213), are fully acquired by the older group and therefore these learners accept both empty subjects and postverbal subjects. Furthermore, Pladevall Ballester claims that properties at the syntax-discourse interface (including topic/focus contexts, coreference in backwards anaphora, and lexical verb class and preverbal/postverbal subject contexts) are late acquired and even in advanced/older learners show a certain amount of optionality. The results of this study reveal that child L2 learners are capable of acquiring most of the syntactic principles of Spanish null and overt subjects at an early age; however, the properties that lie at the lexico-syntactic and discourse-pragmatic interface are more difficult to learn and are problematic even among the more advanced and older age group. Because of the differences between children and adults in terms of both linguistic and general cognitive development, we must be careful in drawing parallels between child and adult L2 learners. Nonetheless, similar findings have been found in adult acquisition of null and overt subjects in L2 Spanish, which suggests that both syntactic and pragmatic properties are at work and are guided by universal constraints. One of the earliest studies in adult L2 Spanish acquisition to consider the interaction of lexico-syntactic and discourse factors is that of Hertel (2003), who examined the development of L1 English learners’ knowledge of lexico-syntactic and discourse-pragmatic constraints on L2 Spanish subject-verb word order. Hertel set out to discover whether or not learners of Spanish acquire the unaccusativeunergative distinction in neutral discourse contexts before they learn the distinction for subject focused contexts. She asked if learners would demonstrate knowledge of SV word order for unergative verbs and VS word order for unaccusative verbs in neutral discourse contexts and at what stage of development the learn-

4.2 The Unaccusative Hypothesis and word order | 97

ers would become sensitive to how the pragmatic feature of focus interacts with word order in Spanish with these two types of verbs. In the study, native Spanish speakers and L2 learners from four levels of proficiency were asked to complete a contextualized written production task designed to manipulate the information structure of participants’ responses, focusing on either the entire sentence (discourse neutral) or the subject (subject focused). Both unaccusative and unergative verbs were included. An example of the types of contexts is illustrated in (80). (80)

You and your friend Sergio are at a party. Sergio leaves to use the bathroom. While he is in the bathroom, Sara, the life of every party, arrives. When Sergio returns he notices that everyone seems much more festive. Sergio asks you: ¿Qué pasó? what happen-3sg-pret ‘What happened?’ What do you answer?

The expected answer for an unaccusative verb in a context neutral discourse context is: (81)

Llegó Sara. (VS word order) arrive-3sg-pret Sara ‘Arrived Sara.’ (Hertel 2003: 287)

Results revealed that for unaccusative neutral contexts the native speakers and the advanced learner group produced a significantly greater number of sentences with the expected VS word order than the other groups; the two intermediate groups had a small percentage of VS and the beginning group had none, most likely due to L1 transfer. Interestingly, the advanced group produced more VS sentences (almost 56 %) than the native speakers (almost 39 %). For the unergative neutral contexts, it was seen that the advanced group produced significantly more of the unexpected VS word order than the other groups, which may have been due to these learners overgeneralizing the expected word order for all intransitive verbs. The advanced group, however, produced fewer inverted sentences for unergative verbs than for unaccusatives, which demonstrates a sensitivity for semantic verb type. Also, despite the fact that the native speakers produced far fewer VS sentences with unaccusatives than expected, in comparison with the percentage of inverted sentences with unergatives (less than 7 %), they also clearly exhib-

98 | 4 The syntax-pragmatics interface in the acquisition of subject pronouns ited knowledge of the distinct syntactic consequences for the two types of verbs. Although there was a slight tendency for the two intermediate groups to produce more VS for unaccusatives than for unergatives, the difference was not statistically significant. Nonetheless, there is observed gradual improvement toward the native speakers’ behavior. For unaccusatives and unergatives in focused subject contexts, where VS word order is expected for both types of verbs, the beginner group again produced no sentences with VS order and the low intermediate group produced very few, demonstrating a lack of sensitivity for both lexical verb type and discourse context. The high intermediate, advanced and native speaker groups all produced significantly more VS word order for both types of verbs than the other lower proficiency groups. Again, the native speakers produced fewer VS sentences than expected (about 36 % for unaccusatives and 33 % for unergatives) and advanced learners behaved statistically similarly to the natives but different from the other learner groups, although they produced an even higher percentage of VS sentences for unaccusatives (54 %) and unergatives (36 %) than the native speakers. The author claims that this is evidence that advanced learners have acquired both the syntactic distinction between the two verb classes and the discourse-related word order preferences of Spanish. However, these results need to be considered more carefully. The fact that they treat both discourse neutral and focused subject contexts about the same reveals that they do not quite understand the syntactic consequences of the discourse-pragmatic principle of focus. Even the native speakers underperformed with VS word order where expected (in neutral contexts for unaccusatives and in focused subject contexts for both semantic types of verbs), which may be the result of the research methodology that explicitly requested complete sentences as responses to focused subject questions, a pragmatically anomalous answer for most native speakers. For example, in response to a question such as ¿Quién llegó? (‘Who just arrived?’), native speakers may prefer to respond with only the focused subject minus the verb, Sara, rather than the complete response, Llegó Sara (‘Arrived Sara’). Nonetheless, for the native speakers, although their production of VS sentences is lower than expected all around, it is significantly higher for unaccusatives than unergatives in neutral contexts and about the same for both types of verbs in focused subject contexts, clearly demonstrating a syntactic consequence for discourse focus. Furthermore, there are clearly important observable differences between the advanced group and the two lower proficiency groups, which seem to point to a gradual development of sensitivity to both lexical-syntactic and discourse-pragmatic principles that govern subject inversion in Spanish. The acquisition of the syntax-discourse interface was also examined by Montrul and Rodríguez Louro (2006), who questioned whether discourse-prag-

4.2 The Unaccusative Hypothesis and word order | 99

matic constraints are acquired at the same time as morphosyntactic properties of the NSP or later on in the acquisition process as has been found in the child L1 and L2 studies. They examined the production of null and overt subjects of 48 English-speaking L2 learners of Spanish from three levels of proficiency (intermediate, advanced and near-native) and 20 native speakers (excluding speakers of Caribbean dialects) in an oral elicitation task where participants were asked to narrate the story of Little Red Riding Hood based on a series of pictures presented to them. The authors included both morphosyntactic aspects of the null subject parameter (i.e. subject-verb agreement, null vs. overt subjects with referential and non-referential subjects, pronominal vs. lexical subjects, and pre vs. postverbal subjects) and discourse-pragmatic aspects (i.e. same vs. switch referent (same or new topic), and emphasis). They had predicted that less proficient learners would start producing null subjects but would show a lower percentage than more advanced learners and native speakers. On the other hand, they expected the more advanced learners and native speakers to produce fewer overt than null subjects. They also expected beginner and intermediate learners to exhibit more agreement errors than advanced learners and native speakers, and to use only preverbal subjects. Their results showed that all three groups of learners had acquired the morphosyntactic properties of the null subject parameter having produced null subjects and no ungrammatical overt expletives; however, the intermediate group produced more agreement errors and more overt than null subjects than the advanced and near-native groups and produced no postverbal subjects, unlike the other two groups who performed quantitatively and qualitatively like the native speakers. The advanced and control groups preferred null subjects more than 56 % and 57 % of the time respectively, whereas the near-native group produced null subjects in more than 68 % of the contexts. The qualitative analysis of the data revealed that although the sentences with preverbal subjects produced by the intermediate learners were not ungrammatical, the lack of preverbal subjects, particularly with unaccusative verbs, appeared stylistically unnatural. For the native speakers and the near-native and advanced learner groups, the majority of the postverbal subjects appeared with unaccusative verbs, signaling that the more proficient learners had acquired the morphosyntactic constraints of the NSP. The analysis of the discourse-pragmatic properties showed an incremental acquisition with the intermediate group producing redundant overt subjects 23 % of the time and the advanced group only 7.6 % of the time; near-native learners produced less than 1 % of redundant overt subjects and the native speakers produced none. Regarding the illicit use (overuse) of null subjects, only the advanced and nearnative learner groups did so in 8.4 % and 5.5 % of the time respectively in contexts of change of referent; the intermediate and native speaker groups produced illicit

100 | 4 The syntax-pragmatics interface in the acquisition of subject pronouns null subjects in these contexts only 1 % – a surprising finding for the least proficient group of the study. However, a look at the contexts in which the intermediate group uses null subjects with accuracy does not reflect native-like pragmatic use but rather is used in a majority of contexts where in English, a null subject is expected such as in coordinating parallel clauses. The study revealed that L2 learners whose L1 is non-pro-drop are capable of acquiring the morphosyntactic properties, but the acquisition of the distribution of null and overt subjects is constrained by the discourse-pragmatic properties of their L1, not the L2, at least in the beginning. These discourse properties are acquirable but emerge only after the morphosyntactic aspects of the structure have been developed and even then, at advanced levels, there are residual deficits. Sensitivity to the syntactic consequences of lexical verb type develops only over time. The fact that even advanced learners overused overt as well as null subjects is evidence both for and against Sorace’s (2004) unidirectional predictions which forecasts overuse only for overt subjects resulting from an underspecification of the feature [topic shift]. Because null subjects lack this feature, the prediction is that they would occur only in non-topic contexts. As Sorace (in a previous study) and Montrul and Rodríguez Louro point out, this could be an artifact of the research methodology or a signal of these learners’ linguistic deficit and needs to be examined more thoroughly in future investigations (2006: 415). If Sorace’s (2004) and Sorace and Filiaci’s (2006) claim holds true that the L1 is responsible for pragmatic transfer and this accounts for the overuse of overt subjects in L2 learners of pro-drop languages whose L1 is non-pro-drop, then we should expect that learners of Spanish whose L1 is also pro-drop would not exhibit the same deficits at the syntax-pragmatic interface. In light of these claims, Margaza and Bel (2006) investigated the L2 acquisition of Spanish subject pronouns in two groups of learners whose native language was Greek (also pro-drop) and a group of native Spanish speakers from Spain. The authors set forth hypotheses that these learners would have positive transfer from their first language either: 1) only in the syntactic domain and not in the pragmatic domain, or 2) both in the syntactic domain and at the syntax-pragmatics interface. They also proposed that they would observe differences in subject pronoun use according to proficiency level. The authors do not assume the Full Transfer/Full Access position of others (Schwartz and Sprouse 1996) but rather the Indirect Access position because “we assume that there is the influence of the L1 in the syntactic domain of grammatical knowledge and not at the syntax-pragmatics interface that is inherently more difficult to acquire in L2” (Margaza and Bel 2006: 90). Although Margaza and Bel state that the study is based on a UG model, their assumptions are contrary to mainstream generative claims that it is the syntactic component that is acquired via UG principles and pragmatic properties are more vulnerable

4.2 The Unaccusative Hypothesis and word order | 101

to L1 interference. Nonetheless, their research sheds light on some compelling findings for Greek learners of L2 Spanish. Nineteen L2 learners studying Spanish in a classroom context in Greece and 10 native Spanish speakers from Spain (Córdoba and Barcelona) participated in the study and were asked to complete two tasks: a cloze narrative test designed to elicit subjects in different syntactic and discursive contexts, and a free written narrative task. The results of both tasks revealed that although both groups of learners were capable of producing null subjects in both matrix and subordinate clauses, thus providing evidence that they know the parametric option for subjects in Spanish, they did not do so for the same percentages as the native speakers and indeed overused overt subjects. The intermediate learners use null and overt pronouns about half and half in matrix clauses for both the cloze test and the free written narrative task. An example from the cloze test (82a) shows the kinds of overused overt pronouns that the intermediate learners produced. (82)

¿A dónde vas? to where go-2sg ‘Where are you going?’ a. Overt pronoun (overuse): #Yo voy a Barcelona. I go-1sg to Barcelona ‘I am going to Barcelona.’ b. Null pronoun (expected): pro Voy a Barcelona. Pro go-1sg to Barcelona ‘[I]’m going to Barcelona.’ (Adapted from Margaza and Bel 2006: 92)

The authors explain that if knowledge of L1 pragmatic knowledge were to transfer, there would be no evidence of the overuse of overt subjects, even among the intermediate learners, and speakers should select the null pronoun option as expected in their native language, Greek. Both learner groups did, however, produce more null subjects in subordinate clauses than in matrix clauses when the subordinate subject maintained the same discourse referent introduced in the previous matrix clause, although again, not to the same degree as the native speakers. Interestingly it was noted in the cloze task, which also tested for knowledge of obligatory subject-verb inversion in quotative contexts, that intermediate learners produced far fewer inversions than the advanced and native speaker groups, as in (83a).

102 | 4 The syntax-pragmatics interface in the acquisition of subject pronouns (83)

a. No inversion ¿A dónde vas? –el señor preguntó. to where go-2sg the man ask-3sg-pret ‘Where are you going? – the man asked.’ b. Obligatory inversion ¿A dónde vas? – preguntó el señor. to where go-2sg ask-3sg-pret the man ‘Where are you going? – asked the man.’ (Adapted from Margaza and Bel 2006: 93)

This is evidence against L1 transfer for a syntactic property that holds for both the L1 and L2. As the authors point out, this may be because the intermediate learners are interpreting the object of the quotative verb as an independent clause. Thus, the non-native-like behavior may be due to a processing difficulty and not due to a lack of knowledge of obligatory inversion in these cases. In the free written narrative task there were few instances of subject-verb inversion for all three groups, although for the native speaker group this was because of a general lack of overt subjects. Nonetheless, for the learners who had a higher proportion of overt subjects, the preferred word order was SVO; this again is evidence against L1 transfer from Greek, which shares the same property with the L2. The authors claim that because these learners use null subject pronouns at least some of the time and also overuse overt subjects in inappropriate contexts, this is evidence that there is positive transfer from the L1 in the syntactic domain but not in the pragmatic one. However, their results mirror many of the results seen in the studies discussed in Chapter 3 and in the current chapter that have examined the acquisition of learners whose L1 is non-pro-drop, mainly English. Not considered in their study was the fact that Greek and Spanish differ in how the notion of focus affects word order in the two languages. Lozano (2006) also looked at the acquisition of word order and focus and nonfocus contexts in several groups of L1 Greek learners of L2 Spanish and native Spanish speakers. He rejects the notion that L2 learners’ deficits can be traced to L1 syntactic transfer due to the fact that in both Greek and Spanish, subjectverb inversion is subject to the same lexical-syntactic properties but differ in the constraints that operate at the syntax-discourse interface for subject focus. For neutral contexts with non-focus, for example in ‘out-of-the-blue’ statements or in response to questions such as What happened?, both Spanish and Greek are constrained by the Unaccusative Hypothesis and tend to place subjects preverbally with unergative verbs and postverbally with unaccusative verbs. However, these two languages show differences at the syntax-discourse interface for focused sub-

4.2 The Unaccusative Hypothesis and word order |

103

ject contexts; whereas Spanish speakers prefer to place subjects postverbally for unergative and unaccusative verbs, Greek speakers prefer the preverbal position for both types of verbs and mark the contrast prosodically. Thus, Lozano attempts to show that the deficits Greek learners of L2 Spanish exhibit in subject-verb/verbsubject word order interpretations derive from their difficulties in encoding the uninterpretable syntactic features of the interpretable features of focus and topic. L2 learners from three different proficiency levels (upper advanced, lower advanced and upper intermediate) and a group of native Spanish speakers were included in Lozano’s study and were asked to complete a contextualized acceptability judgment task which would bias participants’ selection for a pragmatically acceptable response depending on how the question focused the response (neutral or non-focused subject or focused subject) and the type of verb (unaccusative or unergative). Answers were based on a 5-point Likert scale that ranged from −2, which signaled a ‘completely unacceptable’ response, to +2, which expressed a ‘completely acceptable’ response. Below is a sample of the test items. (84)

Vas al cine a ver una película romántica. Durante la película, un niño, que está a tu lado, empieza llorar. Al salir del cine, te encuentras con tu amigo Felipe, que dice que oyó llorar a alguien en el cine pero no sabe quién. ‘You go to the cinema to watch a romantic film. During the film, a little boy, who is sitting next to you, starts crying. After the film, you meet your friend Felipe outside the cinema. Felipe says he heard somebody crying but doesn’t know who.’ Felipe te pregunta ‘¿Quién lloró?’ Tú respondes: ‘So Felipe asks you: “Who cried?’ You reply:’ a. Lloró un niño. −2 −1 0 +1 +2 ‘Cried a boy.’ b. Un niño lloró. −2 −1 0 +1 +2 ‘A boy cried.’ (Lozano 2006: 382–383)

In the sample item, the expected response from a native Spanish speaker would be (a.) because the question, ¿Quién lloró? (‘Who cried?’), sets up the subject to be focused, and thus, despite the fact that the verb is unergative, the preferred position is postverbal. Lozano had predicted that in focused subject contexts like this one, learners at all levels, even at the advanced stage, would fail to show sensitivity to sentence-final focus and would not distinguish between pragmatically anomalous #SV and pragmatically correct VS word orders, regardless of the type of verb (Lozano 2006: 384).

104 | 4 The syntax-pragmatics interface in the acquisition of subject pronouns Results revealed that all participants, both learners and native speakers, preferred the VS word order for unaccusatives in neutral (non-focused subject) environments and rejected the anomalous #SV order more often, although the native speakers rejected this order more often than the learners at a marginally statistically significant difference. Similarly, for the neutral contexts and unergative verbs, all groups preferred the SV word order above the anomalous #VS order, although there is a gradual increase toward the native speaker norm in the acceptance of SV order as proficiency improves, but little difference among learner groups (i.e. a stable performance) for the rejection of pragmatically odd #VS. For the focused subject environments with unaccusative verbs, the upper intermediate group did not distinguish between the acceptable VS and the pragmatically unacceptable #SV and accepted both word orders at the same rate. The lower and upper advanced groups moved increasingly toward the native speaker norm and accepted VS word order at a gradually higher rate and improved on their rejection of anomalous #SV order although not to the same extent as the native speakers. In the focused subject contexts for unergatives, results showed that, in sharp contrast to the native speakers, all learner groups preferred the pragmatically anomalous #SV word order to the pragmatically correct VS order, accepting the former and rejecting the latter, and there appears to be only very slight improvement in this trend as proficiency develops. By examining only the preferences for word order in the neutral contexts of unaccusatives and unergatives, Lozano points out that one could conclude that either these L2 learners of Spanish are sensitive to the universal constraints as set forth in the UH, or that there is positive transfer from the L1, since “the surface syntactic manifestation of unaccusativity is identical in Greek and Spanish” (Lozano 2006: 288). However, the results of the focused subject contexts with unaccusatives show that only upper advanced learners are sensitive to discursive VS word order, which could be explained as overgeneralization of the preferred word order for unaccusatives in neutral contexts. L1 transfer is ruled out here as the upper intermediates and lower advanced groups allow #SV and VS about the same. Lozano claims that the results from the focused subject contexts with unergatives provide the most telling proof that “discursive VS has not been acquired, since learners show optionality at all levels of development, i.e. they equally prefer VS and #SV” (Lozano 2006: 289) and their patterns of acceptability and rejection are opposite to the native speakers’ pattern. That all levels of learners display optionality in their judgments of focused contexts with unergatives might suggest that overgeneralization is a likely explanation for the upper advanced learners’ apparent native-like behavior regarding focused contexts with unaccusatives, but the author rejects this possibility due to the fact that we do not see the same patterns in neutral and focused contexts for this type of verb. Learners may be

4.2 The Unaccusative Hypothesis and word order | 105

aware that neutral and focused contexts are different but are not quite sure how to syntactically mark focused contexts. Lozano summarizes as follows: (R)esults from focused contexts with unaccusatives and unergatives suggest that Greek learners of Spanish show deficits with word order distribution at the syntax-discourse interface, while their intuitions converge with natives’ when word order is constrained by universal properties at the lexicon-syntax interface. (Lozano 2006: 389)

Although learners show these deficits, the author believes that it is not because there is a deficit at the syntax-discourse interface but rather is because L2 learners of Spanish know that there is a difference between neutral and focused contexts, stemming either from universal or L1 knowledge of pragmatic constraints, but are not sure of how to encode this difference syntactically via word order. We could cautiously suggest that this explanation may be valid for English-speaking learners of L2 Spanish as well, who also appear to be sensitive to both the lexicosyntactic and discourse-pragmatic constraints of word order distribution, at least at the advanced level; however, they have not developed complete control over the formal syntactic configuration necessary for distinguishing between neutral and focused contexts in Spanish. Comparisons of the studies reviewed here are difficult not just because of the differing ages and learning environments of the learners, but also because of the different research methodologies employed. Whereas some analyses were based on production data via conversations or oral or written narratives, others used contextualized acceptability judgment tasks (either oral or written) or cloze tests. In addition, not all studies examined the same properties in the same way. What is clear from the results, however, is that L2 learners, both children and adults, acquire the syntactic and lexical properties of subject expression in Spanish before they sort out the discourse constraints responsible for subject distribution. Even L2 learners of Spanish whose L1 is also a pro-drop language display these lingering deficits at the discourse-pragmatic interface. Table 12 summarizes these results. Further evidence of the independence and subsequent development of syntactic and pragmatic properties is taken up in the following section, which reviews research that examines the interaction of syntactic knowledge of the OPC and the discourse constraints of contrastive and non-contrastive focus.

106 | 4 The syntax-pragmatics interface in the acquisition of subject pronouns Table 12. Summary of results for lexical and pragmatic principles in L2 subject acquisition. Study

Syntactic properties

Overt/null constrained by discourse context

Word order constrained by lexical verb class

Word order constrained by focused context

Bel (2003)

Yes

A cautious ‘Yes’

Yes

Yes

Child L1 Spanish and Catalan

Early use of overt and null subjects

Difficult to interpret conversational data

Yes, but not complete

No

Younger children: No

Younger children: No

Older children: Yes, but residual optionality with unergative verbs in neutral contexts

Older children: Yes

Less proficient learners: No (but gradual improvement)

Less proficient learners: No

1;7–2;8 years Pladevall Ballester (2010) Child L1 English– L2 Spanish 5, 10 and 17 years

Some acceptance of overt and null subjects with persistent optionality

All age groups had difficulty judging overt subjects (same referent) in subordinate clauses

Hertel (2003) Adult L1 English– L2 Spanish

Advanced learners: Partial Yes

4 levels of proficiency

Montrul and Rodríguez Louro (2006) Adult L1 English– L2 Spanish 3 levels of proficiency

VS with unaccusative but also with unergative (but less so) Yes.

Partial Yes.

Some deficits and improvement with proficiency

Overt subjects with same referent diminishes with proficiency Null subjects with switch referent persist Null subjects of intermediate learners reflect L1 transfer

Intermediate learners: No Advanced and near-natives: Yes

High Intermediate and Advanced learners: Partial Yes VS for both types of verbs but more so for unaccusative than unergative

4.3 The OPC (again)

| 107

Table 12. (continued). Word order constrained by lexical verb class

Word order constrained by focused context

Study

Syntactic properties

Overt/null constrained by discourse context

Margaza and Bel (2006)

Yes.

Partial Yes.

No.

Some deficits and improvement with proficiency

More null subjects for same referent in subordinate clauses but not like native speakers

Examined word order in quotative contexts – expected VS

Adult L1 Greek– L2 Spanish 2 levels of proficiency Lozano (2006) Adult L1 Greek– L2 Spanish 3 levels of proficiency

Yes.

No.

Residual deficits and improvement with proficiency

Some sensitivity to focused context for unaccusatives and improvement with proficiency No sensitivity to unergatives but slight improvement with proficiency

4.3 The OPC (again) If it can be shown empirically that L2 learners have native-like knowledge of the syntactic properties of the Overt Pronoun Constraint (OPC), as set forth by Montalbetti (1986), while also exhibiting non-native-like deviancies in their judgments and production of null and overt subjects, then this would be further evidence that syntactic competence and discursive-pragmatic knowledge are independent of each other and are acquired in a subsequent manner. As discussed in Section 4.1, null subjects in subordinate clauses are generally co-referential with the subject of the main clause; however, overt subjects of subordinate clauses, depending on the discourse context, can be either co-referential or not with the subject of the main clause but only when the main subject is lexical. However, recall from Chapter 2 that in Spanish, when the subject of the main clause is a variable expression such as a wh- word (e.g. quién ‘who’) or a quantifier determiner phrase (QDP) (e.g. nadie ‘nobody’, muchos profesores ‘many professors’), the OPC is in

108 | 4 The syntax-pragmatics interface in the acquisition of subject pronouns effect and overt pronouns in the subordinate clause can not be co-referential with the matrix subject as the following examples demonstrate. (85)

a. ¿Quiéni dijo que él∗i/j viajaría a México who say-3sg-pret that he travel-3sg-cond to Mexico ‘Whoi said that he∗i/j would travel to Mexico?’ b. Nadiei cree que ellos∗i/j se casarían en enero. nobody believe-3sg that they get marry-3pl-cond in January ‘Nobodyi believes that they∗i/j would get married in January.’

Rothman (2007, 2009) set out to demonstrate that English-speaking learners of L2 Spanish are capable of acquiring the syntactic constraints of subject pronoun use in Spanish as demonstrated by their native-like knowledge of the OPC, but that target-deviant uses of pronouns are due to their insensitivity to the discursive constraints of subject pronoun use in Spanish. He claims that “sensitivity to the syntactic constrains of null/overt subject distribution does not guarantee the same sensitivity to the discursive constraints, at least at the same time” (Rothman 2007: 304). Furthermore, he argues that the interaction of the syntactic and pragmatic modules is a particularly vulnerable interface and causes difficulties for acquisition not just for L2 but L1 and bilingual speakers as well, but is particularly persistent for L2 learners. Like others, Rothman assumes a Minimalist Program feature checking model approach to L2 acquisition, which involves going from the [−D, −interpretable] value in English to the [+D, +interpretable Φ -feature] value in Spanish. The specific syntactic properties that Rothman tested were overt/null referential subjects, overt (ungrammatical)/null expletive subjects and knowledge of the OPC; the discursive-pragmatic constraints under study were contrastive focus and non-contrastive focus. In the (2007) study, the author included thirty intermediate learners of L2 Spanish who were tested following five months of study abroad in Madrid, and 20 native Spanish speakers for control. The participants completed three tests which were similar to (or the same as) those reported on in the Rothman and Iverson (2007a, b, c) studies¹⁸, and comprised: a logical sentence production task (LSPT), a grammaticality judgment and correction task (GJCT), and a co-reference interpretation task based on Kanno’s (1998) test designed for the study of OPC effects in L2 Japanese. For the LSPT, participants were given a brief context and

18 See Chapter 3 of the present volume, Section 3.3, for a discussion of these three studies.

4.3 The OPC (again)

| 109

asked to form relevant logical sentences with uninflected lexical items presented to them. An example appears in (86) below. (86)

Javier y Luisa trabajan para la misma compañía internacional y tienen más o menos el mismo trabajo. Es lógico que deban ganar el mismo sueldo pero hay un desequilibrio entre los sueldos de las mujeres y los de los hombres. ‘Javier and Luisa work for the same international company and have, more or less, the same job. It is logical that they should make equal salaries, but male and female salaries are not always equitable.’ Todos/ saber/ que/ ganar/ más dinero everyone/ to know/ that/ to make/ more money’ (Adapted from Rothman 2007: 306)

For the GJCT, participants were presented with sentences and were asked if they thought they were grammatical or ungrammatical and were requested to correct the ungrammatical sentences or indicate if they didn’t know how to fix them. Examples appear in (87–90). (87)

Grammatical null expletive pro Hace frío afuera hoy. pro make-3sg cold outside today ‘[It]’s cold outside today.’

(88)

Ungrammatical overt expletive *Lo llueve en las montañas. it rain-3sg in the mountains ‘It is raining in the mountains.’

(89)

Grammatical null referential Nadie cuenta con María, pro es mentirosa. nobody trust-3sg with Mary pro be-3sg liar ‘Nobody trusts Mary, [she]’s a liar.’

(90)

Overuse of overt referential #¡No puede ser! María juró que ella lo haría. no able-3sg be-inf Mary swear-3sg-pret that she it do-3sg-cond ‘Impossible! Mary swore she would do it.’ (Adapted from Rothman 2007: 307)

110 | 4 The syntax-pragmatics interface in the acquisition of subject pronouns In the co-reference interpretation task, which tested specifically for knowledge of the OPC, participants were given contextualized sentences as for example in (91) and (92) and were asked to select their interpretation of the embedded subject (null or overt). (91)

Null embedded pronoun with wh-matrix subject ¿Quién no sabe que pro tiene derecho a votar a los 18 años? who no know-3sg that pro have-3sg right to vote-inf at the 18 years ‘Who do you suppose does not know that [he] has the right to vote at 18?’ (a) the same as Quién (b) someone else

(92)

Overt embedded pronoun with a DP matrix subject Vicente afirmó ayer que él le había pedido Vicent state-3sg-pret yesterday that he her-dat have-3sg-imp asked la mano a su novia y que ellos se casarían en julio. the hand to his girlfriend and that they refl-marry-3pl-cond in July ‘Vicente stated yesterday that he had asked for his girlfriend’s hand and that they would marry in July.’ ‘Who do you suppose asked his girlfriend to marry him?’ (a) Vicente (b) someone else (Adapted from Rothman 2007: 307–308)

The author expected, following the OPC, that participants would select both (a) the same as Quién and (b) someone else, as acceptable answers to questions such as (91) since pro has both a bound and disjoint interpretation; whereas for questions like (92), although both bound and disjoint interpretations are possible, Spanish speakers would prefer a disjoint interpretation for the overt pronoun in the embedded clause and should select (b) someone else. This claim is debatable since the discourse context forces a bound interpretation. Results showed that the learners had knowledge of the OPC but performed in a target-deviant manner for both contrastive focus sentences and for overt expletives (never producing them but judging them acceptable). Although learners showed sensitivity to the OPC, they derived a bound variable interpretation for sentences with a null embedded subject with a QDP/wh matrix subject more often that with an overt embedded subject. A subsequent analysis of a subgroup of learners who demonstrated knowledge of OPC at the beginning of the study, however, did show target-like performance of bound variable interpretations of null and overt embedded subjects with QDP/wh matrix subjects. According to the author, the fact that these learners were able to produce null subject sentences

4.3 The OPC (again)

|

111

for both expletive and referential subjects in matrix and embedded clauses is an example of the logical problem of L2 acquisition; that is, learners are capable of acquiring properties of the L2 that are underdetermined in the input. Thus, “(t)he fact that the L2 learners are able to consistently produce null subject sentences in L2 Spanish in contexts not licensed in English must entail that the EPP-feature is checked without merging a subject pronoun to the Spec of AgrSP” (Rothman 2007: 312). However, their acceptability of overt expletives and the overuse of null pronouns (or lack of overt subject pronouns) for contrastive contexts show that they still have not completely acquired the discursive properties that constrain subject expression in Spanish, despite five months of immersion in the target language setting. Further exploring the syntax-pragmatics interface, in his (2009) study, Rothman focused exclusively on classroom English-speaking L2 learners of Spanish in a U.S. university setting. In this study, the author hypothesized that English L2 classroom learners acquire the syntax of the NSP but have problems with the distribution constrained by the discourse pragmatics; he also hypothesized that these problems are not permanent. He sought to verify whether the distribution of null and overt subject pronouns would be unidirectional (overuse of overt pronouns) as predicted by Sorace (2004), Serratrice, Sorace, and Paoli (2004), and Sorace et al. (2009), or bidirectional (overuse of both overt and null pronouns). The participants included intermediate and advanced learners, and a native speaker group for control. The intermediate learners were mostly undergraduate Spanish majors and minors enrolled in content level courses in literature and/or linguistics; the advanced group was comprised of highly proficient L2 speakers who were graduate students and university instructors of Spanish. In order to be included in the study, participants were required to demonstrate knowledge of the OPC which was carried out via a co-reference interpretation task, a modified version of the experiment reported on in Rothman and Iverson (2007a,b) and Rothman (2008), designed to specifically test for the OPC. The native speaker group was comprised of graduate students from a variety of Spanish-speaking countries, including Colombia, Peru, Mexico and Spain. In addition to the co-reference interpretation test, participants completed two other experiments: a pragmatic context-matching felicitousness judgment task and a translation task, both of which were designed to assess knowledge of the discourse-pragmatic constraints on null and overt subject distribution. For the felicitousness judgment task, participants were given background contexts in English and then a related sentence in Spanish. They were asked to judge on a scale from one to five the appropriateness of the Spanish sentences with 1 defined as completely inappropriate and 5 completely natural. Examples included in this task appear as follows:

112 | 4 The syntax-pragmatics interface in the acquisition of subject pronouns (93)

Context supports overt subject (contrastive focus) My friends and I need new shoes. I always buy white shoes, but lately red shoes have become popular. I think about buying red shoes, but when we get to the shoe store I don’t like any and decide to stick with my classic color. My friends still think the red shoes are really cool. Ellos van a comprar los rojos y yo voy a comprar los blancos. they go-3pl to buy-inf the reds and I go-1sg to buy-inf the whites ‘They are going to buy the red ones and I am going to buy the white ones.’ 1 2 3 4 5

(94)

Context supports null subject (no contrastive focus) My girlfriend is studying abroad this semester. I’m very happy for her, but I miss her terribly. I really wish I were able to talk to her more. Mi novia está fuera del país y nunca pro hablo con my girlfriend be-3sg out of-the country and never pro speak-1sg with ella porque siempre pro está ocupada. her because always pro be-3sg busy ‘My girlfriend is out of the country and [I] never talk to her because [she] is always busy.’ 1 2 3 4 5 (Adapted from Rothman 2009: 958)

For the translation task, participants were provided with contextualized stories and asked to translate a dependent sentence related to the story. For this experiment, both contrastive focus and non-contrastive focus contexts were provided as well as answers to topic questions. Examples appear in (95–97) below: (95)

No contrastive focus (expect null subject) My friend Juan is a scientist who is currently in Antarctica. He’s very isolated and doesn’t have a television, radio, telephone or the internet. I recently received a letter from him suggesting that I come stay with him during the warm season. Translate: I hate the cold, but I want to visit him because we are best friends.

4.3 The OPC (again)

| 113

(96)

Topic question (expect overt subject) I tell my mom that I’m going to Antarctica for a while. She tells me to make a list of everything I need and to not forget to pack a little sunscreen since there is a hole in the ozone layer in Antarctica. Just then, my dad, who caught the very end of the conversation, walks into the room. He asks, “Who did you say is going to Antarctica?” I respond: Translate: I’m going to Antarctica to visit Juan.

(97)

Contrastive focus (expect overt subject) My dad starts asking me a million questions about the trip. He wants to know how I’m getting there, how long I’ll be gone, if I have the proper equipment, etc. he asks me when the last time was I went to Antarctica and if Juan has any experience in the cold. Translate: I’ve never visited there, but he has lived there for two years. (Adapted from Rothman 2009: 960)

The results revealed again, as in the (2007) study, that although the intermediate learners had syntactic knowledge of the null subject parameter, there were evident deficiencies in their pragmatic knowledge for overt and null pronoun distribution. Both the advanced learners and a subgroup of intermediate learners showed L2 knowledge of the OPC consistent with the native speakers’ performance more often interpreting null embedded subjects as coreferential with QDP/wh-matrix subjects and overt embedded subjects as disjoint. There was also a significant difference among all groups between the interpretation of null and overt embedded subjects with non-quantified determiner phrase/noun phrase matrix subjects and QDP/wh-matrix subjects. In addition, although all groups judged overt referential subjects in contrastive contexts and null pronouns in non-contrastive environments as highly felicitous, only the advanced and native speaker groups judged sentences with overt pronouns in non-contrastive contexts and null pronouns in contrastive contexts as anomalous. Also, both advanced and native speakers produced null and overt subjects appropriately but the intermediate learners overused null subjects in contexts of contrastive focus or in answer to Topic-wh questions. The intermediate learners consistently showed that they have not acquired the switch-reference or contrastive focus properties of overt subject pronouns in Spanish. By reporting on the results of the experiments that tested for pragmatic knowledge among only participants who had demonstrated knowledge of the syntactic properties of the OPC, Rothman was able to establish that although learners have acquired the formal structural properties of the NSP, they still fail to differentiate the pragmatic contexts that constrain null and overt subject use. Why this is so is still a question to be answered.

114 | 4 The syntax-pragmatics interface in the acquisition of subject pronouns Rothman rejects Sorace’s prediction that the deficits in these learners’ knowledge of the distribution of null and overt subject pronouns in Spanish are due to an English effect. L1 interference could explain the overuse of overt subjects but not the overuse of null subjects. On the other hand, although the Avoid Pronoun Principle could explain the overuse of nulls subjects, it is incapable of explaining why learners would overuse null subjects in ambiguous (i.e. unidentifiable) contexts. Thus, this finding is left unexplained. Rothman suggests that these pragmatic deficits may be due to deficient input from non-native speaking peers and/or instructors in the classroom or even native speakers. Often, when directing speech to learners, native speakers tend to overuse subject pronouns in order to aid comprehension on the part of the non-native speakers. Additionally, some native speakers may tend to simply overuse overt subjects more often than others and exposure to these speakers may have confused some of the L2 learners. This corroborates the findings of Rothman and Iverson (2007a, 2007b, and 2007c) who claimed that increased exposure to input does not help in the resetting of the syntactic properties of the NSP but may have an effect on sorting out the pragmatic features that constrain subject distribution in Spanish. In this case, increased exposure to possibly deviant input may be responsible for the deficits evident at the pragmatic-discourse interface. Finally, as has suggested Sorace et al., processing difficulties may be responsible for “the integration of syntactic and pragmatic information involved in the selection of subject pronouns” (Sorace et al. 2009: 967). If it is indeed the case that pragmatic properties are made evident to learners via discourse patterns, it might also be that L2 learners are not exposed to the amount of natural conversational discourse they need in order to acquire those properties in their L2 Spanish. Furthermore, pragmatic properties constraining null and overt subject pronouns are far more complex than merely detecting topic/focus contexts, as will be explored in Chapter 6. Finally, more recent studies have shown that not all of the phifeatures in the pronominal paradigm are vulnerable at the syntax-discourse interface; in other words, as Lozano (2009) has discovered, these pronominal deficits are selective and appear to affect mostly the third person singular subject pronouns. Research that examines the entire verbal paradigm is the subject of the following section.

4.4 Person and number Because research to date has suggested that the syntax-discourse interface is responsible for the lingering deficits L2 learners of Spanish exhibit, Lozano (2009) sought to determine if these deficits affect the entire pronominal paradigm or are

4.4 Person and number | 115

selective and affect only certain persons and numbers. He argues that the majority of studies has presented evidence of the difficulties L2 learners have acquiring the null/overt distribution only for third person singular subject pronouns and that researchers have implicitly implied that these difficulties extend to all persons and numbers (1st , 2nd , 3rd persons singular and plural). He has suggested that perhaps not all persons and numbers are equally vulnerable and that a certain amount of “selectiveness is a result of how Universal Grammar organizes pronominal features” (Lozano 2009: 128). He bases his analysis on the ‘Feature Geometry Analysis’ for pronouns (subject and object) proposed by Harley and Ritter (2002) (cited in Lozano 2009). In this analysis, UG sets forth a constrained set of pronoun features that are organized hierarchically with three sets of features, including participant, individuation and class. Referring Expression (= Pronoun) participant (= Person) Speaker (= 1st person)

Addressee (= 2nd person)

individuation (= number) Group Minimal Class (= non-sing) (= sing) (= gender) Animate Fem

Inanimate/ Neuter Masc

...

Figure 1. Pronominal feature geometry analysis. (Taken from Harley and Ritter 2002: 508, cited in Lozano 2009: 133)

In this model, the participant node denotes person and individuation represents number, gender and animacy. As can be noted in Figure 1, Speaker corresponds to 1st person and Addressee to 2nd person, both of which are deictic expressions and refer to the participants in any given speech act. In the absence of a participant, the referring expression is underspecified and is interpreted as 3rd person, which is the anaphoric use of a pronoun (as opposed to the deictic use for 1st and 2nd persons). Thus, according to this model “3rd person is the absence of grammatical person” (Lozano 2009: 134). This UG constrained set of pronominal features has been shown to have consequences for child L1 development; e.g. 1st and 3rd person non-animate singular pronouns are acquired earlier than pronouns representing other persons, numbers and 3rd person animate entities.

116 | 4 The syntax-pragmatics interface in the acquisition of subject pronouns Thus, it is logical to suspect that these same features would have an impact on L2 adult acquisition for pronoun acquisition and this is what Lozano set out to investigate. He hypothesized that at the syntax-discourse interface, deficits would “not affect the whole pronominal paradigm,” (Lozano 2009: 137) but rather would be selective, and that he would find “(i) robustness and native-like knowledge with speech-act participants (1st and 2nd person) and with neutrals (3rd person inanimate), but (ii) vulnerability with 3rd person animate.” (Lozano 2009: 138). He examined written narratives of upper and lower-advanced L2 learners and native speakers taken from the Corpus Escrito Del Español L2 (‘L2 Spanish Written Corpus’), a comparative corpus developed at the Universidad Autónoma de Madrid of learners of L1 English-L2 Spanish and native Spanish speakers. The author looked specifically at topic-continuity and topic-shift contexts for all persons and numbers of the subject paradigm (including full noun phrases and overt and null pronouns). He also considered whether subject expressions were animate or inanimate and whether each use was pragmatically appropriate or not. Results revealed that all groups preferred null pronouns for topic-continuity and full noun phrases for topic-shift, the latter a finding that is overlooked in generative L2 studies where it is assumed that prop-drop languages encode topicshift with an overt pronoun most of the time. Nonetheless, the learner groups, especially the lower-advanced level, produced more overt subjects in contexts of topic-continuity than the native speakers and this difference was statistically significant. Conversely, there were fewer unpragmatic uses of null subjects for topicshift contexts, although interestingly the upper-advanced group preferred null pronouns in nearly 10 % of the contexts (although this result was not statistically significant), whereas the lower-advanced only produced 3.3 % null subjects in this context. These findings support those of Montrul and Rodríguez Louro (2006) and Rothman (2007, 2009) who also found an overuse and/or over-acceptance of overt pronouns for non-contrastive (i.e. same reference or topic-continuity) contexts and overuse and/or over-acceptance of null pronouns in contrastive (i.e. switch reference or topic-shift) contexts. The analysis of the pronominal features was revealing, as predicted. The data showed that learners exhibited native-like production in singular 1st person (yo/pro ‘I’), plural 1st person (nosotros/nosotras/pro ‘we’), singular 3rd person inanimate (NP/ello/pro ‘NP/it’) and plural 3rd person inanimate (NP/pro ‘they’) contexts, but a considerable number of pragmatically inappropriate forms in 3rd person animate, particularly in singular (NP/él/ella/pro “NP/he/she’) contexts. Although not statistically significant, there were differences between the two learner groups and native speakers for the pragmatically (in)appropriate uses of plural 3rd person animate subjects (NP/ellos/ellas/pro ‘NP/they’), which lend support to Lozano’s hypothesis that learners would exhibit deficits with 3rd person

4.5 Contributions of findings of the syntax-pragmatic interface model |

117

animate subjects. Finally, there were too few tokens of 2nd person singular and plural for the author to arrive at any conclusions. This reveals that even advanced L2 learners of Spanish show deficits specifically with 3rd person singular animate subjects, but native-like behavior with other features of the pronominal paradigm. Once again, this study points out that English-speaking L2 learners of Spanish are sensitive to the morphosyntactic properties of overt and null subjects and yet exhibit deficits, even at advanced levels, of discursive constraints such as topiccontinuity and topic-shift of pronominal distribution and a residual amount of overuse of null pronouns, more evidence that questions the unidirectionality hypothesis (Sorace 2004; Sorace and Filiaci 2006). However, the contribution Lozano makes to the discussion is his finding that L2 learners demonstrate persistent selective deficits of a UG constrained set of hierarchically organized pronominal features, most importantly 3rd person animate subjects, at least in the written narratives of upper and lower-advanced L2 learners of Spanish. Future research would need to corroborate this finding with a larger L2 population and for oral production and oral and written interpretive tasks as well.

4.5 Contributions of findings of the syntax-pragmatic interface model The research discussed in this chapter reveals several patterns, some inconsistencies and a few remaining unanswered questions regarding the acquisition and distribution of subjects in L2 Spanish. The common finding among all of the studies is that information structure of discourse interacts with morphosyntactic and lexical-semantic properties and that these interfaces are more vulnerable than purely syntactic properties for learners of all ages and acquisition situations. The data also seem to suggest that the syntactic properties of the NSP are acquired before the discourse-pragmatic constraints responsible for subject distribution in Spanish. The studies of bilingual children of English-Italian, English-Spanish and Spanish-Italian languages demonstrate that from a very early age, children allow both null and overt subjects in their pro-drop languages and are sensitive to the discourse-pragmatics that constrain their distribution, although there is evidence of the overuse of overt subjects and, to a lesser extent, overuse of null subjects. Bilingual children appear to take longer than monolingual children of pro-drop languages to integrate the syntactic and pragmatic knowledge necessary in order to resolve the overt/null distinction. The fact that Spanish-Italian children demonstrate the same deficiencies as English-Italian and English-Spanish brings into question the impact of cross-linguistic influence. In addition, the overuse of

118 | 4 The syntax-pragmatics interface in the acquisition of subject pronouns pragmatically anomalous (and ambiguous) null subjects in the pro-drop language also implies that more is involved in the acquisition process than cross-linguistic influence from the non-pro-drop language. Nonetheless, bilingual children of one pro-drop language and one non-pro-drop language accept and produce more pragmatically anomalous overt subjects than monolingual and bilingual children of pro-drop languages. Furthermore, they abandon the overuse of null subjects sooner than the overuse of overt subjects. These facts certainly seem to support a stronger effect of directionality from the non-pro-drop to the pro-drop language in the use of overt subjects, but not necessarily solely unidirectionality. Processing strategies may also partly explain some of the difficulties children exhibit. Faced with two systems: one where pronominal subjects are specified for the features [+topic] and [+focus] and one where subjects are underspecified for these features ([+/−topic] and [+/−focus]), the child may very well opt for the less complex and pragmatically unconstrained system. Indeed, the English data of the bilingual children show that the pro-drop language has little effect on subject use; children at a very early age reject null subjects in English. This further supports the idea that the non-pro-drop system is cognitively less complex to sort out. The question remains as to whether or not bilingual children eventually resolve the differences between their two systems and converge towards the monolingual adult norm for their pro-drop language. The data reported on for adolescent and adult bilinguals provide evidence both for and against eventual independence of the two systems. The older English/Spanish bilingual adolescents in the Pladevall Ballester (2010) study appeared to have acquired most of the discourse properties related to subject distribution in Spanish but showed some lingering deficits. However, the monolingual Spanish group in the study displayed similar deficits, which led the author to suspect a methodological effect. Whereas the advanced group of English-Spanish bilingual heritage speakers in the Montrul (2004) study displayed pragmatically acceptable uses of null and overt subjects, the intermediate heritage speakers exhibited a greater number of inappropriate uses of null and overt subjects. Although Montrul could not specify whether these speakers’ difficulties in null and overt subject distribution were the result of incomplete acquisition or attrition, the data coincide with the child bilingual data that reveals that despite the fact that the syntactic properties are intact early on, it is the discourse-pragmatic interface that is particularly vulnerable in bilingual speakers’ pro-drop language and is responsible for lingering errors. The early parametric studies revealed that learners, even at early stages, are capable of acquiring the formal syntactic properties that license null subjects in L2 Spanish, but could not explain the lingering non-native behavior of the overuse of both overt and null subjects. The focus on the clustering of properties in the pro-

4.5 Contributions of findings of the syntax-pragmatic interface model |

119

drop parameter and the goal of establishing the role of UG in parameter resetting ignored the importance of pragmatic information for explaining the overt/null distinction. More recent studies that examine the syntax-pragmatic interface have recognized information structure as key to understanding the overt/null distribution. These studies have established that although the formal properties are acquired early, the topic/focus distinction develops over time and explains the persistent deficits in L2 Spanish learners’ use of subject expression. The role of learners’ L1 is still unresolved among these studies as evidence for L1 effects is contradictory. The studies reviewed in the following chapter take up the question of how learners interpret pronominal reference when both L1 and L2 strategies are available. “(U)nlike narrow syntax, which is an autonomous and encapsulated system, linguistic discourse is a dynamic open system that interfaces with general cognitive and processing abilities, and as such it is prone to be affected by them” (Sorace et al. 2009: 461). Thus, the next chapter deals more precisely with how cognitive processing interacts with the interpretation and production of overt and null subjects in pro-drop and non-pro-drop languages.

5 Processing studies and L2 subject pronoun acquisition in Spanish Introduction Processing studies in the L2 Spanish acquisition of subject pronouns also assume a Universal Grammar perspective and moreover take into account the Interface Hypothesis discussed in Chapter 4 of this volume. However, processing studies differ from those discussed in the previous two chapters in that the focus is not as much on the product (such as the mental representation or output) of learning but rather on how speakers, including child and adult bilingual speakers and L2 learners, perceive, process and interpret input data (VanPatten and Jegerski 2010). Furthermore, in these studies there is the presupposition that learners have already acquired the formal properties of the L2 but may not yet have acquired the adequate L2 processing strategies for producing and interpreting the specific linguistic forms in a target-like manner. Finally, research discussed in the previous chapter takes the perspective that lingering deficits or ‘optionality’ witnessed in interlanguage grammars is the result of underspecification (due to L1 effects) of interpretable features such as [+topic shift] in the L2 (Sorace and Filiaci 2006), whereas the present chapter focuses on research that seeks to explain L2 deficits as the result of limitations in the processing of different types of information found at the interface between syntax and other cognitive domains (most notably discourse-pragmatic conditions). As related to the L2 acquisition of subject pronouns, investigators seek to determine what types of strategies bilingual speakers and language learners use in assigning antecedents to null and overt pronouns in their pro-drop and non-pro-drop languages. In order to test for the interpretation strategies of subject pronouns, it is assumed that both overt and null subjects exist in the underlying grammar(s) of bilingual speakers and language learners. It is also assumed that the related syntactic properties of at least rich verbal morphology and subject-verb inversion (and perhaps OPC effects) are already present in the L2 Spanish grammar. What is not assumed is that the existence of the null subject parameter will automatically result in native-like processing strategies for interpreting antecedent resolution with different types of pronominal subjects (VanPatten and Jegerski 2010: 8). An important study that has launched research in monolingual, child and adult bilingual, and second language processing of subject pronouns in pro-drop languages is that of Carminati (2002), who proposed the Position of Antecedent Hypothesis (PAH), a processing hypothesis that assumes there is a ‘division of

5 Processing studies and L2 subject pronoun acquisition in Spanish

| 121

labor’ with null and overt subjects: the null pronoun is linked more often than the overt pronoun with a more prominent antecedent, usually the one in preverbal subject position of the antecedent clause and conversely, the overt pronoun (in null subject languages) is linked more often than the null subject with a less prominent antecedent usually in a complement position. No such ‘division of labor’ exists in non-pro-drop languages such as English, which depends on a subject assignment strategy: overt pronouns tend to be linked to the subject antecedent. The majority of studies that have tested the PAH have looked at Italian monolingual children and adults, Italian and English bilingual children, Italian, Spanish and English children and Spanish monolingual adults; a few studies have begun to emerge examining the acquisition of adult L2 Spanish. Studies in second language acquisition from a processing perspective seek to determine whether early Spanish-English bilinguals or heritage speakers and late adult English-Spanish bilinguals (L2 learners) use the same types of processing strategies as native speakers to assign the appropriate antecedent of null and overt pronouns or whether they use a hybrid strategy based on the PAH and subject assignment. In addition, this body of research seeks to explore to what extent innate or universal restrictions and English L1 principles influence the interpretation of ambiguous subject pronouns with intrasentential antecedents. It is important to classify these studies as processing¹⁹ because although they adhere to the claim that syntactic properties available via universal grammar are in principle acquirable in non-native languages (Jegerski, VanPatten, and Keating 2011) and recognize that “properties of the grammar that exist at the interface between the syntax and other areas or modules of linguistic knowledge, such as semantics or discourse, do not reach a stable state of native-likeness” (Jegerski, VanPatten, and Keating 2011: 482), they stress the importance of investigating to what extent innate restrictions and L1 principles influence interface behaviors in L2 acquisition. Ultimately, it is this type of behavior “which is frequently attributed to processing limitations, or a limited capability to integrate multiple sources of knowledge during comprehension” (Jegerski, VanPatten, and Keating 2011: 482) and not a lack of access to the principles of universal grammar. These studies are also to some extent discourse-based and in particular seek to determine which specific

19 VanPatten and Jegerski (2010) distinguish between parsing, which is narrower in scope and refers to “the rapid and automatic assignment of a licit grammatical structure to a sentence” (p. 4), and processing, which is broader in scope and refers to the “moment-by-moment operations during comprehension” (p. 5) and includes the lexicon, semantics, morphology, syntax and interfaces. In this chapter we are interested in the view that certain underlying knowledge is already present in the learners’ grammar and seeks to understand how speakers/learners use this knowledge in interpreting certain features of the language.

122 | 5 Processing studies and L2 subject pronoun acquisition in Spanish features in discourse cue the link between a pronoun and its corresponding antecedent. Work in discourse analysis has shown that anaphoric expressions are most often linked to subject antecedents and thus lead researchers to conclude that the local discourse structure is an important key in processing and interpreting pronoun reference (Hudson-D’Zmura and Tanenhaus 1998: 222). Furthermore, in intrasentential anaphora in complex sentences the discourse structure can play an important role. Anaphoric resolution may depend upon whether or not the two events are parallel (coordinating discourse structure) or a sequence of two events (subordinating discourse structure). Although the methodologies and participant populations are often distinct, the emerging consensus is that there is support for the PAH in the processing of null and overt subjects in pro-drop languages; in cases of bilingualism or L2 acquisition, there is pragmatic interference from the non-pro-drop language which affects anaphoric resolution more according to discourse constraints. In the following sections, the PAH as proposed by Carminati (2002) for Italian monolingual speakers is discussed and compared to anaphoric resolution among Spanish monolingual speakers. Following a brief discussion of the research examining strategies among English-Italian and Spanish-Italian speakers we then turn to the recent work on the study of processing strategies for anaphoric resolution among English-Spanish bilinguals and L2 learners that take into account the PAH and discourse structure.

5.1 Anaphoric resolution in pro-drop languages and the PAH 5.1.1 The PAH in Italian In Carminati’s (2002) study of Italian subject pronouns she claims that in intrasentential anaphora, syntactic position determines antecedent prominence and thus, anaphoric resolution. A pre-verbal subject in the Spec IP position is more prominent than antecedents found in positions lower in the hierarchical syntactic structure, for example in an object position. The PAH predicts that a null subject is more likely to be linked to that position. Specifically, the PAH states “the null pronoun prefers an antecedent which is in the SpecIP position, while the overt pronoun prefers an antecedent which is not in the SpecIP position (Carminati 2002: 33). Carminati tests the PAH for a number of structural positions of possible intra-sentential antecedents that are found at the Spec IP position in Italian, including referential nominative and dative subjects, expletive, and quantified subjects, and found strong support for the PAH for anaphora resolution. Interestingly, she found that the antecedent bias for overt pronouns is less established and

5.1 Anaphoric resolution in pro-drop languages and the PAH |

123

depends more on context than it does for null subjects. In other words, violations of antecedent preferences for overt subjects (i.e. overt subjects referring to preceding subjects) were easier for participants to process than violations of null subjects (i.e. null subjects referring to preceding objects). Furthermore, she found that the PAH was a better predictor of anaphoric resolution for quantified subjects than the OPC. One of the experiments that lead to these conclusions was a reading task with a subordinate clause that introduced both subject and object antecedents and a main clause that could refer to either antecedent. Both antecedents matched in gender and number but the pragmatic context favored co-reference with one or the other antecedents independently of the type of anaphoric pronoun used (null or overt), as seen in the following examples: (98)

a. Quando Vanessai ha visitato Giovannaj in ospedale, leii(/j) when Vanessa have-3sg visited Jean in hospital she le ha portato un mazzo di rose. her-dat have-3sg taken a bunch of roses ‘When Vanessai visited Jeanj in the hospital, shei(/j) took her a bunch of roses.’ b. Quando Vanessai ha visitato Giovannaj in ospedale, proi(/j) when Vanessa have-3sg visited Jean in hospital pro le ha portato un mazzo di rose. her-dat have-3sg taken a bunch of roses ‘When Vanessai visited Jeanj in the hospital, [shei(/j) ] took her a bunch of roses.’

(99)

a. Quando Vanessai ha visitato Giovannaj in ospedale, leii(/j) when Vanessa have-3sg visited Jean in hospital she era già fuori pericolo. be-3sg-past already out danger ‘When Vanessai went to visit Jeanj in the hospital, shei(/j) was already out of danger.’ b. Quando Vanessai ha visitato Giovannaj in ospedale, proi(/j) when Vanessa have-3sg visited Jean in hospital pro era già fuori pericolo. be-3sg-past already out danger ‘When Vanessai went to visit Jeanj in the hospital, [shei(/j) ] was already out of danger.’ (Adapted from Carminati 2002: 361)

124 | 5 Processing studies and L2 subject pronoun acquisition in Spanish The PAH predicts that the null pronouns in (98b) and (99b) will prefer to co-refer to Vanessa (the subordinate clause subject) in both cases and the overt pronoun in (98a) and (99a) will prefer to co-refer to Giovanna (the subordinate clause object); the pragmatic contexts, however, force an interpretation where the main clause pronoun, either null or overt, should co-refer to Vanessa in (98a) and (98b) and to Giovanna in (99a) and (99b). Carminati found that there was a considerable processing penalty, signaled by a delay in reading times, when the participants were pragmatically obligated to violate the PAH and co-refer a null subject in the main clause with a preceding object, more so than when they were forced to corefer an overt subject pronoun with a preceding subject. Carminati concludes that overriding the overt subject bias is less costly than violating the null subject bias and that because these violations do not result in grammatical anomalies but rather pragmatic ones, this is evidence that the PAH is motivated by universal pragmatic principles in terms of the accessibility and/or informativity of referring expressions (Ariel 1990) (see Chapter 6 of the present volume for a more complete discussion of these concepts). That is to say that if a language has null subjects, it is the form that is lowest on the informativity scale and thus will be the most likely candidate to co-refer to the most accessible antecedent. In syntactic terms, an antecedent in subject position is the most prominent and therefore most accessible. That inter- and intrasentential anaphora resolution in a pro-drop language such as Italian is predominantly dependent on the processing principles related to structural and pragmatic properties, specifically that the Spec IP position confers particular prominence on an antecedent, has been examined among Spanish monolinguals by Alonso-Ovalle, Fernández-Solera, Frazier, and Clifton (2002) and Perales and Portillo (2007).

5.1.2 The PAH in Spanish Alonso-Ovalle et al. (2002) designed a series of offline, self-paced written experiments to test the predictions of the PAH among native speakers of Spanish. Although their study did not replicate the tasks carried out in Carminati’s (2002) investigation for Italian, they did include many of the same syntactic contexts but minus the pragmatically forced interpretations. They included tasks that tested for anaphoric resolution for null and overt subjects in Spanish with: 1) intersentential (two sentence) discourse contexts with two referents, one in subject position (Spec IP) and one in direct object position (Compl) positing a potential for ambiguity; 2) intersentential discourse contexts with one referent, thus eliminating the possibility for ambiguity; 3) variable binding contexts (also know as OPC contexts) providing for the possibility of bound variable interpretation or an extra

5.1 Anaphoric resolution in pro-drop languages and the PAH |

125

sentential antecedent; 4) broad and narrow focus contexts; and 5) postverbal and preverbal (non-emphatic and emphatic) subjects in subordinate clauses, which were designed to test the effect of syntactic position of the pronoun regardless of topicality. Examples of intersentential discourse contexts with two referents are as follows: (100) a. Juan pegó a Pedro. pro Está enfadado. John hit-3sg-pret to Peter pro be-3sg angry ‘Juan hit Pedro. [He]’s angry.’ b. Juan pegó a Pedro. Él está enfadado. John hit-3sg-pret to Peter He be-3sg angry ‘Juan hit Pedro. He’s angry.’ (Adapted from Alonso-Ovalle et al. 2002: 3–4) Participants were subsequently asked who was angry (Juan or Pedro, for example). For the intersentential contexts with one referent, participants were asked to rate the naturalness of the null in (101a) or the overt pronoun in (101b) in the second sentence of discourses such as below: (101) a. Teresa llegó al aeropuerto tarde. pro Estaba Teresa arrive-3sg-pret to-the airport late pro be-3sg-imp cansada. tired ‘Teresa arrived at the airport late. [She] was tired.’ b. Teresa llegó al aeropuerto tarde. Ella estaba Teresa arrive-3sg-pret to-the airport late Ella be-3sg-imp cansada. tired ‘Teresa arrived at the airport late. She was tired.’ (Adapted from Alonso-Ovalle et al. 2002: 7) The OPC contexts contained sentences with quantifiers in subject position (102a) and (102b), which were followed by questions regarding the possibilities of coreference interpretation. Sentence (103a) indicates a disjoint interpretation (i.e. an antecedent outside the discourse) and (103b) expresses a bound variable interpretation.

126 | 5 Processing studies and L2 subject pronoun acquisition in Spanish (102) a. Ningún estudiante cree que pro pasó el examen. no student believe-3sg that pro pass-3sg-pret the exam ‘No student believes that [he] passed the exam. b. Ningún estudiante cree que él pasó el examen. no student believe-3sg that pro pass-3sg-pret the exam ‘No student believes that he passed the exam. (103) a. Hay una persona no mencionada de la que ningún estudiante exist-3sg a person no mentioned of the that no student cree que pro pasó el examen. believe-3sg that pro pass-3sg-pret the exam ‘There is a person not mentioned that no student believes [he] passed the exam.’ b. Ningún estudiante sabe que él mismo pasó el no student know-3sg that he himself pass-3sg-pret the examen. exam ‘No student knows that he himself passed the exam.’ (Adapted from Alonso-Ovalle et al. 2002: 8–9) Topic-focus contexts were included to test for the preference of preverbal subjects as sentential topics. Participants were presented with narrow and broad sentence contexts followed by two choices: postverbal and preverbal subjects. In (104), the narrow focus context would favor a postverbal subject response (Vino Juan), whereas in (105), the broad focus context would favor a preverbal subject response (Juan vino). (104) ¿Quién vino? who come-3sg-pret ‘Who came?’ A. Vino Juan. come-3sg-pret John ‘Came John.’ (105) ¿Qué sucedió con Juan? what happen-3sg-pret with John ‘What happened with John?

B. Juan vino. John come-3sg-pret ‘John came.’

5.1 Anaphoric resolution in pro-drop languages and the PAH |

A. Vino Juan. come-3sg-pret John ‘Came John.’

127

B. Juan vino. John come-3sg-pret ‘John came.’ (Adapted from Alonso-Ovalle et al. 2002: 11)

Finally, in a context not examined by Carminati (2002), the authors presented participants with sentences with postverbal and preverbal (non-emphatic and emphatic) subjects in subordinate clauses, which were designed to test for the effect of syntactic position of the pronoun regardless of topicality in (106–108). They wanted to see if the position of the subject pronoun and contrastive focus (108) would override the preferences outlined in the PAH. (106) Pedro piensa que está cansado él. Peter think-3sg that be-3sg tired he ‘Peter thinks that [the one that’s] tired is he.’ (107) Pedro piensa que él está cansado. Peter think-3sg that he be-3sg tired ‘Peter thinks that he is tired.’ (108) Pedro piensa que ÉL está cansado. Peter think-3sg that HE-emph be-3sg tired ‘Peter thinks that HE is tired. (Adapted from Alonso-Ovalle et al. 2002: 13) The sentences were followed by questions asking participants to identify to whom the pronoun referred (either Pedro or Una persona no mencionada ‘Someone not mentioned in the sentence.’). Overall, the authors found support for the PAH for Spanish in the contexts studied; that is, null subjects in the second clauses were preferable for linking to prominent antecedents in subject position [Spec, IP] of the first clauses independently of ambiguous contexts in intersentential contexts or quantifier subjects in [Spec, IP] position in intrasentential contexts. In the former context with ambiguous referents, the null pronoun was linked to the subject of the previous sentence more often (73.2 %) than the overt pronoun (only 50.2 %). In cases of unambiguous contexts, participants consistently rated the sentences with null subjects more natural than those with overt subject pronouns and the difference in ratings was highly significant. In intrasentential contexts, the PAH was better at predicting antecedent resolution than the OPC, which is purported to block a bound interpre-

128 | 5 Processing studies and L2 subject pronoun acquisition in Spanish tation for overt subjects. Null subjects were interpreted as bound variables 86.1 % of the time and overt pronouns received a bound interpretation 63.3 % of the time. Alonso-Ovalle et al. argue that since there is a majority of bound interpretations of overt pronouns for quantifier subjects, the OPC can not be considered a single grammatical principle, but rather should be subsumed under the PAH, which “applies regardless of the referential nature of the antecedent” (Alonso-Ovalle et al. 2002: 8). Finally, in their tests of the topic-focus articulation, they found that for appropriate responses to narrow focus questions, postverbal subjects (nonSpec, IP position) were preferred nearly 93 % of the time, whereas for broad focus questions they appeared postverbally in only 48.5 % of the cases. Postverbal overt pronouns (Pedro piensa que está cansado él [Alonso-Ovalle et al. 2002: 13]) were far less likely to receive a bound interpretation to the matrix referential subject than preverbal overt pronouns, either emphatic or not (Pedro piensa que ÉL/él está cansado [Alonso-Ovalle et al. 2002: 13]). The authors conclude, as does Carminati, that both the form of the pronoun (null and overt) and its syntactic position (Spec IP and non-Spec IP), which results from the topic-focus articulation, work together to resolve anaphoric preferences in Spanish. That Spanish and Italian adhere to the principles of the PAH is evident although it is difficult to compare the Alonso-Ovalle et al. and Carminati studies since the only context in which they overlap is the variable binding (the OPC) cases. Specifically, the Alonso-Ovalle et al. study did not include pragmatic contexts that favored co-reference with one or the other antecedents contra the PAH. In a subsequent study, Perales and Portillo (2007) set out to corroborate the findings of Alonso-Ovalle et al. (2002) and to test for the validity of the PAH in Spanish in guided written and spontaneous oral production tasks of native speakers from Spain. The written data were based on a series of pictures that were presented to speakers at the bachiller (high school) level who were asked to use the pictures to narrate a brief story. The authors subsequently identified all null and overt subject pronouns and analyzed their anaphoric function as either coreferring to a subject or complement antecedent. Although proportionally there were only 5.4 % overt pronouns (the authors did not include nominal subjects in their analysis), all were limited to the demonstrative pronoun, éste (‘this one’, ‘the latter’, the last masculine noun mentioned), which referred 100 % of the time to an object antecedent such as that presented in (109):

5.1 Anaphoric resolution in pro-drop languages and the PAH |

129

(109) pro Acude a un policía para que éste lo ayude a pro go-3sg to a police for that this one him-acc help-3sg-subj to encontrar a su pequeña. find-inf to his little one ‘[He] goes to the police so that this one [he] would help him find his little one.’ (Adapted from Perales and Portillo 2007: 894) The authors claim that this is support for the PAH but it is worth mentioning that the use of éste (‘this one’) is semantically determined as it is in other languages, and must co-refer with the last antecedent mentioned²⁰. The use of an overt personal subject pronoun (él ‘he’) is also syntactically possible in these contexts and the PAH would predict that the tendency is to co-ref such a pronoun to a non-subject antecedent; however, a personal subject pronoun could also corefer to an antecedent in subject position. The PAH only predicts tendencies, not absolutes; therefore the antecedent of an overt personal subject pronoun is potentially ambiguous. With the use of the overt demonstrative pronoun, ambiguity is eliminated – it can only refer to the last named antecedent, which almost always occupies a non-subject position. The authors’ analysis of the null subject pronouns did find support for the PAH; null subjects co-referred to an antecedent in subject position in 78.6 % of the cases, an outcome result remarkably similar to Alonso-Ovalle et al.’s findings for judgments of intersentential contexts with ambiguous referents where the null pronoun was linked to the subject of the previous sentence in 73.2 % of the cases. What is not mentioned in the study is whether the authors distinguished cases of both ambiguous and unambiguous subject pronoun reference in the same manner as did Alonso-Ovalle et al. (2002). To examine the same contexts in spontaneous oral production, Perales and Portillo (2007) consulted the CHILDES (MacWhinney 2000) data base and ana-

20 A demonstrative pronoun in the neutral gender can also refer to a previous verbal situation or an entire discourse situation. For example, (110)

Me dijo que perdió todo su dinero pero luego sacó la me-dat tell-3sg-pret that lose-3sg-pret all his money but later win-3sg-pret the lotería y recuperó todo; ésto no ocurre todos los días. lottery and recover-3sg-pret everything; this no happen-3sg all the days ‘He told me that he lost all his money but later won the lottery and got it all back; this doesn’t happen every day.’

130 | 5 Processing studies and L2 subject pronoun acquisition in Spanish lyzed data from native Spanish speakers of different ages (with an average age of approximately 20) who had narrated a story based on the picture story, Frog, Where are you? (Mayer 1969). The data were analyzed in the same manner as those from their own written production task and they found a similar proportion of null subjects (95.1) of which 76.1 % co-referred to an antecedent in subject position. As before, they found that 4.8 % of subject pronouns were overt but unlike the written data, they found a greater proportion (approximately 63 %) of these pronouns coreferred to an antecedent in a subject position, contrary to the predictions of the PAH. The authors suggest that it is not the anaphoric use that explains the overt subject pronoun, él ’he’, but rather a contrastive function to refer to a (presumably) subject antecedent not mentioned in the immediate discourse such as seen in (111): (111)

Porque de ese árbol sale un búho [. . . ] y bueno pues como él because from that tree exit-3sg an owl [. . . ] and well then as he estaba metiendo la nariz . . . [el niño, no el búho] be-3sg-imp sticking in the nose . . . [the boy not the owl] ‘Because from this tree an owl came out . . . and well, er, since he was sticking his nose in . . . [the boy, not the owl] (author’s translation) (Adapted from Perales and Portillo, 2007: 898)

In this case, the subject of the first clause is the owl and the subject of the second clause is ‘the boy’, presumably a subject mentioned prior in the discourse but not a complement antecedent in the immediate discourse. The results of this study reveal that although the PAH appears to be a strong predictor for overt and null subject pronoun use in spontaneous oral and guided written production among native Spanish speakers, they also highlight the complexity of pronoun use in null subject languages and the difficulty in analyzing natural speech and writing where contexts are not controlled and many more pragmatic functions intervene to determine pronoun use. Furthermore, the study confirms, as does the AlonsoOvalle et al. (2002) study, that the PAH may be valid for explaining anaphora resolution for null subjects in Spanish. However, in contrast to the findings of the Carminati (2002) study for Italian, it appears that more than just the processing principles related to structural properties and the pragmatic principle related to subject prominence is at play in the antecedent resolution of overt subject pronouns in Spanish. These studies confirmed the tendencies of the PAH when pronoun form, syntactic structure and pragmatic context all favor the linking of a null subject pronoun to a subject antecedent and an overt pronoun to a non-subject antecedent. It is difficult, however, to make comparisons for Italian and Spanish

5.1 Anaphoric resolution in pro-drop languages and the PAH |

131

because neither the Alonso-Ovalle et al. nor the Perales and Portillo studies replicated the tasks that Carminati used with her Italian participants. Neither study included tasks that tested for antecedent resolution with pragmatic contexts that override the preferences of the PAH; that is, they lacked the types of tasks that introduce both subject and object antecedents that match in gender and number and where the pragmatic context favors co-reference of the pronoun with one or the other antecedents, regardless of its null or overt form.

5.1.3 The PAH in Spanish and Italian Filiaci (2010) did set out to replicate Carminati’s (2002) study and to test the strength of the PAH for Spanish in monolingual speakers. Although the two languages are similar typologically, Filiaci claims it is important to investigate whether the same pragmatic principles of the PAH that govern the interpretation and distribution of overt and null pronominal subjects in Italian are valid for Spanish as well. In fact, she hypothesizes that overt subject pronouns in Spanish are more flexible (“free(er)” is her term) to select both the prominent and nonprominent antecedents of the second clause of the PAH. There is less of a “division of labor” in Spanish where both null and overt pronouns may link to the subject of the prior clause. In other words, whereas a significant processing penalty is incurred in Italian when an overt pronoun in the second clause is forced to corefer to an antecedent in subject position in the preceding clause, in Spanish no such penalty (in terms of increased processing times) occurs. Sixty-four monolingual adults participated in the study – half were Italian speakers and half were Spanish speakers – and completed the same tasks as those used by Carminati, equivalent versions in both Italian and Spanish. Like the Carminati study, tasks were designed to force a co-reference interpretation with the subject or object of the preceding clause with either an overt subject pronoun or a null subject. Participants were presented with items such as those in Spanish in (112) and (113) below where the two possible antecedents were of the same gender and number. (112)

a. Cuando Anai visitó a Maríaj en el hospital, ellai/j le when Ann visit-3sg-pret to Mary in the hospital she her-dat llevó un ramo de rosas. take-3sg-pret a bunch of roses ‘When Anai visited Maryj in the hospital, shei/j brought her a bunch of roses.’

132 | 5 Processing studies and L2 subject pronoun acquisition in Spanish b. Cuando Anai visitó a Maríaj en el hospital, proi/j le when Ann visit-3sg-pret to Mary in the hospital pro her-dat llevó un ramo de rosas. take-3sg-pret a bunch of roses ‘When Anai visited Maryj in the hospital, [she]i/j brought her a bunch of roses.’ (113)

a. Cuando Anai visitó a Maríaj en el hospital, ellai/j ya when Ann visit-3sg-pret to Mary in the hospital she already estaba fuera de peligro. be-3sg-imp out of danger ‘When Anai visited Maryj in the hospital, shei/j was already out of danger.’ b. Cuando Anai visitó a Maríaj en el hospital, proi/j ya when Ann visit-3sg-pret to Mary in the hospital pro already estaba fuera de peligro. be-3sg-imp out of danger ‘When Anai visited Maryj in the hospital, [she]i/j was already out of danger.’ (Adapted from Filiaci 2010: 176)

Although the subject of the second clause in all four sentences is ambiguous, the PAH predicts that null subject pronouns prefer to co-refer to the antecedent subject and overt subject pronouns are more likely co-refer to the antecedent object. However, the context in (112) forces a co-reference interpretation of the subject pronoun, regardless of whether it is null or overt, to the antecedent subject (because people usually take flowers to others when they visit them in the hospital and not the other way around). In (113), the context forces a co-reference interpretation of the subject pronoun to the antecedent object regardless of the form of the pronoun. Participants were asked to identify the antecedent of the anaphoric subject and reading times for each clause were recorded for the self-paced task. Longer reading times indicate a processing penalty for linking anaphors to their proper antecedents. Results revealed that the Italian speakers performed as predicted as they had in the Carminati (2002) study in support of the PAH; that is, there was a significant reading delay when the null subject was semantically forced to corefer with the antecedent object and when the overt subject was forced to corefer with the antecedent subject (p. 177). The Spanish speakers also revealed a significant delay when forced to link a null subject with an object antecedent; however, for the

5.1 Anaphoric resolution in pro-drop languages and the PAH |

133

opposite context, there was no significant delay for linking the overt subject pronoun to its intended subject antecedent. In short, what Filiaci found is that for both Italian and Spanish the PAH is in effect for the interpretation of null subjects but for the interpretation of overt subject pronouns, the PAH is in effect only for Italian. For Italian speakers there is a significant processing penalty for interpreting overt subjects that go against the bias predicted by the PAH but for Spanish speakers it is easier to retrieve the antecedent in subject position regardless of a null or overt pronoun form. These are important cross-linguistic pragmatic differences concerning the resolution of overt and null subject pronouns in Italian and Spanish monolingual speakers (Filiaci 2010: 179). Filiaci is careful to point out that the PAH does not predict ungrammaticality but rather pragmatic preferences. Furthermore, she states: while Italian OSs (overt subjects) seem to bear a strong ‘Switch Reference’ connotation, indicating to the addressee a switch of reference from the preceding subject, Spanish OSs seem not to carry such a strong connotation and be more compatible with a ‘non-Switch Reference’ reading. So, while in Italian both NS and OS are specialised in retrieving different antecedents, in Spanish only NSs (null subjects) seem to obey a bias and we may argue that this bias alone could be responsible for the patterns of distributions found in corpus studies. (Filiaci 2010: 180)

5.1.4 Summary of effects of the PAH in pro-drop languages The anaphoric resolution of null and overt subject pronouns appears to be guided by the PAH for Italian and thus, when both types of pronouns violate the hypothesis, there is a considerable processing penalty (Carminati 2002; Filiaci 2010), although Carminati did find that the antecedent bias for overt pronouns is somewhat less established and depends more on context than it does for null subjects. For Spanish, however, the evidence suggests that the processing penalty is far stronger for null pronouns than it is for overt pronouns that violate the PAH (Alonso-Ovalle; 2002; Filiaci 2010). As Filiaci points out, Spanish is more flexible in linking overt subject pronouns to prominent (subject) and non-prominent (object) antecedents. This is confirmed by the oral production data examined in the Perales and Portillo (2007) study where the majority of the overt subject pronouns used referred to a subject antecedent although albeit to a lesser extent than null subjects. The difference in adherence to the PAH between Spanish and Italian is subtle but clearly evident and may account for some of the differences seen in the bilingual children’s performances discussed in chapter four of this volume. A summary of the findings in support of the PAH is displayed in Table 13.

134 | 5 Processing studies and L2 subject pronoun acquisition in Spanish Table 13. Monolingual studies testing the Position of Antecedent Hypothesis (PAH). Study

Null subject favors subject antecedent

Overt subject favors non-subject antecedent

OPC effects

Pragmatic context overrides PAH – with processing penalty

Carminati (2002)

Yes

Yes

No

Yes

PAH less established

PAH greater predictor

Overriding overt bias less costly

Yes/No

Yes/No

NA

Overt pronouns evenly split between subj. and non-subj. antecedents

PAH greater predictor

Yes – written data

NA

NA

NA

Ital. – Yes

Italian Alonso-Ovalle et al. (2002)

Yes

Spanish

Perales and Portillo (2007)

Yes

No – oral data

Spanish Filiaci (2010)

Yes

Spanish and Italian

Ital. – Yes Span. – Yes/No

Span. – no significant cost for overriding overt bias

NA = not applicable (not tested)

There is evidence in these studies that the PAH is in effect for anaphoric resolution of null subjects in both languages; however, the results for the PAH for overt subject pronouns is mixed. The bias is less strong, even for Italian, although the Italian participants in the Carminati and Filiaci studies revealed that the PAH is less established for linking overt pronouns to non-subject antecedents. The data for Spanish speakers reveals an even weaker bias for the anaphoric properties of overt subject pronouns. If these cross-linguistic differences exist, then there would be further evidence in bilingual and L2 speakers of both pro-drop and non-pro-drop languages. We would expect bilingual and L2 speakers of two prodrop languages (Italian and Spanish) to exhibit differences in pronoun resolution that are distinct from monolingual speakers of either of these languages and from bilingual and L2 speakers of one pro-drop language (either Spanish or Italian) and one non-pro-drop language (English). The following section is dedicated to examining and discussing the emerging research regarding bilingual and L2 speakers’ strategies for linking subject pronouns to antecedents.

5.2 Anaphoric resolution in bilingual and L2 acquisition

|

135

5.2 Anaphoric resolution strategies in bilingual and L2 acquisition (pro-drop and non-pro-drop languages) Studies of bilingual children and adults and L2 adult learners of null-subject languages such as Italian and Spanish have confirmed that there is a tendency to link null pronouns to the subject antecedent and overt pronouns to non-subject antecedents; whereas in the acquisition of non-null-subject²¹ languages, such as English, the tendency is to interpret overt pronouns with subject antecedents. Thus in bilingual speakers and adult L2 learners with two different parameter settings for their languages, there is the potential for a competition of strategies. Sorace and Filiaci (2006) have suggested that processing difficulties at the syntaxdiscourse interface may explain not just residual deficits among early and late bilingual speakers (i.e. L2 learners), but may help to explain L1 attrition and cross-linguistic influence among fully competent bilingual speakers when one of their native languages is a non-null-subject language (Sorace and Filiaci 2006: 345–346).

5.2.1 Strategies of anaphoric resolution in Italian L2 In their study of monolingual and near-native L2 speakers of Italian (with English as their L1), Sorace and Filiaci (2006) sought to determine whether or not the difficulties L2 speakers have acquiring and using null and overt subject pronouns results from underspecification of L2 knowledge representations or from “insufficient processing resources to integrate the multiple types of information involved at the interfaces between syntax and other cognitive domains” (2006: 340). The authors suggest that there may be evidence in favor of the latter. This is because the non-null-subject language has a more ‘economical’ system (206: 345) and does not rely on discourse-pragmatic factors to constrain the distribution of two types of subject pronouns (null and overt). Thus when speakers are confronted with different strategies from their two languages, the more economical and efficient strategy (from the non-null-subject language) will become the default strategy. Two groups of speakers participated in the study: monolingual speakers of Italian and near-native L2 speakers of Italian whose L1 was English. Both groups took part in a picture verification task that included both production and comprehension of null and overt subject pronouns and preverbal and postverbal nom-

21 The term, non-null-subject, is essentially the same as the term, non-pro-drop, used throughout this volume. Here, I switch to the former term, used in the Sorace and Filiaci (2006) study.

136 | 5 Processing studies and L2 subject pronoun acquisition in Spanish inal subjects in forward and backward anaphora. All sentences included main clauses that contained an animate subject noun phrase and an animate object noun phrase of the same gender and number and a subordinate clause with a subject pronoun (either null or overt) that could refer to either the subject or object of the main clause (or, in the case of the overt pronoun, to a referent outside the discourse). In forward anaphora the main clause introducing the two referents is first and the subordinate clause with the pronoun is second as in (114); in backward anaphora the subordinate clause is first and the main clause is second as in (115). (114) La mammai dà un bacio allá figliak mentre leik/l /proi si the mother give-3sg a kiss to-the daughter while she/pro refl-put mette il cappotto. on-3sg the coat ‘The motheri gives a kiss to her daughterk while shek/l /proi puts on her coat.’ (115)

Mentre leik/l /proi si mette il cappotto, la mammai dà un while she/pro refl-put on-3sg the coat the mother give-3sg a bacio allá figliak . kiss to-the daughter ‘While shek/l /proi puts on her coat, the motheri gives a kiss to her daughterk .’ (Adapted from Sorace and Filiaci 2006: 352)

In both sentences the PAH would predict that the overt pronoun, lei ‘she’, refer either to a non-subject referent, in this case, the object referent, figlia ‘daughter’, or another referent outside the discourse and the null pronoun to the subject referent, mamma ‘mother’. Participants were presented with 20 experimental sentences (as well as 15 distractors) that were accompanied by a set of pictures portraying the different possibilities for interpretation. For the sentences (114) and (115), the accompanying pictures are as seen in Figure 2. The authors predicted that there would be more differences between the monolingual and near-native speakers with the interpretation of overt pronouns than with the interpretation of null subjects, and that the two groups would show more differences in their interpretations of backward anaphora than forward anaphora. Specifically, they had predicted that the cases of backward anaphora would pose more processing difficulties, especially for the English speakers, because the pronoun precedes its potential referents and there are

5.2 Anaphoric resolution in bilingual and L2 acquisition

|

137

La mamma dà un bacio alla figlia, mentre si mette il cappotto.

1

2

3

Figure 2. Forward and backward anaphora picture interpretation task. (Sorace and Filiaci 2006: 366)²²

no discourse constraints on the assignment of antecedents. Results revealed that both groups preferred non-subject antecedents with overt pronouns in forward anaphora contexts, although the near-natives to a lesser extent than the monolinguals, and subject antecedents with null pronouns in backward anaphora following the bias predicted by the PAH. For null subjects in forward anaphora, however, both groups were fairly evenly split between selecting subject and non-subject antecedents. The authors explain that this may be due to the pragmatic plausibility, topicality and accessibility of the object referent, which all worked together to override the bias of the PAH. Thus their findings confirm that near-native speakers have acquired target-like strategies in the interpretation of null pronouns in both forward and backward anaphora. However, for overt pronouns in backward anaphora, native speakers preferred antecedents outside of the discourse context whereas near-native speakers preferred to a significant extent subject antecedents, a processing strategy much like that of their English L1. Sorace and Filiaci concluded that although the near-native speakers had acquired both the null-subject grammar (i.e. the licensing of null subjects) and the discourse constraints that govern null and overt subjects (the discoursepragmatic interface), they still did not possess the necessary processing resources in order to reliably integrate several sources of linguistic information for more native-like outcomes. Although the effects of the PAH tendencies are in tact for the monolingual speakers and for the near-native speakers for null pronouns, the cost of overcoming the PAH bias for backward anaphora is too great and these speakers defer to the L1 for a default processing strategy in assigning a

22 Image by Antonella Sorace and Francesca Filiaci, Second Language Research 22,3; pp. 339– 368, copyright © 2006 by Sage Publications; Reprinted by Permission of SAGE.

138 | 5 Processing studies and L2 subject pronoun acquisition in Spanish subject referent to an overt pronoun rather than searching outside the discourse context for a suitable referent. Another explanation may have to do with universal discourse structure and the notion of topic continuity; in the absence of specific clues for anaphoric interpretation other than the form of the pronoun and its place in the syntactic ordering of elements, the near-native speakers may have simply relied on the pragmatic strategy of assuming topic continuity. Regardless of whether the first subject mentioned is null or overt in the backward anaphora cases, speakers assume the topic is continued and thus link it to the subject in the succeeding clause.

5.2.2 Strategies of anaphoric resolution in Italian-English and Italian-Spanish bilinguals Sorace, Serratrice, Filiaci, and Baldo (2009)²³ also found that both English-Italian and Spanish-Italian bilinguals prefer a more economical, less taxing processing strategy for the interpretation of overt subject pronouns. Testing the effects of cross-linguistic influence on the acceptability of pronominal subjects in both topic continuity (non-topic-shift) and topic shift contexts, the authors extended their previous work to monolingual (English and Italian) children and adults and bilingual (English-Italian and Spanish-Italian) children. They had predicted that there would be more cross-linguistic interference from English than from Spanish for the bilingual children. They found that both the monolingual and bilingual children, regardless of first language, were more likely than the monolingual adults to accept overt pronouns in discourse contexts where they are redundant (i.e. non-topic-shift). However, both groups of bilingual children also accepted more null subjects in contexts that required an overt pronoun in Italian (i.e. topic-shift), although this latter tendency stabilized more quickly than the former and more quickly in the Spanish-Italian bilingual group. The authors suggest, “being bilingual leads to greater difficulty in the acquisition of the discourse conditions governing the distribution of subject pronouns” (Sorace et al. 2009: 473). In other words, in the case of the English-Italian children there is influence from the non-pro-drop language but this is not necessarily unidirectional nor due to the quality and quantity of the input received. Rather, the children select the less taxing interpretation strategy of the non-pro-drop language (English), which has a more economical syntax-pragmatics interface system. For the SpanishItalian children there may also have been crosslinguistic influence due to the

23 See Chapter 4 of this volume for a more complete discussion of this study.

5.3 Strategies of anaphoric resolution in Spanish heritage and L2 speakers |

139

more flexible nature of assigning overt pronouns to non-topic-shift contexts in Spanish than in Italian. In both cases, the bilingual processing strategies lead to persistent, although not ungrammatical, pragmatically inappropriate pronoun use. The evidence, nonetheless, suggests that by adulthood bilinguals’ strategies for anaphoric resolution stabilize for both the pro-drop and non-prodrop languages. The question that arises is whether or not early bilinguals have an advantage over adult L2 learners for interpreting null and overt subject pronouns in the pro-drop language and to what extent they develop independent strategies or a combination of strategies for anaphoric resolution in the two languages. Recent work in Spanish L2 acquisition addresses these issues and is the topic of the subsequent section.

5.3 Strategies of anaphoric resolution in Spanish heritage and L2 speakers 5.3.1 PAH biases in early (heritage) and late (L2) bilingual speakers One major L2 study (Keating, VanPatten, and Jegerski 2011) seeks to investigate whether the quantity and quality of input gives bilinguals an advantage over the L2 learners when resolving intrasentential anaphora in Spanish and examines to what extent the participants’ L1 influences their behavior when interpreting null and overt pronouns. Keating, VanPatten, and Jegerski’s study sought to investigate the different strategies for the resolution of intrasentential anaphora in Spanish among three groups of speakers, early Spanish-English bilinguals or heritage speakers, late adult English-Spanish bilinguals (L2 learners), and monolingualraised Spanish speakers. Specifically, the authors asked whether or not early exposure to Spanish input among the heritage speakers would confer upon them an advantage over the L2 learners for acquiring the grammatical properties of the syntax-discourse interface of pronominal subject reference. Because research has shown that speakers of various languages use different strategies for anaphora resolution in ambiguous contexts when more than one antecedent is available (Keating, VanPatten, and Jegerski 2011: 195–196), the authors expected to discover whether or not optionality of pronoun anaphora was persistent at advanced levels and if early exposure to input helped to resolve this optionality. The authors wanted to investigate whether or not the three groups of speakers studied would prefer a strategy based on Carminati’s (2002) PAH for pro-drop languages, or on a subject assignment strategy as in English. For example, in the following intrasentential contexts,

140 | 5 Processing studies and L2 subject pronoun acquisition in Spanish (116) Juan vio a Carlos cuando pro caminaba en la playa. John see-3sg-pret to Charles when pro walk-3sg-imp on the beach ‘John saw Charles when [he] was walking on the beach.’ (117)

Juan vio a Carlos cuando él caminaba en la playa. John see-3sg-pret to Charles when he walk-3sg-imp on the beach ‘John saw Charles when he was walking on the beach.’ (Adapted from Keating, VanPatten, and Jegerski 2011: 197)

once again the PAH would predict that Spanish speakers would link the null pronoun in (116) with Juan, the subject antecedent, and to associate the overt pronoun in (117) with Carlos, the object antecedent (Keating, VanPatten, and Jegerski 2011: 197). Speakers of English, on the other hand, have been shown to prefer a subject assignment strategy, which links the overt pronoun to the subject antecedent (Keating, VanPatten, and Jegerski 2011: 198). It would be expected that English-speaking learners of L2 Spanish link both the null and the overt pronouns in (116) and (117) with Juan, the subject antecedent. Thus research has shown that there is a supposed universal strategy, the subject rule. This means that for nullsubject languages like Spanish and Italian, speakers will link the null pronoun to the most discourse-prominent antecedent, i.e. the sentential subject, and nonnull-subject languages, like English, will do the same for unstressed overt pronouns. Where languages differ is with the coreference of overt subject pronouns (Keating, VanPatten, and Jegerski 2011: 199). Therefore, bilinguals of both types of languages will have two competing strategies. The research on English-Italian and Spanish-Italian bilinguals revealed few differences on the resolution of null pronouns but found that all bilingual groups were more likely to link overt pronouns with subject antecedents than the monolinguals (Keating, VanPatten, and Jegerski 2011: 199–200). These studies are in accordance with L2 Spanish studies that have observed pragmatically inappropriate uses of overt pronouns in samereferent contexts (Montrul, 2004b; Montrul and Rodríguez Louro 2006; Rothman 2007, 2009; Silva-Corvalán 1994). Thus there is evidence that can provide some clues as to how English-Spanish bilinguals, both early and late, would resolve these strategies. Specifically, Keating, VanPatten, and Jegerski hypothesized that, because the subject rule and the PAH make the same prediction for null subjects, heritage speakers and L2 learners will behave in the same way when faced with a null subject pronoun. When confronting an overt pronoun in Spanish, because the subject rule and PAH conflict, they expected both groups to display a bias toward linking the overt pronoun with the subject antecedent, a distinctly English strat-

5.3 Strategies of anaphoric resolution in Spanish heritage and L2 speakers |

141

egy. Based on previous research and because of the heritage speakers’ contact and dominance of English, they did not expect these speakers to have an advantage over the L2 learners (Keating, VanPatten, and Jegerski 2011: 203–204). Their participants included 19 monolingual-raised Spanish speakers from various countries (excluding Caribbean countries where overt subject pronouns are more frequent), 25 Spanish heritage speakers of Mexican descent, and 19 L2 Spanish learners all of whom were living in the United States at the time of the study and were either professionals or pursing professional degrees not in linguistics-related fields. Participants were asked to complete two tasks: (1) the Language Experience and Proficiency Questionnaire (LEAP-Q) (Marian, Blumenfeld, and Kaushanskaya 2007, cited in Keating, VanPatten, and Jegerski 2011: 205), which elicits self-rated language proficiencies and information about current language experience; and (2) an offline sentence comprehension task containing 40 test and 40 distractor sentences. The test items were globally ambiguous complex sentences that consisted of a main clause introducing two potential referents of the same gender and number, followed by a subordinate clause headed by a temporal conjunction such as when, after, (ever) since, and while. Twenty of the subordinate clauses for the test items contained a null pronoun and twenty had an overt pronoun (él ‘he’ or ella ‘she’). The following are sample test items: (118) Daniel ya no ve a Miguel desde que pro Daniel anymore no see-3sg to Michael since that pro se casó. refl-marry-3sg-pret ‘Daniel no longer sees Michael ever since [he] got married.’ (119) Alicia se encontró con Elena mientras ella corría en el Alice refl-find-3sg-pret with Ellen while she run-3sg-imp in the parque esta mañana. park this morning ‘Alice ran into Ellen while she was running in the park this morning.’ (Adapted from Keating, VanPatten, and Jegerski 2011: 208) General results found that there was a statistically significant difference in the interpretation of antecedent preference between null and overt pronouns only for the monolingual-raised Spanish speakers who selected the subject [Spec, IP] as the antecedent of null subjects almost 73.68 % of the time compared to 53.79 % for overt pronouns. There was no statistically significant difference between the two

142 | 5 Processing studies and L2 subject pronoun acquisition in Spanish choices for the heritage speakers (who linked subject antecedents to null subjects 65.64 % of the time and to overt pronouns 67.12 %) and L2 learners (who preferred subject antecedents for nulls 60.15 % and for overts 54.21 %). Interestingly, the heritage speaker group was more likely than the L2 learners to link an overt pronoun with a subject antecedent, although a subsequent division among the heritage speakers (based on the LEAP-Q) between Spanish-dominant and Englishdominant speakers found that the Spanish-dominant heritage speakers were more likely than the English-dominant heritage speakers to link a null subject with a subject antecedent (Keating, VanPatten, and Jegerski 2011: 211); however, they were also more likely to link an overt pronoun with a subject antecedent as well. Thus the authors concluded that language competency, at least in terms of self-reporting, did not influence the heritage speakers’ preferences. Table 14. Percentage selection of subject [Spec, IP] as antecedent of null and overt pronouns. Group Native speakers (n = 19) Heritage speakers (n = 25) L2 learners (n = 19)

Null

Overt

73.68 (19.71) 65.64 (21.37) 60.15 (20.42)

53.79 (29.66) 67.12 (17.70) 54.21 (22.44)

Note: Standard deviations in parentheses. (Adapted from Keating, VanPatten, and Jegerski 2011)

Despite the fact that there was a statistically significant bias in antecedent assignment for the native speaker group, the authors could not corroborate the same kind of clear complementary distribution for null and overt pronouns as Carminati (2002) found for Italian (Keating, VanPatten, and Jegerski 2011: 213).²⁴ The authors explain that this difference may be due to the fact that the native speakers were influenced by English, the language of their host country. Another and more likely explanation is that this difference is due to micro variations across null-subject languages for the pro-drop parameter. “Spanish and Italian are often assumed to pattern similarly with respect to the distribution of null and overt pronouns, but some recent experimental evidence suggests this may not be the case”

24 Although Carminati (2002) found a robust preference to link null pronouns to subject antecedents she also found more flexibility in linking overt pronouns to both subject and nonsubject antecedents. Sorace and Filiaci (2006) suggest there is a lower processing cost for an overt subject to be linked to a subject antecedent than for a null subject to take a non-subject antecedent. This may explain the Spanish monolinguals’ choices here as well.

5.3 Strategies of anaphoric resolution in Spanish heritage and L2 speakers |

143

(Keating, VanPatten, and Jegerski 2011: 214).²⁵ In fact, the native speakers in the Keating, VanPatten, and Jegerski study perform very much like the monolingual Spanish speakers of the Alonso-Ovalle et al. (2002) and Perales and Portillo (2007) studies who were all living in Spain. Moreover, the oral data examined in the Perales and Portillo study show less of a differentiation for anaphoric resolution of null subject pronouns where, it is to be recalled, 76.1 % of the null subjects were linked to subject antecedents and 63 % to object antecedents²⁶. There is clearly not the division of labor that was witnessed in the Carminati (2002) study for Italian. Nonetheless, there appears to be more influence from English for the heritage speakers than for the native Spanish speakers living in the U.S. Thus, it can be concluded that because Spanish does not clearly divide the functions for overt subject pronouns as it does for null subjects, there is more room for English strategies to take hold and persist among bilinguals who have developed English and Spanish from an early age. Of interest for language acquisition and bilingualism are the results for the heritage speakers who did not interpret the overt pronouns as +topic shift (switch reference) in Spanish. Whereas a null pronoun in switch-reference contexts would result in ambiguity, an overt pronoun in same-reference contexts would result only in redundancy, and these heritage speakers prefer to avoid ambiguity.²⁷ Despite abundant naturalistic input from an early age, when confronted with two conflicting subject assignment strategies for overt pronouns, these speakers appear to adopt the subject rule for English rather than sort through the two systems for resolving anaphoric reference. As for the L2 learners, for a while at least null and overt pronouns appear in a type of free variation as they sort out the pragmatic rules that constrain subject use in Spanish (Keating, VanPatten, and Jegerski 2011: 215). The authors conclude that heritage speakers and L2 learners “differ from each other and from monolingual raised Spanish speakers and that early input and language use do not confer advantages to heritage speakers relative to L2 learners in this particular domain” (2011: 96) perhaps because “properties situated at the syntax-discourse interface are particularly vulnerable to persistent and poten-

25 See section 5.1 of this chapter for a detailed discussion of the differences between anaphoric resolution resolution and PAH biases in Italian and Spanish. 26 Although it is to be remembered there were very few instances of overt subject pronouns (4.8 %) in the spontaneous oral production data examined. 27 Indeed, both Carminati (2002) and Sorace and Filiaci (2006) suggest that the flexibility with overt pronouns and antecedent assignment in native speakers and the over-production of overt subjects attested in bilingual speakers and L2 learners may be due to the same ‘Avoid Miscommunication’ principle (Sorace and Filiaci 2006: 348–349).

144 | 5 Processing studies and L2 subject pronoun acquisition in Spanish tially permanent optionality in bilingual grammars” (2011: 194). As Sorace et al. (2009) discovered with bilingual children of two pro-drop languages, there appears to be a persistent preference for a more efficient, less strenuous processing strategy that, although may result in perhaps a pragmatically unusual system, is not syntactically deviant. Thus it is difficult to tease apart to what extent this preference is due to cross-linguistic differences, universal processing strategies or some other factors related to discourse structure. In the following sub-section we discuss research designed to examine how discourse structure may effect antecedent assignment of null and overt subject pronouns in monolingual Spanish speakers and L2 learners.

5.3.2 PAH biases and discourse structure in L2 learners Jegerski, VanPatten, and Keating (2011) examined the interpretation strategies of both native Spanish speakers and English-speaking L2 learners of Spanish in order to pinpoint possible cross-linguistic influence (2011: 483). As in their previous study, this research also focused on intrasentential anaphora in complex sentences, but in addition asked to what extent the discourse relation between two clauses affects pronominal reference in a group of L2 learners. The study is based on both Carminati’s (2002) PAH and Asher and Vieu (2005)’s Segmented Discourse Representation Theory (SDRT). SDRT is a hierarchical model of discourse structure that specifies the relations between a series of sentential or clausal constituents (Jegerski, VanPatten, and Keating 2011: 484). In this model, discourse coordination and subordination (not to be confused with syntactic coordination and subordination), are seen as fundamentally different and influence how new information is integrated into discourse which, in turn, affects processes such as reference assignment (Jegerski, VanPatten, and Keating 2011: 485). Specifically, discourse coordination evokes parallel events as in (120a) and topic continuity; whereas discourse subordination evokes subordinating events with the possibility of topic shift as in (120b): (120) Coordinating events: a. Jeffrey saw Ricky while he was hunting for coins in the fountain. Subordinating events: b. Anita talked to her sister after she had the baby. (Jegerski, VanPatten, and Keating 2011: 489)

5.3 Strategies of anaphoric resolution in Spanish heritage and L2 speakers |

145

SDRT would predict that speakers interpret the two clauses in (120a) as having the same topic and therefore, the pronoun he of the second clause should be interpreted as co-referential with the subject antecedent Jeffrey of the first clause. On the other hand, the subordinated discourse event in (120b) would be more likely to be interpreted as a change of topic, making the she of the second clause conceivably more co-referential with the non-subject antecedent her sister. The questions that guided the Jegerski, VanPatten, and Keating study included asking how the language learner of a non-pro-drop language deals with anaphora resolution in the non-native pro-drop language; whether or not nativelike L2 strategy can be acquired over time; and finally, where behavior is divergent from the native target, if there is there any similarity to L1 behavior (2011: 488). The authors conducted four experimental tasks with 43 university students, all English-speaking L2 learners of Spanish, and 26 native speakers from a variety of Spanish-speaking countries, excluding the Caribbean region. All participants were undergraduate or graduate students or professionals at U.S. public universities. The first experiment was conducted only with the English speakers and was designed to see if coordinate and subordinate discourse structure affected their interpretation of overt pronouns in intrasentential contexts in their L1 English. Test sentences included stimuli with ambiguous pronouns in the second clause, half of which displayed a discourse coordination relationship and half of which expressed a discourse subordination relationship, such as those presented in (120a–b). Sentences were then followed by a binary choice interpretation question; for example, following a sentence such as (120b), the choices were presented as in (121). (121)

Who had a baby? a. Anita b. Her sister (Jegerski, VanPatten, and Keating 2011: 490)

The results revealed that participants selected an antecedent in the subject position (SpecIP) more often for the coordinate relation (64.19 % of the time) than for the subordinate relation (53.26 % of the time). Subsequent binominal tests revealed that the results for the coordination stimuli was significant (p < .001), whereas for the subordination stimuli they were only marginally significant (p = .091). The findings establish how these participants interpret ambiguous pronouns in their L1 and provide support for the SDRT, which predicts, “coordinate and subordinate discourse structure influenced intrasentential reference assignment for ambiguous pronouns in English” (Jegerski, VanPatten, and Keating 2011: 491).

146 | 5 Processing studies and L2 subject pronoun acquisition in Spanish Table 15. Percentage of overt subject pronouns interpreted as subject antecedent in English according to discourse structure – English speakers. Coordinate relation English speakers (N = 43)

Subordinate relation

64.19 %

53.26 %

(Adapted from Jegerski, VanPatten, and Keating 2011: 490)

The second experiment was the same as the first except the task was in Spanish, included both null and overt ambiguous pronouns, and only Spanish native speakers participated. This experiment was conducted in order to establish how native speakers interpret ambiguous pronouns in both coordinate and subordinate discourse contexts in Spanish. For null pronouns, participants selected Spec IP antecedents approximately 70 % of the time in discourse subordinating contexts and almost 73 % of the time in discourse coordinating contexts. For overt pronouns, in subordinating and coordinating contexts SpecIP antecedents were selected approximately 51 % and 52 % of the time respectively. Thus, results showed a main effect for pronoun use but none for discourse structure, and no significant interaction between the two factors. Whether the discourse was coordinate or subordinate did not influence referential ambiguity but type of pronoun, null or overt, did. There is a clear preference for linking null pronouns to subject antecedents but the preference for interpreting the antecedents of overt pronouns is fairly evenly distributed between subject and complement antecedents. The differences in the L1 English and L1 Spanish results suggest that syntactic principles (i.e. the null/overt distinction) apply to reference assignment for subject pronouns in Spanish and discourse structure is less important; whereas in English with two referents of the same number and gender, the processor must rely on discourse principles alone for interpreting ambiguous antecedents. For the third and fourth experiments, two L2 groups (intermediate and advanced) completed the Spanish task used in the second experiment. Results revealed that for null pronouns, intermediate participants selected Spec IP antecedents approximately 55 % of the time in discourse subordinating contexts and Table 16. Percentage of null and overt subject pronouns interpreted as subject antecedent in Spanish according to discourse structure – Spanish speakers. Spanish speakers (N = 26) Null Overt

Coordinate relation

Subordinate relation

73 % 50 %

70 % 51 %

(Adapted from Jegerski, VanPatten, and Keating 2011: 492)

5.3 Strategies of anaphoric resolution in Spanish heritage and L2 speakers |

147

slightly more than 65 % of the time in discourse coordinating contexts. For overt pronouns, in subordinating and coordinating contexts Spec IP antecedents were selected 50 % and 60 % of the time respectively. This finding for the intermediate group revealed that the factor of discourse structure was marginally significant overall and significant for certain items. More importantly, for type of pronoun there was no significant main effect; and there was no interaction for the two factors combined. There is a slight tendency to link null and overt pronouns to subject antecedents in coordinate relations but anaphoric assignment in subordinate relations is fairly evenly distributed between null and overt pronouns. These outcomes demonstrate that for these learners anaphora resolution was not influenced by the pronoun used (overt vs. null) but is more likely influenced by the kind of discourse structure involved (discourse coordination or subordination), which is the same strategy used in their L1. Table 17. Percentage of null and overt subject pronouns interpreted as subject antecedent in Spanish according to discourse structure – L2 learners. Coordinate relation

Subordinate relation

Intermediate group (N = 22) Null Overt

66 % 60 %

55 % 50 %

Advanced group (N = 21) Null Overt

70 % 57 %

52 % 55 %

(Adapted from Jegerski, VanPatten, and Keating 2011: 496 and 498)

In discourse subordinating contexts advanced participants selected Spec IP antecedents for null pronouns approximately 52 % of the time and for overt pronouns approximately 55 % of the time. Like the intermediate learners, the advanced group also appears to depend on an L1 discourse strategy for assigning subject antecedents, although the data reflect that pronoun interpretation in subordinating contexts is more likely due to chance. In coordinating contexts they selected Spec IP antecedents for null pronouns nearly 70 % of the time and for overt pronouns approximately 57 % of the time. Results of the more advanced group showed that for discourse structure there was a marginally significant effect; for type of pronoun there was no main effect; but there was a significant effect for the interaction between the two factors in coordinate discourse, but not in subordinate discourse contexts. These learners reveal a combination of L1 and L2 strategies for resolving pronoun reference.

148 | 5 Processing studies and L2 subject pronoun acquisition in Spanish Graph 1 displays a summary of the results of all three groups for the Spanish tasks tested in experiments 2, 3 and 4. Native Spanish speakers clearly depend on a syntactic strategy and interpret antecedents according to the type of pronoun (i.e. the null/overt), regardless of the discourse structure (i.e. coordinate/discourse), in accordance with the biases of the PAH. 80 70 60 Null pronoun

50 40

Overt pronoun

30 20 10 0 Subordinate

Coordinate

Native Speakers n = 26

Subordinate

Coordinate Subordinate

Intermediate Spanish n = 22

Coordinate

Advanced Spanish n = 21

Graph 1. Summary of SpecIP antecedents by discourse structure and pronoun type.

On the other hand, the two learner groups showed two different strategies. Whereas the low intermediate level learners of Spanish tended to apply the same L1 English discourse-based strategy for their L2 Spanish in the interpretation of null and overt pronouns, the more advanced learners appear to be using both L1 and L2 strategies. However, a preference for a subject antecedent only came about when both the L1 discourse strategy and the L2 PAH overlapped with null pronouns in a coordinate structure, not in the subordinate structure. The authors concluded that English-speaking L2 learners can acquire the Spanish position of antecedent-like strategy for assigning anaphoric reference but in order to do so, they must somehow deactivate the L1 interpretation strategy of relying on discourse properties and learn the importance of interpreting the syntactic properties of the L2.

5.3.3 Conclusions of anaphoric resolution in bilingual and Spanish L2 speakers The findings of the bilingual heritage speakers and L2 learners bear some resemblance to the findings of the monolingual Spanish speakers in the Alonso-Ovalle et al. (2002), Perales and Portillo (2007) and Filiaci (2010) studies and the Italian-

5.3 Strategies of anaphoric resolution in Spanish heritage and L2 speakers |

149

English and Italian-Spanish bilingual speakers of the Sorace et al. (2009) study. The PAH appears to be a strong predictor of anaphoric resolution for null subject pronouns in Spanish for monolingual and bilingual speakers and more advanced L2 learners, especially when the form of the pronoun (null), syntactic position (preverbal subjects in Spec IP) and pragmatic context all favor the bias predicted by the PAH. The evidence suggests that this tendency stabilizes in adult bilingual speakers and strengthens as L2 proficiency improves. The predictability of the PAH is weaker for anaphoric resolution of overt subject pronouns. Although the tendency is stronger in monolingual Italian speakers, among monolingual Spanish speakers overt subject pronouns are less likely to link to non-subject antecedents; in fact, Spanish has proven to be more flexible and the evidence from several studies reveals that overt subject pronouns link equally to subject and object antecedents. It has also been seen that this weaker (or more flexible) bias of the PAH has effects among bilingual speakers and L2 learners. Among bilingual speakers, anaphoric resolution for overt subject pronouns takes longer to stabilize and among L2 learners takes longer to sort out. The lack of a clear distinction in the functions of overt subject pronouns may explain bilingual speakers’ and L2 learners’ continued dependence on the less taxing and more efficient English processing strategy of the subject rule for anaphoric assignment rather than the PAH. This in part explains the persistent over-use of overt subject pronouns in bilingual speaker and L2 learner production. It is clear, however, that L2 learners differ from monolingual Spanish speakers in their dependence on the discourse structure of coordination and subordination (an L1 strategy) for determining anaphoric assignment of null and overt pronouns in Spanish. Whereas the native Spanish speakers interpret anaphoric assignment based on pronoun form independently of discourse structure, L2 learners assign subject antecedents to both null and overt subject pronouns according to the discourse structure. It is only the more advanced learners who have begun to develop an L2 processing strategy based on type of pronoun as per the PAH, but only when both the L1 and L2 strategies conjoin to link a subject antecedent with a null pronoun in a coordinate discourse structure. They still have not abandoned their L1 strategy for assigning subject antecedents in subordinating structures where the L1 discourse strategy goes against the L2 strategy predicted by the PAH. For some time, at least, there are two competing strategies. Thus it has been shown that antecedent assignment of null and overt subject pronouns in Spanish among bilingual heritage speakers and L2 learners depends on a myriad of factors including cross-linguistic differences in the adherence to the PAH or the subject rule, universal processing strategies, and language specific processing strategies related to discourse structure.

150 | 5 Processing studies and L2 subject pronoun acquisition in Spanish

5.4 Chapter summary The processing perspective of subject pronoun use and acquisition focuses on how linguistic knowledge is processed; in other words, how speakers interpret the antecedent resolution of potentially ambiguous subject pronouns in discourse. Although the perspective assumes that the formal morphosyntactic properties of the pro-drop parameter are in place (i.e. speakers ‘know’ that both null and overt subject pronouns are possible in the grammar and most probably that a rich verbal morphology and subject-verb inversion are also in place), the central questions guiding research include what strategies are available to speakers and learners and how these are activated in language processing and production. Several hypotheses have been launched and investigated to explain the distinct strategies and how speakers and learners make use of these. A major hypothesis set forth to explain antecedent assignment in pro-drop languages is the position of antecedent hypothesis, the PAH. The PAH predicts that speakers are more likely to assign subject antecedents to null pronouns and non-subject antecedents or objects to overt pronouns. Research in Italian monolingual speakers has provided strong support for the predictions of the PAH and furthermore has shown that there is a high processing penalty for discourse contexts that force an interpretation that violates the biases of the PAH. However, it has also been shown that the predictions of the PAH are strongest for null pronouns and that the processing cost is higher for contexts that force speakers to violate the PAH for null pronouns than for those that violate the PAH for overt pronouns. This latter tendency is even greater among speakers of Spanish, which apparently is more flexible in the anaphoric assignment of overt subject pronouns – an important finding in cross-linguistic differences between at least two closely related pro-drop Romance languages. The evidence suggests that both language specific processing strategies and local discourse structure are important elements for the interpretation of pronoun reference. Nonetheless, important universal strategies have also been uncovered in the research. Monolingual and bilingual children of both pro-drop and non-prodrop languages accept overt pronouns for non-topic shift contexts (i.e. subject antecedents) and null subjects in topic shift contexts (that require an overt subject), but by adulthood these children appear to adhere to the biases predicted by the PAH, although the rejection of null subjects for topic shift stabilizes more quickly. The linking of overt pronouns to non-topic shift contexts and subject antecedents continues into adulthood and research reveals that for Spanish monolingual speakers only approximately half of the time are overt pronouns linked to topic shift and/or object antecedents, contra the predictions of the PAH. The preference for redundancy over ambiguity appears to be a universally more

5.4 Chapter summary

| 151

economical and less taxing interpretation strategy for assigning pronouns to appropriate antecedents. The ramifications for the L2 acquisition of pro-drop languages are clear. When faced with two processing strategies for anaphoric resolution, the L1 strategy of subject assignment is reinforced by the universal strategy. Advanced L2 learners of pro-drop languages eventually sort out the language specific strategy of the PAH for null subjects but for overt subject pronouns continue to rely on the nonpro-drop L1 strategy for anaphoric assignment – a hybrid strategy that reflects both L1 and universal tendencies. A more complex situation occurs when learners are confronted with two distinct discourse structures, one that favors topic continuity as in the coordinate relation and one that favors topic shift as in the subordinate relation. Lower level learners depend on the L1 discourse strategy for assigning pronouns to subject antecedents regardless of the form of the pronoun (null or overt). Advanced learners use a hybrid strategy that reflects both L1 and L2 processing strategies; they depend on the L1 processor for subordinate relations and the L2 for coordinate relations. This hybrid strategy is also universal in the sense that native speakers as well are fully capable of violating the predictions of the PAH when the discourse context demands it. Although there is a processing penalty for contexts that force a contrary interpretation, especially for null subjects, the evidence points to the prominence that information structure plays in understanding anaphoric interpretation in discourse and in second language acquisition.

6 Discourse-pragmatics approaches in the acquisition of subject expression Introduction The theoretical models and the studies examined in the previous chapters provide evidence that information structure in discourse is key to understanding the acquisition of subject pronouns in L2 Spanish. Nonetheless still unanswered is the question of which specific pragmatic constraints of discourse structure affect both use and acquisition. Chapter six presents a model from a non-generative, discourse-pragmatic perspective that has examined discourse structure, pragmatic constraints or pragmatic rules that guide subject use in Spanish L2 learners. Studies from this framework are based on discourse notions of narrative cohesion and coherence (Saunders 1999; López Ortega 2002); referent saliency, cognitive focus, contrastive focus, pragmatic weight (Davidson 1996); and cognitive status (Gundel, Hedberg, and Zacharski 1993). Studies reviewed include those of Saunders (1999); López Ortega (2002); Lubbers Quesada and Blackwell (2009); and Blackwell and Lubbers Quesada (2012). Saunders (1999) investigates to what extent the use of subject forms as predicted by the anaphoric hierarchy is determined by the role, familiarity and relative importance of a referent in a narrative (as the central protagonist, the secondary character, or as a background figure); the function of that referent within the narrative (whether the character is being introduced, maintained or reintroduced); and the syntactic position of the referent in a clause (whether the character appears in a subject or nonsubject position). López Ortega (2002) examines discourse conditions beyond the traditional pragmatic notions of contrast, emphasis and disambiguation and looks at discursive aspects as they relate to subject use, including speaker identity, topic involvement, semantic features of referents, narrative structure, verb type, conjunctions and adverbials. She seeks to trace the development of L2 learners’ discourse patterns for marking contrast, emphasis or disambiguation, in an effort to create cohesive discourse. Lubbers Quesada and Blackwell (2009) identify pragmatic rules that appear to constrain overt and null first person singular subject pronoun use and analyze narratives of native speakers and L2 learners to ascertain to what extent L2 speakers of Spanish develop these largely untaught pragmatic rules. They examine whether or not focus of attention or saliency, new/non-presupposed information, and speaker commitment and evaluation can predict subject use and acquisition. Although focus and presupposition are pragmatic notions that affect subject pronoun use among all persons (first, second

6.1 Discourse structure and the acquisition of subject expression

|

153

and third person) and number (singular and plural), speaker commitment and evaluation are pragmatic notions only available to speakers themselves and thus can determine only first person (and mostly first person singular) subject pronoun use. The only two options for first person subject expression are overt and null pronouns. In a subsequent study, Blackwell and Lubbers Quesada (2012) examine the extent to which speakers take into account the cognitive saliency of multiple possible referents in discourse when selecting subjects. The authors examine the use of all subject expressions, including null subjects, subject and determiner pronouns, full noun phrases and proper nouns, in order to test the Givenness Hierarchy (GH), proposed by Gundel, Hedberg, and Zacharski (1993). The GH sets forth that a set of implicationally-related cognitive statuses constrain the form and distribution of referring expressions in discourse. The cognitive statuses are based on underlying assumptions cooperative speakers make about the informativeness and saliency of referents in discourse and assume that greater referent saliency corresponds to more minimal referring expressions (Levinson 2000). Like the generative studies discussed previously, these studies from a pragmatic perspective provide further evidence that learners at an early stage are capable of selecting the null pronoun option, that discourse factors affect use, and that overuse of both null and overt pronouns continue at advanced levels. In addition, they point to universal properties, in this case, general cognitive abilities related to the notions of textual cohesion and coherence, saliency, focus, memory, attention state, and speaker involvement that all appear to guide subject expression for both native speakers and L2 learners alike.

6.1 Discourse structure and the acquisition of subject expression Saunders (1999) proposes that the pro-drop parameter within a Universal Grammar framework is insufficient for the analysis of the L2 acquisition of subject expression in Spanish due to the fact that the pro-drop parameter is designed to explain the binary distinction between null and overt subject pronouns and does not consider the entire range of linguistic forms that speakers use to refer to referents in discourse, including null and overt pronouns, proper nouns, definite noun phrases and indefinite noun phrases. Furthermore, research based on the UG framework strives to determine whether or not learners are capable of acquiring native-like judgments regarding the grammaticality of null subjects at the sentence level and assume that if they accept null subjects and use them at least some of the time in their production, then this is adequate evidence that learners

154 | 6 Discourse-pragmatics approaches in the acquisition of subject expression have reset the syntactic parameter for null pronouns. Both Saunders and López Ortega (2002) argue, however, that choices of linguistic forms occur at the discourse level and thus a more complete explanation of what determines knowledge and acquisition of subject expression must include an analysis of the entire range of possible forms used to refer to subjects. Saunders claims that, “unless learners understand the disambiguating and emphatic functions of overt pronouns in Spanish, as well as the distinctive functions of pronouns and noun phrases, it cannot be claimed that null subjects have been acquired” (Saunders 1999: 11). Because different linguistic forms serve distinct discourse functions, an examination of the selection of pronouns should not be separated from that of the choice of lexical noun phrases (Saunders 1999: 13). Saunders proposes that the ideal level for the examination of the acquisition of subject expression is the narrative structure, and most ideally, oral narratives. A typical complete narrative includes several components, including an abstract, an orientation, complicating actions, a resolution, a coda, and evaluating remarks. A minimum of components would include the setting, complicating actions and a resolution (Saunders 1999: 44). The complexity of the narrative can have an effect on the number and range of subject expressions included. Furthermore, narrative structure requires cohesion and for this, depends on the continuity of reference, which is preserved principally through subject expressions (i.e. pronouns and noun phrases) (Saunders 1999: 48). López Ortega (2002) argues that it is worth examining L2 subject expression in conversational discourse because “grammatical features that are considered acquired at the sentence level . . . [are] still problematic for students at the discourse level” (López Ortega 2002: 3).

6.2 Cohesion theory and the anaphoric hierarchy Although cohesion manifests itself differently from language to language, its general principles appear to be universal. Linguistic markers (overt or otherwise) that indicate associations between referents, propositions or clauses form cohesion in discourse. This can be accomplished via conjunctions, proforms and ellipsis. Pronouns and noun phrases work to form cohesive ties between and among referents. Saunders (1999) points out that a network of cohesion includes a number of referring expressions that include a cohesive element and a node, which together form a cohesive tie. The anaphoric expressions of a narrative are the cohesive elements and the nodes are the entities to which they refer. Therefore, a node is used to introduce a new character, reintroduce a character or to express emphasis. The cohesive element is used to maintain a character introduced with a node in the narrative. For example, in sentences (122 a–c), the underlined referent is the

6.2 Cohesion theory and the anaphoric hierarchy

|

155

cohesive element that is linked to a previous referent, which is the node, indicated in bold. The relationship between the two is the cohesive tie. (122)

a. Arthuri left the party because hei wasn’t feeling well. b. Once there was a boyi who lived on Maple Street. His name was Sami c. A six-year-old boyi fell from a window yesterday . . . The boyi was taken to the hospital with minor injuries. (Saunders 1999: 49)

When several cohesive elements refer to the same node, the result is a network of cohesion, exemplified in (123): (123) My friend Georgei went to the dry cleaners yesterday to pick up his suit. When hei got home, hei realized that hei had the wrong suit. Hei immediately returned to the dry cleaners and 0i informed them of the mistake. (Saunders 1999: 50) Saunders is careful to point out that in examining cohesion what is really investigated is cohesive potential. Speakers use linguistic devices in order to create the potential for cohesion, which is successful only if the listener interprets the signals as such. In fact, it is the choice of linguistic form (null or overt pronoun, definite or indefinite noun phrase) that “reflects the speaker’s presuppositions of the listener’s knowledge of the topic” (Saunders 1999: 50). Cohesive ties can be either intratextual, when the cohesive element refers to a node previously mentioned in the immediate narrative, or they can be extratextual, when the cohesive element refers to an external node for which the listener/reader can infer its referent. Saunders proposes that in narrative discourse the use of pronouns and noun phrases follows a universal anaphoric hierarchy (presented in (124)), which is “based on the amount of information that the speaker/writer assumes the listener/reader already to possess” (Saunders 1999: 51). (124) indefinite NP > definite NP > proper noun > overt pronoun > null pronoun (Saunders 1999: 51) In oral narrative structure, the higher up in the hierarchy (to the left) a referring expression is, the less information the speaker assumes the listener has regarding that referring expression’s potential referent. Therefore the indefinite noun phrase (NP) presupposes the listener has no information regarding the referent,

156 | 6 Discourse-pragmatics approaches in the acquisition of subject expression whereas the null pronoun presupposes the maximum amount of information. Saunders states An entity typically is introduced into a narrative with a linguistic form that is high on the hierarchy. Once a referent has been mentioned in the discourse, subsequent reference to that entity generally should be lower on the hierarchy until an intervening referent is introduced. When a referent is reintroduced after an intervening referent, the speaker reverts to an element higher on the anaphoric hierarchy. (Saunders 1999: 51–52)

For both Spanish and English, the highest linguistic form on the anaphoric hierarchy is the indefinite noun phrase, which is usually employed when introducing a new character in a narrative. All other subsequent mentions of the character should be via forms lower on the hierarchy. A definite noun phrase is used when reintroducing a character, after another character(s) has been introduced into the narrative, although definite noun phrases can also be used to introduce a new entity when the character is part of a shared knowledge schema between speaker and listener. For example, in (125), (125) We saw a great film last night. The leading actor was terrific. the definite noun phrase, the leading actor, has no previously introduced entity but both speaker and listener share the schematic knowledge that films usually have actors, and if so, they have leading and supporting actors. Proper nouns are lower on the hierarchy than indefinite and definite noun phrases and assume greater knowledge on the part of the listener and thus can be used anaphorically. Both speaker and listener must know that the name refers to the same referent. Pronouns are the forms that are lowest on the hierarchy and assume the greatest amount of shared knowledge between speaker and listener. Once a character has been introduced into a narrative, it is maintained by means of a pronoun; in nonpro-drop languages, such as English, an unstressed overt pronoun is used and in pro-drop languages, such as Spanish, a null pronoun is expected. Because their referents are recoverable from the discourse context, pronouns are the most cohesive of referring expressions and in Spanish it is null subjects that perform this function. Finally, Saunders highlights the fact that a narrative can be coherent but not cohesive which is what happens when L2 learners unnecessarily create new nodes to refer to the same character, overusing overt subject pronouns, noun phrases and proper nouns when a null subject would be appropriate in order to provide the cohesiveness required of the text. An example of a narrative text that is coherent

6.2 Cohesion theory and the anaphoric hierarchy |

157

but lacking in cohesion is the following, in which the non-native speaker overuses proper nouns and overt subject pronouns (indicated in bold): 21 años; Scotti (126) Mi amigo mejor se llama Scotti ; Scotti tiene my friend best refl-call-3sg Scott Scott have-3sg 21 years Scott es muy gordo y proi tiene lentes; éli tiene pelo castaño be-3sg very fat and pro have-3sg glasses he have-3sg hair brown y muy largo; éli juega la guitarra muy bien. and very long he play-3sg the guitar very well ‘My best friend’s name is Scotti . Scotti is twenty-one; Scotti is very chunky and [he] has glasses; hei has dark and very long hair; hei plays the guitar very well.’ (Lubbers Quesada and Blake 2003) Native speakers of Spanish assume that the maintenance of the same subject will be signaled via a null subject; when the speaker violates the anaphoric hierarchy and continually creates new nodes with proper nouns and overt subject pronouns to form a cohesive tie, the result is a text that requires greater processing effort on the part of the listener.

6.2.1 Testing the anaphoric hierarchy and cohesion theory Saunders (1999) tests to what extent L2 learners of Spanish acquire and use null and overt subjects²⁸ that are consistent with both the anaphoric hierarchy and cohesion theory. She predicts that both native speakers of Spanish and L2 learners will be constrained by the predictions of the anaphoric hierarchy and by the need to create a cohesive narrative; more specifically, speakers will select subject forms along the hierarchy in order to create a narrative where several characters (including a main and two secondary protagonists) are introduced, maintained and reintroduced throughout the story line. She also predicts, however, that native speakers and learners will display some differences in their choices of subject forms, that lower level learners will not use null subjects to the same degree as native speakers and more advanced learners, and that all learners will tend to overuse overt subjects.

28 Saunders examines pronominal and anaphoric reference in both subject and non-subject positions. Since this volume is mainly concerned with subject reference, only the results of her study that deal with subjects are discussed.

158 | 6 Discourse-pragmatics approaches in the acquisition of subject expression The author examined 135 oral narratives of 60 learners of L2 Spanish from five levels of proficiency ranging from novice to advanced, and eight native speakers of Spanish. All participants were asked to narrate two stories based on a series of pictures that depicted three characters involved in a mix-up: 1) a customer whose purchase becomes mixed up with, 2) another customer’s and, 3) a clerk who is in charge of resolving the problem. The main character is the first customer who appears in the initial picture and completes several actions before the second and third characters appear. The two stories allow for the full range of subject forms included in the anaphoric hierarchy and Saunders predicts that each character will be introduced, maintained and reintroduced with different linguistic forms according to their role and their actions in the story and their appearances at different points throughout the narrative.

6.2.2 Native speakers’ use of subject forms according to the anaphoric hierarchy The results reveal that native speakers introduce the main character with either an indefinite noun phrase or a proper noun (since he is introduced as a proper noun in the task instructions) whereas the secondary character, the second customer, is only introduced as an indefinite noun phrase. The fact that some participants introduce the main and secondary characters differently is significant and reflects assumptions the speakers have about the knowledge state of the listener. The main character, introduced in the instructions, is familiar to both speaker and listener and therefore can be introduced with a proper noun, a subject form lower on the anaphoric hierarchy. The clerk is introduced in a significantly different manner than the other two characters with a variety of forms and more often in non-subject positions. The majority of speakers use a definite noun phrase (la vendedora ‘the saleswoman’, la señorita en la caja ‘the woman at the cash register’) to refer to the clerk. This is in sharp contrast to the other two characters who are introduced mainly with an indefinite noun phrase. However, definite noun phrases can be used to introduce a new character when the entity forms part of a shared knowledge schema between speaker and listener. This is based on the collective assumption that clerks work in stores and one must interact with them when making purchases. More than a third of the respondents in the study do not mention the clerk explicitly but rather imply her presence with several linguistic expressions. Some imply the clerk by referring to the store as in (127) and (128); others indirectly refer to the clerk with a null pronoun expressing a generic ‘they’, as in (129), or with an indirect object pronoun, as in (130). (All bold forms refer to ‘the clerk’.)

6.2 Cohesion theory and the anaphoric hierarchy |

159

(127) Pero la tienda le dijo que pro tenía que but the store him-dat say-3sg-pret that pro have-3sg-imp that regresar el siguiente día. return-inf the next day ‘But the store told him that [he] had to return the next day.’ (128) Entonces pro regresó a la tintorería para quejarse. so pro return-3sg-pret to the dry-cleaners to complain-refl-inf ‘Then [he] returned to the dry cleaners to complain.’ (129) pro Se encontró que pro le habían dado ropa pro find-refl-3sg-pret that pro him-dat have-3pl-imp given clothes interior femenina. interior feminine ‘[He] found that [they] had given him women’s underwear.’ (130) él fue . . . otra vez . . . a regresar su traje y pro le he go-3sg-pret another time to return-inf his suit and pro her-dat dijo, “Qué pro hicieron con mi traje?” say-3sg-pret what pro do-3pl-pret with my suit ‘he went . . . back . . . to return his suit and [he] said to her, “What did [you] do with my suit?” (Adapted from Saunders 1999: 104) Saunders claims that this variety of forms suggests that for many of the native speakers, the clerk is not viewed as a character in the story but rather is seen as part of the background or the setting for the narrative. It is evident that native speakers take into account different knowledge states on the part of the listener for each of the three characters. There are also differences in how the native speakers maintain the three characters in their narratives. The main character is maintained in subject position via null pronouns almost exclusively (nearly 93 % of the time), whereas the secondary character (the second customer) is maintained with a null subject slightly more than half (55 %) of the time. The fact that the main character is referred to more often with the null subject, the lowest element on the anaphoric hierarchy, leads to the greatest amount of narrative cohesion (Saunders 2009: 95). The only time this is overridden is in one case where the speaker uses a proper noun to maintain the main character in order to emphasize a contrast, as in (131), or to ease linguis-

160 | 6 Discourse-pragmatics approaches in the acquisition of subject expression tic processing between the same verb and verb form used in main and subordinate clauses as in (132). (The bold forms are used to maintain the character.) (131)

Pero pro resulta que cuando proi llegó a su casa para su but pro result-3sg that when pro arrive-3sg-pret at his house for his sorpresa, proi encontró un brasier en el paquete. ¿Cómo era surprise pro find-3sg-pret a brassier in the package how be-3sg-imp posible que Joséi se fuera a poner un brasier? possible that José refl go-3sg-imp-subj to put-inf a brassier ‘But [it] turns out that when [he] got home to his surprise, [he] found a bra in the package. How was it possible that José would wear a bra?’

(132) Pues el paquetei que élj tenía, tenía ropa interior. well the package that he have-3sg-imp have-3sg-imp clothes interior ‘Well the package that he had, had underwear.” (Adapted from Saunders 1999: 87) The secondary character is maintained with null subjects, overt subject pronouns and definite noun phrases. The tertiary character, the clerk, is maintained only three times in the native speaker narratives and only in non-subject position. Saunders explains that these differences in how the three characters are maintained in the narrative may be because the second customer is introduced late in the narrative and the clerk is often viewed as part of the setting. The fact that the main character is introduced early and maintained more often is not incidental to the structure of the narrative and the selection of anaphoric subjects. Null subjects are reserved mostly for main characters in narrative structure precisely because the story is about the main character who engages in more of the action and thus, the use of null subjects assures that the focus is continually on that character. The way in which each character is reintroduced into the narrative is also different. The main character is reintroduced only with proper nouns (68 % of the time) and overt pronouns (32 % of the time) and the secondary and tertiary characters are reintroduced exclusively with definite noun phrases. Again, because the focus is on the main character, it is possible to reintroduce him into the narrative with a subject form lower on the hierarchy because it presupposes a greater amount of shared assumptions regarding his role in the actions of the story. The story is not about the secondary and tertiary characters and thus, their reintroduction must be realized via a subject form higher on the hierarchy.

6.2 Cohesion theory and the anaphoric hierarchy |

161

Saunders found that indeed the native Spanish speakers select subject forms according to the role of the referent in the narrative (main, secondary or tertiary character) and their function at different points throughout the narrative (introduction, maintenance, reintroduction) and that these selections are in accord with the anaphoric hierarchy. Introductions of characters with little shared background knowledge are done with forms high on the hierarchy and those of the main character, less high on the hierarchy. Maintenance of the characters also depends on their role and null subjects are reserved more for the main character than for the others. The null subject is lowest on the hierarchy and assures the greatest degree of narrative cohesion and allows for keeping the focus on the main character. Finally, the reintroduction of the different characters is also in line with the predictions of the hierarchy, which dictates that forms used for reintroductions be no higher on the anaphoric hierarchy than the original introduction (p. 100). Since the main character can be introduced with a proper noun, then the available forms are proper nouns, overt pronouns and null pronouns. Because the secondary and tertiary characters are originally introduced with indefinite or definite noun phrases, then the available forms include also definite noun phrases. In Table 18 is a summary of the use of subject forms according to character and the anaphoric hierarchy based on the data presented in the Saunders (1999) study. Even a quick glance at Table 18 reveals that not all characters in the narrative are referred to with the same type of subject. Clearly, the character’s role in the narrative plays an important part in how that entity is represented linguistically. Table 18. Summary of native speaker use of forms occurring in subject position according to character role and anaphoric hierarchy. Indefinite NP Main character

introduce

Secondary character

introduce

Tertiary character

>

Definite NP

>

Proper noun introduce

maintain and reintroduce

>

Overt pronoun

>

Null pronoun

reintroduce

maintain

maintain

maintain

introduce²⁹ and reintroduce (Taken from Saunders 1999)

29 When explicitly mentioned.

162 | 6 Discourse-pragmatics approaches in the acquisition of subject expression Subject forms used to refer to the main character are more heavily weighted toward the right-hand side of the hierarchy that signals greater focus and cohesion. The story is about this character and there is a need to focus on that character. Subject forms used to refer to the secondary and tertiary characters are more heavily weighted toward the left-hand side revealing less focus and less cohesion. In fact, the tertiary character (the clerk) is never brought into focus in the narrative; she is introduced mostly in non-subject positions and maintained only in non-subject positions. Because the story is not about these characters, more elaborate forms are needed to signal that they are being talked about. Using forms located higher on the hierarchy indicates to the listener that these characters are not the focus of the narrative; that is to say, the story is not about them. Conversely, as Saunders indicates, forming cohesive ties by using anaphoric forms lower on the hierarchy for the main character allows the listener to focus in on that character who is, after all, the topic of the narrative (Saunders 1999: 106).

6.2.3 L2 Learners’ use of subject forms according to the anaphoric hierarchy The results of the learners, as reported by Saunders, show significant differences among the proficiency levels in terms of narrative complexity and use of subject forms according to the anaphoric hierarchy. For example, at the novice level, only four of the 24 narratives include references to all three characters and seven include only the main character. As proficiency improves, the narratives becomes more complex, including more elements of narrative structure (abstract, orientation, complicating action, resolution, coda and evaluation), a greater number of clauses per narrative, more characters are mentioned, and the use of subject forms adheres more closely to the principles of the anaphoric hierarchy. All groups are much more likely to introduce the main character with a proper noun (between 67 and 96 %) except for the advanced group, which like the native speaker group, used both proper nouns and indefinite noun phrases. Other forms used include indefinite noun phrases and overt pronouns. For maintaining the principle character, the novice group uses a null pronoun less than one-third of the time, whereas the intermediate low and high groups prefer a null pronoun but only a little more than half of the time. The only group that performs more like the native speakers again is the advanced group who uses a null subject a little over 80 % (compared to the native speaker use of over 93 %)³⁰. All learner groups

30 Another group, the intermediate mid also uses null subjects to maintain the main character almost 78 % of the time, close to the advanced learners but only eight of the twelve participants

6.2 Cohesion theory and the anaphoric hierarchy

|

163

overuse overt subjects; the novice group depends mostly on proper nouns. The second most preferred form for the intermediate low group is also a proper noun and for the intermediate high, the second most preferred form is an overt pronoun, él ‘he’. Saunders’ data reveal that even though beginning learners are capable of producing null subjects, the fact that they overuse overt subjects (more than 68 % of the time) indicates that the null subject parameter has not been acquired since they do not know that a null subject pronoun is the required form for maintaining a referent, specifically the main character, in narrative discourse (Saunders 1999: 125). It could be argued that syntactically, the participants in this study are not committing errors when they produce overt subject forms for maintaining the main character. An analysis based on Universal Grammar would suggest that since null subjects are used some of the time, then this is evidence that the formal properties of the null subject parameter have been reset. An analysis based on the Interface Hypothesis would claim that the overuse of overt subjects in non-topic shift contexts is due to deficits in the underspecification in the L1 of interpretable features such as +topic shift (Sorace and Filiaci 2006) and this knowledge transfers to the L2. Saunders, however, rejects the claim by some that L1 transfer is the cause of learners’ errors (White 1986). If learners are simply transferring the L1 setting, then it would be expected that they would overuse overt subject pronouns, not proper nouns, which the anaphoric hierarchy predicts is the form to use for maintaining referents in discourse in English. It is in fact, not until more advanced levels that learners begin to overuse overt pronouns, more in line with the English system and the anaphoric hierarchy, which predicts that cohesion is obtained by maintaining a character with a form that is lower on the hierarchy. Because of these facts, Saunders claims that Cohesion Theory better explains the deficits in the L2 learners’ production. The overuse of the proper noun detracts from the cohesion of the narrative . . . cohesion is attained by maintaining a character with a referent that is lower on the anaphoric hierarchy than the introduction of the character. Doing so creates a cohesive tie with the original mention of the character. Each time the speaker uses a proper noun, the same form used

use a null more than 80 % of the time. Saunders considers 80 % accuracy in the use of a form as evidence of acquisition. She attributes the greater apparent success of the intermediate mid group (over the intermediate high group) to their tendency to construct simpler narratives and include fewer characters. The intermediate high group produces more complex narratives and tends to include all three of the characters. In addition, because the intermediate mid group also tends to overuse proper nouns, Saunders concludes that it cannot be claimed that the intermediate mid group has acquired the appropriate forms for the function of maintaining the main character.

164 | 6 Discourse-pragmatics approaches in the acquisition of subject expression for the introduction of the main character, a new node is created unnecessarily. The result is a lack of cohesion, which often leads to difficulty in comprehension. (Saunders 1999: 126)

Like the native speakers, most of the learners use proper nouns and overt subject pronouns in order to reintroduce the main character into the narrative, although they use a much higher proportion of proper nouns and a lower proportion of overt pronouns than the native speakers. Thus, although they use the same forms as the native speakers, they do not use the same pattern for reintroducing the main character. The differences in patterns for all groups can be seen in Table 19. Table 19. Summary of native speaker and learner use of forms occurring in subject position for main character according to anaphoric hierarchy. Indefinite NP >

Definite NP

>

Proper noun

>

Overt pronoun

>

Null pronoun

Native speakers

introduce

introduce

reintroduce

maintain

Upper proficiency levels

introduce

introduce and reintroduce

maintain

maintain

Intermediate proficiency levels

introduce, maintain, and reintroduce

maintain

maintain

Lower proficiency levels

introduce, maintain, and reintroduce (Taken from Saunders 1999)

Thus, it is clear that native speakers, except for the function of introducing (where they vacillate between indefinite noun phrases and proper nouns), have a clear division of labor for the use of subject forms for the different functions of referring to the main character in narrative structure. Upper proficiency learners follow the same pattern for introducing but overuse proper nouns for reintroducing and overuse overt pronouns for maintaining the main character; intermediate proficiency learners follow a similar pattern as the more advanced learners for reintroducing the main character but overuse both proper nouns and overt pronouns for maintaining this character; and finally, lower proficiency learners for the most

6.2 Cohesion theory and the anaphoric hierarchy |

165

part do not differentiate and use proper nouns for all three functions of the main character. As Saunders reports, the analysis of the secondary character in the narrative is rendered slightly more difficult due to the fact that only about two-thirds of the participants in the lower levels mention the second customer. In the upper levels this character is introduced, like the native speakers, most often with an indefinite noun phrase. Lower level learners use indefinite noun phrases but also use a large proportion of definite noun phrases and some use of definite noun phrases with an accompanying description (el otro hombre con el traje de Arturo ‘the other man with Arturo’s suit’) and even proper nouns. The use of a definite noun phrase again negates the possibility that L1 transfer is responsible due to the fact that the anaphoric hierarchy also predicts an indefinite noun phrase for introducing characters in English. Although there are few instances of maintaining the secondary character, in general, learners reflect more the native speaker production and use definite noun phrases, overt subject pronouns and null subjects. When reintroducing the secondary character, the majority of the learners use a definite noun phrase like the native speakers, although beginning learners also use indefinite noun phrases and proper nouns – a pattern of use also seen with these learners for the overuse of proper nouns to refer to the main character in all of its functions. A summary of how participants referred to the secondary character according to the anaphoric hierarchy can be seen in Table 20. Table 20. Summary of native speaker and learner use of forms occurring in subject position for secondary character according to anaphoric hierarchy. Indefinite NP >

Definite NP

Overt pronoun >

Null pronoun

Native speakers

introduce

maintain and reintroduce

maintain

maintain

Upper proficiency levels

introduce

maintain and reintroduce

maintain

maintain

Intermediate proficiency levels

introduce

introduce, maintain, and reintroduce

maintain

maintain



Lower proficiency levels

introduce and reintroduce

introduce, maintain, and reintroduce

maintain

maintain



>

Proper noun

introduce and reintroduce

>

No mention

(Taken from Saunders 1999)

166 | 6 Discourse-pragmatics approaches in the acquisition of subject expression Finally, it is to be recalled that the native speakers in Saunders’ study introduce the tertiary character (the clerk) into their narratives in a significantly different way from the other two characters revealing that the participants view this character as part of the setting of the story. The clerk is introduced mainly with a definite noun phrase, not with a definite noun phrase. This is because part of speakers’ and listeners’ shared knowledge schema includes knowing that clerks are found in stores. In other cases, the clerk is implied by other linguistic means but not explicitly mentioned. The learners in this study also tend to introduce the clerk with a definite noun phrase, when the clerk is included in their narratives³¹. Advanced learners also imply the presence of the clerk with other linguistic means. The native speakers do not maintain the clerk in any subject positions and when reintroducing this character, use only definite noun phrases. Only the advanced learners maintain the clerk to any significant degree in subject position and the majority of these references are made with null subjects. The learners who reintroduce the clerk do so with a majority of definite noun phrases but also with a number of overt subject pronouns, contrary to the pattern of production used by the native speakers. Table 21 summarizes the subject forms used for the clerk according to anaphoric hierarchy by all groups. The lack of a variety of forms to refer to this character is due to the fact that she appears late in the narrative and there are fewer opportunities for her to be maintained and reintroduced. In addition, this tertiary character is viewed at least by the native speakers as part of the background. Nonetheless, all levels of learners use definite noun phrases to refer to her when introducing and reintroducing.

6.2.4 Predictions and universality of the anaphoric hierarchy The data from the Saunders (1999) study reveal that native speakers adhere to rules that dictate the use of subject forms in narrative discourse and that the choice of the form is determined by the role, familiarity and relative importance of the character in the story; by the function of that character during different points throughout the narrative, whether the character is being introduced, maintained or reintroduced; and by the syntactic position of the referent. Although in the present volume we are concerned mainly with the use, distribution and acquisition of subject expression, the data presented by Saunders reveals that the role

31 The lowest proficiency group only made reference to the clerk on three occasions whereas the lower and middle intermediate groups made no mention of her one-third and one-fifth of the time respectively.

6.2 Cohesion theory and the anaphoric hierarchy |

167

Table 21. Summary of native speaker and learner use of forms occurring in subject position for tertiary character according to anaphoric hierarchy. Indefinite NP >

Definite NP

>

Native speakers

introduce and reintroduce

Upper proficiency levels

introduce and reintroduce

Intermediate proficiency levels

introduce and reintroduce

Lower proficiency levels

introduce

Proper noun >

Overt pronoun

>

reintroduce

Null pronoun

No mention

maintain

maintain



✓ (mostly)

(Taken from Saunders 1999)

and function of the character will also determine the syntactic position of the referent throughout the narrative. The main and secondary characters appear more often in subject position than the tertiary character; and the main character is mentioned more than five times as often as the secondary character, attesting to his prominence in the story, both syntactically and cognitively. The learners do not show the same pattern of use for subject forms until they are in an advanced proficiency level, although all learners in general use different forms to refer to the three characters depending on their role in the narrative. The learners in the beginning and intermediate levels overuse proper nouns for all three functions of the main character, which is evidence against L1 transfer that would have resulted in the use of overt pronouns for maintaining the character. Those in the advanced level pattern in a similar manner to the native speakers, although they overuse proper nouns to reintroduce the character and overuse overt pronouns to maintain him. In reference to the secondary character, the beginning and intermediate learners do not distinguish between indefinite and definite noun phrases for introducing the character and overuse definite noun phrases to maintain him. This pattern of use continues with all learners up until the very advanced level where the pattern of use is like that of the native speakers. The tertiary character is referred to seldom by the beginning group but for all learner groups who do mention this character, definite noun phrases are preferred, a characteristic in line with the native speakers and the anaphoric hierarchy. This reveals to some

168 | 6 Discourse-pragmatics approaches in the acquisition of subject expression extent that learners are also constrained by the role and function of the characters in narrative structure, just not in the same ways (until advanced levels) as native speakers. Furthermore, as Saunders points out, it appears that the learners do not violate the order of the hierarchy for maintaining referents; that is, they do not use a form that is higher on the hierarchy to maintain a character than the form they have used to introduce that character. That both native speakers and learners are constrained by the predictions of the anaphoric hierarchy, albeit in different ways until advanced levels, attests to the universality of the hierarchy. Learners do not violate the order of the hierarchy for maintaining characters although they will overuse redundant forms. This lack of distinction in the beginning and intermediate levels does not lead to ungrammaticality but it does lead to a lack of cohesion in their narrative structure. Only by analyzing the data from a discourse perspective does this deficit in L2 language become evident and possible to explain.

6.3 Emphasis, contrast and disambiguation Functional studies of native speaker subject expression have focused mainly on the capacity of varying forms to create cohesive discourse by means of emphasis, contrast and disambiguation of ambiguous verbal forms (e.g. first and third person imperfect forms in Spanish: hablaba ‘[I] would speak/was speaking’ vs. hablaba ‘[s/he] would speak/was speaking’); accessibility of topic (e.g. null subjects are most accessible and lexical noun phrases are least accessible) (Givón 1983); and grounding (e.g. overt subjects tend to be used more in the foreground and mark actions or a change in narrative structure) (Fox 1987). In addition, it has been revealed that subject expression is sensitive to the specificity, animacy and humanness of the referent and the semantic context of the verb (i.e. whether the verb marks an action or a belief, opinion, emotion, etc.). In studies of native speakers from Madrid, Enriquez (1984) found that for third person explicit subjects to appear, a necessary condition is for these subjects to be [+specific] and [+human/animate] and Davidson (1996) discovered that first and second person singular forms are more frequent in conversational discourse because speakers tend to add ‘pragmatic weight’ to their utterances revealing a greater commitment to certain propositions. For Puerto Rican Spanish speakers, Cameron (1993) revealed that switch reference is a strong predictor for overt subject pronoun use. It is these findings that lead López Ortega (2002) to examine L2 subject expression in conversational discourse in order to identify which features are acquired at the discourse level.

6.3 Emphasis, contrast and disambiguation |

169

6.3.1 A longitudinal case study In her longitudinal case study of four English-speaking college students studying abroad in Spain for a four-and-a-half month period, López Ortega found that subject expression did not change dramatically quantitatively but did so qualitatively. In fact, had she limited her study to the quantitative data she would have concluded that after four and a half months of living and studying in an immersion program in the target language country, learners’ development of Spanish subjects showed very little, if any, development at the morpho-syntactic level. It was only through her qualitative analysis of the learners’ conversations that she detected important gains over time among her participants. Thus, she claims that at “intermediate to advance levels of proficiency, development of subjects occurs mainly at the discursive level” (López Ortega 2002: 188). Specifically, she revealed that L2 learners are capable of developing specific discourse principles that permit them to create cohesion in their conversations and that these principles become more native-like, although not completely so, by the end of their study abroad period. Furthermore, these principles go beyond the traditionally held claims that native speakers use overt subjects merely for contrast, emphasis and disambiguation (Real Academia Española 1983; Gili Gaya 1969) and also use them to express their own emotions, desires, beliefs and attitudes. The study included four English-speaking intermediate level students who were participating in a semester abroad program in Seville, Spain where they lived with Spanish host families and took all of their coursework in Spanish. Language level at the beginning and end of the program was determined by two Oral Proficiency Interview (OPI)-style interviews and self-assessment background questionnaires. The data presented for analysis was collected via six onsite informal non-structured conversational interviews between the author (a native Spanish speaker) and participants at three-week intervals (although the data reported on come from the first two and the last two interviews). Conversational topics were related to the learners’ lives and included personal narratives which involved the speaker and others with enough actions and complication to assure a variety of overt verbal morphology and thematic roles of subjects (López Ortega 2002: 98). The types of subjects analyzed were full lexical subjects, pronominal subjects, null subjects and ‘other’ (López Ortega’s ‘other’ category included obligatory null expletive subjects, impersonal verb phrases such as se dice que . . . ‘it is said that . . . ’, and generic non-specific subjects such as Uno come bien aquí ‘One eats well here’). She examined each subject form according to several structural, semantic and discourse variables. Structural variables included person and number of the subject form and verbal ending (specifically if the verb was

170 | 6 Discourse-pragmatics approaches in the acquisition of subject expression ambiguous for person and number or was an incorrect form)³². The semantic variables included specific reference, human (animate) reference and hearer knowledge (if the referent is known or unknown to the hearer). The discourse variables consisted of a variety of features that, because López Ortega defines in a different way from those outlined in Chapter 2 and examined in research in Chapter 4 of the present volume³³, are presented (with her definitions) as follows: (133) Discourse driven features Topic discontinuity (New topic): Whether the referent has already been established as the topic of the discursive piece. Narrativity (new narrative): Whether the utterance establishes a new narrative line. Emphasis: [P]ronouns were coded as emphatic when the environment around them (estimative verbs, marked stress, appearance of a discourse connector) gave cues for emphasis. Contrast: [C]ontrast is found in cases of intervening subjects, sometimes in adversative and disjunctive coordination (but, on the other hand, neither), as well as ambiguous verbal morphology that required the need to mark who the actor of the action was when more than one referent in the discourse occurred. Linguistic vs. contextual reference (deixis): Contextual referents are first and second person . . . where at least one of the parties is contextual [i.e. present] . . . a linguistic referent is never present in the immediate physical space where the conversation is taking place . . . [and] . . . is necessarily anaphoric. Antecedence: A referent may be anteceded by a previous mention, whether as subject or as verbal complement of the immediately preceding utterance. New information: Whether the topic presents new versus old information. A new topic can present information already known in the context of the

32 López Ortega also coded for type of clause (main, dependent or relative) and whether the verb was a dative verb where the experiencer is syntactically an indirect object. None of these factors were found to be significant for subject use. 33 In fact, in her classification there appears to be overlap in some cases, e.g. contrast, topic discontinuity and switch reference all refer to the introduction of a referent not referred to in the immediately previous utterance. She admits that there is overlap in many of the categories. For example, a new referent may be also a new topic and/or new information for the hearer, but not necessarily so.

6.3 Emphasis, contrast and disambiguation |

171

discourse; however, new information most times implies a new topic being introduced in the discourse. Switched reference: [W]hen an utterance introduces a new subject referent, whether it is a new topic or new information, or not. (López Ortega 2002: 133–135)

6.3.2 Quantitative findings In the quantitative findings of the four participants as a group, López Ortega found a statistically significant decrease in the use of full lexical noun phrase subjects over time but no significant change in the use of null subjects. She also found a statistically significant fluctuation in the use of overt pronominal subjects but neither an increase nor a decrease over time. For lexical subjects there was a significant interaction with many of the semantic and discourse variables but in general, the author found that lexical subjects marked new information, unknown referents and new topics. Overt pronominal subjects correlated positively with the positive values of all three semantic features ([+specific, +human, +hearer unknown]), but there was no statistically significant change over time. Null subjects correlated positively with ambiguous and incorrect verb forms, dative verbs, and third person referents along with the semantic variables of [+specific] [+hearer unknown] – all structural and semantic features that should correlate positively only with overt subjects. Finally, there was little development over time for the use of overt pronominal subjects for emphatic and contrastive purposes. Especially revealing are the results that point to the lack of a significant interaction between overt subject pronouns and first person, which among native speakers has been shown to have a strong correlation; native speakers tend to use overt first person subject pronouns to increase the speaker’s involvement, identity, stance, or to establish intent to take the floor (Davidson 1996). Also the author found that there was a correlation between both null and pronominal subjects and ambiguity of verb forms and incorrect verb forms, which surprisingly is in line with the studies among native speakers and against the morphological compensation hypothesis, which suggests that overt subject pronouns are used to compensate for possible referent ambiguity of ambiguous morphological forms (Hochberg 1986). Finally, contrary to the expected findings, contrast and emphasis did not correlate positively with overt pronominal subjects nor did antecedence (i.e. topic/referent continuity) correlate positively with null subjects, and there was no statistically significant improvement over time for the four learners as a group. The lack of clear significantly measurable results may lead one to conclude that there is little development in the use of subject expression in this group of learners and may,

172 | 6 Discourse-pragmatics approaches in the acquisition of subject expression indeed, be disheartening for study abroad program directors. The qualitative analysis, however, reveals more promising results.

6.3.3 Qualitative findings 6.3.3.1 Intermediate high learners López-Ortega (2002) analyzed each learner’s conversational discourse individually and found that there were important differences among the learners and evidence of convergence towards native speaker use of Spanish subjects in different ways. The two more proficient participants who began their study abroad experience at the intermediate high level and were evaluated as advanced low by the end of their program were found to have converged towards some native speaker discourse strategies, but also exhibited others that are attributed to their individual interlanguages. One learner’s (BG) pronominal subject production increased over time and the use of first person subjects, which were completely absent in the first two interviews, had slightly increased by the last two interviews and were used to mark contrast and emphasis. The other learner (AO), who had exhibited use of overt pronominal first and third person subjects from the beginning, decreased use of these types of subjects. BG had used few overt subjects to disambiguate or create contrast in the beginning, but by the end was using more overt subjects, both pronominal and lexical, to create the necessary cohesion and clarification of referents. López Ortega describes this earlier narrative pattern a ‘centrality versus periphery’ strategy (2002: 192–193). It is a simplified strategy for marking different referents in discourse, much like children’s ‘egocentric speech’ (Ingram 1978, cited in López Ortega 2002: 192). Basically, the speaker uses null subjects (in bold) to refer to the main referent (usually the speaker or the speaker and others) of the story line and overt subjects (in underlined bold) to mark secondary or peripheral referents. BG appears to use egocentric pronoun marking (null for first person/overt for third person) even when he is the peripheral referent and the central referent is not present at the time of the interview.

6.3 Emphasis, contrast and disambiguation

| 173

muy simpática y eh . . . pro pienso que ellos (134) y mi familia es and eh pro think-1sg that they and my family be-3sg very nice les gusta a enseñar la lengua a los estudiantos them-dat please to teach-inf the language to the students ‘and my family is very nice and uh [I] think that they like to teach the language to students’ (Adapted from López Ortega 2002: 195)³⁴ Although this was the dominant discourse pattern in BG’s earlier interviews, over time his discourse structure became more target-like with null subjects used to mark topic/referent/narrative continuity (indicated in bold) and overt subjects used to mark contrast/switch reference (indicated in underlined bold). In the following excerpt, the speaker is explaining why the students are going to have a party earlier than planned. The only time overt subject pronouns are used is for contrast (tú ‘you’ and él ‘he’) at the end of the utterance. (135) pro no querían tener dos fiestas y por eso pro íbamos a pro no want-3pl-imp have-inf two parties and for that pro go-1pl-imp to celebar los dos en la semana que viene, pero . . . porque tú celebrate-inf the two in the week that come-3sg but because you ibas a venir . . . él quería hacer todo go-2sg-imp to come-inf he want-3sg-imp do-inf all ‘[they] didn’t want to have two parties and that’s why [we] were going to celebrate both next week but . . . because you were going to come . . . he wanted to do everything’ (Adapted from López Ortega 2002: 212) AO, on the other hand, showed a more variable use of discourse patterns. From the beginning she used overt lexical and pronominal subjects for emphatic and contrastive purposes, but also overuses subject pronouns (marked in bold), which creates redundancy and repetition when not necessary.

34 For ease of reading I have eliminated many of the transcription conventions of the original examples. Transcription conventions include the marking of pauses, ellipsis, intonation, overlap, etc. I have also shortened some examples for illustrative purposes. I use pro to indicate null pronouns where López Ortega used the symbol: 0. As previously mentioned, I highlight the crucial element or elements under discussion in bold or underlined bold.

174 | 6 Discourse-pragmatics approaches in the acquisition of subject expression (136) Mi madre siempre me repite las cosas mil veces. Como my mother always me-dat repeat-3sg the things thousand times since ella es mayor, ella no suele tener buena memoria. she be-3sg older she no tend-3sg have-inf good memory ‘My mother always repeats things to me a thousand times. Since she is older, she tends not to have a good memory.’ (Adapted from López Ortega 2002: 221) In (136), the mother is introduced with a full lexical noun phrase and the speaker herself is referenced with the indirect object pronoun, me ‘to me’; the next mention of her mother is via an overt pronoun. It could be argued that since there are two possible referents in the discourse that this use is for contrastive purposes, even though the verbal morphology makes clear who the referent is. The next time the speaker mentions her mother via an overt pronoun, it is clearly redundant. This redundant use continues into AO’s final interviews. Furthermore, the use of overt subjects to mark emphasis and contrast actually decreases over time. This may be due to the fact that overall, her overt subject use decreases from the earlier interviews to the final ones. She also exhibits a ‘centrality to periphery’ pattern but it is not as strong and is more variable than that seen in BG’s discourse. Where AO shows the greatest convergence towards native-speaker-like strategies is in her use of overt first person pronouns for expressing an opinion, belief, or evaluation or to take a stand, especially after adversative conjunctions (aunque ‘although’, pero ‘but’), a common strategy in native speaker discourse. This use is illustrated in (137). (137) aunque yo vivo en un país diferente, la vida es más o although I live-1sg in a country different the life be-3sg more or menos la misma . . . no pro es lo mismo, pro estoy en Sevilla y less the same no pro be-3sg the same pro be-1sg in Seville and pro hace sol y me encanta el programa y la gente y pro make-3sg sun and me-dat love-3sg the program and the people and la cultura . . . pero también yo sé que pro necesito hacer the culture but also I know-1sg that pro need-1sg make-inf cosas things ‘although I live in a different country, life is more or less the same . . . [it]’s not the same, [I] am in Seville and [it]’s sunny and I love the program and the people and the culture . . . but I also know that [I] need to do things’ (Adapted from López Ortega 2002: 222)

6.3 Emphasis, contrast and disambiguation |

175

Inexplicably, this native-like strategy appears to give way to the ‘centrality to periphery’ pattern by AO’s final interview.

6.3.3.2 Intermediate low learners The two learners who began their study abroad experience at the intermediate mid level did not exhibit the ‘centrality to periphery’ strategy of pronoun use in their interviews, but rather tended to mark predominantly first person singular subjects, a ‘deictic pronoun strategy’, even when no intervening referents were introduced into the conversation. One of these learners (HM) is the one that showed the most linguistic progress over the four and a half month stay in Seville.³⁵ Her frequent use of overt first person pronouns, especially with verbs of belief, knowing, desire, emotion, etc., throughout the four interviews leads López Ortega to suggest that this speaker has a strong sense of identity and the greatest need to distinguish herself from other referents in her conversations using contrasting overt subjects. In the following passage (138), HM has been talking about what she likes to eat and she is already the topic of the conversation. This is further confirmed by the (obligatory) first person dative clitic pronoun, me ‘to me’, at the beginning of the excerpt. In the following utterance there is a switch of referent and she marks this with the full lexical noun phrase, mi señora ‘my (host) mom’, but when she switches back to herself, she uses the null pronoun since she is already the topic of the narrative and the verb morphology is unambiguous. However, in the final clause, which appears to be an aside to the interviewer, she uses the overt first person subject pronoun, yo ‘I’, for emphasis and clarification – a strategy much like native speakers use. This is also a strategy to identify and set herself apart from the others. (138) me gusta la comida porque mi señora cocina mucho de lo me-dat please-3sg the food because my lady cook-3sg much of it que pro quiero, pro es que yo soy judía that pro like-1sg pro be-3sg que I be-1sg Jewish ‘I like the food because my “señora” cooks much of what [I] want, [it] is that I am Jewish.’ (Adapted from López Ortega 2002: 250)

35 López Ortega believes that this is partly attributed to the fact that HM became very close to her host mother and daughter and was involved in and participated fully in her host family’s and the community’s activities.

176 | 6 Discourse-pragmatics approaches in the acquisition of subject expression HM exhibits the most complex narratives of all of the learners by the end of her program of study – narratives that include several intervening entities (both human and inanimate) interacting in various activities and to which the speaker expresses multiple reactions. Like AO’s earlier interviews, she also makes use of other linguistic means to express contrast, including conjunctions (también ‘also’, pero ‘but’, sin embargo ‘nevertheless’), temporal adverbials (ahora ‘now’, después ‘later’), and discourse markers (es que . . . ‘it is that . . . ’). These are often accompanied by overt subject pronouns. Her first two interviews are also marked by the overuse of redundant third person subject pronouns even when there is no need to disambiguate or contrast the referent with others. López Ortega explains that this is because her narratives are richer in the use of linguistic (i.e. third person anaphoric) referents unknown to the interlocutor. The other informants’ interviews tended to be marked for more frequent contextual (i.e. first person deictic) referents. Nonetheless, despite the fact that HM’s grammatical accuracy is still at the intermediate to advanced low level (according to her two OPI-style interviews), her discourse shows many of the features of native speaker discourse for using explicit pronouns for contrast and emphasis and shows a high level of discourse cohesion. The other intermediate low learner (JC) showed very little linguistic progress in terms of subject expression throughout her study abroad experience.³⁶ Overall she exhibits a high use of overt subject pronouns and this increases over time. Most common are first person singular pronouns, used mostly with verbs of opinion, belief, emotion, desire, etc., but over time spreads to dynamic verbs as well as exhibited in (139). (139) y mi señora anoche . . . ella sabía que yo estaba un poco and my lady last night she know-3sg-imp that I be-1sg-imp a little enferma . . . y para la cena yo comí arroz sick and for the dinner I eat-1sg-pret rice ‘and my host mom last night . . . she knew that I was a little sick . . . and for dinner I ate rice’ (Adapted from López Ortega 2002: 302) One peculiar feature of JC’s discourse is the use of overt subject pronouns in syntactic contexts which in native speaker discourse are not generally acceptable,

36 López Ortega explains that this learner was the least integrated into sevillano life and interacted mostly with the other American students. She also experienced a traumatic event when her host father suddenly passed away in the middle of the semester.

6.3 Emphasis, contrast and disambiguation |

177

such as subordinate conditional clauses (yo no sé si yo quiero continuar ‘I don’t know if I want to continue’), indirect interrogative clauses (yo no sé qué yo quiero hacer ‘I don’t know what I want to do’) (López Ortega 2002: 316), and relative clauses (edificios que todos están vacíos ‘buildings that all are empty’) (López Ortega 2002: 291). Of the four learners she is the one who seems to have developed a strategy of associating first person deictic uses with explicit pronouns, most likely due to L1 transfer but also because deictic referents are cognitively easier to access than third person referents.

6.3.4 Comparing the quantitative and qualitative findings The qualitative findings reveal results that had been masked in the group quantitative results. Whereas some learners decreased their use of overt pronominal subjects, others (specifically JC) increased their use. This explains the lack of a statistically significant change overt time. The expectation that learners tend to mark [+specific] and [+human/animate] referents and referents unknown to the hearer with overt subjects, and [−specific] and [−animate] referents with null subjects is confirmed in both the quantitative and qualitative data. Furthermore, Lopez Ortega found no evidence either in the quantitative or the qualitative analyses to support the claim that learners use overt subjects to disambiguate ambiguous and incorrect verb forms. In fact, she found that, like native speakers, for the learners “the semantic content of the verb, as well as pragmatic needs such as convergence or divergence, identification with the topic or the interlocutor appear to have more weight than morphological ambiguity” (López Ortega 2002: 322). Also, whereas in the quantitative data she found little development in the use of explicit pronominal subjects for emphatic and contrastive purposes, the qualitative data showed this to be a strong developmental pattern for three of the four participants. Finally, regarding the lack of a statistically significant interaction between overt pronominal subjects and first person in the quantitative analysis, in the qualitative study she found that this is where all participants coincided the most and where the most development occurred for three of the four learners. Only JC did not show improvement; her overall strategy was to use overt pronouns at all times for first person reference regardless of the need for emphasis, contrast or stance-taking. López Ortega claims that this development reveals that “learners seem to be learning the rules that govern explicit pronoun use for emphatic and contrastive purposes” and that “these two aspects are intimately connected to pragmatics of discourse and the speaker’s point of view” (2002: 322–323). Thus, while the learners had arrived in Spain with the morphological forms and syntactic rules of subject expression under control, the development that she docu-

178 | 6 Discourse-pragmatics approaches in the acquisition of subject expression mented revealed a convergence towards native speaker discourse rules that are also necessary for a coherent and cohesive discourse structure.

6.4 Pragmatic constraints of saliency, focus, and pragmatic weight López Ortega (2002) found irregular but important qualitative development among study abroad learners in the use of explicit pronominal subjects for emphatic and contrastive purposes and the use of first person for adding pragmatic weight to their utterances via verbs of perception, knowing, belief, among others. Lubbers Quesada and Blackwell (2009) delve further into the pragmatic constraints that condition the use of first person subject pronouns for marking referent saliency, contrastive focus and pragmatic weight. Specifically, they propose five pragmatic rules for Spanish that appear to constrain first person subject pronoun use in native speaker discourse defined by these pragmatic notions and then test them with L2 learners of Spanish. Their definitions of these notions are based on speakers’ perceived cognitive status of referents. For example, referent saliency refers to the cognitive status or accessibility of an entity that is in focus in the discourse structure. This is not to be confused with topic continuity, which can be in focus but is not necessarily so. The pragmatic notion of focus refers to the idea that new and non-presupposed entities and contrastive information carry a distinct status and must be prominent in the discourse. Finally, the notion of pragmatic weight refers to a speaker’s desire to increase their stake in an utterance whether it is an argument or a statement of belief. The authors analyze spontaneous guided oral narratives of native speakers and L2 learners to confirm whether native speakers are constrained by these rules for first person subject use, and examine whether L2 speakers of Spanish develop these pragmatic rules, which are not generally taught in the foreign language classroom.

6.4.1 The rule of salient referent The first rule proposed, salient referent, dictates that a null subject should be used whenever the intended referent is in focus, and is based on the work of Gundel, Hedberg, and Zacharski (1993) who explain subject expression as partially determined by the notion of focus of attention or saliency. A referent is salient if it can be assumed to be the focus of attention of both speech participants at a certain point in the discourse. Gundel, Hedberg, and Zacharski maintain that the in focus status correlates cross-linguistically with the appropriate use of minimal,

6.4 Pragmatic constraints of saliency, focus, and pragmatic weight | 179

unstressed and/or unmarked forms, including null subjects and unstressed clitic pronouns. In Spanish the in focus status most appropriately correlates with null forms. In examples (140) and (141), taken from the narrative of a young woman describing her high school boyfriend, the use of the null subject (in bold) is not due to a switch in referent but rather is the result of the entities already being in focus in the discourse and therefore highly salient. (140) Su nombre era, es Fernando López Aguirre y pro his name be-3sg-imp be-3sg Fernando López Aguirre and pro era el más guapo de la escuela . . . be-3sg-imp the most handsome of the school ‘His name was, is Fernando López Aguirre and [he] was the most handsome guy in the school . . . (141)

pro Creo que en quinto semestre, las chavas de sexto lo pro think-1sg that in fifth semester the girls of sixth him-acc acosaban mucho porque, como pro era muy guapo, pro bother-3pl-imp much because as pro be-3sg-imp very handsome pro lo acosaban demasiado him-acc bother-3sg-imp too much ‘[I] think in fifth semester, the sixth-semester girls would bother him a lot, because, since [he] was so handsome, [they] would bother him too much’ (Lubbers Quesada and Blake 2003)

In (140) the null pronoun, pro, does not refer to the subject of the prior clause, su nombre (‘his name’). Rather, the null pronoun in this example refers to the subject complement, the boy himself whose name is Fernando López Aguirre and who has been brought into focus in the discourse and is now clearly salient, thus warranting the null subject pronoun. In (141), the continuation of the same narrative, the young woman refers to herself with a null pronoun, pro, since both she and the boyfriend have been in focus since the beginning of the narrative. When she introduces a new entity, las chavas de sexto (‘the sixth semester girls’), they are marked with a full noun phrase; when she switches back to the boyfriend she uses a null subject pronoun again since he has been brought back to the focus of attention via an overt direct object pronoun, lo (‘him’). There is again a switch of topic back to the sixth-semester girls, but since they are also still in focus, a null subject is warranted.

180 | 6 Discourse-pragmatics approaches in the acquisition of subject expression 6.4.2 The rule of switch focus of attention The second rule, switch focus of attention, specifies that unstressed overt subject be used to switch the focus of attention from one referent/topic to another. This rule follows from Levinson’s (2000) neo-Gricean M(anner)-Principle, which suggests that speakers use a more marked, less minimal or more complex form to implicate a referent that a null would not. Blackwell points out that the “use of marked expressions reflects the influence of the relative saliency of the referents in the discourse on the speaker’s choice of anaphoric expressions; it also shows the speaker’s adherence to the M-principle . . . ” (Blackwell 2002: 254).³⁷ Although the notion of switch focus of attention is similar to switch reference or topic switch, defined by Cameron (1995) as “a relationship of same or different reference between two sequentially ordered subjects” (Cameron 1995: 1), Lubbers Quesada and Blackwell’s (2009) rule gives greater importance to the saliency of different referents. Whereas null subjects can be used in cases of non-topic continuity due to the fact that different referents can be in focus at the same time, a switch in the focus of attention to another entity requires an overt subject. In the following example, the speaker is describing her first boyfriend to the interviewer whom she met when he came into the classroom to greet one of her best friends. (142) se llama Jonathani . . . loi pro conocí porque proi refl-call-3sg Jonathan him-acc pro meet-1sg-pret because pro llegó un día a saludar a uno de mis mejores amigos y mi arrive-3sg-pret one day to greet-inf to one of my best friends and my amigoj entró . . . a una clase y éli se quedó friend enter-3sg-pret to a class and he refl-stay-3sg-pret haciéndome compañía y pro empezamos a platicar. making-me-acc company and pro start-1pl-pret to talk-inf ‘his name is Jonathani . . . [I] met himi because [hei ] came in one day to say hi to one of my best friendsj and my friendj came in . . . to the class and hei stayed to keep me company and [we] started to talk.’ (Lubbers Quesada and Blake 2003) She begins the discourse mentioning the name of the boyfriend and explains how she met him. When she mentions a new referent, she uses a full noun phrase to

37 See Blackwell 2002, Chapter 2 for a complete discussion of Levinson’s neo-Gricean M-principle, and Chapter 5 of the same volume for a revision of the M-principle as it applies to Spanish conversational and narrative discourse.

6.4 Pragmatic constraints of saliency, focus, and pragmatic weight | 181

introduce the referent and again uses a full noun phrase, mi amigo ‘my friend’, to disambiguate the referent (of the same gender and number), but when she switches the attention back to Jonathan, she uses an overt pronoun, él ‘he’, to signal this. In cases of first person reference, speakers can use either an overt as in (143) or a null subject pronoun as in (144). Morphology in these cases can also signal a switch focus of attention. (143) aunque en un principio éli no quería que proj although in a beginning he no want-3sg-imp that pro estudiara esto, sino que proi quería que proj study-1sg-imp-subj this but that pro want-3sg-imp that pro fuera médico como él, de todas maneras proi aceptó be-1sg-imp-subj doctor like he of all manners pro accept-3sg-pret lo que yoj decidí. it-acc that I decide-1sg-pret ‘although in the beginning he didn’t want that [I] studied this, but rather [he] wished that [I] were a doctor like him, anyway [he] accepted what I decided.’ (144) Un día él anduvo con una amiga que también le one day he be-3sg-pret with a friend that also her-dat gustaba él y solamente pro duraron un día, pero esto sí, like-3sg-imp he and only pro last-3pl-pret one day but this yes me lastimó. Y ya no le pro volví a me-acc hurt-3sg-pret and no longer him-dat pro return-1sg-pret to tener mucha confianza have-inf much trust ‘One day he was with a girlfriend that also liked him and [they] only lasted a day but, that really hurt me. And [I] didn’t trust him very much anymore.’ (Lubbers Quesada and Blake 2003) In (143), the speaker is talking about how his father didn’t want him to study arts but rather wished that he had become a doctor. The focus of attention is on the father, on what the father wants for the son. The speaker alternates back and forth between his father and himself and only the discourse context makes clear who the subject of each clause is, not the verbal morphology, which is ambiguous. However, when the speaker wishes to bring the focus of attention back to himself, he uses the overt pronoun, yo ‘I’, even though the morphology of the verb, decidí

182 | 6 Discourse-pragmatics approaches in the acquisition of subject expression ‘(I) decided’, makes that clear. In (144), the speaker is focusing on her boyfriend and a girl that also liked him. When she switches the focus of attention back to herself, she uses a null pronoun (although she has used an overt object pronoun, me ‘me’, in the previous clause which also serves to bring the focus of attention back to the speaker herself).

6.4.3 The rule of contrastive focus The third pragmatic rule proposed by Lubbers Quesada and Blackwell (2009) is that of contrastive focus, which conditions the use of a stressed overt subject to introduce new/non-presupposed information in discourse, or when a subject is in opposition to another referent. The notion of contrastive focus has been defined in several ways. Lambrecht (1994) maintains that focus conveys new information and involves the notions of presupposition and assertion; whereas Zubizarreta (1998) holds that the presupposed part of the utterance is the assumption that the speaker and hearer share and the nonpresupposed information is the prosodically prominent or focused element; finally, Hidalgo-Downing (2003) refers to focus as the most essential and most salient information in the discourse which can either be new information or information already introduced into the discourse. Lubbers Quesada and Blackwell define a focused pronoun as prosodically prominent and contrastive when its referent is in opposition to another (either expressed or implied) referent in the discourse. In (145), the speaker is discussing how he and his father spend their days; both referents have been introduced into the discourse and the discourse is mostly about the father and why he is so important to the speaker. In this particular discourse, the speaker is talking about how he and his father are different. The verbal morphology on the verbs in both clauses makes clear who each of the referents are but they are in opposition to each other so the speaker uses overt subject pronouns. In this particular discourse, there is also a case of switch reference but in addition is the notion of a contrast between the two referents and their opposing likes and activities. The same situation is observed in (146) where the speaker is describing her best friend and how they are no longer able to spend as much time together as she would like because one friend is in one location and the other is in a different location. (145) Él se la pasa todo el día en sus actividades deportivas, y he refl it-acc spend-3sg all the day in his activities sports and yo siempre estoy en la casa. I always be-1sg in the house ‘He spends all day on his sports activities and I am always at home.’

6.4 Pragmatic constraints of saliency, focus, and pragmatic weight

| 183

(146) Me gusta pasar mucho tiempo con ella aunque a veces pro me-dat like-3sg spend-inf much time with her although at times pro no tenemos tiempo ya que . . . yo estoy acá y ella está allá. no have-1pl time now that . . . I be-1sg here and she be-3sg there ‘I like to spend a lot of time with her even though sometimes [we] don’t have time now that . . . I am here and she is there.’ (Lubbers Quesada and Blake 2003) Finally, in (147) there is only one referent and the verbal morphology makes clear the reference is to the speaker herself. It is contrasting because the implication is that either the speaker does not like poetry or does not read Latin (or both), and she is contrasting herself (implicitly) with those who do appreciate poems in Latin. (147) ¿Qué voy a hacer yo con un poema en latín? what go-1sg to do-inf I with a poem in Latin ‘What am I going to do with a poem in Latin? (Lubbers Quesada and Blake 2003) The implicit contrast is seen in the following example from Luján (1999) where the null pronoun in (148a) would not be an acceptable answer to the question in (148) asking who is going to pay the bill, despite the fact that the verbal morphology makes clear who the referent is. Only the answer in (148b) would be acceptable. (148) –¿Quién va a pagar la factura? a. –*Pago b. – Yo pago – Who is going to pay the bill? a. –*pro Pay b. – [I] pay. (Adapted from Luján, 1999: 128) The implication is that there is an implicit contrast with someone else who may want or feel obligated to pay the bill.³⁸

38 Lubbers Quesada and Blackwell (2009) point out that word order also plays a role in either highlighting new information introduced into the discourse or implying contrastive focus. For example, in answer to the same question ¿Quién paga?, a preverbal overt pronoun as in exam-

184 | 6 Discourse-pragmatics approaches in the acquisition of subject expression 6.4.4 The rule of pragmatic weight Rule four is grounded is Davidson’s (1996) definition of the term pragmatic weight which explains that speakers use topicalized overt subject pronouns, usually in preverbal position, when they wish to express greater personal involvement or commitment to an utterance, take a stand or establish a taking of the floor at the beginning of a turn. This rule, pragmatic weight, for Spanish describes how speakers use overt subject pronouns with verbs of opinion, belief, claiming (creer ‘to believe’, pensar ‘to think’, decir ‘to say’), emotion (sentir ‘to feel’, querer ‘to want’), or knowledge (saber ‘to know’) to add pragmatic weight to an utterance. (149) La primera vez que pro me enamoré, y yo the first time that pro me-refl fall in love-1sg-pret and I creo que pro ha sido la única, . . . believe-1sg that pro have-3sg been the only, ‘The first time that [I] fell in love, and I believe that [it] has been the only time, . . . ’ (150) pro fue una amistad muy bonita y de hecho yo siento pro be-3sg-pret a friendship very beautiful and of fact I feel-1sg que a veces pro me hacía más caso a mí que a su that at times pro me-dat make-3sg-imp more case to me than to her novio. boyfriend ‘[it] was a really nice friendship and in fact I feel that at times [she] paid more attention to me than to her boyfriend.’ (Lubbers Quesada and Blake 2003) In both of the examples (149) and (150), there is no morphological ambiguity on the verb forms that justifies the use of the overt subject pronouns (yo ‘I’), nor is there a shift of attention or a contrastive focus due to the fact that the speakers, in both examples, are already talking about themselves. In fact, in (149) the immediately preceding clause is also in first person and the speaker uses a null pronoun;

ple (148b) implies a contrastive focus (Yo pago, no tú ‘I’ll pay, not you’); whereas a postverbal overt stressed pronoun (Pago YO lit. ‘Pay I’) introduces new, non-presupposed information. This follows from Zubizarreta (1998) that in Romance languages new or non-presupposed information receives nuclear stress and is placed in utterance-final position (Lubbers Quesada and Blackwell 2009: 121).

6.4 Pragmatic constraints of saliency, focus, and pragmatic weight | 185

therefore the subject of creo ‘[I] believe’ is a continuation of the same topic and the same referent. In (150), the speaker is discussing his friendship with a girl and the focus has been on his feelings for her; again there is no shift in topic or focus (although there is a shift in the syntactic subject of the previous clause which is a null expletive pronoun). In both cases, the speakers use an overt subject pronoun to express a greater stake or emotional commitment to their utterances. This pragmatic function is only available in the first person, and most commonly in the singular. If the purpose is to take a stand one cannot take a stand for another referent.

6.4.5 The rule of epistemic parentheticals The final rule proposed by Lubbers Quesada and Blackwell (2009), is the one that dictates how pronouns are used in expressions of epistemic parentheticals. Most research on subject pronouns does not take epistemic parentheticals into account due to the fact that they function outside the discourse. Lubbers Quesada and Blackwell include them in their study because subject pronouns used in epistemic parentheticals, although metapragmatic in function, appear to be constrained by the same pragmatic principles of focus and topic and the prosodic feature of emphasis. Specifically in this case, the use of null pronouns for epistemic parenthetical contexts lack both focus and topic and prosodic emphasis. Lubbers Quesada and Blackwell’s rule indicates that speakers use null subjects in referring to themselves in epistemic parentheticals to evaluate their previous or subsequent utterances. This rule also follows from Davidson (1996) who explains that certain verbs of perception (knowing, seeing, believing) can be used to mitigate or modify a speaker’s position on an utterance, or to clarify or reword an utterance. These types of verbs, when used as metapragmatic comments on speaker’s utterances, lose their truth-functional meaning. Spanish expressions such as no sé (‘I don’t know’ or ‘I dunno’) or digo (‘I say’ or ‘I mean’) serve as asides or as evaluative commentary on an utterance, which explains why they would occur most often with null pronouns and, like the pragmatic weight function, only occur with first person, usually singular verbal morphology as in (151) and (152), although first person plural is also a possibility as in (153).

186 | 6 Discourse-pragmatics approaches in the acquisition of subject expression (151)

pro Le agradezco a mi mamá y . . . no solamente por mí, pro her-dat appreciate-1sg to my mom and no only for me también pro hablo por mis hermanos que también, pro digo . . . mi also pro speak-1sg for my brothers that also pro say-1sg my mamá es enfermera, una profesión más o menos remunerada mom be-3sg nurse a profession more or less well-paid ‘[I] appreciate my mom and . . . not only for me, [I] also speak for my brothers who also, [I] mean . . . my mom is a nurse, a profession more or less well-paid’

(152) La persona con la que yo estoy, sí pro la quiero pero, pro no the person with her that I be-1sg yes pro her-acc love-1sg but pro no sé, como que esa vez fue muy especial para mí. know-1sg like that that time be-3sg-pret very special for me ‘The person that I am with, yes, [I] love her but, [I] dunno, like that time was very special for me.’ (153) pro me comentó, que cuando pro nos pro me-dat comment-3sg-pret that when pro 3-sg-refl enamoramos, pro digamos, ella sintió algo muy fell in love-3pl-pret pro say-1pl-subj she feel-3sg-pret something very extraño. strange ‘[she] told me that when [we] fell in love, as [we] say, she felt something very strange.’ (Lubbers Quesada and Blake 2003) Lubbers Quesada and Blackwell point out that epistemic parentheticals demonstrate speakers’ evaluation or a reflection of their own utterances and are therefore not prosodically emphasized (reflected more accurately by the English translation of no sé as ‘I dunno’) in the discourse or used with overt subject pronouns. Furthermore, epistemic parentheticals do not have the function of bringing a referent back into focus. In (154) this is evident where the first person singular verb of the epistemic parentheticals, pro no sé ’[I] dunno’ (marked in bold) are null subjects and in the clause immediately following is expressed with an overt first person singular subject pronoun, yo ‘I’ (underlined bold) is expressed to bring the referent, the speaker himself, back into focus.

6.4 Pragmatic constraints of saliency, focus, and pragmatic weight | 187

(154) tal vez en 10 o 15 años pro me veo ya casado, con uno o maybe in 10 or 15 years pro refl-1sg see-1sg already married with one or dos hijos . . . y pro me veo, pro no sé, trabajando en two children and pro refl-1sg see-1sg pro no know-1sg working in el extranjero con mi esposa, una casa bonita, con un jardín enorme, the abroad with my wife a house nice with a yard huge pro no sé, yo me veo en Manhattan o en Londres. pro no know-1sg I refl-1sg see-1sg in Manhattan or in London ‘maybe in ten or fifteen years [I] see myself already married, with one or two kids . . . and [I] see myself, [I] dunno, working abroad with my wife, a nice house, with a huge yard, [I] dunno, I see myself in Manhattan or in London.’ (Lubbers Quesada and Blake 2003)

6.4.6 Saliency, focus, and pragmatic weight in L2 learners’ narrative structure After identifying and proposing the five pragmatic rules that determine overt and null first person subject use, Blackwell and Lubbers Quesada (2009) examined 140 oral narratives of L2 learners from three different levels of proficiency and five native speakers of Mexican Spanish to ascertain to what extent the four different groups were constrained by these pragmatic rules for first person singular subject pronoun use. The oral narratives came from the data base generated by the UAQUCD Project for the Study of Spanish as a Foreign/Second Language in Mexico and the U.S.: The Acquisition of Discourse Competence (Lubbers Quesada and Blake 2003) and included retell narratives of a 3-minute portion of a Charlie Chaplin silent film, personal narratives, descriptions of an important person in their lives, and descriptions of the participants’ plans for the future. The authors identified each referential context according to one of the five pragmatic rules proposed and operationalized the contexts in the following manner. A first person subject pronoun, either overt or null, was identified as salient referent if it referred to a referent in focus and was prominent in the discourse. In other words, in a sentence such as (155), (155) Cuando yo tenía 15 años, pro jugaba al fútbol. when I have-1sg-imp 15 years pro play-1sg-imp to-the football ‘When I was 15 years old, [I] would play football.’ (Lubbers Quesada and Blackwell 2009: 125)

188 | 6 Discourse-pragmatics approaches in the acquisition of subject expression the subject pro of jugaba ‘would play’ is the same as the subject yo ‘I’ of tenía ‘had’ and is therefore prominent in the discourse. When a referent was different from the previous referent in focus in non-parallel constructions, that is, in a subordinate construction as in (156), it was considered switch focus. (156) Pero él sabe que yo no estaba felíz. but he know-3sg that I no be-1sg-imp happy ‘But he knows that I wasn’t happy. (Lubbers Quesada and Blake 2003) Subject pronouns that appeared in parallel constructions involving at least one first person referent were identified as contrastive focus as in (157). (157) Yo vivo en Detroit y él vive en Boston. I live-1sg in Detroit and he live-3sg in Boston ‘I live in Detroit and he lives in Boston.’ (Lubbers Quesada and Blake 2003) Subject pronouns that functioned to add pragmatic weight were identified as such with verbs of opinion, belief, claiming, knowing, or emotion, as in (158). (158) o sea, verla así, muy muy débil, muy vulnerable, . . . yo that is see-inf-her-acc like that very very weak very vulnerable I lo sentí muy feo. it-acc feel-1sg-pret very ugly ‘that is, to see her like that, very very weak, very vulnerable, . . . I felt it really bad’ (Lubbers Quesada and Blake 2003) And finally, first person subject pronouns were categorized as epistemic parentheticals when used with verbs of perception or speaking, as in (159). (159) pro No sé, muy cómodo. pues pro era pro no know-1sg well pro be-3sg-imp very comfortable ‘[I] dunno, well, [it] was very comfortable.’ (Lubbers Quesada and Blake 2003)

6.4 Pragmatic constraints of saliency, focus, and pragmatic weight | 189

6.4.7 Results of analysis of pragmatic constraints The overall results (displayed in Table 22) showed that in general all groups, even the lower proficiency learners, used more null subjects than overt first person subject pronouns in their narratives and, as in previous studies, the learner groups used a higher percentage of null subjects although chi-square tests did not reveal significant differences among the groups. Table 22. Overall pronoun distribution for first person singular subject pronoun referential contexts. Percentage Overt Group 1 – beginners Group 2 – intermediate Group 3 – advanced Group 4 – native speakers

Number Null

33 20 29 35

67 80 71 65

Word counts

1s SP referential contexts

7 643 9 315 16 118 10 281

123 199 377 289

df = 3, χ 2 = 5.4, p = 0.1449, Cramer’s V = 0.0759 (Lubbers Quesada and Blackwell 2009: 125)

Graphs 2 and 3 present the distribution of overt and null subject pronouns respectively for all groups according to pragmatic context. The solid bars represent contexts (switch focus, contrastive focus and pragmatic weight) where the overt subject pronouns are expected and the patterned bars represent contexts (salient referent and epistemic parentheticals) where a null subject is expected. Results reveal that the native speaker group performed as predicted by the pragmatic rules proposed. In fact, the use of overt and null subject pronouns was practically categorical and showed that pronoun use for first person subjects is constrained by pragmatic context. Overt subject pronouns were preferred for switch focus in 90 % of the instances, 100 % for contrastive focus, and 100 % of the time for pragmatic weight; and null subjects were used for salient referent contexts 93 % of the time and for epistemic parentheticals 100 %. The chi-square (χ 2 = 227.74, p < .0001, Cramer’s V = 0.8877, df = 4) confirmed that there was a statistically significant difference for type of pronoun used depending on pragmatic context. The results for the learner groups were far more varied according to proficiency level and pragmatic context but the figures reveal that as proficiency improves, so does the categorical distribution of null and overt subject pronouns according to pragmatic constraint. The beginning learners showed a preference

190 | 6 Discourse-pragmatics approaches in the acquisition of subject expression

100 90 80 70 OSR

60 50

OSF

40

OCF

30

OPW

20 OEP

10 0 G1

G2

G3

G4

(OSR = overt salient referent, OSF = overt switch focus, OCF = overt contrastive focus, OPW = overt pragmatic weight, OEP = overt epistemic weight) Graph 2. Distribution of overt first person subject pronoun by pragmatic rule – all groups.

100 90 80 70 NSR

60 50

NSF

40

NCF

30

NPW

20 NEP

10 0 G1

G2

G3

G4

(NSR = null salient referent, NSF = null switch focus, NCF = null contrastive focus, NPW = null pragmatic weight, NEP = null epistemic weight) (Adapted from Lubbers Quesada and Blackwell 2009: 126–128) Graph 3. Distribution of null first person subject pronoun by pragmatic rule – all groups.

only for the use of null subjects with salient referent, 70 to 30 %, and overt pronouns to express contrastive focus, but to a lesser degree: 61 to 39 %. They had the most difficulty with the switch focus context where they chose null subjects 44 % of the time compared to overt pronouns 56 % of the time. There were no instances of overt yo ‘I’ used for pragmatic weight. The chi-square analysis (χ 2 = 13.93,

6.4 Pragmatic constraints of saliency, focus, and pragmatic weight | 191

p < .003, Cramer’s V = 0.3365, df = 3) did reveal a statistically significant distribution of null and overt pronouns according to context for all pragmatic constraints except switch focus, whose distribution of pronouns was indicated at chance levels. Despite this, the distribution of pronouns showed only tendencies and was not to any extent categorical (except for the few instances of the epistemic parenthetical contexts, which were most likely unanalyzed phrases such as no sé ‘I dunno’). The authors reject the claim by Liceras and Díaz (1998) that the overuse of overt subjects is due to a production strategy that helps learners produce the correct verbal morphology. Lubbers Quesada and Blackwell found that morphological errors were as common with overt subjects as they were with null subject pronouns. They also found an overuse of null subjects for contexts that required overt pronouns (switch focus and contrastive focus) and the most morphological errors for the context of switch focus. The intermediate group showed improvement over the beginning learners with their categorical use of overt subject pronouns (100 %) for the contrastive focus contexts and a greater tendency for the use of overt subject pronouns (68 %) in switch focus contexts, although at 32 %, like the beginners, the intermediate learners also display an overuse of null subjects when the pragmatic rule dictates that an overt pronoun be used. There was, however, improvement among these participants for the use of null pronouns in salient referent contexts (84 % of the time) and for the epistemic parenthetical function (98 %), although again for the latter these were used mostly as unanalyzed phrases. There were also no instances of pragmatic weight contexts. The chi-square and Fisher’s Exact tests showed that the intermediate learners selected null and overt subject pronouns according to pragmatic constraint and that this was statistically significant (χ 2 = 78.29, p < .0001, Cramer’s V = 0.6272, df = 3, Fisher’s p = 3.357E−16). Finally, for some of the pragmatic constraints, the more advanced group behaved like the other two learner groups, but for other contexts they performed more like the native speakers. This group patterned like the other L2 groups for salient referent and contrastive focus contexts with 80 % preference for null subjects for salient referent contexts, and 96 % use of overt subject pronouns for contrastive focus. For the switch focus context, however, this group used overt subject pronouns almost categorically (in 96 % of the cases), a behavior very similar to that of the native speakers. Another divergence from the lower proficiency groups and convergence toward the native speaker performance of this group was the use overt subject pronouns for pragmatic weight, a function not observed for the other groups. The statistical analysis confirmed that there was a significant difference in how the more advanced learner group selected pronouns according to pragmatic constraint (χ 2 = 174.69, p < .0001, Cramer’s V = 0.6807, df = 4).

192 | 6 Discourse-pragmatics approaches in the acquisition of subject expression Even in beginning stages, L2 learners are not using null and overt subject pronouns indiscriminately but appear to be constrained by at least some of the pragmatic contexts that constrain native speaker use. The participants from the beginning group produce null subjects for all contexts but prefer overt subject pronouns most notably for contrastive focus and null pronouns for salient referent contexts. By the intermediate level, learners were still not producing overt subject pronouns to express switch focus like advanced learners and native speakers and both the beginning and intermediate groups had not learned to use overt subject pronoun use for expressing pragmatic weight; only the advanced group had. Curiously, the advanced learners were still overusing overt subject pronouns to express a referent that was already salient in the discourse. The authors conclude that for both native speakers and L2 learners the use of overt/null subject pronouns in Spanish is neither optional nor random but rather depends on discourse constraints of referents to determine subject pronoun use, although the strength of the pragmatic constraints is different for native speakers and for learners depending on the proficiency level at different stages of learning.

6.5 Cognitive status and the Givenness Hierarchy In a subsequent study, Blackwell and Lubbers Quesada (2012) examined the use and acquisition of null and overt third person subjects, both singular and plural. The pragmatic constraints examined in their previous work, most notably the evaluative functions of pragmatic weight and epistemic parentheticals, are not valid for other persons. Furthermore, in their (2012) study they wanted to ascertain to what extent discourse-pragmatic constraints (saliency and topic/focus) interact with cognitive constraints (memory, attention and knowledge) to determine subject use in story-telling narrative structure, which focuses mostly on third persons. Their objective was to test the predictions of Gundel, Hedberg, and Zacharski’s (1993) Givenness Hierarchy, which proposes six cognitive statuses that constrain the form and distribution of referring expressions in discourse. The cognitive statuses (see 160) are a set of implicationally related statuses based largely on underlying assumptions cooperative speakers make about the informativeness and saliency of referents in discourse, and about speakers’ presumed assumptions about their addressee’s knowledge and attention state where referential subject expressions are used. Claims by Ariel (1990), Givón (1983), and Levinson (1987, 2000) suggest that greater referent saliency corresponds to the use of more phonetically minimal referring expressions and lesser referent saliency is consistent with more phonetically elaborate expressions.

6.5 Cognitive status and the Givenness Hierarchy

|

193

(160) The Givenness Hierarchy (Gundel, Hedberg, and Zacharski 1993: 275) in focus > activated > {it}

uniquely familiar > identifiable > referential >

{that, this, this N} {that N}

{the N}

type identifiable

{indefinite this N} {a N}

Gundel, Hedberg, and Zacharski explain that when using a referring expression, it is assumed that it corresponds to the cognitive status for that category while the status(es) lower on the hierarchy are also met. That is to say that if a speaker uses a demonstrative determiner with a noun, that dog, then it is assumed that the addressee is ‘familiar’ with that referent and that it is also ‘uniquely identifiable’, ‘referential’ and ‘type identifiable’, but is not ‘activated’ or ‘in focus’. Thus, the hierarchy predicts an implicational relationship among the cognitive statuses such that any particular referent entails statuses lower on the hierarchy but not the other way around.

6.5.1 Cognitive status of type identifiable and referential Type identifiable is a necessary cognitive status for any nominal expression and in English is sufficient for the use of the indefinite article a. For example in (161), the use of a dog is sufficient for the addressee to be able to link the expression to the type of thing to which it refers, assuming the addressee knows the meaning of the word dog. (161) I couldn’t sleep last night. A dog (next door) kept me awake. (Gundel, Hedberg, and Zacharski 1993: 276) In English, the cognitive status, referential, is necessary for the corresponding use of all definite expressions and specifically for the use of the indefinite this N in informal English, as in (162). An addressee must access both the appropriate type of representation and recover a previous representation of the speaker’s intended referent or form a new representation as soon as the utterance has been processed. In Spanish, there are no referring expressions that distinguish it from the cognitive status of type identifiable; therefore, Blackwell and Quesada subsumed these two statuses into one. (162) I couldn’t sleep last night. This dog (next door) kept me awake. (Gundel, Hedberg, and Zacharski 1993: 277)

194 | 6 Discourse-pragmatics approaches in the acquisition of subject expression 6.5.2 Cognitive status of uniquely identifiable Uniquely identifiable is a necessary cognitive status for the use of definite referents and corresponds to the use of a definite article and noun. In (163), the phrase the dog could be used if in the addressee’s memory there is a previously existing representation, but by adding next door, the referent is sufficiently identified by the speaker. Thus, it is possible for the addressee to recover or build a new representation of the referent based on the definite noun phrase. (163) I couldn’t sleep last night. The dog (next door) kept me awake. (Gundel, Hedberg, and Zacharski 1993: 277)

6.5.3 Cognitive status of familiar The cognitive status, familiar, is required for the use of personal pronouns and definite demonstratives, but is also sufficient for the use of the demonstrative determiner that, as in (164) if the addressee knows that the speaker’s neighbor has a dog. The addressee can therefore identify the intended referent due to the fact that there is a representation of it in either in short-term memory if the referent was mentioned recently in the discourse, or in long-term memory if not. (164) I couldn’t sleep last night. That dog kept me awake. (Gundel, Hedberg, and Zacharski 1993: 278)

6.5.4 Cognitive status of activated The status of activated is necessary for a referent to be referred to via all pronominal forms, both stressed and unstressed, and is necessary for the use of the demonstrative pronouns. An activated referent is represented in short-term memory, which can be recovered from long term-memory or from the immediate discourse or the extra-linguistic context. For example, in (165) the use of that is only appropriate if a dog’s barking had been mentioned in the discourse or had been barking during that the time of the speech event. (165) I couldn’t sleep last night. That kept me awake. (Gundel, Hedberg, and Zacharski 1993: 278)

6.5 Cognitive status and the Givenness Hierarchy

|

195

6.5.5 Cognitive status of in focus Finally, for the use of null and unstressed pronouns the cognitive status of in focus is necessary. That is because the referent is in short-term memory due to the fact that it is the current center of attention during the discourse. Therefore, referents that are in focus usually are the topic of the immediately previous utterance or other topics that are highly salient or relevant in the discourse. Gundel, Hedberg, and Zacharski point out that although topics are generally also in subject position, this is not necessarily so. They use the term ‘in focus’ “to refer to the psychological notion of focus of attention” (Gundel, Hedberg, and Zacharski 1993: 279) which often, but not always, intersects with the linguistic notion of focus as the most prominent part of a sentence precisely because “elements tend to be linguistically focused because the speaker wants to bring them into the focus of attention” (1993: 279). The authors point out that because both prosodic form and syntactic structure encode topic-comment structure and emphasize referents that the speaker wishes to focus on, linguistic form determines which elements belong to the in-focus set. Thus in (166a), the bull mastiff has been introduced in the subject position and is therefore a candidate for in-focus status in (166b) and can be referred to with the pronoun it. But since the same referent has been introduced as the object of a preposition in (167a), the use of it in (167b) is infelicitous. (166) a. My neighbor’s bull mastiff bit a girl on a bike. b. It’s the same dog that bit Mary Ben last summer. (167)

a. Sears delivered new siding to my neighbors with the bull mastiff. b. #It’s the same dog that bit Mary Ben last summer. (Gundel, Hedberg, and Zacharski 1993: 280)

Pragmatic considerations can, however, override syntactic considerations for what can bring a referent into focus. In (168b) it also refers to the object of the prepositional phrase in (168b), a large wind energy project, but because of its importance to the entire discourse, it is clear from sentence (168b) that the prepositional object has been brought into focus and can therefore be referred to with the minimal referring expression, it. (168) a. However, the government of Barbados is looking for a project manager for a large wind energy project. b. I’m going to see the man in charge of it next week. (Gundel, Hedberg, and Zacharski 1993: 280)

196 | 6 Discourse-pragmatics approaches in the acquisition of subject expression 6.5.6 A revised hierarchy for Spanish The hierarchy proposed by Gundel, Hedberg, and Zacharski is expected to be universal although not all statuses are necessary for all languages. Of the five languages examined by the authors, only English has a form, this N, which requires a referential status. Chinese has forms that require uniquely identifiable and type identifiable, whereas Japanese and Russian subsume uniquely identifiable, referential and type identifiable for bare nouns (with no preceding determiner). Spanish subsumes referential and type identifiable for bare nouns and indefinite noun phrases. In addition, Blackwell and Lubbers Quesada (2012) suggested a revised hierarchy for Spanish adding to the status of activated two more subtypes: activated and recoverable and activated and non-recoverable. They proposed the substatus of activated and recoverable to account for instances where an intended referent is not the focus of attention and yet a null subject can be used to refer to the referent based on other linguistic and extra-linguistic factors. The sub-status of activated and non-recoverable accounts for referents that are activated in the immediately preceding clause but are not recoverable with a pronominal form due to the fact that there are two or more competing activated referents. These include more semantically general noun phrases such as el otro ‘the other’, los dos ‘the two’, uno de ellos ‘one of them’. The revised hierarchy appears in (169). (169) Revised Givenness Hierarchy for Spanish in focus > activated >

familiar >

uniquely referential/ identifiable > type identifiable

{pro/él} {ÉL/éste/ése/aquél/este N} {ese/aquel N} {el N} activated + recoverable {pro} activated + non-recoverable {el/los/uno de N/pronoun}

{0 N/un(a) N}

(Adapted from Gundel, Hedberg, and Zacharski 1993: 284; Blackwell and Lubbers Quesada 2012: 149–150)

6.5.7 The revised hierarchy and predicting subject use in L2 learners’ narratives Based on the revised hierarchy, Blackwell and Lubbers Quesada (2012) set forth hypotheses to predict subject expression in oral narratives of native speakers and L2 learners according to the five cognitive statuses and two sub-statuses of en-

6.5 Cognitive status and the Givenness Hierarchy |

197

tities to which the expressions might refer. They predicted that null pronouns would be used to refer to entities that were either in focus, or activated and recoverable (entities not in focus but understandable from the discourse context with a null subject); definite (including demonstrative) pronouns correspond to the activated status; semantically general noun phrases (los dos ‘both’, el otro ‘the other’) would be used with activated and non-recoverable referents (not retrievable with a personal overt pronoun alone); definite noun phrases or proper names would be used for both the familiar and uniquely identifiable statuses (el sol); and indefinite noun phrases would refer to referential and type identifiable referents. The authors analyzed the use of null and overt subjects in the third person, both singular and plural, for all subject expression. Because their interest is in testing the Givenness Hierarchy for all cognitive statuses, they include null and overt subject pronouns, definite and indefinite noun phrases and proper names. They examined 40 oral film retell narratives of the same L2 learners (Group 1 = beginning, Group 2 = intermediate, Group 3 = advanced) as their previous study and 10 native speakers of Mexican Spanish (Group 4). The film retell of a short clip of a silent Charlie Chaplin film, A Woman, was selected for this study in order to control for the narrative genre, to assure a similar discourse structure and third person subject reference, and because there are several protagonists participating in numerous activities, interacting with each other and with several objects throughout the film clip. The participants included a variety of possible human discourse referents, including: a family consisting of a husband, a wife, and a daughter; Charlie Chaplin; a flirty woman; two other men; and a policeman. There are also several possible inanimate referents, including: a sprinkler, a cane, a bench, a bottle of soda, a blindfold, and a lake. In the end, their analysis was based on a total of 1902 third person subject forms among the four groups studied. Graphs 4–10 exhibit a summary of their findings for each cognitive status. In general, as they claim, for both native speakers and learners the selection of subject forms appears to be constrained by the cognitive status of referents as predicted by the Givenness Hierarchy.

6.5.8 Results of testing the revised hierarchy The clearest patterns of use that emerged were for the extremes of the hierarchy where speakers selected forms as predicted by the hierarchy; that is, they selected null pronouns for entities that are in focus and indefinite noun phrases for referents that are type identifiable. Native speakers showed the strongest adherence to the predictions of the hierarchy and selected null subjects more than 90 % for in focus entities, whereas the learner groups ranged from 61 to 65 % in their selec-

198 | 6 Discourse-pragmatics approaches in the acquisition of subject expression tion of null subjects for this cognitive status (see Graph 4). The logistic regression showed a statistically significant difference between learners and native speakers for the use of null subjects for the in focus status. For the use of indefinite noun phrases for type identifiable referents there were no statistically significant differences among the four groups (see Graph 5). All groups produced indefinite noun phrases for referents that are type identifiable between 77 and 91 % of the time. Although the hierarchy predicts that type identifiable referents would be referred to with an indefinite noun phrase, occasionally learners and native speakers introduced known entities for the first time with definite noun phrases such as el policía ‘the police’ or la mujer ideal ‘the ideal woman’.

100 80 G1 60 G2 40

G3 G4

20 0 Null

Overt Pro

GenNP

Def NP

PrpN

IndNP

df = 3, χ 2 = 100.2005, p < .0001 (Adapted from Blackwell and Lubbers Quesada 2012: 154) Graph 4. In focus status (expect null pronouns).

There were also tendencies to select subject forms according to cognitive status for referents that are activated, familiar and uniquely identifiable, however, the patterns of use for these statuses were less strong for all participant groups. The authors did, however, find that the learner groups behaved in a similar manner to the native speaker groups for these three cognitive statuses. For the activated status, the hierarchy predicted the use of overt pronouns and all groups selected overt pronouns for this status but they also selected definite noun phrases and to a lesser extent proper nouns (in this case, only Charlie Chaplin or Chaplin). Although the authors observed that learners diverged from native speakers and selected fairly evenly between both overt pronouns and definite noun phrases for activated entities, they reported no statistically significant differences between learners and native speakers (see Graph 6).

6.5 Cognitive status and the Givenness Hierarchy

|

199

100 80 G1 60 G2 40

G3 G4

20 0 Null

Overt Pro

GenNP

Def NP

PrpN

IndNP

df = 1, χ 2 = 1.1602, p < .2814, Fisher’s p = .3674 (Adapted from Blackwell and Lubbers Quesada 2012: 161) Graph 5. Type identifiable status (expect indefinite NP).

100 80 G1 60 G2 40

G3 G4

20 0 Null

Overt Pro

GenNP

Def NP

PrpN

IndNP

df = 5, χ 2 = 24.9720, p < .0001, Fisher’s p = .00012 (Adapted from Blackwell and Lubbers Quesada 2012: 157) Graph 6. Activated status (expect overt pronouns).

For the familiar and uniquely identifiable statuses, all participant groups tended to select either definite noun phrases or proper nouns as predicted by the hierarchy. Although the learner groups followed the same tendencies, the native speakers’ results were more robust and more closely aligned with the predictions of the hierarchy. The logistic regression revealed that there were statistically significant differences between learners and native speakers for the use of subject forms for familiar and uniquely identifiable statuses (see Graphs 7 and 8).

200 | 6 Discourse-pragmatics approaches in the acquisition of subject expression

100 80 G1 60 G2 40

G3 G4

20 0 Null

Overt Pro

GenNP

Def NP

PrpN

IndNP

df = 3, χ 2 = 19.0302, p < .0003 (Adapted from Blackwell and Lubbers Quesada 2012: 159) Graph 7. Familiar status (expect definite NP).

100 80 G1 60 G2 40

G3 G4

20 0 Null

Overt Pro

GenNP

Def NP

PrpN

IndNP

df = 1, χ 2 = 5.7237, p < .0167, Fisher’s p = .0183 (Adapted from Blackwell and Lubbers Quesada 2012: 160) Graph 8. Uniquely identifiable status (expect definite NP or proper noun).

Where the native speakers and the learner groups differed the greatest were with the activated and recoverable and activated and non-recoverable statuses. For the prior status, all participant groups used a wide range of subject forms but the native speakers used mostly null subjects (82 % of the time) as predicted by the revised hierarchy (see Graph 9). Whereas only the advanced group preferred null subjects above other forms for the activated and recoverable status, they did so to a lesser degree than the

6.5 Cognitive status and the Givenness Hierarchy

|

201

100 80 G1 60 G2 40

G3 G4

20 0 Null

Overt Pro

GenNP

Def NP

PrpN

IndNP

df = 4, χ 2 = 45.2529, p < .0001, Fisher’s p = .00015 (Adapted from Blackwell and Lubbers Quesada 2012: 155) Graph 9. Activated and recoverable status (expect null subjects).

native speakers (only 59 % of the time). The intermediate group chose mostly overt pronouns and beginners selected mostly proper names to refer to referents that were activated and recoverable. Native speakers and to some degree more proficient L2 learners are able to assign reference to cognitively activated entities by means of null subjects when the discourse context and semantics makes it clear who the referent is. Lower proficiency learners tend to use a more elaborate form for this cognitive status as they have not yet learned how to decode the semantics and pragmatics of the discourse. The hypothesis that speakers would use a semantically general noun phrase (e.g. uno ‘one’, uno/a de ellos/ellas ‘one of them’, los dos ‘the two’ or ‘both’) to refer to referents that are activated and non-recoverable was confirmed only for the native speakers and intermediate learners, although for this latter group there were very few tokens; the beginning and advanced groups used more proper nouns (see Graph 10). The logistic regressions for the use of subject expressions for these two cognitive statuses (activated and recoverable and activated and nonrecoverable) revealed that there were significant differences between learners and native speakers. The graphs (9 and 10) that summarize the use of different subject forms for these two cognitive statuses also present visual confirmation of the differences among the four groups. Thus, the authors conclude that although all participant groups tended to associate subject forms with the appropriate cognitive status of the hierarchy, with the exception of the type identifiable status, there were significant differences between the learner groups and the native speakers for assigning referents to a

202 | 6 Discourse-pragmatics approaches in the acquisition of subject expression

80 70 60 G1

50 40

G2

30

G3

20

G4

10 0 Null

Overt Pro

GenNP

Def NP

PrpN

IndNP

2

df = 2, χ = 7.6051, p < .0223 (Adapted from Blackwell and Lubbers Quesada: 158) Graph 10. Activated non-recoverable status (expect semantically general NPs).

variety of competing subject forms. In general, the native speakers opted for the less specific forms (null or overt pronouns) on more occasions than the learners who selected more specific forms (definite noun phrases or a proper name). If the Givenness Hierarchy is universal, as Gundel, Hedberg, and Zacharski maintain, then it appears that L2 learners have access to this knowledge, although there may be an interaction of both L1 and universal knowledge. In fact, the learners performed as the Givenness Hierarchy, and specifically as the revised hierarchy, predicted but also because their L1 is constrained by some of the same statuses and similar subject forms. Nonetheless, as the authors point out, when the hierarchy does not predict learner use for native speakers and learners, native speakers tend to use the next, more minimal form and learners tend to use the next, more elaborate form, going against both the predictions of the Givenness Hierarchy and their L1. Learners had the most difficulty with assigning an appropriate subject form to the activated status(es), which is a more cognitively challenging task because there are more referents to keep track of and speakers are confronted with a wider range of linguistic choices to express these cognitive notions. It appears that the learners resort to more elaborate forms when unsure; whereas native speakers tend toward the more minimal forms. This study demonstrates that third person subject expression for both learner groups and native speakers appears to be constrained by the cognitive status of referring entities and that the L2 acquisition of subject referential marking is a process of replacing more elaborate forms with more minimal ones within the cognitive status hierarchy.

6.6 Contributions of findings of the discourse-pragmatics approach |

203

6.6 Contributions of findings of the discourse-pragmatics approach What research from a discourse pragmatic perspective has revealed is that it is not sufficient to examine subject expression at just the clausal level but that it is necessary to consider the entirety of narrative structure in order to make claims regarding L2 learners’ acquisition of subject expression. We have seen that the complexity of the narrative and the number and role of the intervening referents can have an effect on when and which type of subject expression is used. Furthermore, we saw that narrative structure is preserved mainly via an entire range of subject forms. Finally, it was seen that the choice of a subject form reflects a speaker’s presuppositions regarding the topic or status of the referents or what they believe to be their listener’s knowledge of the referents. López Ortega (2002) discovered that like native speakers, learners mark specific and human/animate referents, as well as referents unknown to the hearer, with overt subjects, and nonspecific, non-animate and known referents with null subjects. Saunders (1999) and Blackwell and Lubbers Quesada (2012) both proposed that native speakers and learners are guided by a type of universal hierarchy for subject expression. Saunders’ proposed an anaphoric hierarchy based on what the speaker supposes the listener already knows about a topic and referents in discourse. Blackwell and Lubbers Quesada followed Gundel, Hedberg, and Zacharski’s (1993) Givenness Hierarchy (and their revised hierarchy based on the GH) and proposed that the selection of subject forms depends on underlying assumptions cooperative speakers make about the informativeness and saliency of referents in discourse, and about speakers’ presumed assumptions not only about their addressee’s knowledge of a topic and referents, but also about their addressee’s attention state regarding potential referents. Both hierarchies suppose an implicational relationship among forms. For Saunders the higher up in the hierarchy a referring expression is, the less information the speaker assumes the listener has regarding that referring expression’s potential referent; thus indefinite noun phrases presuppose little or no information regarding the referent and a null form presupposes a maximum amount of information. A similar proposal is set forth by Blackwell and Lubbers Quesada (2012), among others (Ariel 1990; Givón 1983; Levinson 1987, 2000). In other words, greater referent saliency corresponds to the use of more phonetically minimal referring expressions and lesser referent saliency is consistent with more phonetically elaborate expressions. Furthermore, whereas native speakers of Spanish assume that the maintenance of the same subject is signaled via a null subject, L2 learners will often violate the anaphoric hierarchy and redundantly use overt subjects. Although

204 | 6 Discourse-pragmatics approaches in the acquisition of subject expression not grammatically inappropriate, from the point of view of cohesion theory this overuse of subjects creates new nodes to unnecessarily form cohesive ties, which results in a narrative structure that requires greater processing effort on the part of the listener. The research reveals that native speakers and L2 learners adhere to rules that condition the use of subject forms in narrative discourse. Saunders showed that the choice of form is determined by the role, familiarity and relative importance of the character in the story; by the function of that character during different points throughout the narrative, whether the character is being introduced, maintained or reintroduced; and by the syntactic position of the referent. López Ortega and Lubbers Quesada and Blackwell found the use of overt first person subject pronouns was linked to verb semantics and the speaker’s commitment to certain beliefs, knowledge and emotions. The need to contrast two or more intervening referents was also a strong predictor for the use of overt subjects. Furthermore, Blackwell and Lubbers Quesada revealed that the choice of subject forms in narrative discourse is largely determined by discourse-pragmatic constraints such as saliency and topic and focus and that these interact with cognitive constraints that have to do with memory, attention and knowledge. What was found was that the hierarchies proposed were accurate at predicting native speaker use almost categorically and they were good at predicting L2 learners’ tendencies use of subjects in many cases. Throughout the studies, it was evident that learners, especially in the lower proficiency levels, tend to use more elaborate forms than dictated by the hierarchies. Ultimately, they tend to overuse overt subject pronouns when a null subject would be more appropriate and they tend to overuse lexical subjects when a pronominal subject would be more suitable for the discourse context. In addition, Saunders found that her lower level learners do not distinguish between indefinite and definite noun phrases for introducing a character and overuse definite noun phrases to maintain that character. Similarly, Blackwell and Lubbers Quesada found that for the activated status, learners tended to use more definite noun phrases and proper nouns than the native speakers who used overt pronominal subjects. Both Saunders and Blackwell and Lubbers Quesada found that learners do not violate the order of the proposed hierarchies for subject expression; that is, they do not use a form that is higher on the hierarchy to maintain a character than the form they have used to introduce that character. A general pattern found in the Blackwell and Lubbers Quesada study was that learners tend to use the next, most elaborate form on the hierarchy. For example, when a null subject would be the preferred option for a native speaker, the learner will choose an overt pronominal subject but not a lexical noun phrase or proper name. When an overt pronominal

6.6 Contributions of findings of the discourse-pragmatics approach |

205

subject is the most appropriate form, then they will tend to use a lexical noun phrase or proper name. Finally, Lubbers Quesada and Blackwell found that saliency of referent, contrastive focus, and switch focus were predictors for subject expression for both native speakers and L2 learners, although the strength of the pragmatic constraints were different for native speakers and learners and different for lower proficiency and higher proficiency learners. Beginning learners prefer overt subject pronouns only for contrastive focus contexts and null pronouns only for salient referent contexts; intermediate learners tend to use more overt pronouns for switch focus contexts but only advanced learners and native speakers use overt subject pronouns for expressing pragmatic weight. All learner groups overused overt subject pronouns for same referent, i.e. a referent that was already salient in the discourse. This research shows again that learners are capable of mastering early the most evident distinctions, such as contrastive focus and salient referent. It is when they must deal with the less evident distinctions such as switch focus, which is neither contrastive nor salient, that learners have difficulties learning the finer grained distinctions for subject expression. Research in L2 acquisition and use of subject expression in Spanish has revealed that it is not only knowledge of syntactic properties that leads to competence in an L2, but also the capacity to use this knowledge in discourse appropriate ways. Linguistic competence includes knowledge of syntactic properties of language, which may be universal, but also knowledge of discourse constraints or pragmatic rules that guide appropriate language use and acquisition. Specifically this knowledge includes knowing how sequential organization and coherence in discourse structure is produced and understood (Levinson 1983) and how addressees’ cognitive states are considered. To what extent this type of linguistic knowledge of discourse constraints is universal will be explored further in Chapter 8.

7 A variationist approach to L2 subject pronoun acquisition Introduction Like the studies from a discourse-pragmatic perspective, variationist research into the L2 acquisition of Spanish subjects examines a wider range of forms than the generative and processing studies and takes into account the functions of subject forms within different discourse contexts or narrative genres. Therefore, these two approaches could very easily be subsumed into a single functional approach. However, the research objectives and methodological designs and analyses of the variationist approach are distinct from the discourse-pragmatic one in several ways. Whereas the discourse-pragmatic approach focuses on speakers’ presuppositions, perceived intentions and cognitive state of mind, the variationist approach attempts to objectively operationalize the (dependent and independent) variables under study. Thus, whereas the discourse-pragmatic approach at times straddles the fence between qualitative and quantitative data collection and analysis, the variationist approach is, according to Labov (1969), inevitably quantitative. This is because large samples of language are necessary for identifying patterns in discourse. Therefore, another difference between the two approaches is that variationist studies in second language acquisition (SLA), like those in native speaker research, make use of multivariate regression analyses that are capable of handling large numbers of tokens for variable structures. Multivariate statistical tests provide a probabilistic weight that can predict when a speaker is likely to select between one of two possible structures when certain independent variables are present. These statistical tools help researchers establish the effect of a broad range of linguistic and social factors that interact to influence the acquisition and use of particular forms (Bayley and Tarone 2012). More recent work in L2 studies has used a multinomial probit model, which is a multivariate statistical test that permits the analysis of dependent variables with more than just binary categories (Gudmestad, House, and Geeslin 2013). The fact that little research from this perspective has been carried out in the second language acquisition of subject expression in Spanish is surprising given that sociolinguistic and variationist models have been the principle theoretical basis for much of the research in subject pronoun use among native and heritage Spanish speaker populations (e.g. Abreu 2009, 2012; Ávila-Shah 2000; Bayley and Pease-Álvarez 1997; Bentivoglio 1987; Cameron 1994, 1995; Cameron and FloresFerrán 2004; García Salido 2008; Hochberg 1986; Hurtado 2005; Otheguy and Zen-

7 A variationist approach to L2 subject pronoun acquisition

|

207

tella 2007; Ranson 1991; Silva-Corvalán 1982, 1994; Torres Cacoullos and Travis 2011; Travis 2007; Travis and Torres Cacoullos 2014; among others). Furthermore, as Gudmestad points out, “variationism is suitable for SLA, in part because of three tenets that first-language (L1) and L2 research share: A linguistic system is rule governed, undergoes change and is impacted by social factors” (Gudmestad 2013: 81). The emerging body of work on the L2 acquisition of subject expression in Spanish from a variationist perspective has revealed that both the frequency of a variant and its conditioning linguistic and social factors are important to our understanding of the L2 acquisition process. The assumption among sociolinguistic studies is that because variability of non-native speaker language is most likely seen in contexts where native speaker language does not vary, analyses of learner language must include all the contexts for subject expression and not just a subset of contexts. The term variation was first applied to the description of second language acquisition by Tarone (1982, 1983, 1985a, 1988, 1990) and Ellis (1985, 1988, 1989, 1990), who claimed that models of second language acquisition necessarily had to explain learners’ apparent variable competence. Sharply criticized by L2 acquisition researchers from a generativist perspective (Gregg 1990; Jordan 2004), this early model of variable competence sought to determine to what extent learners’ intuitions of the second language and their production vary at different stages of development. Both the L2 variable competence models of Tarone and Ellis and the sociolinguistic/variationist approach discussed in this chapter hold that learners often vary in their use of forms in ways that native speakers do not. The model put forth by Tarone, however, predicts that learners at different stages of development have different representations of forms depending on the amount of attention they are paying to their speech; they could even have several representations at one time that are in conflict with each other. Ellis, on the other hand, suggests that learners adjust their speech along a continuum of styles or registers, although these registers differ from those used by native speakers. Preston (1989, 1993) early on advocated for a sociolinguistic model for analyzing variation in interlanguage, but at the same time, admits that social factors alone cannot account for this variability. He mentions that in as much as “. . . a developing interlanguage is a natural language, it will require the sorts of descriptive apparatuses made available by a variationist grammar” (Preston 1989: 198), but that the factors involved in native speaker variation are not necessarily the same for a non-native speaker. Preston (1989) proposes analyzing interlanguage variation at different levels (phonological, syntactic, lexical, etc.) from a variety of sociolinguistic issues: social context, task, topic, interlocutors, attention to task or form, ethno linguistic background, etc. and how these factors interact with purely linguistic factors, such as linguistic environment (phonological or syntactic) and function-form relationships. In

208 | 7 A variationist approach to L2 subject pronoun acquisition reviewing data from other studies he concludes, however, that at the early stages of second language acquisition “linguistic rather than social environments are more powerful influences” (1989: 256). The sociolinguistic/variationist model that is examined in this chapter follows from Preston (1993, 2002) and Bayley and Preston (2008) and examines the development and production of linguistic forms in terms of probability. According to Gudmestad, “each linguistic and social factor that conditions the use of a variable structure has a probabilistic weight that aids in determining the likelihood that a learner will select a given form whenever there is a context of use for that variable structure” (Gudmestad 2014: 81). Furthermore, this model claims that learners’ production varies in many of the same ways and is constrained by the same range of linguistic and discourse factors for the selection of a particular form as it is for native speakers. What research has revealed is that learners and native speakers differ in the degree to which a variant of a form is used in different contexts and functions. Research from this approach is based on the general findings of sociolinguistic research for subject expression which has shown that both monolingual and bilingual speakers’ choices between overt and null pronominal forms is constrained by multiple factors, including morphosyntactic and semantic factors of verb form and semantics, morphological and contextual ambiguity, specificity, polarity, among others; and discourse-level factors of switch or same referent, continuity of topic, perseveration or priming of subject form, and discourse connectedness. Like the pragmatic studies of the previous chapter, some L2 variationists claim that because L2 language is a system, it is necessary to take into account all of the forms that L2 speakers produce for subject expression and not only the binary options of null and overt pronoun forms (Geeslin and Gudmestad 2008; Gudmestad and Geeslin 2010; among others). Most recently Gudmestad, House, and Geeslin (2013) found that highly advanced L2 learners differ significantly in their use of lexical noun phrases and demonstrative, interrogative and indefinite pronouns but not in their use of overt personal subject pronouns and null subjects. Others, following the model for investigations of native and bilingual speakers’ language use, focus exclusively on the alternation on null and overt subject pronouns (Abreu 2009; Linford and Shin 2013). Nonetheless, what these studies have in common is the claim that our knowledge of the acquisition of subject expression is incomplete without considering how learners acquire “the ability to vary language use according to both social and linguistic factors” (Geeslin and Gudmestad 2008: 1) and that “the second language acquisition [. . . ] of variation, as well as contrasts between native speakers and non-native speakers, can be described by the frequency of selection of a variant and the linguistic and social factors that condition the use of that variant” (Gudmestad

7.1 Constraining factors in variationist studies |

209

and Geeslin 2010: 1). Thus the variationist model for L2 acquisition highlights the need to examine other factors which have not been considered in the generative/processing and even pragmatic paradigms but which can shed light on the complex process of L2 subject expression.

7.1 Constraining factors in variationist studies Although much variationist research has examined sociolinguistic and linguistic factors that favor either null or overt subject pronouns in switch or same reference contexts, in recent years studies have included a wide range of factors that potentially affect subject expression including: subject forms (lexical noun phrases, overt, null, demonstrative, interrogative, and indefinite pronouns); a variety of specificity contexts (specific, non-specific, and group referents); person (first, second or third) and number (singular or plural) of the subject; verb forms (tense, mood and aspect); semantic class of verb (mental activity verbs, stative verbs, external activity verbs); clause type (main or dependent); polarity (negative and affirmative clauses) and effects of high and low frequency verbs occurring in the local discourse. In addition, research has focused on several discourse constraints such as same or switch reference, perseveration and priming, and the level of discourse cohesiveness or connectedness of reference relations. In the following section we present some of the major linguistic and discourse factors as they relate to subject expression in Spanish and as they are specifically defined in L2 variationist studies.

7.1.1 Linguistic variables 7.1.1.1 Subject form, person, number and specificity As exemplified in previous chapters of the present volume, subject expressions in Spanish can take on a variety of linguistic forms as they do in other languages. Spanish has both overt forms as in (170-172) and (174-176) and a null form as in (173). Lexical noun phrases and proper nouns (170) El niño corrió hacia la playa. the boy run-3sg-pret towards the beach ‘The boy ran towards the beach.’

210 | 7 A variationist approach to L2 subject pronoun acquisition (171)

Juan descansa los jueves. John rest-3sg the Thursdays ‘John rests on Thursdays.’

Overt pronoun (172)

Ella quiere más café. she want-3sg more coffee ‘She wants more coffee.

Null pronoun (173) Nos pro llamaron a las seis de la mañana. us-dat pro call-3pl-pret at the six of the morning ‘[They] called us at six in the morning.’ Demonstrative pronoun (174) Ése no me cree. that no me believe-3sg ‘That one doesn’t believe me.’ Interrogative pronoun (175)

¿Quién no sabe la respuesta? who no know-3-sg the answer ‘Who doesn’t know the answer?

Indefinite pronoun (176) Alguien tiró la leche. someone spill-3sg-pret the milk ‘Someone spilled the milk.’ The examples above show subjects and verb forms in the third person singular (170–172) and (174–176) or third person plural (173). Spanish also distinguishes a first person singular and plural, a second person singular formal and informal, and a second person plural formal, and in some varieties a second person plural informal (see chapter two of the present volume for examples of the complete

7.1 Constraining factors in variationist studies |

211

verbal paradigm for a verb in the present and preterit indicative forms). Specificity is a notion that defines whether referents can be distinguished by name or not. Examples (170–172) show referents that are specific; example (176) makes reference to a non-specific entity. The category ‘group’ refers to entities for which a group is named but individual members of the group are not specified to the extent that they can be identified individually, as in (177). Specific and group referents can be referred to by lexical noun phrases, overt, null, and demonstrative pronouns and non-specific referents by indefinite pronouns. Group referent (177) Los niños no entienden límites. the children no understand-3pl limits ‘Children don’t understand limits.’ For the third person plural subject pronoun, ellos ‘they’, Geeslin and Gudmestad (2008) and Abreu (2009) distinguish between referential (specific) and nonspecific uses, and within the category of non-specific, Abreu further distinguishes between non-referential and semi-referential. Referential third person plural subject pronouns can be overt or null depending on the discourse context (see 178 below); non-referential subjects are infelicitous with overt pronouns (see example 179); and semi-referential subjects are generally null (example 180). Referential (and specific) ellos (178)

a. Mis papás viven en un pueblo chico porque pro prefieren la My parents live-3pl in a town small because pro prefer-3pl the tranquilidad. tranquility ‘My parents live in a small town because [they] prefer tranquility.’ b. Mis papás viven en un pueblo chico. Yo prefiero la ciudad my parents live-3pl in a town small I prefer-1sg the city grande pero ellos prefieren la tranquilidad. big but they prefer-1pl the tranquility ‘My parents live in a small town. I prefer the big city but they prefer tranquility.’

212 | 7 A variationist approach to L2 subject pronoun acquisition Non-referential ellos (179)

a. #Ellos dicen que el hambre agudiza el ingenio they say-3pl that the hunger sharpen-3sg the ingenuity ‘They say that necessity is the mother of invention.’ b. pro dicen que el hambre agudiza el ingenio pro say-3pl that the hunger sharpen-3sg the ingenuity ‘[They] say that necessity is the mother of invention.’

Semi-referential ellos (180) a. pro Llamé al banco y ?ellos me dijeron que [I] call-1sg-pret to-the bank and they me-dat tell-3pl-pret that el depósito ya está listo. the deposit already be-3sg ready ‘[I] called the bank and they told me that the deposit is now ready.’ b. pro Llamé al banco y pro me dijeron que el pro call-1sg-pret to-the bank and pro me-dat tell-3pl-pret that the depósito ya está listo. deposit already be-3sg ready ‘[I] called the bank and [they] told me that the deposit is now ready.’ Second person singular tu ‘you’ can also be specific or non-specific. The deictic function (as in 181) can be overt or null, depending on the discourse function, but the non-specific function is infelicitous with the overt, as in (182). Deictic (and specific) tú (181)

a. pro Quiero que tú me ayudes, no ella. pro want-1sg that you me-acc help-2sg-subj no her ‘[I] want you to help me, not her.’ b. pro Quiero que pro me ayudes. pro want-1sg that pro me-acc help-2sg-subj ‘[I] want [you] to help me.’

7.1 Constraining factors in variationist studies |

213

Non-specific tú (182) a. Ya no pro aceptan trámites los sábados; #tú tienes anymore no pro accept-3pl paperwork the Saturdays you have-2sg que ir entre semana. that go-inf between week. ‘[They] don’t accept paperwork on Saturdays anymore; you have to go during the week.’ b. Ya no pro aceptan trámites los sábados; pro tienes anymore no pro accept-3pl paperwork the Saturdays pro have-2sg que ir entre semana. that go-inf between week. ‘[They] don’t accept paperwork on Saturdays anymore; [you] have to go during the week.’ To what extent person, number and specificity impact the use and acquisition of subjects in Spanish has been studied by Geeslin and Gudmestad (2008), Abreu (2009), and Gudmestad, House, and Geeslin (2013). Linford and Shin (2013) also examined first and third person singular verbs but did not include the variable of specificity.

7.1.1.2 Verbal form and potential ambiguity Spanish verbs are marked morphologically not only for person and number, which agree with the syntactic subject, but are also marked for tense (present, past, future), aspect (perfect, perfective, imperfective, progressive) and mood (indicative, subjunctive and imperative). The first and third-person singular forms in the conditional, imperfect, and present subjunctive are the same and thus, potentially ambiguous; and second-person singular, first-person plural, and third-person plural verbs in the periphrastic future, present indicative, present perfect, and preterit are all contrastive. Thus, within the verbal paradigm, there is the potential for morphological ambiguity in some persons and numbers as illustrated in (183) and (184).

214 | 7 A variationist approach to L2 subject pronoun acquisition (183) pro Voy a describir a mi mejor amigo; pro vivía en pro go-1sg to describe-inf to my best friend pro live-1sg/3sg-imp in México en ese entonces. Mexico in that then [I] am going to describe my best friend. [I/he] (ambiguous) was living in Mexico at that time.’ (Lubbers Quesada and Blake 2003) (184) A: ¿Cuándo pro se va? when pro refl-leave-2sg/3sg B: ¿Quién? who A: Usted. you A: ‘When are [you] leaving?/When is [he/she] leaving?’ (ambiguous) B: ‘Who?’ A: ‘You.’ (Personal conversation) In (183) the verb imperfect indicative form vivía ‘was living’ could potentially refer to either the speaker (first person singular) who is the subject of the previous clause (referred to with a null subject but with unambiguous verb morphology); or it could refer to mi mejor amigo ‘my best friend’ (third person singular). In (184), the present indicative form se va ‘leave’ could potentially refer to either the addressee (second person singular formal); or it could refer to a third person singular referent outside of the discourse. Thus, it has been suggested that speakers use overt subjects in order to resolve these types of potential morphological ambiguity (Hochberg 1986; García Salido 2008). However, research has shown that this is not the case for native speakers (Bentivoglio 1987; Ranson 1991; Silva-Corvalán 1994; Hurtado 2005; Torres Cacoullos and Travis 2011; Travis and Torres Cacoullos 2014) who depend more on discourse context, or a combination of verbal morphology and discourse context, to resolve such potential subject-verb ambiguity. Therefore, the question remains as to what extent non-native speakers rely on overt subjects to help them define the subject of a sentence, especially in light of the fact that their control over verbal morphology in their L2 Spanish may still be developing. It has also been suggested that when the tense, aspect or mood of a verb switches from one clause to the next, that there is a greater likelihood that a speaker will use an overt subject to signal this change as in (185a), where the

7.1 Constraining factors in variationist studies |

215

speaker uses imperfect verb morphology to describe an ongoing action in the past and uses an overt subject pronoun to refer to herself. Then in the very next clause she changes verb tenses to the present to describe a habitual situation and repeats the use of the overt subject pronoun for first person singular, despite the fact that both the discourse context (continuity of reference or same referent) and the verbal morphology, tengo ‘[I] have’, make clear who the subject of the present tense verb is. (185) a. al siguiente día él se sentó junto a mí . . . y todo lo to-the next day he refl-sit-3sg-pret next to me and all it que yo escribía, yo tengo una libreta dónde pro anoto that I write-1sg-imp I have-1sg a booklet where pro note-1sg pensamientos o cosas que suceden en el día . . . thoughts or things that happen-3pl in the day ‘the next day he sat down next to me and everything that I would write, I have a little book where [I] write down thoughts or things that happen in the day . . . (Lubbers Quesada and Blake 2003) Research in L2 acquisition that has examined the effect of verbal form and potential ambiguity include Abreu (2009); Gudmestad and Geeslin (2010); Linford and Shin (2013); and Gudmestad, House, and Geeslin (2013).

7.1.1.3 Semantic verb class Verbs can be divided into different categories according to semantic class and research in subject pronoun use of native and bilingual speakers of Spanish has shown that mental activity verbs (e.g. creer ‘to believe’, pensar ‘to think’, ver ‘to see/understand’, saber ‘to know’) tend to be used more often with overt pronouns in the first person (Bentivogio 1987; Silva-Corvalán 1994; Travis 2007; Torres Cacoullos and Travis 2011 for L1 speakers; López Ortega 2002; Lubbers Quesada and Blackwell 2009 for L2 learners). The extent to which semantic class and how this interacts with the frequency of a verb to affect pronoun use in L2 learners is examined from a variationist perspective by Abreu (2009) and Linford and Shin (2013).

216 | 7 A variationist approach to L2 subject pronoun acquisition 7.1.2 Discourse variables 7.1.2.1 Switch or same reference Variationists also take into account the factor of switch or same reference and how it impacts the use of null and overt subjects. Geeslin and Gudmestad define this variable as distinguishing “contexts where a given pronoun (overt or null) refers to the same referent as the subject of the preceding tensed verb form from those where the referent is different from one tensed verb to the next” (Geeslin and Gudmestad 2011: 17). Cameron’s (1985) definition for same reference is the context in which a subject is the same as the preceding subject of a tensed verb, including datives of gustar-type experiencer verbs, focused noun phrases of existential haber (‘there is’) and preposed noun phrases of se passive verbs. Switch reference is the context in which a different subject is found from the preceding subject of any tensed verb. Research in bilingual and native Spanish speakers has shown that overt subjects are more likely to be used in contexts of switch reference (Silva-Corvalán 1994; Bayley and Pease-Álvarez 1997; Cameron 1994, 1995). Again focusing on the passage in (185a) above, the subject of the first clause is expressed with an overt pronoun, él ‘he’ and even though the speaker brings herself back into the discourse via an object pronoun mí ‘me’, when referring to herself in the succeeding clause as a subject, she uses an overt pronoun to signal the switch of reference. Some would explain that the speaker uses an overt pronoun in this case because there is also the possibility of ambiguity; the verb escribía ‘was writing’ could refer to the subject of the previous clause or to the speaker herself. However, in the final two clauses of the passage, continued in (185b) below, there are again two cases of switch reference with potential verbal ambiguity but the speaker does not use an overt pronoun in either case. (185) b. yo tengo una libreta dónde pro anoto pensamientos o cosas I have-1sg a booklet where pro note-1sg thoughts or things que suceden en el día, siempre pro trataba de ver lo que that happen-3pl in the day always pro try-3sg-imp of see-inf it that pro escribía pro write-1sg-imp ‘I have a little book where [I] write down thoughts or things that happen in the day, always [he] would try to see what [I] was writing’ (Lubbers Quesada and Blake 2003) In the clause, siempre trataba de ver ‘would always try to see’, the verb form could refer to first or third person singular but the intended referent is él ‘he’ and, in lo que escribía ‘what (I) was writing’, also ambiguous morphologically

7.1 Constraining factors in variationist studies |

217

for first or third person singular, the intended referent is the speaker herself. It is the discourse context that disambiguates the intended referent. The effects of same reference, as Cameron has noted, can “linger” over two or more intervening subjects in tensed clauses (1995: 12). It is for this reason that variationists will consider how far back in the discourse a referent has been explicitly or implicitly mentioned. Same or switch reference has been examined by Abreu (2009), Geeslin and Gudmestad (2011), Linford and Shin (2013), and Gudmestad, House and Geeslin (2013).

7.1.2.2 Perseveration or priming The term perseveration (Labov 1994) or priming (Bock and Griffin 2000) refers to whether or not a word or structure is likely to trigger the same form or a different form in the following utterance. According to Abreu (2012), priming “is a psycholinguistic process found to occur in the production of many languages in the domains of syntax and the lexicon . . . [and the] process consists of the repetition of an element or linguistic structure that has been recently experienced” (Abreu 2012: 1). Travis claims that the effects of priming or perseveration illustrate that language structures are not constructed independently but are shaped by what comes before in discourse, which has ‘profound implications for our view of grammar, as it indicates that the grammar of discourse is developed on-line and in real time as discourse is constructed” (Travis 2013: 667). In subject expression, one possible hypothesis predicts that it is more likely that a null subject pronoun will follow an overt subject pronoun whereas a competing hypothesis predicts that a null subject is more likely to follow a null subject and an overt subject pronoun is more likely to follow an overt subject pronoun (Geeslin and Gudmestad 2011, p. 18). Perseveration and priming have been examined most recently in native speaker and bilingual communities by Flores-Ferrán (2002), Cameron and FloresFerrán (2004), Travis (2005, 2007), Torres Cacoullos and Travis (2011), and Abreu (2012); and in L2 speakers of Spanish by Abreu (2009), Geeslin and Gudmestad (2011), and Gudmestad, House, and Geeslin (2013).

218 | 7 A variationist approach to L2 subject pronoun acquisition 7.1.2.3 Discourse cohesiveness or connectedness Discourse cohesiveness or discourse connectedness³⁹ is a variable that appears to impact the use of null or overt subjects in same or switch reference contexts. This variable has been examined by variationists in Mexican-American bilingual children (Bayley and Pease-Álvarez 1997), the written production of Brazilian Portuguese speakers (Paredes Silva 1993), Caribbean Spanish speakers (Avila-Shah 2000), monolingual and bilingual Puerto Rican Spanish speakers (Abreu 2009), and L2 learners of Spanish (Abreu 2009; Geeslin and Gudmestad 2011). Contexts that provide the maximum amount of discourse cohesiveness or connectedness are 1) same subject, same verbal tense, aspect and mood; and, 2) same subject, different verbal tense, aspect and mood. Examples of these contexts based on Paredes Silva’s (1993) categories are as follows. Same subject, same verbal tense, aspect and mood (186) regresa el hombre viejoi con la bebida y proi se enoja, return-3sg the man old with the drink and pro refl-get angry-3sg ‘the old man returns with the drink and [he] gets mad,’ (Lubbers Quesada and Blake 2003) In (186), the subjects of both clauses, el hombre viejo ‘the old man’ and pro, refer to the same person and both verbs, regresa ‘returns’ and se enoja ‘gets mad’, share the same tense, aspect and mood. This context allows for the greatest level of discourse cohesiveness or connectedness and is the context that maximally promotes the use of the null subject in the succeeding clause. It is in these contexts in English that also allows for the use of a null subject in the subsequent clause. Another context that allows for a great deal of discourse cohesiveness, but to a slightly lesser extent, but also favors the use of a null subject is the following:

39 As in the generativist studies there are discrepancies in terminology. Different authors use different terms to refer to similar concepts. Paredes Silva (1993), Bayley and Pease Álvarez (1997), and Geeslin and Gudmestad (2011) use the term referent cohesiveness and Ávila-Shah (2000) and Abreu (2009) use the term discourse connectedness to refer to the relationship of referents from clause to clause.

7.1 Constraining factors in variationist studies |

219

Same subject, different verbal tense, aspect and mood y ya después, como que proi se quiere (187) proi se sentó pro refl-sit-3sg-pret and then later like that pro refl-want-3sg ir go-inf ‘[he] sat down and then later, like [he] wants to leave’ (188) Y éli sabía que proi era muy guapo, éli sabe and he know-3sg-imp that pro be-3sg-imp very handsome he know-3sg que proi es muy guapo, that pro be-3sg very handsome ‘And he knew that [he] was very handsome; he knows that [he] is very handsome’ (Lubbers Quesada and Blake 2003) The same subject is maintained in both of the previous examples but there is a switch in tense from past to present in the subsequent clauses. In (187) a null subject is used to maintain the same referent and in (188) an overt subject pronoun is used. In (188), the speaker is clarifying the previous utterance and the overt pronoun is not used so much to maintain the referent but to emphasize the clarification. In this case, emphasis or contrast overrules same subject. Contexts that provide lower levels of discourse cohesiveness or where the connectedness is weakened are those of switch reference (also based on Paredes Silva’s 1993 categories). Depending on the number of competing referents in the discourse, both null and overt subjects are used, although there will be less of a tendency for the use of null subjects in contexts of switch reference than for same reference. These categories (again based on Paredes Silva) include: 1) switch in referent from the immediately preceding clause but mentioned two or more clauses back and the referent is not ambiguous; 2) different subject but mentioned in immediately preceding clause in different syntactic position (e.g., direct object, object of preposition); and 3) new discourse topic.

220 | 7 A variationist approach to L2 subject pronoun acquisition Switch in referent in one or more intervening clauses (189) las chavas de sextoi loj acosaban mucho porque, como the girls from sixth him-acc bother-3pl-imp much because like proj era muy guapo, proi loj acosaban demasiado pro be-3sg-imp very handsome pro him-acc bother-3pl-imp too much ‘the sixth-grade girls would bother him a lot because, since [he] was so handsome, [they] would bother him too much’ (Lubbers Quesada and Blake 2003) In (189), the subject of the first clause, las chavas de sexto ‘the sixth-grade girls’, is also the subject of the third clause, with a switch in subject referent (a third person singular pro) in the intervening clause. There are no other competing referents and thus, a null subject can be used to refer to this subject despite the fact that the cohesiveness has been weakened by the intervening clause⁴⁰. Different subject, mentioned in previous clause in different syntactic function (such as a direct object): (190) pro es que no, yoi no puedo fijarme en esa personaj pro be-3sg that no I no able-1sg notice-refl-1sg in that person-obl porque proj les agarraba a las muchachas de la mano. because pro them-dat grab-3sg-imp to the girls of the hand ‘[it]’s not that, I cannot notice that person because [he] would grab the girls by the hand.’ (191) y yoi lej agradezco mucho que élj esté conmigo. and I him-dat appreciate-1sg much that he be-3sg-subj with-me ‘and I appreciate him a lot that he is with me.’ (Lubbers Quesada and Blake 2003) In the first clause of (190), a referent, esa persona ‘that person’, is mentioned as an object of a preposition and in the succeeding clause is the subject but is

40 Some researchers (see Abreu 2009 and Avila-Shah 2000) divide this category into three, distinguishing among: switch in referent but no other competing referents (intervening clauses have inanimate subjects), switch in referent with other competing referents but unambiguous due to other contextual or morphological clues, and switch in referent with several competing referents (also may be unambiguous due to contextual or morphological clues).

7.1 Constraining factors in variationist studies |

221

referred to with a null subject. Again, there is no other competing referent and although the discourse connectedness is weaker than in same reference contexts, a null subject is unambiguous and therefore acceptable. In (191), the subject of the second clause is mentioned in the previous clause as an indirect object. Again, there is no other competing referent and a null would have been acceptable but the speaker chooses an overt subject pronoun (perhaps for emphasis). New discourse topic (192) Entonces, éli siempre ha buscado en que ocuparse; le gusta estar en la casa con mi mamáj , y ayudar en hacer tareas de la casa, hacer el mandado y todo eso, y a parte, como que proi tiene el trabajo de llevar a mis sobrinos a la escuela. Entonces, pues en la mañana proi se levanta, prok nos levantamos, entonces proi los lleva a la escuela, y proi los lleva a su casa, de regreso. Y este, en general, mi papái es una persona muy positive. ‘So, he always has looked for things to keep himself busy; he likes to be in the house with my mom and help to do chores in the house, do the shopping and all that, and another thing, like [he] has the job of taking my nephews to school. So, well, in the morning [he] gets up, [we] get up, then [he] takes them to school, and [he] takes them back home. And, uhm, in general, my dad is a very positive person.’ (Lubbers Quesada and Blake 2003) In (192) the speaker is describing his father as the most important person in his life and after several intervening clauses where the father as a single referent is not mentioned, the speaker refers to him with an overt subject pronoun. Throughout the text, subsequent mentions of the father as a subject are referred to via a null pronoun. When there is a change in topic, from the father’s daily activities to the speaker’s assessment of his father, an overt subject, in this case a lexical noun phrase, mi papa (‘my dad’), is used to signal this switch in discourse topic. First mention, not identifiable and not in the preceding 10 clauses The first mention of a referent, which is not identifiable in the discourse context, is usually referred to with a full lexical noun phrase, as seen in (189) above (part of which is repeated below as [193]) with the noun phrase las chavas de sexto ‘the sixth grade girls’. As a group referent they have not been introduced into the discourse previously.

222 | 7 A variationist approach to L2 subject pronoun acquisition (193) las chavas de sexto lo acosaban mucho the girls from sixth him-acc bother-3pl-imp much ‘the sixth-grade girls would bother him a lot’ (Lubbers Quesada and Blake 2003) Research on native and bilingual speakers has revealed that the greater the discourse connectedness, the more likely speakers are to use null subjects and the weaker the discourse connectedness the more likely speakers are to use overt subjects, although as seen in the examples above, these are only tendencies and other discourse factors can intervene and override them. Abreu (2009), Geeslin and Gudmestad (2011) and Gudmestad, House, and Geeslin (2013) have investigated the impact of discourse connectedness on L2 subject expression and acquisition.

7.2 Conditioning factors in L2 variationist studies For an examination of a number of the multiple linguistic factors that condition the use and acquisition of subject expression, we turn to research into the L2 Spanish acquisition of subject expressions. L2 research has focused on many of the same variables that have been shown to determine variation in sociolinguistic studies of native speakers and bilingual speakers of Spanish. They seek to examine which linguistic forms learners produce and to what extent their choice of form depends on the person, number and specificity of the referent, the morphological form of the verb, the discourse context, perseveration, and the level of cohesiveness or connectedness of the discourse relations (Geeslin and Gudmestad 2008; Abreu 2009; Gudmestad and Geeslin 2010; Geeslin and Gudmestad 2011; Gudmestad, House, and Geeslin 2013; and Linford and Shin 2013).

7.2.1 Linguistic variables in L2 studies 7.2.1.1 Person, number, and specificity of the referent In an attempt to establish a descriptive database of L2 subject expression, Geeslin and Gudmestad (2008) set out to examine the production of subject forms among advanced L2 learners and to be able to compare their use with that of native speakers. By means of an oral interview task, the authors sought to determine which subject forms L2 learners produce and what the relationship is between subject forms and verbal person and number and specificity of the referent. The authors argue that in contrast to current sociolinguistic research, which has focused on the contrast between overt and null pronouns, research in L2 should examine all

7.2 Conditioning factors in L2 variationist studies |

223

forms produced so as to better understand the L2 learners’ system. Participants included 16 advanced English-speaking L2 Spanish learners, who were graduate students and also language instructors, and 16 native speakers from a variety of Spanish-speaking countries, including Argentina, Chile, Colombia, Ecuador, Mexico, Spanish, the United States (Puerto Rico and the Southwest), and Uruguay. The learners had spent between three months and 11 years in a Spanish-speaking country and had studied Spanish between three and 17 years; all of the native speakers were living in the U.S. at the time of data collection. In order to assure that participants were comparable, each completed a 25-item discrete-point grammar test. The scores for the learner group varied from 16 to 25 points and scores for the native speaker group ranged between 22 and 25 points. The dependent variable for the study was the subject form used and included lexical noun phrases, overt subject pronouns, null subject pronouns, demonstrative pronouns (e.g., éste ‘this one’), interrogative pronouns (e.g., quién ‘who’), and indefinite pronouns (e.g., alguien ‘someone’). The independent linguistic variables included person and number of the verb and specificity of the referent, that is, if the human referents could be identified by name or not. For this last variable, four categories were identified: specific, non-specific, group, and not applicable. Overall, the L2 learners’ and the native speakers’ use of null subject pronouns was very similar: 70.2 % for the learners compared to 67.6 % for the native speakers. The analysis revealed that subject forms varied for person and number of the verb for both native speakers and learners and that the distribution of forms was most similar for the two groups in the use of first person singular forms. But they also discovered important differences between the two groups; they found that the learners used more null subjects than the native speakers for all forms except first person singular and third person plural forms and that the learner group used more lexical noun phrases and subject pronouns for the third person plural forms than the native speaker group. In the analysis of the specificity variable, they found statistically significant differences and observable qualitative differences in the distribution of forms for both groups across categories. The distribution of forms for specific referents was almost the same for both groups; however, the learners used more null subjects and fewer lexical noun phrases when the referent was a group. For non-specific referents, the learners produced more lexical noun phrases and fewer overt pronouns. One of the greatest differences found in the data was for the specific second person singular contexts, where the learners’ use of null subjects was 94.3 % compared to only 75.7 % for the native speakers, and the use of overt pronouns was 5.7 % compared to 24.3 % for the native group. Verbal morphology here is never ambiguous and thus there is no grammatical reason to employ an overt pronoun, and yet the native speakers use an overt pronoun almost 20 % more often than the

224 | 7 A variationist approach to L2 subject pronoun acquisition learners. This is clear evidence that learners do not necessarily limit variation in their production to the same contexts as native speakers and that learners even at high levels still need to acquire the appropriate constraints for subject expression (Geeslin and Gudmestad 2008: 10). For specific, group, and non-specific third person singular referents, the learners used more lexical noun phrases than the native speakers and fewer overt subject pronouns for the third person singular specific referents. These results show that specificity reveals different patterns of use for each type of referent according to person and number of the verb (Geeslin and Gudmestad 2008: 11). In general, the authors found that learners tended to use more null subject forms for all persons except for first person singular and third person plural contexts, where learners preferred overt forms and native speakers used more null forms. This shows “that the acquisition problem for English-speaking learners of Spanish is not merely one of allowing or disallowing null subjects (in which case a lower frequency of null subjects would be expected among learners), but rather acquisition involves learning to respond to the additional factors in the context that determine when use of null subjects is appropriate” (Geeslin and Gudmestad 2008: 13). Abreu (2009), in her comparative study of native and heritage speakers of Puerto Rican Spanish and L2 Spanish learners also found that learners tended to use more null subject forms in general (74 % overall) than both native (62 %) and heritage speakers (51 %); however, in contrast to Geeslin and Gudmestad, she found that null subjects were most common among learners for all persons except third person singular where overt subject pronouns were most common⁴¹ . Abreu’s study was designed to compare and contrast null and overt subject pronouns among three groups of speakers: 1) monolingual speakers of PR Spanish living in the western town of Isabela, Puerto Rico, 2) bilingual heritage PR Spanish-English speakers living in northern Florida, and 3) intermediate to low advanced L2 learners (native speakers of English). This final group was recruited from fourth-year Spanish classes in a major U.S. institution and had not studied abroad for more than six weeks. The data were collected by means of informal oral interviews based on a wide range of topics, coded for a number of factors, including person,

41 Abreu’s (2009) and Geeslin and Gudmestad’s (2008) results for third person singular are not completely comparable because the latter included a wider range of subject forms for these persons and numbers, including full lexical noun phrases, demonstratives, etc. Thus, for null subjects, Geeslin and Gudmestad found null subjects were used for third person singular in nearly 60 % of the instances for learners and 52 % of the instances for native speakers, they are comparing these numbers with a wider range of forms.

7.2 Conditioning factors in L2 variationist studies |

225

number, and specificity, among others (to be discussed in subsequent sections) and statistically analyzed using the GoldVarb X program (Sankoff, Tagliamonte, and Smith 2005, cited in Abreu 2009) to see which factors are significant for overt pronoun use. Person and number were identified as significant in their impact on the use of overt subject pronouns for all three participant groups and was one of the strongest constraints for each group, although there were differences among the three participant groups. The monolingual and bilingual groups patterned together for first and second person singular (both specific and nonspecific tú ‘you’), and used more overt subject pronouns, whereas the L2 learners produced more null subjects for both first and second person singular (only specific – there were no instances of nonspecific tú⁴²) (see Table 23 for a summary of Abreu’s results). Monolinguals’ and bilinguals’ use of first and second person singular overt subject pronouns was approximately double than that of the L2 learners. For third person singular, on the other hand, the bilingual and L2 learners patterned together and distinctly from the monolingual group. Abreu suggests that perhaps bilingual speakers and L2 learners tend to use more overt pronouns for third person singular subjects because of the greater potential of morphological ambiguity in third person singular verb forms; however, she discards this hypothesis because third person plural verb forms also present the possibility of ambiguity and overt pronouns are lowest for this type of subject. All three groups showed a similar preference for null pronouns for plural subjects for all persons, in general. Where null use is highest for all three groups is for nonspecific third person plural ellos ‘they’, although Abreu reports that the number of instances was low for all groups. It is difficult to assess the impact of specificity due to the low number of tokens for nonspecific tú and ellos. Although the L2 learners in both the Abreu (2009) and Geeslin and Gudmestad (2008) studies pattern in the same way in their preference for null subjects in second person singular specific contexts, the monolingual speakers of these two studies are in stark contrast: whereas Abreu’s monolingual speakers preferred overt subject pronouns, the native speakers in the Geeslin and Gudmestad study preferred null subjects. This might possibly be attributed to the fact that Abreu’s monolingual speakers were all of the PR variety. The differences in the L2 learners in the two distinct studies in their preferences for third person singular could possibly be accounted for by the different levels: Abreu characterized her L2 partici-

42 There were four tokens of the discourse marker tú sabes (‘ya know’), which were not considered in the analysis.

226 | 7 A variationist approach to L2 subject pronoun acquisition Table 23. Distribution (%) of subject pronouns by person and specificity. Subject pronoun form

monolinguals

bilinguals

L2 learners

overt

null

overt

null

overt

null

yo ‘I’ tú ‘you-sg-inform’ tú (nonspecific) usted ‘you-sg-form’ él/ella ‘he/she’ nosotros ‘we’ ustedes ‘you-pl’ ellos/ellas ‘they-m/they-f’ ellos (nonspecific)

50 56 54 60 31 19 41 10 5

50 44 46 40 69 81 59 90 95

59 65 60 – 57 8 40 31 9

31 35 40 – 43 92 60 69 91

24 29 – – 56 5 0 19 6

76 71 – – 44 95 100 81 94

Total

38

62

49

51

26

74

(Adapted from Abreu 2009: 169)

pants as intermediate to low advanced, whereas the L2 participants in the Geeslin and Gudmestad study were graduate students and language instructors. Linford and Shin (2013) examined the effects of subject expression and a number of independent variables in two groups of L2 learners. Although the principle objective of the study was to examine the effects of lexical frequency on subject pronoun expression, the authors also included other linguistic variables that possibly interact with the lexical frequency of verb forms, including person and number, specifically first and third person singular subject pronouns (null and overt). The authors did not include other persons and numbers due to the low number of examples. Participants included 12 native speakers of English from two levels of proficiency. The lower proficiency group was comprised of students enrolled in a second-year Spanish courses and the upper proficiency group was recruited from a fourth-year Spanish course. All participants completed both a grammar test to verify proficiency level and a semi-directed sociolinguistic interview that included topics ranging from sociodemographic information to narratives regarding memories, future plans and social and political topics (Linford and Shin 2013: 177). The interviews were recorded and transcribed. The authors identified 980 tokens in total, of which 765 were included in the analysis of person. Analyses were conducted via chi-square tests and binary logistic regressions. Both groups preferred null pronouns and the more advanced group preferred null pronouns to a greater extent (87 % of the tokens were null for the advanced compared to 72 % for the intermediate level participants). The authors found that person did not significantly affect pronoun expression for the lower proficiency

7.2 Conditioning factors in L2 variationist studies |

227

group but it did so for the more advanced proficiency group and at a highly significant level (p < .01) regardless of the frequency of verbs (frequent verbs: χ 2 (1) = 8.01; infrequent verbs: χ 2 (1) = 7.23). Overt subjects for third person singular were approximately two and half times more frequent than first person singular for both frequent and infrequent verbs. Overall overt subject pronoun use was 13 % for this group of learners and pronoun use for first person singular frequent and infrequent verbs was 12 % and 14 % respectively. For third person singular frequent and infrequent verbs, however, the use of overt pronouns was 29 % and 33 % respectively. Although not necessarily in line with native speaker use (which is more frequent with first person singular subjects), Linford and Shin’s results are in line with recent findings of Spanish speakers’ use of overt and null subject pronouns (Erker and Guy 2012). Gudmestad, House, and Geeslin (2013) examined the relationship between the use of third person singular subject forms and person and number of the verb and specificity of the referent expression (among other variables) in advanced L2 learners and a group of native speakers. The authors used a Bayesian multinomial probit model, which allowed them to examine the relationship between subject expression and multiple independent variables at the same time. In traditional statistical analyses, such as Goldvarb, the dependent variable (in this case, subject expression) can have only two categories and relies on independent observations. The Bayesian multinomial probit model can rely on unobserved measures of preference for a variety of categories and can include random effects for each participant in an estimate of the mean. Therefore, as the authors explain, the model “represents the degree to which (unmeasured) disparities in the participants may have influenced their choices of subject expression” (Gudmestad, House, and Geeslin 2013: 385). Participants for the study included 16 graduate students who were instructors of Spanish and native speakers of English, and 16 native Spanish speakers (also graduate students) from a variety of Spanish-speaking countries but living in the U.S. at the time of data collection. Each speaker participated in a sociolinguistic interview conducted by two native speakers of Spanish on a range of topics involving future, past experiences and viewpoints; the interviews were recorded and later transcribed. They identified and based their analysis on a total of 6 342 tokens of subjects of third-person verb forms, including null subjects, personal pronominal subjects, lexical noun phrases and other pronominal forms (including demonstrative, indefinite and interrogative pronouns). The multinomial probit model that the authors fit to their data comprised three submodels, each of which portrayed the preference of a personal pronominal subject, lexical noun phrase or other pronominal form over a null subject. Results revealed that overall both groups preferred null subjects over personal pronouns, which in turn were preferred over lexical noun phrases and these

228 | 7 A variationist approach to L2 subject pronoun acquisition were preferred over other pronouns. In addition, both groups preferred this same order of subject expression for the variable number and preferred the order of null subjects, lexical noun phrase, personal pronoun and other pronoun for the variable specificity (group). However, the sensitivity to subject expression and number and specificity was different for native speakers and L2 learners. L2 learners were more sensitive to the variables of switch reference (ranked 1st ), referent cohesiveness (ranked 2nd ) and perseveration (ranked 3rd ) and native speakers were more sensitive to the variables specificity (ranked 1st ), TAM (ranked 2nd ), object pronoun (ranked 3rd ) and number (ranked 4th ). For the learners, number and specificity were ranked 11th and 6th respectively (see Table 24). Table 24. Summary of Gudmestad, House, and Geeslin’s (2013) rank for number and specificity (order of estimated posterior probabilities for subject expression). Variable

Category Most probable

Second most probable

Third most probable

Least probable

Euclidean Rank distance⁴³

Native speakers Specificity Group Number Sing.

Null Null

NP PP

PP NP

OP OP

0.174 0.059

1 4

L2 learners Specificity Group Number Sing.

Null Null

NP PP

PP NP

OP OP

0.270 0.140

6 11

NP = lexical noun phrase; PP = personal pronoun; OP = ‘other’ pronoun (demonstrative, indefinite, interrogative) (Adapted from Gudmestad, House, and Geeslin 2013: 392–393)

Nonetheless, it can be observed that specificity is a higher ranked, and therefore more important, variable in the native speaker and learner groups. Thus, although there are subtle and important differences between how L2 learners and native speakers select subject forms, it is evident that learners are sensitive to the same constraints as native speakers.

43 The authors explain that variables “with large Euclidean distances impact the expected choice of subject expression more than those with small distances” (Gudmestad, House, and Geeslin 2013: 391).

7.2 Conditioning factors in L2 variationist studies |

229

7.2.1.2 Verbal form and potential ambiguity in L2 studies Abreu (2009) investigated the impact of tense, aspect and mood (TAM) and ambiguity on the use of null and overt subject pronouns among the three participant groups of monolingual PR Spanish speakers living in Puerto Rico, bilingual PR Spanish-English speakers living in the U.S., and a group of intermediate to low advanced L2 learners. Only for the monolingual group was TAM found to significantly interact with subject pronoun use. The subjunctive was found most frequently with overt subject pronouns, followed by imperfect and present tenses and the preterit disfavored overt subject use. These results are in line with SilvaCorvalán’s (2001) claim that because imperfect and subjunctive tend to express background information, they tend to combine more often with overt pronouns, and preterit is used more often in foregrounding events and combines more often with null subjects. Although the effects for TAM for both the bilingual and L2 learner groups were not found to significantly correlate to subject expression, mostly likely due to low frequencies in the use of all tenses other than present indicative, preterit and imperfect, Abreu noted certain trends among the two groups which mirror the findings of the monolingual group. For the bilingual group, present subjunctive, imperfect and present tenses all showed higher rates of overt subject pronouns and preterit and future tenses appeared to disfavor overt pronouns. For the learner group, overt subject pronouns occurred most often with present and imperfect tenses but also with preterit, a finding that is in contrast with both the native and bilingual speakers. For all other tenses, there were too few tokens to detect important patterns. On the other hand, the factor of ambiguity did not condition subject expression for either the monolingual or the bilingual groups but it did for the L2 learners. In fact, for the monolingual group there was no statistically significant difference in the use of overt subject pronouns among three conditions: 1) no morphological ambiguity, 2) morphological ambiguity but no contextual ambiguity, and 3) both morphological and contextual ambiguity⁴⁴. This is clearly against the functional hypothesis that proposes speakers use overt subjects in order to avoid potential morphological ambiguity (Hochberg 1986; García Salido 2008). Furthermore, as Abreu points out and in line with Silva-Corvalán, native speakers appear to use overt subject pronouns for distinguishing discourse functions and not for resolving potential morphological ambiguity. The bilingual group displayed similar results with no contextually ambiguous cases and only one instance of a case with both potential morphological and contextual ambiguity. The learner group was, however, affected for morphological ambiguity where learners tended to use

44 There were only two instances of this latter case among 457 finite verbs analyzed.

230 | 7 A variationist approach to L2 subject pronoun acquisition more overt pronouns with morphologically ambiguous verbs (42 % with a factor weight of .62) than with morphologically unambiguous verbs (19 % with a factor weight of .45). One example of morphological and contextual ambiguity occurred in the data and the L2 learner attempts to resolve the ambiguity with an overt subject pronoun, as illustrated in (194). (194) pero ahora ellai está con mi tíaj que vive en Kentucky porque but now she be-3sg with my aunt that live-3sg in Kentucky because ellai/j no está muy bien ahora she no be-3sg very well now ‘. . . but now she is with my aunt that lives in Kentucky because she (ambiguous) isn’t well now’ (Abreu 2009: 155) Abreu questions why L2 learners would be more sensitive to potential contextual ambiguity triggered by verb morphology than monolinguals and bilinguals and suggests that this may be an effect of classroom learning that focuses a great deal of time on the learning of verbal morphology and the need to identify distinct subjects with the same verb forms (Abreu 2009: 156). Gudmestad and Geeslin focused on the effects of TAM and verb-form ambiguity and the use of subject expression in native and L2 learner production in their (2010) study. This subsequent analysis was based on the same data set as their (2008) study, which included native speakers from several Spanish-speaking countries and advanced L2 learners, and sought to investigative to what extent the relationship of the TAM of the verb would be to the forms for subject expression produced by native speakers and learners. They also asked to what extent the differences found across the categories of the TAM variable correspond to potential morphological ambiguity (of the verb) and/or to switch reference for one or both participant groups (the latter factor will be discussed in 7.2.4.). In line with their previous study, they again included a wide range of subject forms, but this time collapsed demonstrative, indefinite and interrogative pronouns into one category (i.e. ‘other pronouns’) due to the low frequency of these forms. Thus their categories included: null subjects, lexical noun phrases, personal pronouns, and other pronouns. As in the (2008) study, the researchers found that the L2 learners produced more null subjects and fewer other pronouns than native speakers but both groups were similar in their use of lexical noun phrases and personal pronouns. The researchers found a significant relationship between TAM and verb form for each group. There were similarities between groups, which produced more null subjects for all verb forms ranging from 51.4 % (with the conditional

7.2 Conditioning factors in L2 variationist studies |

231

among learners) to 78.6 % (for the periphrastic future among native speakers). The researchers did detect differences between the two groups, however. The L2 learners produced more null subjects than the native speakers for imperfect, present, present perfect and preterit tenses and the native speakers used more null pronouns for conditional, periphrastic future and present subjunctive forms. There were also greater differences in the use of present indicative and present subjunctive between the two groups; the learners produced more present indicative forms and fewer present subjunctive than the NSs. The authors had expected that the variable of ambiguity would explain differences in the use of different subject forms as reported for bilingual speakers in previous studies (Silva-Corvalán 1982). They specifically examined the relationship between subject forms and the conditional, imperfect and present subjunctive tenses, which have ambiguous verb forms in the first and third person singular. However, their results did not confirm this expectation. They did not find a significant relation between these potentially ambiguous verb forms and subject expression. This finding is in accord with Abreu’s results for monolingual and bilingual PR Spanish speakers who do not appear to use subject expression to resolve verbal ambiguity, but against Abreu’s findings for L2 learners who use more overt subject pronouns in instances of verbal ambiguity. Finally, although Gudmestad and Geeslin found that TAM influences the use of subject forms, their results revealed that the effect of TAM disappears in redundant contexts, a factor that will be explored further in section 7.2.4. when we discuss the impact of discourse factors, specifically same and switch reference, on subject expression. Linford and Shin (2013) investigated the impact of TAM (as well as verb frequency) on the production of first and third person singular subject pronouns in two groups of L2 learners. They included only simple present indicative, preterit and imperfect indicative tenses, which comprised 94 % of the 980 tokens included in the study for analysis. The binary logistic regression analysis revealed that TAM was not significant for either the intermediate or the more advanced group when all the verbs were grouped together. However, the variable was found to be significant for the more advanced groups’ production of frequent verbs (χ 2 (1) = 9.654, p < .01). Specifically they found that the learners used more overt subject pronouns with imperfect forms, which have the potential for greater referent ambiguity in the first and third person singular forms. These findings corroborate those for Abreu’s learners, although the latter also found that the subjunctive was more frequent with overt pronouns, followed by the imperfect. The Bayesian analysis employed by Gudmestad, House, and Geeslin in their (2013) study of third person singular subject expression included TAM as another of their multiple independent variables. They found important similarities and differences between the native speaker and L2 learner groups. For both native

232 | 7 A variationist approach to L2 subject pronoun acquisition speakers and L2 learners, subject expression was sensitive to the periphrastic future of TAM; however, for native speakers the variable was ranked second and for the L2 learners it was ranked twelfth (among 12 variables). Both groups preferred null subjects above personal pronouns, lexical noun phrases and other pronouns (in that order) for the periphrastic future. In addition, the learners were more sensitive to the present and imperfect indicative tenses (ranked 8th and 10th ), confirming what Abreu and Linford and Shin found for intermediate and low advanced L2 learners.

7.2.1.3 Semantic verb class in L2 studies Studies among monolingual and bilingual Spanish speakers have revealed that certain semantic classes of verbs, specifically verbs of knowledge, perception and opinion, express greater speaker subjectivity and tend to trigger greater first person overt subject use (Enríquez 1984; Hurtado 2005; Travis 2007; Travis and Torres Cacoullos 2014). L2 studies from a pragmatic perspective found this to affect learners’ use of overt and null subject pronouns as well (López Ortega 2002; Lubbers Quesada and Blackwell 2009). Abreu’s (2009) study examined the effect of semantic verb class and found that it was not significant in the multivariate analysis for the monolingual PR Spanish speakers and the L2 intermediate/low advanced learners, but it was for the bilingual PR Spanish-English group. Abreu included in the analysis verbs of cognition (e.g. pensar ‘think’, creer ‘believe’), communication (e.g. decir ‘say’, contar ‘tell’), statives (ser and estar ‘be’), possession (tener ‘have’, quedarse ‘keep’), volition (querer ‘want’) and other verbs of action. Although not significant for the monolinguals, she did find a greater use of overt subject pronouns with the verbs of cognition (54 %) and communication (46 %) than with other verbs (all less than 45 % use of overt pronouns). Overt pronouns were used the least with verbs of motion such as ir ‘go’ and caminar ‘walk’. For the bilingual speaker group verb class was found to be significant; the verbs that were found to favor overt subject pronouns were cognitive verbs, verbs of volition and stative verbs. All other verb types (communication, motion, possession, etc.) disfavored overt subjects. One finding that Abreu had to eliminate from the multivariate analysis because it was an invariable context, was the use of estar ‘be’ for reactions or reported speech, much like the English phrase, ‘I was like . . . ’, as illustrated in (195).

7.2 Conditioning factors in L2 variationist studies |

233

(195) Y yo estaba “Qué clase de loquera es esto”? and I be-1sg-imp what class of craziness be-3sg this ‘And I was like, “What kind of craziness is this?” (Taken from Abreu 2009: 132) For the L2 learners, Abreu reports that verb class was not found to influence pronoun use in a statistically significant way and therefore, does not include a discussion of any notable trends. Linford and Shin (2013) also included the variable of semantic verb class in their analysis of verb frequency and its impact on first and third person singular subject expression (null and overt) in two groups of L2 learners. They divided their verbs into three categories: 1) verbs of mental activity (e.g. saber ‘know’, creer ‘believe’, suponer ‘suppose’, recordar ‘remember’), 2) stative verbs (e.g. estar ‘be’, ser ‘be’, tener ‘have’) and verbs of external activity, either physical, social or behavioral (e.g. jugar ‘play, salir ‘leave’). All 980 of their tokens were coded for one of these three categories and were analyzed via a binary logistic regression in order to examine the impact of lexical frequency and a measure the authors define as ‘repetition’, which included contexts where participants repeated a verb, and thus, simulated a frequency effect. Linford and Shin found that the semantic class of the verb was statistically significant only for the more advanced learners and only in contexts with frequent verbs (frequent: χ 2 (2) = 14.46, p < .001 vs. infrequent verbs: (χ 2 (2) = 3.503, p < .174). What is unusual about Linford and Shin’s findings is that overt pronouns were more frequent with activity verbs and least frequent with verbs of mental activity, in direct contrast with the findings of native and bilingual speakers who tend to use more overt pronouns with verbs of mental activity. Linford and Shin attribute these findings to the fact that the learners in their study tended to use the form sé ‘I know’, a mental activity verb that occurred with great frequency in their data without the overt subject pronoun, yo ‘I’, possibly skewing the results. Nonetheless, these findings are more in line with Abreu who found that intermediate learners’ use of overt and null subject pronouns is not impacted by semantic verb class but is also in contrast with the behavior of native speakers, who tend to use the overt yo with verbs such as sé and creo ‘I believe’ (Travis and Torres Cacoullos 2013; Lubbers Quesada and Blackwell 2009).

234 | 7 A variationist approach to L2 subject pronoun acquisition 7.2.2 Discourse variables in L2 variation studies 7.2.2.1 Switch or same reference in L2 Several studies have examined the factor of same or switch reference in L2 learner production based on Cameron’s (1985) definition and have found findings similar to that found for native and bilingual speakers. Abreu established that switch reference was significantly related to overt subject pronoun use, as she had hypothesized, for the monolinguals and bilinguals of her (2009) study. In contexts of switch reference, overt pronouns were favored more often than in contexts of same reference, which favored more the use of null subjects. The L2 learners preferred overt subjects only half of the time in contexts of switch referent with other possible candidates for subject but not in other switch referent contexts (more specific results for several different same and switch contexts as they relate to discourse cohesiveness are discussed in section 7.2.2.3). In their (2010) study of native speakers from a variety of regions and advanced L2 learners, Gudmestad and Geeslin sought to investigate whether or not the effect of verb type is annulled in same and switch reference contexts. In general, they found that switch and same reference was a strong predictor of subject form: both the native speaker and learner groups produced more null subjects in the context of same reference and more overt subjects for switch reference. Furthermore they documented that for both groups, the difference in the distribution of subject forms (null, pronominal, lexical noun phrases and other pronominal forms) was significant across categories of verb forms for switch reference but not for same reference. In same reference contexts, the null subject was undoubtedly the dominant form within each category of tense, aspect and mood for both groups. Furthermore, not only did the null subject occur less frequently in contexts of switch reference, but the span of its use varied across categories of tense, aspect and mood to a greater extent than in contexts of same reference, where lexical noun phrases, personal pronouns and ‘other’ pronoun forms were used more frequently. It is for this reason the authors claim that in same reference contexts, the effect (discussed more fully in 7.2.1.2) of verbal form is eliminated. In Linford and Shin’s (2013) study of two groups of L2 learners’ production of first and third person singular subject pronouns and the impact of several independent variables, including switch reference, the authors found that switch reference significantly predicted pronoun use (i.e. overt subject pronouns). This significance was greater for the more proficient group (significant at the < .01 value) than for the lower level group (significant at the < .05 value). In fact, for the lowerlevel group only switch reference and lexical frequency and repetition were significant; for the higher-level group, switch reference, person and semantic verb class were all significant and switch reference reached the highest level of sig-

7.2 Conditioning factors in L2 variationist studies |

235

nificance. Linford and Shin conclude that as L2 learners reach higher levels of linguistic proficiency, they become “increasingly sensitive to the various linguistics factors that influence pronoun expression among native speakers” (Linford and Shin 2013: 181). Gudmestad, House and Geeslin’s (2013) study of third person singular subject expression using a Bayesian multinomial probit model included switch reference as one of the independent variables. Specifically, their statistical model included an interaction of switch reference with referent cohesiveness (the latter of which will be discussed further in 7.2.2.3) following the results of Bayley and Pease-Álvarez (1997). In general, the authors found that switch reference explained the variability in the selection of lexical noun phrases and other pronouns (demonstrative, indefinite and interrogative pronouns) over null subjects but did not explain the variability in the selection of personal pronouns over null subjects. The model predicted that for the category of switch reference, speakers are more likely to use a lexical noun phrase, next a personal pronoun followed by a null subject and least likely to use a subject from the ‘other’ category. Most importantly, the model showed that switch reference was the most important variable for predicting the distribution of subject expressions among L2 learners (ranked 1st ) but not for native speakers, for whom the most important variable was specificity, as mentioned in 7.2.1.1.

7.2.2.2 Perseveration or priming in L2 subject expression The tendency to repeat a word or a general syntactic pattern of an utterance that has just been produced is a psycholinguistic process known as priming or perseveration. In the case of subject expression, research has focused on to what extent a certain subject form (usually null or overt) triggers the same form in the immediately following utterance. Abreu (2009) had hypothesized that priming would play a significant role in overt expression for all three groups of participants: the monolingual PR Spanish speakers, bilingual PR Spanish-English and L2 learners. She found that for each group priming was a significant factor for intraspeaker priming.⁴⁵ See Table 25 for a summary of Abreu’s results for the effect of priming on subject expression in the three groups.

45 Although Abreu examined both intraspeaker priming (within the same speaker) and interspeaker priming (between another speaker and the speaker under study), she reports that the tokens for the interspeaker data were too few to include in her multivariate analysis. Therefore, only the results for the former are discussed here.

236 | 7 A variationist approach to L2 subject pronoun acquisition Table 25. Summary of Abreu’s (2009) results for the effect of priming on overt subject pronouns (three participant groups). Participant group Monolingual speakers

Bilingual speakers

L2 learners

Factor weight

% overt

N

% of data

Previous mention of same referent overt Previous mention of same referent null Range Previous mention lexical noun phrase (not included in multivariate analysis)

.65 .40 25

54 25

181 129

40 60

28

19

–*

Previous mention of same referent overt Previous mention of same referent null Range Previous mention lexical noun phrase (not included in multivariate analysis)

.66 .31 49

68 28

256 101

36 34

39

29

7

Previous mention of same referent overt Previous mention of same referent null Range Previous mention lexical noun phrase (not included in multivariate analysis)

.72 .41 31

49 17

137 128

27 71

37

23

–*

* Data not provided. (Adapted from Abreu 2009: 99, 125, and 146)

For the monolingual speakers, 54 % (with a factor weight of .65) of subjects were overt pronouns when they followed a previous mention of the same referent that was also overt, and only 25 % (factor weight of 40) were overt when the previous mention was a null subject. The bilingual group showed an even stronger preference, producing overt pronouns 68 % (with a factor weight of .66) of the time following a previous overt subject pronoun and producing overt pronouns only 28 % (factor weight of .31) of the time following a null subject.⁴⁶ The L2 learners, it was hypothesized, would also reveal an effect of priming on their subject expression and this was borne out by the data. When the previous mention was an overt subject pronoun, the following subject was an overt pronoun in 49 % (with a factor weight of .72) of the cases and when the previous mention was a null subject, the following subject as overt in only 17 % of the cases (with a factor weight of .41).

46 Abreu also included previous mention of a referent that was in English for both bilingual and L2 speakers and found that bilingual participants produced an overt subject pronoun (in Spanish) 74 % of the time in these contexts but L2 learners used an overt pronoun only 24 % in these contexts. Thus English showed an effect for priming for only the bilingual speakers.

7.2 Conditioning factors in L2 variationist studies |

237

Abreu also examined the effect of lexical noun phrase subjects on the following subject expression although there were too few tokens for each participant group to include in the multivariate analysis. Nonetheless she found that lexical subjects did not favor overt pronouns in the following utterance. Geeslin and Gudmestad (2011) examined the interaction of subject expression and perseveration, as well as discourse cohesiveness. The participants for this study included two groups: 1) 16 advanced L2 learners of Spanish and 2) 16 native speakers of Spanish from a variety of countries of origin but who were residing in the U.S. at the time of data collection. All participants completed a sociolinguistic interview conducted by native Spanish speakers on a range of topics and the L2 learners, in addition, completed a grammatical proficiency exam. The interviews were recorded and subsequently transcribed. The authors identified all subjects of tensed verbs. For perseveration, they considered three categories, if the previous mention of the referent was a null subject, an overt pronoun, or a lexical noun phrase. Cross-tabulation and chi-square tests were performed and it was found that the distribution of the different subject forms for the three categories mentioned was significant for the variable of perseveration. Results showed similar general trends for both groups for an effect of perseveration. In general, both groups preferred null subjects for all three categories; however, both groups by far preferred null subjects following null subjects (92.3 % for learners and 84.8 for native speakers). Nulls were also preferred following pronouns and lexical noun phrases but to a lesser extent (60 % and 60.3 % following pronouns, and 72.2 % and 65.7 % following lexical noun phrases for the learner and native speaker groups respectively). In Table 26, the percentages in bold highlight that for each type of previous mention of referent, the percentage for a following subject is highest for the same type. Thus, in the case of perseveration, production varied importantly depending on the form of the previously mentioned referent. The authors point out that “although no form ever surpassed the use of the null subject, the direction of these changes across categories is consistent with the results for perseveration in other sociolinguistics studies (e.g., Cameron, 1994)” (Geeslin and Gudmestad 2011: 26). Gudmestad, House, and Geeslin (2013) included perseveration of null subjects, overt personal pronouns, lexical noun phrases and other pronouns (i.e. interrogative, indefinite and demonstrative) as variables in their Bayesian multinomial probit analysis. The authors found a preference for null subjects above overt personal pronouns, lexical noun phrases and other pronouns overall. In addition, for the L2 learner group, they found that perseveration was an important variable, which adjusted the predictive distribution for subject the most importantly in relation to the base-case (overall) distribution. In other words, for L2 learners, there was a strong preference for participants to select null subjects following

238 | 7 A variationist approach to L2 subject pronoun acquisition Table 26. Summary of Geeslin and Gudmestad’s (2011) results⁴⁷ for the effect of perseveration (priming) on overt subject pronouns (three participant groups). Participant group

Native speakers

Previous mention of referent

Null Overt pronoun Lexical NP

Subject expression form Null

Overt pronoun

Lexical NP

%

N

%

N

%

N

84.8 60.3 65.7

2910 482 161

12.4 26.4 3.7

424 211 9

1.4 0.9 25.7

49 7 63

290 202 14

0.9 0.7 19.4

41 4 42

(χ 2 = 903.743, df = 6, Cramer’s V = 0.318, p < 0.001, no small cells) L2 learners

Null Overt pronoun Lexical NP

92.3 60 72.2

4224 333 156

6.3 36.4 6.5

(χ 2 = 1022.135, df = 6, Cramer’s V = 0.309, p < 0.001, 3 small cells) (Adapted from Geeslin and Gudmestad 2011: 25)

null subjects and overt pronouns following overt pronouns. They found a similar pattern for the native speaker group but subject expression was not found to be as sensitive to the variable of perseveration for the latter group as it was for the L2 learner group. Thus, among the studies that have examined perseveration or priming, it has been seen that this process produces a strong effect for subject expression for monolingual native speakers, bilingual speakers and both intermediate, advanced and high advanced L2 learners. The results reported on so far lend support to Abreu’s claims that priming is a universal process that manifests itself in both the L1 and L2 (Abreu 2009: 151).

7.2.2.3 Discourse cohesiveness or connectedness in L2 Building on the concepts of same or switch reference, L2 variationists have examined how discourse cohesiveness or connectedness impacts the use of null or overt subjects in contexts with varying degrees of same or switch reference. Previously, it was mentioned that contexts that provide the maximum amount of discourse

47 Geeslin and Gudmestad (2011) also report on the results of other pronouns (demonstratives, interrogative and indefinite). Although also significant, for the purposes of comparing different studies’ results in terms of perseveration to the extent possible, I leave these data out of the summary.

7.2 Conditioning factors in L2 variationist studies |

239

cohesiveness have the same subject and the same verbal tense, aspect and mood as the previous clause and those with the least amount of cohesiveness are contexts with subjects that are first mentioned in the discourse, are not identifiable, or are a new discourse topic. There are, of course, contexts varying to different degrees between these two extremes (see 7.1.2.3 for a complete discussion and examples). Abreu (2009) investigated the impact of discourse connectedness on the speech of monolingual PR Spanish speakers, bilingual PR Spanish-English speakers and L2 learners. She identified eight levels of discourse connectedness as follows: Level 1: Same referent, same verb TAM Level 2: Same referent, change in TAM Level 3: Change in referent, intervening subject impersonal or [−human] Level 4: Change in subject, referent performs other syntactic function in previous clause Level 5: Change in referent, referent is unambiguous Level 6: Change in referent with another possible candidate for subject Level 7: Same referent maintained, change in topic Level 8: Both discourse topic and referent are changed (Abreu 2009: 87) The author had hypothesized that all three groups of speakers would tend to use more overt subject pronouns the greater the discourse disconnectedness of the contexts (levels 6–8) and more null subjects the greater the connectedness (levels 1–3). For the monolingual speakers the author found that the hypothesis was upheld. The context that triggered the greatest percentage of overt pronouns was level 5, with a change in referent but with an unambiguous referent. This was the context with the greatest number of tokens as well. As the contexts moved up the scale of connectedness (towards same referent, same TAM), the percentage of overt pronouns diminished. Abreu states that “monolingual Spanish speakers respond to the continuity of discourse; when the discourse is disrupted by a change in verb TAM or by a switch in subject, speakers use more overt pronouns than when subjects and verb TAM remain the same across clauses” (2009, 104–105). The bilinguals in her study showed a significant effect for connectedness only at level 3, a switch in subject with an intervening (although not a rival) subject, although when the author collapsed the data into two levels (one where there was a change in subject and one where there is no change), she found that overt pronouns were favored with a weight of .63 in contexts of switch-reference at levels 3, 4, and 5. Overt pronouns were disfavored at levels 1 and 2 (with factor weights of .37 and .46 respectively) (Abreu 2009: 135). Part of the explanation for

240 | 7 A variationist approach to L2 subject pronoun acquisition the bilinguals’ preferences for overt pronouns in contexts at level 3 may be due to their overall greater use of overt pronouns, much in line with Flores-Ferrán’s (2002) findings for New York PR Spanish-English bilingual speakers. The L2 learners in Abreu’s study showed the same general tendencies as the bilingual speakers except for level 3, which showed the least amount of overt subject use among all of the levels. In general, for this group there is greater use of null subjects overall but there does appear to be a very gradual increase in producing overt pronouns in contexts of decreasing discourse connectedness, except at level 3, although the differences are not statistically significant. The only significant difference is between levels 3 and 4; there is no significant difference between levels 1, 2 and 3 or between levels 4 and 5. Thus it is difficult to claim that these intermediate L2 learners respond to discourse connectedness in anything but a very incipient manner. This is not the case, however, for more advanced learners. Geeslin and Gudmestad (2011) also examined the interaction of subject expression and referent cohesiveness in advanced L2 learners and in fact expanded on the model set forth by Bayley and Pease-Álvarez (1997) used in their own previous study. Whereas in their 2010 study they considered only contexts where the subject of the preceding finite verb was the same or different, for their (2011) study they included nine different categories of referents, some of which were the same as those examined in Abreu (2009), and are as follows: Categories 1–4: (same as Abreu’s Levels 1–4) Category 5: subject performed other syntactic function of a preceding verb, one or more intervening clauses Category 6: subject not in the preceding 10 clauses, first mention, not identifiable Category 7: same subject as preceding verb, different person/number, same TMA Category 8: same subject as preceding verb, different person/number, different TMA Category 9: same subject of a previous verb, different person/number, one or more clauses intervene (Geeslin and Gudmestad 2011: 21) Examining the same data set used in their previous two studies, the authors found for the variable of referent cohesiveness that both groups produced more null subjects for all nine categories, but like the Abreu (2009) study, the L2 learners used even more null subjects than the native speakers. However, unlike the Abreu L2 learners, the more advanced learners in the Geeslin and Gudmestad (2011) study were constrained by the same levels of discourse connectedness as the native speaker group. Both groups produced more null subjects in contexts of same

7.2 Conditioning factors in L2 variationist studies |

241

reference (categories 1 and 2), between 84.8 % and 87.9 % for learners and between 83.2 % and 84.3 % for native speakers, than for contexts of switch reference (categories 3–5). The authors, as in their previous studies, included more than just null and overt subject pronouns and considered lexical noun phrases and ‘other’ (demonstrative, interrogative and indefinite) subject pronouns. For the categories of changes in the person or number of the referent, personal pronouns were more frequent than ‘other’ pronouns and lexical noun phrases, and in contexts where the referent had not been mentioned in the previous ten clauses (category 6), the lexical noun phrase was the most common choice following null subjects: 38.1 % for learners and 39.1 % for native speakers. The authors’ findings demonstrate that although there are differences in the rates of production for different contexts, in general, the patterns of use for the L2 learners is similar to those of the native speakers. In other words, advanced learners are constrained by the same discourse-level factors as native speakers when selecting among the various choices available for subject expression forms. In terms of referent cohesiveness, the choice of subject forms depends on the distance of a referent from its original mention and changes in its grammatical function. In fact, in their multinomial probit model of the same data set, Gudmestad, House, and Geeslin (2013) found that the advanced L2 learners’ subject expression is more sensitive to the variable of referent cohesiveness than native speakers’ subject expression. The authors found that, among the highest ranked variables of switch reference, perseveration, and referent cohesiveness, the latter was ranked second with the second highest Euclidean distance of 0.630 (compared to 0.733 and 0.626 for switch reference and perseveration respectively)⁴⁸ for the L2 learners. Although research on native and bilingual speakers has revealed that connectedness or cohesiveness is an important factor that impacts subject expression in natural discourse, the two L2 studies that have examined this variable in learners suggest that paying attention to the form and placement of referents in diverse discourse contexts is a complex linguistic and cognitive activity that perhaps only develops gradually as learners gain greater control over the use of diverse subject forms and are better able to work out their different discourse functions. The need to use forms to connect different parts of the discourse to one another can lead to either redundancy or ambiguity and, as we have seen in Chapter 6, less proficient learners will often opt for greater redundancy, and this can explain why

48 The authors used Euclidean distance to assume that categories with greater distances (closest to 1) influence the projected choice of subject expression more than categories with smaller distances.

242 | 7 A variationist approach to L2 subject pronoun acquisition the learners in Abreu’s (2009) study select more overt subjects even when there are no competing referents. At more advanced levels, as seen in the Geeslin and Gudmestad (2011) study, learners are more adept at producing different forms and can control their selections for more complex discourse relations.

7.3 Chapter summary The work on the L2 acquisition of subject expression from a variationist/sociolinguistics perspective has shown that patterns of use for both native speakers and L2 learners are similar in many ways. That is to say, learners, at least at the advanced level, use subjects in a similar manner as native speakers and this use appears to be constrained by the same linguistic and discourse-level factors as native speakers when selecting subject expression forms. The interaction between subject expression and person, number and specificity of the referent is a significant factor for native and bilingual speakers and intermediate and advanced L2 learners. All of the studies reviewed here found that L2 learners of different proficiency levels preferred null subjects above all other forms even more than native speakers and quite a bit more than bilinguals, who tend to use more overt subjects than monolingual native speakers. Other important differences between native speaker and L2 learner groups are seen among the different persons and number. Whereas native speakers tend to use more overt subject pronouns for first person singular and second person singular specific, L2 learners tend to prefer null subjects for these persons and overt pronouns for third person singular except among the most advanced learners, who were shown to use more first person overt subjects, like native speakers. However, these same advanced learners, unlike native speakers, tend to use more null subjects for second person specific singular and more overt subjects for first person plural. Finally, several studies showed that native speakers of Spanish are more sensitive to specificity and subject expression than learners. Thus, the studies provide evidence that some persons in the paradigm are more problematic than others at different levels of proficiency, a finding in line with Lozano (2009) and which demonstrates that although learners’ subject expression is conditioned by person, number and specificity, their use of subjects is selective and affects the pronominal paradigm in different ways and at different levels of development. For the factor tense, aspect, mood (TAM) and potential ambiguity, again the results are mixed. In general, native speakers used a wider variety of verbal forms in the studies that included them. Furthermore, it was found that native speakers are more impacted by TAM but not by potential ambiguity and L2 learners, except at the more advanced levels, are not impacted by TAM, but are by possibly poten-

7.3 Chapter summary

| 243

tial ambiguity. Native speakers, it appears, are more likely to use overt subjects with subjunctive, imperfect, and present tenses but not because of the potential ambiguity involved, but rather because they tend to occur in subordinate clauses where overt subjects and imperfective verbal morphology are used for expressing background information (Silva-Corvalán 2001). Native speakers are inclined to use null subjects for conditional and periphrastic future tenses, whereas L2 learners, on the other hand, tend to use more null subjects for present, imperfect, preterit and present perfect. Therefore, TAM does appear to interact with subject expression for both native speakers and L2 learners but in different ways. On the other hand, only for the lower proficiency group in the Abreu study was it found that ambiguity interacted with overt pronoun use; for all other studies, for both native speakers and L2 learners, ambiguity was not found to impact subject expression in a significant way. Abreu suggested that perhaps lower proficiency learners are more dependent upon their classroom learning and more attuned to potential ambiguity for different morphological forms. Only two studies from a variationist perspective (to my knowledge) have examined the impact of semantic class on subject pronoun use and the findings are contradictory. Abreu found that semantic class is not significant for either native speakers (a surprising finding and contra previous studies of native and bilingual populations) or L2 learners, although the native speakers are apt to use more overt pronouns with verbs of cognition and communication than any other verb class and tend to use the least amount of overt pronouns with verbs of motion. The use of overt pronouns in the bilingual speaker group in her study, on the other hand, was found to be significant and speakers tend to use more overt pronouns with cognitive, stative and volition verbs and null pronouns with verbs of communication, motion and possession. Linford and Shin (2013) found for the more advanced learner group a contrary finding; semantic class was found to significantly interact with pronoun use but overt pronouns are more common with verbs of activity and least common with verbs of mental activities. These contradictory findings may be attributed to the different ways in which semantic class is defined among different studies and in identifying, tabulating and analyzing subject forms. Clearly future studies in the L2 acquisition of subject expression will need to define semantic class in a more stringent manner. The factor of switch or same reference was found in all of the studies to significantly impact the selection of overt and null subjects for both native speaker and L2 learner groups, and even more so for more advanced learners. Abreu (2009) found that the monolingual and bilingual groups clearly prefer overt subject pronouns in contexts of switch reference, whereas the L2 learner group only do so in 50 % of the cases. All the other studies reviewed here found switch reference to be a significant factor in predicting overt subject use. In the Linford and Shin (2013)

244 | 7 A variationist approach to L2 subject pronoun acquisition study, the factor was found to be significant for both groups but even greater for the upper level proficiency group. In the Gudmestad, House, and Geeslin (2013) study, switch reference was found to be the most important variable for the L2 learners who are more likely to select a lexical noun phrase in these contexts, followed by an overt subject pronoun. In terms of perseveration or priming, it was seen that production of subject forms varied in important ways according to the form of the prior mentioned referent. In fact, of all the factors examined, it could be said that with perseveration or priming, there is the greatest convergence among the different speaker groups examined. There are very clear and similar patterns of use for subject expression for native speakers, bilingual speakers and both intermediate, advanced and high advanced L2 learners: either learners are very good at picking up on a language specific discourse property in Spanish or the effects of priming can be attributed to a universal process in natural discourse. It is most likely the latter. Finally it was shown that native speakers, bilingual speakers, and L2 learners are all sensitive, although to differing degrees and at different levels of proficiency for the L2 learners, to the degree of discourse cohesiveness or connectedness when choosing subject forms. This is to say, the selection of subjects is contingent on the distance of a referent from its first mention in the discourse, on changes in that referent’s grammatical function, on changes to the tense, aspect and mood of that subject’s verbal form, and on possible competing referents. Native speakers and more advanced learners respond to the demands of discourse cohesiveness and use more overt subject pronouns the greater the disconnectedness of the discourse and more null subjects the greater the connectedness of the discourse. Among lower level learners there is not as clear a pattern of use but in the Abreu study, the author reported a gradual increase in producing overt pronouns in contexts of decreasing discourse connectedness. Finally, Gudmestad, House and Geeslin (2013) found that very advanced L2 learners’ subject expression is even more sensitive to the variable of referent cohesiveness than native speakers’ subject expression. Because discourse cohesiveness involves more clausal distance and more and competing referents and/or topics in many cases, and is determined by several linguistic factors such as tense, aspect, mood, number and person of the referent, this factor, and how it interacts with subject choices is by far a more cognitively demanding task. Learners appear to respond well to local discourse factors, related to the immediately previous clause, such as switch or same referent or priming, and thus their choices for subject expression mirror more closely those of native or bilingual speakers. However, it appears that responding to cohesiveness in discourse takes time to develop and refine. These studies confirm the general findings of sociolinguistic research for subject expression, which has shown that monolingual and bilingual speakers’

7.3 Chapter summary

| 245

choices between overt and null pronominal forms, are constrained by multiple factors. The research reviewed in this chapter reveals that the linguistic factors of person, number, specificity, TAM, and semantic verb class, all constrain subject expression in all groups of speakers but in different ways for the L2 learners at different stages of development. Furthermore, the complex discursive factor of cohesiveness or connectedness involves the juggling of multiple linguistic and discursive variables and also appears to develop over time. However, the local discursive factors of switch or same reference and perseveration or priming appear to produce the same effects among native and bilingual speakers and L2 learners of Spanish. Future research will need to take a closer look at these findings in order to ascertain why this should be so. Finally, the variationist model for L2 acquisition highlights the need, apart from same and switch reference, to examine multiple factors, both linguistic and discursive which, to date, have not been explored thoroughly in other approaches to the acquisition of L2 subject expression in Spanish.

8 Summary of findings Introduction This final chapter summarizes the findings of the research on L2 learners’ acquisition, use and interpretation of subject expression in Spanish from the multiple paradigms discussed throughout the book. The purpose of this chapter is to point out some of the more common and noteworthy findings among the studies, but also to highlight some of the contradictory evidence among the different perspectives. As mentioned in the first chapter, the purpose of this volume is to come to a more complete understanding of the complexity of the acquisition process in general and of the L2 acquisition of Spanish subjects in particular. The overriding objective is to find common ground among the distinct theoretical perspectives. I firmly believe that, although I work within a functional discourse-pragmatic perspective myself, each of the theoretical models has valuable and irrevocably profound contributions to the field of second language acquisition. From each chapter, I present a brief summary of the objectives of the model and of the findings of the various research studies. Subsequently, I return to the major research questions/issues that I believe define the research agenda for each model. An attempt to answer each of the questions based on the evidence from the various studies is then made. In the final section I discuss the implications of the multiple approaches for our understanding of the L2 acquisition of Spanish subjects.

8.1 The parametric model Research in the L2 acquisition of subject pronouns in Spanish from a parametric perspective has in common the two-fold goal of establishing the role of Universal Grammar (UG) in acquisition and of identifying the specific properties of the null subject parameter (NSP), one well-defined parameter claimed to form part of UG. Recall (from Chapters 2 and 3 of the present volume) that early studies from this perspective examined four properties of the NSP as identified by Rizzi (1982, 1986), including overt and null referential subjects in tensed clauses, obligatory null expletive subjects, subject-verb inversion, and that-trace effects. Rich verbal morphology in Spanish marks person and number and thus, referential subjects can be null, whereas expletive subjects that have no real world referents are obligatorily null in Spanish. Overt subjects can appear pre- or post-verbally and Spanish allows object and subject extraction of wh- words with overt complementizers, an apparent violation of the that-trace filter. Although not identified by Rizzi, an

8.1 The parametric model |

247

additional principle of null subject languages proposed by Montalbetti (1986) is the Overt Pronoun Constraint (OPC), which states that overt subject pronouns of subordinate clauses cannot be co-referenced with subjects of main clauses when the main clause subject is variable, as in the cases of interrogative pronouns (e.g. quién ‘who’) or indefinite pronouns (e.g. alguien ‘someone’). It is assumed that L1 child learners set a parameter either one way or another according to the options allowed by UG. If it could be documented that adult L2 learners are capable of acquiring the NSP in the face of underdetermined input, since some of the properties of the NSP are not necessarily derivable from the input, and specifically if it could be shown that learning one property of the NSP automatically leads to the acquisition of all or part of the properties associated with the parameter, then this would be evidence that the initial state for L2 learners is knowledge of UG and that adults have access to this knowledge in the same way that children do. If there is evidence of violation of the properties of the NSP in the early stages of acquisition but indications of adhering to the NSP in later stages, then this is evidence that adult L2 learners start the learning process with their L1 setting and gradually reset the parameter to the L2 option, perhaps evidencing an implicational path of a hierarchy of properties of the NSP. In the latter case, this would confirm that L2 learners have access to UG but only via their L1. Since not all of the properties of the NSP are evident in the input, researchers from this perspective are also interested in examining the role of input in the triggering of the various parameters in order to explain the learnability problem. In sum, the parametric study of the acquisition of subject pronouns in L2 Spanish, as discussed more fully in Chapter 2, has been concerned with four basic issues: 1. What is the initial state of the adult L2 learner: UG or the L1? 2. Do L2 learners acquire the properties of the NSP as a cluster (acquisition of one property leads to “automatic” learning of other properties)? 3. Or does acquisition occur in a hierarchical manner along an implicational path? 4. What is the role of input in the triggering or “re-setting” of parameters? A secondary issue related to these four main ones is whether or not the four properties as proposed by Rizzi and the OPC as proposed by Montalbetti form part of the NSP. We turn now to discussing each issue in light of the evidence.

8.1.1 The initial state Perhaps the most important question in the parametric model is what is the initial state of the language learner? Research that has examined one of the properties

248 | 8 Summary of findings that is never taught in the language classroom and is not frequent in the input and therefore a good candidate to answer this question is learners’ knowledge of the OPC. Attesting to knowledge of the OPC would be evidence that learners have direct access to Universal Grammar. In general, research has revealed that many learners acquire the OPC, even at early stages. This is a formal syntactic property in Spanish that is neither taught in the foreign language classroom nor evident from the input (e.g. complex sentences where it operates are infrequent). There is, however, also conflicting evidence regarding knowledge of the restrictions of the OPC and the data examined offered variable results among the learners studied regarding knowledge and access to the OPC. Whereas Lozano (2002b) indicated that Greek (also a null subject language) learners of Spanish L2 performed better than English speakers on determining OPC effects, a finding that suggests the L1 plays an important role in the acquisition process, Liceras (1989) and PérezLeroux and Glass (1997, 1999), who examined OPC effects of English and French speakers (both non-pro-drop languages), found no support for L1 effect in their studies. In fact, they found that the OPC appears to operate at early stages of development, although as language proficiency improved among the learner groups, null pronoun use increased in both bound variable and referential contexts. An examination of the OPC contexts used in the Pérez-Leroux studies indicated discourse constraints were also at work, suggesting that the OPC is susceptible to discourse constraints as well as syntactic ones. Rothman and Iverson (2007a, 2007b, 2007c), in their studies of English speakers at home and studying abroad in Spain, assume that the L2 initial state is the L1 grammar and that adults can reset the L2 features via direct access to UG. Furthermore, the fact that not all the learners in their studies had knowledge of the OPC and additional input did not lead to resetting the constraint is further proof that L1 is the initial setting but sheds doubt on whether learners, at least all of them unconditionally, have access to UG.

8.1.2 Automatic learning of a cluster of properties Research from a parametric perspective has not provided convincing evidence that the four properties of the NSP as proposed by Rizzi are learned as a cluster. Liceras (1989) and Isabelli (2003) documented that learners reset the parameter for null subjects and subject-verb inversion but not for that-trace effects. However, conflicting evidence is provided by Rothman and Iverson (2007a, b, c), who demonstrated that subject-verb inversion and that-trace effects continued to cause problems at late stages of acquisition although learners appeared to be native-like with null referential subjects. Regarding null expletive and referential pronouns, Lozano (2002a) and Al-Kasey and Pérez-Leroux (1998) both found that

8.1 The parametric model |

249

learners apparently treat these two types of pronouns differently. But whereas Al-Kasey and Pérez-Leroux claim that these expletive and referential subject pronouns form part of a cluster of properties in the NSP, Lozano maintains that learners have distinct mental representations of them, which the data appear to support.

8.1.3 Acquisition along a hierarchical/implicational path Because Liceras (1989) had found that L2 learners were capable of resetting the supposed parameter for allowing both null and overt referential subjects and subject-verb inversion but had difficulties in learning the constraint for thattrace effects, this led her to propose that there is a hierarchy in the learning of the clustered properties and that-trace is the most difficult and perhaps lastlearned property. That it is late-learned has also been attested by Rothman and Iverson (2007a, b, c). We again have a situation (as in the case of expletive and referential subjects) where the data coincide but researchers’ explanations are conflicting. Liceras, at this point in time (1989), was not willing to abandon the idea that properties clustered together in the NSP and were responsible for language acquisition and proposed the hierarchical/implicational path as a possible solution to the data. Rothman and Iverson, on the other hand, reject this hypothesis. They point out that input triggers the settings of the parameter, as in the case of OPC, and therefore if that-trace is a part of the NSP, then the evidence would show learners are capable of resetting that parameter as well. Rothman and Iverson claim that that-trace and possibly subject-verb inversion, also shown to be late-learned among their participants, is not part of the NSP.

8.1.4 The role of input in resetting parameters Despite the fact that the role of input in resetting parameters is one of the major issues of the generative model and of the parametric perspective in particular (and indeed an on-going debate in all of L2 acquisition), few researchers have specifically examined the role of input in the acquisition of the NSP in Spanish. Rothman and Iverson (2007a, 2007b, 2007c) investigated whether learners exposed to more abundant and naturalistic input in the study abroad context were at an advantage for resetting the NSP, specifically the OPC, but found that additional input via study abroad was not beneficial in helping to reset the constraint. Thus, the role of input for acquiring the syntactic properties of subjects as they relate to UG is still not clear.

250 | 8 Summary of findings 8.1.5 Conclusions of findings of the parametric approach Table 27 below summarizes the findings of the parametric approach in terms of answering the research questions proposed in Chapter 1 of the present volume and discussed in section 8.1. of this chapter. It can be seen that despite an abundance of research carried out over two and a half decades, there is conflicting evidence and interpretations and little consensus regarding what the data reveal. Furthermore, and more importantly, the vast amount of research has provided few definitive answers to the questions proposed. In fact as early as 1997, Pérez-Leroux and Glass suggested that the difficulties learners had in acquiring subject pronouns in Spanish could be due to a lack of knowledge of the discourse-pragmatic constraints responsible for null and overt pronoun distribution. Table 27. Summary of findings of parametric approach. Research question or issue

Finding

Comments

Is L1 the initial state of the L2 learner?

Yes

Most studies confirm this.

Do L2 learners have access to UG?

Yes/No

Some properties constrained by UG are learned early; others learned late; some appear constrained by discourse properties.

Do L2 learners acquire properties of the NSP as a cluster?

No

None of the studies could provide support for this claim.

Do learners acquire properties of the NSP in a hierarchical/implicational manner?

No

Although learners acquire different properties of the NSP at different stages of development, there is no evidence that this is implicational.

What is the role of input?

?

Too few studies and not enough evidence to make claims.

What we have found out is that the L1 plays a larger role in L2 acquisition than previously believed and that learners are capable of learning very early the fact that Spanish allows both null and overt referential pronouns. They also appear to learn early on that expletive subjects are obligatorily null. Also, apparently learned early for some learners, although not all, is the OPC, but several researchers have argued that knowledge of the OPC is also constrained by discourse properties. Subject-verb inversion and that-trace appear to be learned late and it has been argued that these properties are not related directly to the NSP. Thus, although the parametric model has not been able to give definitive answers to all

8.2 The syntax-pragmatics interface model

|

251

of the questions posed, it has made a contribution to our knowledge of subject acquisition in L2 Spanish and has reignited interest in the role of the L1 and input and how these interact with the principles of UG and has highlighted the need to examine more closely the discourse-pragmatic properties constraining subjects in Spanish.

8.2 The syntax-pragmatics interface model The syntax-pragmatics interface model is generative because syntactic domains are also considered modular although determined by functional features. As discussed in Chapters 1 and 4, the model diverges from the parametric model because it takes into account the integration of both modular linguistic domains and domains of general cognition to explain residual optionality (lingering errors) among L2 learners (Rothman and Pascual y Cabo 2014). Thus, although purely syntactic grammatical properties are learnable by non-native speakers, those like subject pronouns of pro-drop languages that require an integration of both types of domains may never become native-like. This model has been proposed as a way to explain why the end result of L2 acquisition is non-uniform, unlike the uniform attainment of children’s L1 acquisition. The parametric studies hinted at the possibility, and more recent studies have confirmed, that subject expression in Spanish is dependent on the pragmatic properties of topic and focus in discourse. Thus generative studies turned to discourse pragmatics in order to search for answers regarding the L2 acquisition of subject pronouns in pro-drop languages. Recall from Chapters 2 and 4 that it has been found that native Spanish speakers have a preference for marking non-focus and same reference (i.e. topic continuity) with null pronouns and focus and switch reference (i.e. topic shift) with overt pronouns or full lexical noun phrases. If new information is presented into discourse, the tendency is to use a full noun phrase to bring the referent into focus. When this information continues as a topic, it is known or shared information and native speakers will generally use a null subject because it is topic information or topic-continuity (also ‘same reference’). A change of referent signals topic-shift or ‘switch reference’ and in these cases in Spanish, speakers will normally use an overt subject, either a full noun phrase or an overt pronoun. Furthermore, the transitivity of a verb interacts with focus and can affect word order in sentences with overt subjects; speakers tend to produce preverbal subjects with unergative verbs and postverbal subjects with unaccusative verbs in unfocused declarative sentences. In focused sentences, the effects of ergativity/accusativity disappear and postverbal subjects are expected with both types of verbs. Thus topic and

252 | 8 Summary of findings focus interact with syntax and verb semantics to govern subject expression in Spanish and other pro-drop languages. The question now becomes, in the interface model, how do non-native speakers acquire syntactic structures that lie at the interface of non-syntactic properties necessary for their interpretation and use and, furthermore, whether it is possible for non-native speakers’ knowledge and use of subjects to become native-like. It has been shown that L2 learners at an early stage are capable of acquiring the nullovert parameter for referential subjects, perhaps also obligatory null expletive subjects and the properties of the OPC. The research examined in this volume has shown that bilingual speakers and L2 learners of pro-drop and non-pro-drop languages have higher cognitive demands placed on them when navigating their two languages and this causes difficulties and may explain the lingering deficits in their representational structures of their L2 Spanish. In order to understand how learners integrate the two domains, syntactic and discourse-pragmatic, the syntax-pragmatics interface model, discussed more fully in Chapter 2, attempts to address the following basic issues: 1. Do L2 learners acquire syntactic properties before discourse-pragmatic properties or simultaneously? 2. Why do discourse-pragmatic deficits linger even in very advanced learners? 3. Why is the end result of L2 acquisition non-uniform among adult learners? Additional questions addressed specifically in the L2 acquisition of subject pronouns include the following: 4. Is subject pronoun use in L2 learners constrained by discourse context? 5. To what extent is transfer from the non-pro-drop to the pro-drop language unidirectional (evidenced by the overuse of overt subjects)? 6. Is word order constrained by lexical verb class and focus? The extent to which these questions/issues are addressed and answered in the research from a syntax-pragmatics interface model and to what extent there is consensus among studies is the topic of the following sections. We begin by addressing the more specific questions to subject pronouns and end with the more general, theoretical questions guiding the research agenda of the syntax-pragmatics interface model.

8.2.1 Subject pronoun use constrained by discourse context The studies discussed in the present volume presented results that suggest that bilingual speakers and L2 learners, except for very young bilingual children and

8.2 The syntax-pragmatics interface model |

253

very low-level learners, are sensitive to discourse context in selecting null and overt subject pronouns. The bilingual studies showed that bilingual children of various age groups tended to reject ambiguous null subjects more than redundant overt subject pronouns. This finding was also reflected among adult bilingual (heritage speakers), where the intermediate speakers had more difficulties than the more advanced group, but both groups were more accurate pragmatically in the use of overt pronouns than null subject pronouns. In the L2 studies, younger children and lower level learners had more difficulties, but as proficiency improved, all learners became more sensitive to the discourse constraints and selected null and overt subjects accordingly. Nonetheless, there were lingering deficits for even advanced learners. The general tendency observed in the L2 studies is that the use of overt subjects with same referent diminishes with proficiency, as it should, but the use of null subjects for switch referent persists even up to advanced levels. These results conflict with the child bilingual data where all groups, with increasing age and proficiency, eventually rejected more often the ambiguous null subjects in contexts of switch reference than they did the redundant overt subjects in contexts of same reference.

8.2.2 Unidirectional transfer There appears be partial support for the claim that transfer from the non-prodrop to the pro-drop language is unidirectional as evidenced by the overuse of overt subjects (Sorace 2004; Serratrice, Sorace, and Paoli 2004; Sorace, Serratrice, Filiaci, and Baldo 2009) but not completely. The use of anomalous null subjects in the pro-drop language contradicts the unidirectional hypothesis although ambiguous null subjects are discarded by bilingual children before redundant overt subject pronouns are. L2 learners appear to overuse both null and overt pronouns, again evidence against the unidirectional hypothesis.

8.2.3 Word order constrained by lexical verb class and focus In the studies that examined the impact of the interaction of lexical verb class and focus on word order (with overt subjects only) the results are mixed. In the younger children of the Pladevall Ballester (2010) study, neither lexical class nor focused context affected word order: the English-Spanish bilingual children appeared to prefer the canonical S–V word order regardless of verb class and focus. However, the Spanish-Catalán bilingual children of the Bel (2003) study, even as young as two years of age, were sensitive to both constraints, although caution

254 | 8 Summary of findings should be taken when making claims as it was difficult to interpret these very young children’s conversational data. Furthermore, both Catalán and Spanish are pro-drop languages. Regarding the adult L2 learners, the consensus seems to be that lower level learners are not sensitive to either verbal class or discourse context but there is gradual improvement as proficiency develops, however, with residual deficits even among advanced learners. The implications of these findings for the three research issues that are specific to the L2 acquisition of subject pronouns in Spanish will be discussed as we address the more general theoretical issues of interest to those working within the syntax-pragmatics interface model.

8.2.4 Subsequent vs. simultaneous acquisition of syntax and discourse-pragmatic constraints One of the main questions in research in L2 acquisition based on the interface model is to ask to what extent learners acquire syntactic and pragmatic properties at the same time or in a subsequent manner. The general consensus is that formal morphosyntactic properties are acquired before discourse-pragmatic constraints and the evidence is that bilingual children, bilingual adults and adult L2 learners very early on in the acquisition process all use both null and overt referential subject pronouns, although they also overuse ambiguous null subjects and redundant overt subject pronouns. Furthermore, bilingual children of a pro-drop language and a non-pro-drop language take longer to sort out the prodrop system than monolingual children and bilingual children of two pro-drop languages. In addition, in the studies that included them, it was seen that L2 learners learn fairly early that expletive subjects are obligatorily null. Finally, there is increasing evidence that at least some learners are capable of acquiring the properties of the OPC, despite the lack of input that would make this evident. Therefore, the evidence seems to support the claim that syntax is acquired before the discourse-pragmatic constraints, which are acquired late and develop more slowly over time, although Lozano claims that at least Greek learners of L2 Spanish have difficulty grammaticalizing syntactically the discursive feature of focus (Lozano 2006: 392–393).

8.2.5 Why do discourse-pragmatic deficits linger? Once established that bilingual speakers and L2 learners have acquired the formal morphosyntactic properties of a given structure, the logical subsequent question to pose is why do discourse-pragmatic deficits linger even in very advanced learn-

8.2 The syntax-pragmatics interface model |

255

ers? It has been suggested that the difficulties evidenced in the distribution of null and overt subjects for bilingual and L2 speakers of non-pro-drop and pro-drop languages is due to how speakers process the information status of subjects in the two languages. In pro-drop languages distribution of null/overt and preverbal/postverbal subjects interfaces with léxico-syntactic and discourse-pragmatic properties, whereas in non-pro-drop languages such as English it does not. Thus in pro-drop, overt subjects are specified for the features [+topic shift] and [+focus] and in non-pro-drop languages subjects are underspecified for these features. In English, lexical subjects express high informative status and pronouns express low informative status. Prosodic stress as well is used to indicate topic shift and/or focus in non-pro-drop languages such as English. When speakers are faced with these two competing systems bilinguals and L2 learners will select the less pragmatically constrained option, which is the non-pro-drop option. The effect is that overt subject pronouns in the pro-drop lose their association with the features of [+topic shift] and [+focus] and are interpreted as coreferential with non-focus and/or topic continuity. That explains why speakers of non-prodrop learning pro-drop languages overuse overt pronouns. When we consider the lexical-semantic properties of verbs and how they interact with the feature of [+focus] to determine word order, the use and distribution of subjects becomes increasingly complex. Each discourse-pragmatic and lexical-semantic constraint adds an additional layer of complexity to the acquisition process. Thus, we can see how the use and distribution of a structure that requires the integration of syntactic and pragmatic knowledge is understandably more vulnerable than structures with purely syntactic properties. Even learners whose L1 is also pro-drop have been attested to have difficulties with the discourse-pragmatic and lexicalsemantic constraints (Lozano 2006).

8.2.6 Non-uniformity of L2 acquisition Why is the end result of L2 acquisition non-uniform among adult learners, unlike the uniform attainment of children’s L1 acquisition? The observable variability in the development and end result of L2 acquisition, unlike the uniform attainment of L1 acquisition, is explained once we understand the complex (and therefore, late-learned) discourse-pragmatic properties responsible for the use and distribution of subject pronouns in pro-drop languages such as Spanish. Furthermore, as has been pointed out by Rothman (2009), because pragmatic properties are made evident to learners through discourse patterns, it could possibly be that L2 learners are not exposed to the amount of natural conversational discourse they need in order to acquire those properties early in the L2 acquisition process. Add

256 | 8 Summary of findings to this the fact that the use of overt pronouns in redundant contexts is not syntactically anomalous and learners have an even more confounding task and may intuitively not pay attention to the more complex discourse-pragmatic patterns of their L2.

8.2.7 Conclusions of findings of the syntax-pragmatics interface model Table 28 displays the summary of the findings for the acquisition and use of null and overt subject pronouns as they relate to each of the questions and issues of the syntax-pragmatics interface model. It is not surprising that the first three, more general theoretical questions have been answered in the affirmative, since the syntax-pragmatics interface model was proposed precisely to respond to the observable variability in the development and end result of L2 acquisition, unlike the uniform attainment of L1 acquisition. Furthermore, researchers set out to explain this variability by examining the apparently complex and late-learned discourse-pragmatic properties of subject distribution in the acquisition of prodrop languages by L1 speakers of both pro-drop and non-pro-drop languages. The empirical research has confirmed what was observed in previous studies from a generative perspective but not systematically documented. And although the answers to the three, more specific research questions probed by the studies discussed here are both affirmative and negative, the pattern that is observed is that the discourse-pragmatic constraints are difficult for L2 learners to navigate, but with greater language proficiency, and presumably contact with natural input of the target language, these constraints are also potentially learned. The questions that still remain are why deficits linger and why full native-like acquisition appears to be beyond the reach of the adult L2 learner?

8.3 Processing models The processing model examined in this volume is also generative and attempts to understand which innate restrictions and/or L1 principles influence how learners process or interpret L2 structures that are similar to or different from their L1; thus, researchers working within this model are interested in exploring more fully the relationship between language structure and cognitive processing. Whereas the interface models could not explain why there are lingering deficits and nonuniform attainment of subject expression in Spanish, the processing models suggest that L2 deficits are the result of processing difficulties with the information found at the interface between syntax and other cognitive domains. The perspec-

8.3 Processing models

| 257

Table 28. Summary of findings of syntax-pragmatics interface model. Research question or issue

Finding

Comments

Are syntactic properties acquired before discourse-pragmatic properties or simultaneously?

Yes

All studies confirm this.

Do discourse-pragmatic deficits linger even in very advanced learners? Why?

Yes

Discourse-pragmatic and lexicalsemantic constraints add additional complexity; L2 learners select less pragmatically constrained option.

Is end result of L2 acquisition nonuniform in adult learners? Why?

Yes

Complex discourse properties learned late and developed through natural conversational discourse.

Are subject pronouns in L2 learners constrained by discourse context?

Yes/No

Early learners, no; advanced learners, yes but with lingering deficits.

Is transfer unidirectional in L2 learners?

Yes/No

Evidence of overuse of null and overt pronouns – more overt pronouns in early levels; more null pronouns in advanced levels.

Is word order constrained by lexical verb class and focus in L2 learners?

Yes/No

Early learners, no; advanced learners, yes but with lingering deficits.

tive adopted in the studies examined in Chapter 5 of the present volume assumes that certain underlying knowledge is already present in the learners’ grammar. However, the assumption that knowledge of the syntactic properties of Spanish subjects exists in the underlying grammar of speakers does not suppose that this knowledge necessarily results in native-like processing strategies (VanPatten and Jegerski 2010). This is especially so for bilingual speakers and L2 learners from one non-prodrop language, such as English, learning a second language (either as a child or an adult) that is pro-drop. These languages have two different strategies for assigning antecedents to pronouns. Speakers of pro-drop languages are more likely to link a null subject pronoun with a more prominent antecedent (usually the subject of the preceding clause) and an overt subject pronoun with a less prominent antecedent (usually in a complement position). This so-called ‘division of labor’ has been proposed as the Position of Antecedent Hypothesis or PAH (Carminati 2002). Speakers of non-pro-drop languages such as English, on the other hand, depend on a subject assignment strategy in which overt pronouns tend to be linked to the

258 | 8 Summary of findings subject antecedent. The studies discussed in Chapter 5 of this volume all examined to some extent how speakers resolve these two competing strategies. The four major questions or issues guiding research in processing studies that have been posited in the examination of the acquisition and use of subject pronouns in L2 Spanish are summarized below. I present them in the opposite order from their presentation in Chapter 1 of this volume, so as to discuss them in descending order from more specific to more general theoretical issues. 1. What strategies do bilingual speakers and L2 learners use for resolving conflicting (L1 and L2) morphosyntactic cues? 2. How do bilingual speakers and L2 learners process different types of information at the interface between syntax and other cognitive domains? 3. How does processing mediate between input and universal linguistic properties in L2 acquisition? 4. Do innate restrictions or L1 principles influence bilingual speakers’ and L2 learners’ interpretations of similar and distinct L2 structures? The evidence from research that responds to these questions/issues is examined in the following sections.

8.3.1 Strategies for resolving conflicting cues The first more specific research question addresses how bilingual speakers and L2 learners resolve the conflicting L1 and L2 cues for antecedent assignment. Recall that the research revealed that native speakers of pro-drop (Italian and Spanish) languages adhere to the PAH for the anaphoric resolution of antecedents; this tendency is stronger in Italian than it is in Spanish, highlighting subtle but important cross-linguistic differences between two closely related typological languages. Furthermore, it has been documented that the predictions of the PAH are most reliable for null pronouns and that the processing cost is higher for speakers when interpreting contexts that force speakers to violate the PAH for null subject pronouns. The research indicates that low proficiency-level L2 learners clearly prefer a subject assignment strategy regardless of the type of pronoun (null or overt) for all contexts. However, among Spanish-English bilingual speakers and English-speaking learners of L2 Spanish at the advanced level, the PAH appears to be a strong predictor for the anaphoric resolution of null subjects and this tendency appears to strengthen as L2 proficiency increases. The PAH is less predictable for the anaphoric resolution of antecedents with overt subject pronouns. Bilingual speakers and L2 learners seem to depend on a subject assignment strategy, much like the one that operates in their L1, and it

8.3 Processing models

| 259

takes longer for them to acquire the L2 syntactic strategy proposed by the PAH, for overt subject pronouns.

8.3.2 Processing at the interface The second research question deals with how bilingual speakers and L2 learners process different types of information at the interface between syntax and other cognitive domains. The research examined two different types of information that would either reinforce or go against speakers’ strategies for the anaphoric resolution of antecedents. Some considered discourse contexts that would either favor or violate the tendencies predicted by the PAH and others investigated pronoun resolution in subordinating and coordinating discourse structures. Again, among monolingual speakers of pro-drop languages, and especially Italian, it was found that there is a high processing cost for anaphoric resolution in contexts that force an interpretation that violates the biases of the PAH and this cost is greater for contexts that violate the PAH for null subjects. The trend for bilingual speakers and L2 learners is similar. The PAH strategy for null subjects appears to be strongest for L2 learners when both the syntactic position (preverbal) of the antecedent and the pragmatic context favor the bias predicted by the PAH. Recall also that native Spanish speakers interpret anaphoric assignment based on pronoun form (null or overt) independently of discourse structure as predicted by the PAH. For L2 acquisition, however, it was found that only very advanced L2 learners adopt the L2 processing strategy and only for the anaphoric resolution of null subjects in coordinating discourse structures. They continue to rely on the L1 strategy for null subjects in subordinating structures and for overt subject pronouns in both types of discourse structure, thus leading researchers to suggest that advanced L2 learners of Spanish have a hybrid strategy for resolving anaphoric assignment (Jegerski, VanPatten, and Keating 2011).

8.3.3 Mediation between input and universal linguistic properties The third research question of interest to researchers that work within a processing model deals with the extent to which processing mediates between input and universal linguistic properties in L2 acquisition. The evidence here is conflicting. On the one hand it was found in the Keating, VanPatten and Jegerski (2011) study that, despite abundant naturalistic input from an early age, when confronted with two conflicting subject assignment strategies for overt pronouns, early bilingual or heritage speakers adopt the subject rule for English rather than

260 | 8 Summary of findings sorting through two competing systems for resolving anaphoric reference. The authors attribute this to a “persistent and potentially permanent optionality in bilingual grammars” (Keating, VanPatten and Jegerski 2011: 194). On the other hand, very advanced L2 learners appear to sort out the pragmatic rules that constrain subject use in Spanish, at least for null subjects (Jegerski, VanPatten, and Keating 2011). Finally Sorace et al. (2009) discovered that bilingual children of two pro-drop languages continue to prefer the more efficient, less strenuous subject processing strategy, which, although is not syntactically deviant, may result in pragmatically anomalous structures. Eventually, however, with time the pro-drop system stabilizes for bilingual children.

8.3.4 Innate or L1 principles The final question of interest to researchers working from a processing model involves understanding whether innate restrictions or L1 principles influence bilingual speakers’ and L2 learners’ interpretations of pronouns. The answer appears to be, both. Although the discussion regarding whether or not the pro-drop option of the NSP is the unmarked one is far from settled, research has shown that even monolingual children of pro-drop languages have a subject strategy and interpret overt pronouns as co-referential with subject antecedents. However, by adulthood, both monolingual and bilingual children exhibit biases for anaphoric interpretation as predicted by the PAH. Thus it appears that the strategy for interpreting overt subjects as subject antecedents is a universal one and the one that adheres to the PAH is a language specific strategy, which monolingual children eventually sort out. A similar although less satisfying end-result is documented for L2 learners of pro-drop languages. L2 learners, when faced with two competing systems for anaphoric resolution, prefer the L1 strategy for overt pronominal resolution, which is reinforced by the universal strategy observed in young monolingual children. They do seem to acquire the L2 strategy for null subjects as predicted by the PAH, and thus L2 learners, even at the advanced stages, end up with a kind of hybrid strategy that reflects a combination of L1, L2 and universal principles.

8.3.5 Conclusions of findings of the processing model The findings of the processing model are summarized in Table 29. We can confirm that antecedent assignment of null and overt subject pronouns in Spanish among bilingual heritage speakers and L2 learners depends on a number of factors in-

8.4 Discourse-pragmatic approach |

261

cluding cross-linguistic differences in the adherence to the PAH or the subject strategy, universal or innate processing strategies, and language specific processing strategies related to pragmatic context and discourse structure. Table 29. Summary of findings of the processing model. Research question or issue

Finding

Comments

What strategies do L2 learners use for resolving conflicting L1 and L2 cues?

PAH and Subject strategy

Early learners – subject strategy Advanced – PAH for null/subject strategy for overt

Does pragmatic context and discourse structure interface with syntax for pronoun resolution in L2 learners?⁴⁹

Yes/No

PAH strategy strongest when it coincides with pragmatic context and discourse structure, but only for null subjects

Does processing mediate between input and universal linguistic properties in L2 acquisition?

Yes/No

Conflicting evidence for bilingual and L2 learners

Do innate restrictions or L1 principles influence anaphoric interpretations in L2?

Both

When L1 reinforces innate tendencies, these interpretations linger longer

By considering the information structure of discourse and the interface of syntactic and discourse-pragmatic features, we have been able to come closer to an understanding of the distribution and acquisition of null and overt subject expression in L2 Spanish and the lingering problems learners display. The research shows that learners are constrained by the same discursive features as native speakers but in different ways at different levels of proficiency.

8.4 Discourse-pragmatic approach The discourse-pragmatic approach holds that linguistic competence also includes the knowledge of discourse constraints or pragmatic rules that guide appropriate language use and acquisition, and how notions such as coherence and sequential organization are produced and understood in discourse structure. We saw that re-

49 Based on the original question of “How do L2 learners process information at the interface between syntax and other cognitive domains?”

262 | 8 Summary of findings search from a discourse-pragmatic approach considers language beyond the level of the sentence and examines larger portions of discourse or text to identify a wide variety of discourse notions that affect language use, including: cohesion and coherence, saliency, focus, presupposition, pragmatic weight, and cognitive status. In addition, researchers have taken into account the role, familiarity, relative importance and function of referents in narrative structure and consider what shared knowledge the speaker expects the addressee to already possess. This knowledge is reflected in proposed universal hierarchies, which predict that the less information shared regarding a potential referent, the higher in the hierarchy a referring expression is. Thus, a null pronoun presupposes the maximum amount of information and a full lexical noun phrase presupposes the least amount of information shared between speaker and listener. This is in line with Ariel (1990), Givón (1983) and Levinson’s (1987) claims that greater referent saliency and informativeness correspond to more phonetically minimal referring expressions and lesser referent saliency is consistent with more phonetically elaborate expressions. The research presented in Chapter 6 of the present volume takes into account many of these discourse-pragmatic notions in attempting to understand the L2 acquisition of subject expression. The major questions (discussed more fully in Chapters 1 and 6) guiding this body of research include the following: 1. What are the discourse constraints or pragmatic rules that guide language use and L2 acquisition? 2. How does knowledge of information structure in discourse contribute to appropriate use and acquisition in L2 learners? 3. To what extent do implicational hierarchies related to speaker presupposition and cognitive state account for subject forms in L2 discourse? 4. How do universal cognitive abilities and language specific principles constrain L2 acquisition and use? We now turn to summarizing the research that has attempted to respond to these questions.

8.4.1 Discourse constraints/pragmatic rules Needless to say these are big questions and not ones that can be answered by a handful of studies. Nonetheless, what has been uncovered so far is that in many ways L2 learners are sensitive to many of the same constraints as native speakers, although in different ways and at different stages of proficiency. In general terms, all of the studies revealed that the need to contrast two or more potential referents turned out to be a strong predictor for the use of overt subjects, and this did in-

8.4 Discourse-pragmatic approach |

263

deed, reflect the predictions of both the anaphoric (Saunders 1999) and Givenness (Gundel, Hedberg, and Zacharski 1993; Blackwell and Quesada 2011) hierarchies tested. Furthermore, it was seen that verb semantics and the level of speaker commitment appear to correspond to greater use of overt first person subject pronouns (López-Ortega 2002; Lubbers Quesada and Blackwell 2009). Finally, the majority of the studies reviewed determined that the choice of subject forms in L2 narrative discourse appears to be determined by constraints such as saliency (informativeness), topic and focus and these interact with more abstract properties of memory, attention and presupposition (i.e. presupposed knowledge). Specifically, Lubbers Quesada and Blackwell (2009) found that saliency of referent, contrastive focus, and switch focus were predictors for subject expression for both native speakers and L2 learners, but the strength of the pragmatic constraints were different for native speakers and learners and also different for lower proficiency and higher proficiency learners. Furthermore, it was observed that all L2 learners overused overt subject pronouns for same referent, i.e. a referent that was already salient in the discourse. Lower proficiency learners are constrained only by contrastive focus contexts and are not sensitive to the discourse constraints of switch focus or referent saliency, whereas learners at the more advanced proficiency level responded to all three of these constraints. Only higher advanced learners and native speakers use overt subject pronouns for expressing pragmatic weight. This research shows again that learners are capable of mastering early the most evident distinctions, such as contrastive focus and salient referent. The fact that learners had difficulties with the constraint of switch focus, which is neither contrastive nor salient, reveals that learners have difficulties learning the finer grained distinctions for subject expression.

8.4.2 Impact of information structure in discourse This issue has been dealt with most completely by Saunders (2003) and LópezOrtega (2002). Saunders determined that the choice of form (i.e. null form, overt pronoun, definite noun phrase, indefinite noun phrase, etc.) among native speakers and L2 learners is determined by the role, familiarity and relative importance of the character in the story; by the function of that character during different points throughout the narrative, whether the character is being introduced, maintained or reintroduced; and by the syntactic position of the referent. Main and secondary characters appear more often in subject position than less prominent characters which confirms, both syntactically and cognitively, the importance of the main characters to the story line. Nonetheless, L2 learners do not show the same pattern of use until the advanced level when they start to converge on the

264 | 8 Summary of findings pattern observed in the native speakers. This reveals to some extent that learners are also constrained by the role and function of the characters in narrative structure and use different subject expressions for maintaining cohesion and coherence throughout their narratives. López-Ortega also revealed that L2 learners are capable of developing specific discourse principles that permit them to create cohesion in their conversations and that these principles become more native-like with exposure to the target language. Specifically she found that L2 learners are sensitive to the role of referents in discourse. L2 learners tended to mark specific, human referents and unknown (to the hearer) referents with overt subjects, and nonspecific and inanimate referents with null subjects. Furthermore, she found the use of explicit pronominal subjects for emphatic and contrastive purposes was a strong developmental pattern.

8.4.3 Predictions of implicational hierarchies The extent to which implicational hierarchies (related to speaker presupposition and cognitive state) predict subject expression in L2 discourse was addressed most thoroughly by Saunders (2003) and Blackwell and Lubbers Quesada (2012). Both the anaphoric (Saunders 2003) and the givenness (Gundel, Hedberg, and Zacharski 1993) hierarchies were found to predict native speaker subject expression almost categorically and they were fairly accurate at predicting L2 learners’ use. L2 learners, however, especially in the lower proficiency levels used more elaborate forms than expected by the hierarchies. In general, L2 learners overused both overt subject pronouns and lexical noun phrases. In addition, Saunders found that lower level learners do not distinguish between indefinite and definite noun phrases for introducing characters and overuse definite noun phrases to maintain characters. Both Saunders and Blackwell and Lubbers Quesada found that learners do not violate the order of the proposed hierarchies for subject expression; that is, they do not use a form that is higher on the hierarchy to maintain a character than the form they have used to introduce that character.

8.4.4 Universal cognitive abilities vs. language specific principles That both native speakers and learners are constrained by the predictions of implicational hierarchies, albeit in different ways until advanced levels, attests to the universality of such proposed hierarchies. Cooperative speakers of all languages make underlying assumptions regarding the informativeness and saliency

8.5 Variationist model

| 265

of referents in discourse and assume that greater referent saliency corresponds to more minimal referring expressions (Levinson 2000). Languages differ, however, in how they encode informativeness and saliency. In Spanish the most minimal referring expression is the null form whereas in English, it is an overt pronominal form. Thus learners must distinguish between two new forms for expressing their underlying assumptions. Until they do, this leads to an overuse of redundant overt forms. In addition, all languages mark contrast, emphasis or disambiguation in an effort to create cohesive discourse. For English and Spanish this is usually done via overt forms, although Spanish can also rely on null forms, especially when the referent is particularly salient in the discourse. Finally, notions such as speaker commitment, intention and evaluation are universal pragmatic notions that are highly subjective. They may communicate a certain kind of contrast or disambiguity, but not necessarily so. Spanish speakers tend to use overt subject pronouns, usually in the first person, to portray a greater commitment or evaluation towards a proposition. These are subtle pragmatic uses of subject expression that are lost on the L2 learner, even at very advanced stages.

8.4.5 Conclusions of findings of the discourse-pragmatics model Thus, we can conclude that although the discourse-pragmatic notions that constrain and guide subject expression are found in all languages, and certainly it can be confirmed that both native speakers and learners are constrained by the same constraints, it does not necessarily translate into the same distribution of the different subject forms available for creating cohesive discourse. It is only by analyzing data beyond the level of the sentence at the discourse level that certain deficits in L2 language become evident and possible to explain. Table 30 shows the summary of the findings for the acquisition and use of null and overt subject pronouns from the discourse-pragmatics perspective as they relate to each of the questions and issues related to the approach.

8.5 Variationist model The variationist model, like the discourse-pragmatic approach, also examines language beyond the sentence level; however, variationist research can also consider extra-linguistic social factors, such as social or economic status, social distance, and social context. Researchers who work within this model claim there is a need to establish the effects of a broad range of linguistic and social factors and the assumption among sociolinguistic studies is that because non-native

266 | 8 Summary of findings Table 30. Summary of findings of the discourse-pragmatics model. Research question or issue

Finding

Comments

Are the discourse constraints that guide language use and L2 acquisition the same as those for native speakers?

Yes/No?

Same constraints but this does not necessarily translate into the same subject forms; deficits remain until advanced levels.

Does knowledge of information structure in discourse contribute to appropriate use and acquisition in L2 learners?

Yes/No?

Choices of subject forms determined by information structure but different from native speakers; deficits remain.

Do implicational hierarchies related to speaker presupposition and cognitive state account for subject forms L2 discourse?

Yes/No?

Predict subject expression; L2 learners tend to use more elaborate forms on hierarchy.

Do universal cognitive abilities and language specific principles constrain L2 acquisition and use?

Yes

But question remains as to how these universals translate into a particular language.

speaker language varies in contexts where native speaker language does not, analyses of learner language must include all the contexts for subject expression. The distinct research studies that have been discussed in the present volume from a variationist perspective all examine the development of linguistic forms in terms of probability, where linguistic and extralinguistic factors that are likely to condition the selection of one variable over another has a probabilistic weight. This weight predicts a statistically significant level for the likelihood that any given form will be selected by a speaker when the independent variables for use are present. L2 variationist studies are based on sociolinguistic research of monolingual and bilingual speakers’ choices between overt and null pronominal forms and take into account a wide variety of factors likely to constrain these choices, including verb form and semantics, morphological and contextual ambiguity, referent specificity, switch or same reference, topic continuity, perseveration or priming of subject form, and discourse connectedness. The main hypothesis of many of these studies is that L2 variability is constrained by many (if not all) of the same linguistic and discourse constraints responsible for L1 variability. The leading questions guiding research of subject expression in Spanish from a variationist perspective include the following: 1. Does L2 learners’ language exhibit variability in the same linguistic contexts where native speaker language varies?

8.5 Variationist model | 267

2. 3.

Does L2 learners’ language exhibit variability in the same discourse contexts where native speaker language varies? Does the lexical semantics of the verb affect variability of subject expression?

The extent to which the answers to these questions are addressed, and answered, is the subject of the following sections.

8.5.1 Linguistics contexts/factors and variability affecting subject expression in L2 For all of the studies reviewed from a variationist perspective it was found that the interaction between subject expression and person, number and specificity of the referent is a significant factor for native and bilingual speakers and L2 learners alike. Nonetheless, it was found that L2 learners preferred null subjects above all other forms and whereas native speakers use more overt subject pronouns for first person singular and second person singular specific, L2 learners use null subjects. Also, whereas native speakers and advanced learners use more null subjects for third person singular, lower level learners tend to use more overt subjects. In many ways, the patterns of use are inverted. This shows that L2 learners know that both null and overt subjects are options, but have not completely acquired the nativespeaker constraints that dictate their use. Finally, the studies revealed that native speakers respond more to referent specificity than learners. Thus, although person, number and specificity all constrain subject expression for native speakers and L2 learners, they do not result in the same production. In the same vein, it was shown that tense, aspect, mood (TAM) and potential ambiguity produced mixed results. In general, it was found that native speakers are more affected by TAM but not potential ambiguity and L2 learners, except at the more advanced levels, are not impacted by TAM, but are by possibly potential ambiguity. Native speakers are more likely to use overt subjects with verbal morphology that is more apt to be used in subordinate clauses, whereas L2 learners do not make these distinctions. Therefore, TAM does appear to interact with subject expression for both native speakers and L2 learners but, again, in different ways.

8.5.2 Discourse contexts/factors and variability affecting subject expression in L2 An important finding in all of the variationist studies was the fact that the discourse context of switch or same reference significantly impacts the selection of

268 | 8 Summary of findings overt and null subjects for both native speaker and L2 learner groups and this trend becomes stronger as L2 proficiency develops. Also, for the factor of perseveration or priming, it was seen that the choice of subject forms varied significantly according to the form of the prior mentioned referent. In fact, among all of the factors examined, priming appears to be the one where the greatest convergence between native speakers and L2 learners occurs. Finally it was shown that native speakers, bilingual speakers, and L2 learners are all sensitive, although to differing degrees and at different levels of proficiency for the L2 learners, to the degree of discourse cohesiveness or connectedness when choosing subject forms. Native speakers and more advanced learners respond to the demands of discourse cohesiveness and use more overt subject pronouns when there is the potential for greater disconnectedness in the discourse, and conversely, use more null subjects the greater the connectedness of the discourse. Among lower level learners there is not as clear a pattern of use but gradual increase in producing overt pronouns in contexts of decreasing discourse connectedness was reported. Learners appear to respond well to the demands of the local discourse context for switch or same referent or priming because their choices for subject expression are more like those of native speakers. However, responding to longer distance relations with more intervening clauses and referents requires greater proficiency.

8.5.3 Effects of verbal lexical semantics and variability of subject expression in L2 The findings for the effects of verbal semantics on the use of subjects in L2 discourse have been conflicting. Abreu (2009) found that semantic class was not significant for either native speakers or L2 learners, although native speakers tended to use more overt pronouns with verbs of cognition and communication. The use of overt pronouns with verbs of cognitive in the bilingual speaker group, on the other hand, was found to be significant. Linford and Shin (2013) found the opposite effect where a group of advanced learners used more overt pronouns with verbs of activity and fewer with verbs of mental activities. Because this is a factor that has been proven to have an impact on the use of overt subject pronouns in native speakers, future studies will need to take a closer look at this factor among L2 learners of Spanish. The summary of the major findings of L2 variationist studies on subject expression reveals that both linguistic and discourse contexts and factors impact L2 learners’ discourse production and that learners are constrained by the same variables and in many of the same ways as native speakers. The evidence from the linguistic factors shows greater differences. It is clear that L2 learners are con-

8.6 Common findings and concluding remarks |

269

Table 31. Summary of findings of the variationist model. Research question or issue

Finding

Comments

Does L2 learners’ language exhibit variability in the same linguistic contexts where native speaker language varies?

Yes

Person, number and specificity of referent, and TAM and ambiguity affect subject expression but in different ways.

Does L2 learners’ language exhibit variability in the same discourse contexts where native speaker language varies?

Yes

Same, switch reference, and priming affect subject expression in L2 discourse like native speakers; connectedness in different ways.

Does the lexical semantics of the verb affect variability of subject expression?

No

Too few studies and not enough evidence to make strong claims.

strained by the same linguistic variables but in different ways, but as proficiency develops there is greater convergence toward the types of variability evidenced in native speaker discourse production. In sum, the research examined from a variationist perspective has shown that L2 learners’ language does vary in the same ways but to differing degrees compared to native speaker language.

8.6 Common findings and concluding remarks After reviewing a number of important and influential studies from several theoretical perspectives and discussing at length their findings and contributions to our knowledge of how subject expression is acquired in L2 Spanish, it is now time to take stock of what we know and what we still need to find out. It is apparent that L2 learners begin the acquisition process from the vantage point of their first language. Many of the characteristics of the L2 learner’s system for expressing subjects in Spanish reflect this, whether they be syntactic, semantic, discursive, pragmatic, strategic or some combination of or all of these. Furthermore, when universal constraints and or processing strategies reinforce the L1 traits, these can linger until very advanced stages of learning and perhaps remain permanently. Nonetheless, it is quite remarkable what the L2 learner is capable of acquiring in a relatively short period of time, considering that the majority of the studies examined learners with an average of 4–5 years of foreign language study and many of whom had not ever spent any considerable time abroad in a Spanishspeaking country. Compare this to the child, who is immersed in her language for the entirety of her language acquisition process (also unequivocally a remarkable feat). But the difference is that the child becomes a native speaker of her language

270 | 8 Summary of findings and the adult L2 learner does not. Researchers would like to know why this is so and will continue to search for more satisfactory explanations to this query. It is evident that the L2 learner is capable of learning the more basic, more prominent characteristics or distinctions of subject expression in Spanish in both the syntactic and discourse-pragmatic domains and when interpreting or producing subjects, depends on a less cognitively demanding processing strategy. The research attests to learners’ early acquisition of null and overt referential subject pronouns, of a rich verbal morphology (although with deficits at the early stages), and most possibly a relatively early acquisition of obligatorily null expletive subjects. These findings appear to be uncontroversial. There also seems to be a considerable body of evidence that some learners, at least by the intermediate stage, acquire properties of the OPC, although this is far from a settled manner. Not all learners display knowledge of the OPC, even from the same proficiency levels or after having spent time abroad. If knowledge of the OPC is evidence of access to UG despite little linguistic input to trigger that knowledge (as well as impervious to greater quantities of input), the question remains as to why some adult learners have access to the principles of UG where others do not. Furthermore, researchers have claimed that the OPC is a syntactic property that depends on the discursive constraint of focus for its interpretation. A more logical explanation then, is that some learners’ apparent knowledge of the OPC is tied to their understanding of the discourse properties related to the constraint. Other syntactic properties that appear to be late learned are subject-verb inversion, which is dependent upon an interaction of verbal semantics and the discourse property of focus, and thattrace effects which result from complex syntax like the OPC but which are not dependent upon interfacing with the discourse property of focus. The discourse-pragmatic properties of subject expression in Spanish that seem to be learned at early stages are also the more basic, prominent distinctions among referents in discourse. Referents that are emphatic and contrastive are expressed with overt subjects and those that are highly salient in the discourse are marked with null subjects and these are clear distinctions that learners acquire early. However, referents that are in a context of switch reference are neither contrastive nor salient and thus, learners must look to other clues in order to determine their discourse status and this causes problems for them. In terms of discourse cohesiveness, learners do well distinguishing referents in contingent clauses but when the reference context involves greater clausal distance and more competing referents, the research has shown that responding to this factor in discourse is problematic and only develops over time. Keeping track of referents in longer stretches of discourse is a cognitively demanding task for L2 learners. Finally, it has been demonstrated that adult L2 learners depend on a subject strategy in assigning an antecedent to a pronominal subject in the pro-drop lan-

8.6 Common findings and concluding remarks |

271

guage regardless of the subject’s form (null or overt), and gradually adopt the L2 PAH strategy but only for null subject pronouns and only when the strategy coincides with the pragmatic context and discourse structure. This same strategy is observed in monolingual and bilingual children of pro-drop languages. Thus, it appears that the subject strategy is a cognitively less taxing strategy for children and adult learners alike. When the non-pro-drop L1 reinforces the strategy, it takes a longer time to sort out the pro-drop L2 scheme for anaphoric resolution. Although learners appear to learn the most prominent contrasts first, they also eventually acquire the myriad of more subtle factors that interact in the use and distribution of subject expression in discourse. Research reveals that learners, like native speakers, are constrained by the same linguistic features of subjects, including person, number, tense, aspect, mood, and specificity, and of the lexical properties of their verbs. They are constrained by the same discoursepragmatic features of topic, focus, same and switch reference, cohesiveness, and connectedness, and are subject to implicational hierarchies that predict subject expression based on the perceived saliency and informativeness of different referring expressions. To comprehend the acquisition of subjects in L2 Spanish and the multiple levels of knowledge that are necessary for the learner to integrate, it is necessary to examine all of the factors (although not all at the same time) that contribute to the understanding and use of this complex structure. The common finding among the different theoretical perspectives is that universal syntactic, discourse-pragmatic, and cognitive properties, as well as apparent universal processing strategies, all guide subject expression in Spanish for native speakers and learners alike. Generativists argue that evidence pointing to L2 learners’ knowledge of the null/overt option, rich verbal morphology, obligatory null expletives, and possibly the OPC, lends support to the claim that adult learners have access to the properties of UG, but the lingering deficits in subject pronoun use, even at the advanced level, are due to difficulties in integrating the modular syntactic domain with the more general cognitive domain of discourse pragmatics (Rothman and Pascual y Cabo 2014). Researchers working within a processing model have sought to determine whether L2 learners rely on their L1 strategy, an L2 strategy or a combination of L1, L2 and universal processing strategies to assign the appropriate antecedent of anaphoric subjects. The evidence seems to suggest that both children and adults, when confronted with two systems, one complex and the other less complex, will choose the less cognitively demanding of the two, which in and of itself, is a universal strategy. The discoursepragmatic approach examines the extent to which subject expression depends on the universal properties of sequential organization, coherence and saliency in discourse structure and links these properties to general cognitive abilities. The variationist model demonstrates that speakers are guided by the interaction of

272 | 8 Summary of findings numerous conditioning linguistic and contextual factors that are common to all languages. Most importantly, each of the studies considered here reveal that the information status of a referent is crucial. And this can only be determined in discourse, not in isolated clauses, because language structure is constructed out of discourse and interaction in the real world. This understanding can only cause us to consider that the properties of human cognition that lead to language are far more general than narrow syntactic principles, and acquisition of these properties is not determined by the interaction of L1 knowledge and L2 input alone. It is a complex enterprise that interweaves all of cognition, human interaction and perception to acquire and produce meaningful linguistic structure.

References Abreu, Laurel. 2009. Spanish subject personal pronoun use by monolinguals, bilinguals and second language learners. Gainesville, FL: University of Florida dissertation. Abreu, Laurel. 2012. Subject pronoun expression and priming effects among bilingual speakers of Puerto Rico. In Kimberly Geeslin & Manuel Díaz-Campos (eds.), Selected proceedings of the 14th Hispanic Linguistics Symposium, 1–8. Somerville, MA: Cascadilla Proceedings Project. Al-Kasey, Tamara & Ana Teresa Pérez-Leroux. 1998. Second language acquisition of Spanish null subjects. In Suzanne Flynn, Gita Martohardjono, & Wayne O’Neil (eds.), The generative study of second language acquisition, 161–183. Hillsdale, NJ: Lawrence Erlbaum. Alonso-Ovalle, Luis, Susana Fernández-Solera, Lyn Frazier & Charles Clifton, Jr. 2002. Null vs. overt pronouns and the topic-focus articulation in Spanish. Rivista di Linguistica 14. 1–19. Ariel, Mira. 1990. Accessing noun-phrase antecedents. New York: Routledge. Ariel, Mira. 1994. Interpreting anaphoric expressions: A cognitive versus a pragmatic approach. Journal of Linguistics 30. 1–42. Asher, Nicholas & Laure Vieu. 2005. Subordinating and coordinating discourse relations. Lingua 115. 591–610. Ávila-Shah, Bárbara I. 2000. Discourse connectedness in Caribbean Spanish. In Ana Roca (ed.), Research on Spanish in the United States: Linguistics issues and challenges, 238–251. Somerville: Cascadilla Press. Ayoun, Dalila. 2003. Parameter setting in language acquisition. London & New York: Continuum. Bayley, Robert & Lucinda Pease-Álvarez. 1997. Null pronoun variation in Mexican-descent children’s narrative discourse. Language Variation and Change 9. 349–371. Bayley, Robert J. & Elaine Tarone. 2012. Variationist perspectives. In Susan M. Gass & Alison Mackey (eds.), The Routledge handbook of second language acquisition, 41–56. London: Routledge. Barron, Anne. 2003. Acquisition in interlanguage pragmatics. Amsterdam & Philadelphia: John Benjamins. Bel, Aurora. 2003. The syntax of subjects in the acquisition of Spanish and Catalan. Probus 15. 1–26. Bentivoglio, Paola. 1987. Los sujetos pronominales de primera persona en el habla de Caracas. Caracas: Universidad Central de Venezuela. Bialystok, Ellen. 1990. The competence of processing: Classifying theories of second language acquisition. TESOL Quarterly 24. 635–648. Blackwell, Sarah E. 1998. Constraints on Spanish NP anaphora: The syntactic versus the pragmatic domain. Hispania 81. 606–618. Blackwell, Sarah E. 2003. Implicatures in discourse: The case of Spanish NP anaphora. Amsterdam: John Benjamins. Blackwell, Sarah E. & Margaret Lubbers Quesada. 2012. Third-person subjects in native speakers’ and L2 learners’ narratives: Testing (and revising) the Givenness Hierarchy for Spanish. In Kimberly Geeslin & Manuel Díaz-Campos (eds.), Selected proceedings of the 14th Hispanic Linguistics Symposium, 142–164. Somerville, MA: Cascadilla Proceedings Project.

274 | References Bley-Vroman, Robert. 1990a. What is the logical problem of foreign language learning? In Susan M. Gass & Jacquelyn Schachter (eds.), Linguistic perspectives on second language acquisition, 41–68. Cambridge, New York, Port Chester, Melbourne & Sydney: Cambridge University Press. Bley-Vroman, Robert. 1990b. The logical problem of foreign language learning. Linguistic Analysis 20. 3–49. Bock, J. Kathryn & Zenzi M. Griffin. 2000. The persistence of structural priming: Transient activation or implicit learning? Journal of experimental psychology: General 129(2). 177–192. Burzio, Luigi. 1986. Italian Syntax: A Government-Binding approach. Dordrecht: Reidel. Camacho, José A. 2013. Null subjects. New York & Cambridge: Cambridge University Press. Cameron, Richard. 1993. Ambiguous agreement, functional compensation, and nonspecific tú in the Spanish of San Juan, Puerto Rico, and Madrid, Spain. Language Variation and Change 5. 305–334. Cameron, Richard. 1994. Switch reference, verb class and priming in a variable syntax. In Katherine Beals, Jeannette Denton, Robert Knippen, Lynette Melmar, Hisami Suzuki & Erica Zeinfeld (eds.), Papers from the 30th regional meeting of the Chicago Linguistic Society: Volume 2: The parasession on variation in linguistic theory, 27–45. Chicago: Chicago Linguistic Society. Cameron, Richard. 1995. The scope and limits of switch-reference as a constraint on pronominal subject expression. Hispanic Linguistics 6/7. 1–27. Cameron, Richard & Nydia Flores-Ferrán. 2004. Perseveration of subject expression across regional dialects of Spanish. Spanish in Context 1. 41–65. Carminati, Maria N. 2002. The processing of Italian subject pronouns. Amherst, MA: University of Massachusetts dissertation. Chomsky, Noam. 1975. Reflections on language. New York: Pantheon. Chomsky, Noam. 1980. Rules and representations. New York: Columbia University Press. Chomsky, Noam. 1981. Lectures on government and binding: The Pisa lectures. Dordrecht, The Netherlands: Foris. Chomsky, Noam. 1982. Some concepts and consequences of the Theory of Government and Binding. Cambridge, MA: MIT Press. Chomsky, Noam. 1986. Knowledge of language: Its nature, origin, and use. New York: Praeger. Chomsky, Noam. 1991. Some notes on economy of derivation and representation. In R. Freidin (ed.), Principles and parameters in comparative grammar, 417–54. Cambridge, MA: MIT Press. Chomsky, Noam. 1995. The minimalist program. Cambridge, MA: MIT Press. Chomsky, Noam & Howard Lasnik. 1977. Filters and control. Linguistic Inquiry 8. 425–504. Davidson, Brad. 1996. Pragmatic ‘weight’ and Spanish subject pronouns: The pragmatic and discourse uses of ‘tú’ and ‘yo’ in spoken Madrid Spanish. Journal of Pragmatics 26. 543–565. Dowens, Margaret Gillon & Manuel Carreiras. 2014. Psycholinguistic approaches to second language Spanish. In Kimberly L. Geeslin (ed.), The handbook of Spanish second language acquisition, 64–79. Malden, Oxford & West Sussex: Wiley Blackwell. Ellis, Nick C. 1996a. Sequencing in SLA: Phonological memory, chunking, and points of order. Studies in Second Language Acquisition 18. 91–126. Ellis, Nick C. 1996b. Analyzing language sequence in the sequence of language acquisition. Studies in Second Language Acquisition 18. 361–368. Ellis, Rod. 1985. Sources of variability in interlanguage. Applied Linguistics 6. 118–131.

References |

275

Ellis, Rod. 1988. The effects of linguistic environment on the second language acquisition of grammatical rules. Applied Linguistics 9. 257–274. Ellis, Rod. 1989. Sources of intra-learner variability in language use and their relationship to second language acquisition. In Susan Gass, C. Madden, Dennis Preston & Larry Selinker (eds.), Variation in second language acquisition, Volume II: Psycholinguistics Issue, 22–45. Clevedon: Multilingual Matters. Ellis, Rod. 1990. A response to Gregg. Applied Linguistics 11. 384–391. Enríquez, Emilia V. 1984. El pronombre personal sujetos en la lengua española hablada en Madrid. Madrid: Consejo Superior de Investigaciones Científicas. Instituto Miguel de Cervantes. Erker, Daniel & Gregory Guy. 2012. The role of lexical frequency in syntactic variability: Variable subject personal pronoun expression in Spanish. Language 88(3). 526–557. Escobar, Anna María. 2010. Variación lingüística en español. In José Ignacio Hualde, Antxon Olarrea, Anna María Escobar & Catherine E. Travis (eds.), Introducción a la lingüística hispánica, 391–444. Cambridge: Cambridge University Press. Filiaci, Francesca. 2010. Null and overt subject biases in Spanish and Italian: A cross-linguistic comparison. In Claudia Borgonovo, Manuel Español-Echevarría & Philippe Prévost (eds.), 171–182. Selected proceedings of the 12th Hispanic Linguistics Symposium. Somerville, MA: Cascadilla Proceedings Project. Flores-Ferrán, Nydia. 2002. Subject personal pronouns in Spanish narratives of Puerto Ricans in New York City: A sociolinguistic perspective. Munich: Lincom Europa. Flynn, Suzanne. 1989. The role of the head-initial/head-final parameter in the acquisition of English relative clauses by adult Spanish and Japanese speakers. In Susan M. Gass & Jacquelyn Schachter (eds.), Linguistic perspectives on second language acquisition, 89–108. Cambridge, New York, Port Chester, Melbourne & Sydney: Cambridge University Press. Fox, Barbara. 1987. Anaphora in popular written English narratives. In Russell S. Tomlin (ed.), Coherence and grounding in discourse, 157–174. Philadelphia: John Benjamins. García Salido, Marcos. 2008. Influence of referential accessibility in the usage of personal subject and object pronouns in conversational Spanish. Boletín de Filología 43. 83–108. García Miguel, José María. 2005. Aproximación empírica a la interacción de verbos y esquemas construccionales, ejemplicada con los verbos de percepción. Estudios lingüísticos de la Universidad de Alicante 19. 1–24. Gass, Susan M. & Alison Mackey. 2007. Input, interaction, and output in second language acquisition. In Bill VanPatten & Jessica Williams (eds.), Theories in second language acquisition: An introduction, 175–199. Mahwah, NJ & London: Lawrence Erlbaum. Geeslin, Kimberly L. (ed.). 2014. The handbook of Spanish second language acquisition. Malden, Oxford & West Sussex: Wiley Blackwell. Geeslin, Kimberly L. & Aarnes Gudmestad. 2008. Variable subject expression in secondlanguage Spanish: A comparison of native and non-native speakers. In Melissa Bowles, Rebecca Foote, Silvia Perpiñán & Rakesh Bhatt (eds.), Selected proceedings of the 12th Hispanic Linguistics Symposium, 69–85. Somerville, MA: Cascadilla Proceedings Project. Geeslin, Kimberly L. & Aarnes Gudmestad. 2011. Using sociolinguistic analyses of discourselevel features to expand research on L2 variation in forms of Spanish subject expression. In Luke Plonsky & Maren Schierloh (eds.), Selected proceedings of the 2009 Second Language Research Forum, 16–30. Somerville, MA: Cascadilla Proceedings Project.

276 | References Gili Gaya, Samuel. 1969. Curso superior de sintaxis española. Barcelona: Bilograf. Givón, Talmy. 1983. Topic continuity in discourse: A quantitative cross-language study. Amsterdam: John Benjamins. Gregg, Kevin R. 1990. The variable competence model of second language acquisition, and why it isn’t. Applied Linguistics 11. 364–383. Gudmestad, Aarnes. 2014. Variationist approaches to second language Spanish. In Kimberly L. Geeslin (ed.), The handbook of Spanish second language acquisition, 80–95. Wiley Blackwell. Gudmestad, Aarnes & Kimberly L. Geeslin. 2010. Exploring the roles of redundancy and ambiguity in variable subject expression: A comparison of native and non-native speakers. In Claudia Borgonovo, Manuel Español-Echevarría & Philippe Prévost (eds.), Selected proceedings of the 12th Hispanic Linguistics Symposium, 270–283. Somerville, MA: Cascadilla Proceedings Project. Gudmestad, Aarnes, Leanna House & Kimberly L. Geeslin. 2013. What a Bayesian analysis can do for SLA: New tools for the sociolinguistic study of subject expression in L2 Spanish. Language Learning 63 (3). 371–399. Gundel, Jeanette K., Nancy Hedberg & Ron Zacharski. 1993. Cognitive status and the form of referring expressions in discourse. Language 69. 274–307. Harris, Randy Allen. 1993. The linguistics wars. Oxford & New York: Oxford University Press. Hertel, Tammy Jandrey. 2003. Lexical and discourse factors in the second language acquisition of Spanish word order. Second Language Research 19. 273–304. Hildalgo-Downing, Raquel. 2003. La tematización en el español hablado. Madrid: Gredos. Hochberg, Judith G. 1986. Functional compensation for /s/ deletion in Puerto Rican Spanish. Language 62. 609–621. Hualde, José Ignacio, Antxon Olarrea & Erin O’Rourke (eds.). 2014. The handbook of Hispanic linguistics. Malden, Oxford & West Sussex: Wiley Blackwell. Hudson-D’Zmura, Susan & Michael K. Tanenhaus. 1998. Assigning antecedents to ambiguous pronouns: The role of center of attention as the default assignment. In Marilyn A. Walker, Aravind K. Joshi & Ellen F. Prince (eds.), Centering theory in discourse, 199–226. Oxford & New York: Oxford University Press. Hulstijn, Jan H. 1989. A cognitive view on interlanguage variability. In Miriam R. Eisenstein (ed.), The dynamic interlanguage: Empirical studies in second language variation, 17–31. New York: Plenum Press. Hurtado, Luz Marcela. 2005. Syntactic-semantic conditioning of subject expression in Colombian Spanish. Hispania 88. 335–348. Hyams, Nina. 1986. Language acquisition and the theory of parameters. Dordrecht, The Netherlands: Reidel. Hyams, Nina. 1994. V2, null arguments and COMP projections. In Teun Hoekstra & Bonnie D. Schwartz (eds.), Language acquisition studies in generative grammar, 21–55. Amsterdam: John Benjamins. Hymes, Dell H. 1972. On communicative competence. In J. B. Pride & Janet Holmes (eds.), Sociolinguistics, 269–293. Harmondsworth: Penguin. Isabelli, Casilde A. 2003. The acquisition of the null subject parameter properties in SLA: Some effects of positive evidence in a naturalistic learning context. Hispania 87(1). 150–162. Jaeggli, Osvaldo. 1982. Topics in Romance syntax. Dordrecht: Foris. Jaeggli, Osvaldo & Nina Hyams. 1988. Morphological uniformity and the setting of the null subject parameter. In James Blevins & Juli Carter (eds.), Proceedings of NELS 18, 238–253.

References |

277

Amherst, MA: Graduate Linguistics Student Organization, Department of Linguistics, South College, University of Massachusetts, Amherst. Jaeggli, Osvaldo & Kenneth J. Safir. 1989. The null subject parameter and parametric theory. In Osvaldo Jaeggli & Kenneth J. Safir (eds.), The null subject parameter, 1–44. Dordrecht, The Netherlands: Reidel. Jegerski, Jill, Bill VanPatten & Gregory D. Keating. 2011. Cross-linguistic variation and the acquisition of pronominal reference in L2 Spanish. Second Language Research 27. 481–507. Jordan, Geoff. 2004. Theory construction in second language acquisition. Amsterdam & Philadelphia: John Benjamins. Judy, Tiffany. 2011. L1/L2 parametric directionality matters: More on the null subject parameter in L2 acquisition. In EUROSLA Yearbook, 165–190. Amsterdam: John Benjamins. Kanno, Kazue. 1998. Consistency and variation in second language acquisition. Second Language Research 14. 376–388. Kasper, Gabriele & Kenneth R. Rose. 2002. Pragmatic development in a second language. Oxford: Blackwell. Keating, Gregory D., Bill VanPatten & Jill Jegerski. 2011. Who was walking on the beach? Studies in Second Language Acquisition 33. 193–221. Labov, William. 1969. Contraction, deletion, and inherent variability of the English copula. Language 45. 715–762. Labov, William. 1994. Principles of linguistic change: Internal factors, Vol. 1. Oxford: Blackwell. LaFond, Larry. 2003. Putting the pieces together: Second language learning of null subjects, inversion, and that trace. In Juana M. Liceras, Helmut Zobl & Helen Goodluck (eds.), Proceedings of the 6th Generative Approaches to Second Language Acquisition Conference (GASLA 2002): L2 Links, 168–175. Somerville, MA: Cascadilla Proceedings Project. Lambrecht, Knud. 1994. Information structure and sentence form: Topic, focus, and the mental representations of discourse referents. Cambridge: Cambridge University Press. Lantolf, James P. & Steven L. Thorne. 2007. Sociocultural theory and second language learning. In Bill VanPatten & Jessica Williams (eds.), Theories in second language acquisition: An introduction, 201–224. Mahwah, NJ & London: Lawrence Erlbaum. Levin, Beth & Malka Rappaport-Hovav. 1995. Unaccusativity at the syntax-lexical semantics interface. Cambridge, MA: MIT Press. Levinson, Stephen. 1983. Pragmatics. Cambridge: Cambridge University Press. Levinson, Stephen C. 1987. Minimization and conversational inference. In Jef Verschueren & Marcella Bertuccelli-Papi (eds.), The pragmatic perspective, 61–129. Amsterdam: John Benjamins. Levinson, Stephen. 2000. Presumptive meanings: The theory of generalized conversational implicature. Cambridge, MA: The MIT Press. Liceras, Juana M. 1989. On some properties of the pro-drop parameter: Looking for missing subjects in non-native Spanish. In Susan Gass & Jacqueline Schacter (eds.), Linguistic perspectives on second language acquisition, 109–133. Dordrecht: Foris. Liceras, Juana M. & Lourdes Díaz. 1998. On the nature of the relationship between morphology and syntax: Inflectional typology, f -features and null/overt pronouns in Spanish interlanguage. In Maria-Luise Beck (ed.), Morphology and its interfaces in second language knowledge, 307–338. Amsterdam: John Benjamins. Liceras, Juana M. & Lourdes Díaz. 1999. Topic-drop versus pro-drop: Null subjects and pronominal subjects in the Spanish L2 of Chinese, English, French, German, Japanese and Korean speakers. Second Language Research 15. 1–40.

278 | References Lightfoot, David. 1991. How to set parameters: Arguments for language change. Cambridge, MA: MIT Press. Linford, Bret & Naomi Lapidus Shin. 2013. Lexical frequency effects on L2 Spanish subject pronoun expression. In Jennifer Cabrelli Amaro, Gillian Lord, Ana de Prada Pérez & Jessi Elana Aaron (eds.), Selected proceedings of the 16th Hispanic Linguistics Symposium, 175–189. Somerville, MA: Cascadilla Proceedings Project. Lipski, John. 1996. Patterns of pronominal evolution in Cuban-American bilinguals. In Ana Roca & John B. Jensen (eds.), Spanish in contact, 159–186. Somerville, MA: Cascadilla Press. Lipski, John. M. 2014. Geographical and social varieties of Spanish: An overview. In José Ignacio Hualde, Antxon Olarrea & Erin O’Rourke (eds.), The handbook of Hispanic linguistics, 1–26. Malden, Oxford & West Sussex: Wiley Blackwell. Lipski, John M. (no date). Subject pronoun usage among Spanish dialects. http://www. personal.psu.edu/jml34/pronouns.pdf Long, Michael H. 1981. Input, interaction and second language acquisition. In Harris Winitz (ed.), Native language and foreign language acquisition. Annals of the New York Academy of Sciences Vol. 379. 259–278. López-Ortega, Nuria Rocío. 2002. The development of discourse competence in study abroad learners: A study of subject expression in Spanish as a second language. Ithaca, NY: Cornell University dissertation. Lozano, Cristóbal. 2002a. Knowledge of expletive and pronominal subjects by learners of Spanish. ITL Review of Applied Linguistics 135. 37–60. Lozano, Cristóbal. 2002b. The Interpretation of overt and null pronouns in non-native Spanish. Durham Working Papers in Linguistics 8. 53–66. Lozano, Cristóbal. 2006. The development of the syntax-discourse interface: Greek learners of Spanish. In Vincent Torrens & Linda Escobar (eds.), The acquisition of syntax in Romance languages, 372–399. Amsterdam/Philadelphia: John Benjamins. Lozano, Cristóbal. 2009. Selective deficits at the syntax-pragmatics interface: Evidence from the CEDEL2 Corpus. In Neal Snape, Yan-kit Ingrid Leung & Michael Sharwood Smith (eds.), Representational deficits in second language acquisition, 127–166. Amsterdam: John Benjamins. Lubbers Quesada, Margaret & Sarah E. Blackwell. 2009. The L2 acquisition of null and overt Spanish subject pronouns: A pragmatic approach. In Joseph Collentine, Maryellen Garcia, Barbara Lafford & Francisco Marcos Marín (eds.), Selected proceedings of the 11th Hispanic Linguistics Symposium, 117–130. Somerville, MA: Cascadilla Proceedings Project. Lubbers Quesada, Margaret & Robert Blake. 2003. The UAQ-UCD project for the study of Spanish as a foreign/second language in Mexico and the U.S.: The acquisition of discourse competence, University of California Institute for Mexico and the United States (UCMEXUS) and the Consejo Nacional de Ciencia y Tecnología de México (CONACYT), Grant No. CN0255. Luján, Marta. 1999. Expresión y omisión del pronombre personal. In Ignacio Bosque & Violeta Demonte (eds.), Gramática descriptiva de la lengua española, Vol. I, 1275–1315. Madrid: Espasa Calpe. MacWhinney, Brian. 2000. The CHILDES Project: Tools for Analyzing Talk (3rd Edition). Hillsdale, NJ: Lawrence Erlbaum. Margaza, Panagiota & Aurora Bel. 2006. Null subjects at the syntax-pragmatics interface: Evidence from Spanish interlanguage of Greek speakers. In Mary Grantham O’Brien,

References |

279

Christine Shea & John Archibald (eds.), Proceedings of the 8th Generative Approaches to Second Language Acquisition Conference (GASLA 2006), 88–97. Somerville, MA: Cascadilla Proceedings Project. Mayer, Mercer. 1969. Frog, where are you? New York: Dial Books for Young Readers (a division of Penguin Putnam). Montalbetti, Mario. 1986. How Pro Is It? In Osvaldo Jaeggli & Carmen Silva-Corvalán (eds.), Studies in Romance linguistics, 137–152. Dordrecht: Foris Publications. Montrul, Silvina A. 2004a. The acquisition of Spanish: Morphosyntactic development in monolingual and bilingual L1 acquisition and adult L2 acquisition. Amsterdam & Philadelphia: John Benjamins. Montrul, Silvina. 2004b. Subject and object expression in Spanish heritage speakers: A case of morphosyntactic convergence. Bilingualism: Language and Cognition 7. 125–142. Montrul, Silvina & Celeste Rodríguez Louro. 2006. Beyond the syntax of the null subject parameter: A look at the discourse-pragmatic distribution of null and overt subjects by L2 learners of Spanish. In Vincent Torrens & Linda Escobar (eds.), The acquisition of syntax in Romance languages, 401–418. Amsterdam/Philadelphia: John Benjamins. Morales, Amparo. 1997. La hipótesis functional y la aparición de sujetos no nominal: El español de Puerto Rico. Hispania 80. 153–165. Otheguy, Ricardo & Ana Celica Zentella. 2007. Apuntes preliminares sobre el contacto lingüístico y dialectal en el uso pronominal del español en Nueva York. In Kim Potowski & Richard Cameron (eds.), Spanish in contact: Policy, social and linguistic inquiries, 275–298. Amsterdam: John Benjamins. Paredes Silva, Vera Lúcia. 1993. Subject omission and functional compensation: Evidence from written Brazilian Portuguese. Language Variation and Change 5. 35–49. Paradis, Johanne & Samuel Navarro. 2003. Subject realization and crosslinguistic interference in the bilingual acquisition of Spanish and English: What is the role of the input? Journal of Child Language 30. 371–393. Perales, Susana & Rosario Portillo. 2007. Sobre las propiedades referenciales de los sujetos nulos y pronominales del español oral y escrito. In Enrique Balmaseda Maestu (ed.), Las destrezas orales en la enseñanza del español L2-LE: XVII Congreso Internacional de la Asociación del Español como Lengua Extranjera (ASELE), Vol. 2, 889–900. Logroño, Spain: Universidad de la Rioja. (http://dialnet.unirioja.es/servlet/articulo?codigo=2470113) Pérez-Leroux, Ana Teresa & William R. Glass. 1997. OPC effects on the L2 acquisition of Spanish. In Ana Teresa Pérez-Leroux & William R. Glass (eds.), Contemporary perspectives on the acquisition of Spanish, Vol. 1: Developing grammars, 149–165. Somerville, MA: Cascadilla Press. Pérez-Leroux, Ana Teresa & William R. Glass. 1999. Null anaphora in Spanish second language acquisition: Probabilistic versus generative approaches. Second Language Research 15. 220–249. Perlmutter, David M. 1978. Impersonal passives and the unaccusative hypothesis. In Proceedings of the 4th Annual Meeting of the Berkeley Linguistics Society, 157–189. Berkeley, CA: BLS. Pienemann, Manfred. 1998. Language processing and second language development: Processability theory. Amsterdam: John Benjamins.

280 | References Pienemann, Manfred. 2007. Processability theory. In Bill VanPatten & Jessica Williams (eds.), Theories in second language acquisition: An introduction, 137–154. Mahwah, NJ & London: Lawrence Erlbaum. Pinker, Steven. 1994. The language instinct. New York: William Morrow. Pladevall Ballester, Elisabet. 2010. Child L2 development of syntactic and discourse properties of Spanish subjects. Bilingualism: Language and Cognition 13. 185–216. Preston, Dennis. 1989. Sociolinguistics and second language acquisition. Oxford: Basil Blackwell. Preston, Dennis. 1993. Variation linguistics and SLA. Second Language Research 9. 153–172. Preston, Dennis. 2002. A variationist perspective on SLA: Psycholinguistic concerns. In Robert B. Kaplan (ed.), Oxford handbook of applied linguistics, 141–159. Oxford: Oxford University Press. Ranson, Diana L. 1991. Person marking in the wake of /s/ deletion in Andalusian Spanish. Language Variation and Change 3. 133–152. Real Academia Española. 1983. Esbozo de una nueva gramática de la lengua española. Madrid/Barcelona: Espasa Calpe. Rizzi, Luigi. 1982. Issues in Italian syntax. Dordrecht: Foris. Rizzi, Luigi. 1986. Null objects in Italian and the theory of pro. Linguistics Inquiry 17. 501–57. Rizzi, Luigi. 1994. Early null subjects and root null subjects. In Teun Hoekstra & Bonnie D. Schwartz (eds.), Language acquisition studies in generative grammar, 151–76. Amsterdam: John Benjamins. Rosengren, Per. 1974. Presencia y ausencia de los pronombres personales sujetos en español moderno. Stockholm: Acta Universitatis Gothoburgensis. Rothman, Jason. 2007. Pragmatic solutions for syntactic problems: Understanding some L2 syntactic errors in terms of pragmatic deficits. In Sergio Baauw, Frank A.C. Dirjkoningen & Manuela Pinto (eds.), Romance languages and linguistic theory 2005, 299–320. Amsterdam: John Benjamins. Rothman, Jason. 2009. Pragmatic deficits with syntactic consequences?: L2 pronominal subjects and the syntax–pragmatics interface. Journal of Pragmatics 41. 951–973. Rothman, Jason & Michael Iverson. 2007a. On parameter clustering and resetting the nullsubject parameter in L2 Spanish: Implications and observations. Hispania 90. 328–341. Rothman, Jason & Michael Iverson. 2007b. Input type and parameter resetting: Is naturalistic input necessary? IRAL 45. 285–319. Rothman, Jason & Michael Iverson. 2007c. The syntax of null subjects in L2 Spanish: Comparing two L2 populations under different exposure. RESLA 20. 185–214. Rothman, Jason & Diego Pascual y Cabo. 2014. Generative approaches to Spanish second language acquisition. In Kimberly L. Geeslin (ed.), The handbook of Spanish second language acquisition, 46–63. Malden, Oxford & West Sussex: Wiley Blackwell. Saunders, Joy Kathleen. 1999. Null and overt references in Spanish second language acquisition: A discourse perspective. Austin, TX: University of Texas dissertation. Schachter, Jacquelyn. 1990. Testing a proposed universal. In Susan M. Gass & Jacquelyn Schachter (eds.), Linguistic perspectives on second language acquisition, 73–88. Cambridge, New York, Port Chester, Melbourne & Sydney: Cambridge University Press. Schwartz, Bonnie D. & Rex A. Sprouse. 1996. L2 cognitive states and the full transfer/full access hypothesis. Second Language Research 12. 40–72. Serratrice, Ludovica & Antonella Sorace. 2003. Overt and null subjects in monolingual and bilingual Italian acquisition. In Barbara Beachley, Amanda Brown & Frances Conlin (eds.),

References |

281

Proceedings of the 27th Annual Boston University Conference on Language Development, 739–750. Somerville, MA: Cascadilla Press. Serratrice, Ludovica, Antonella Sorace & Sandra Paoli. 2004. Crosslinguistic influence at the syntax-pragmatics interface: Subjects and objects in English-Italian bilingual and monolingual acquisition. Bilingualism: Language and Cognition 7. 183–205. Silva-Corvalán, Carmen. 1982. Subject expression and placement in Mexican-American Spanish. In Jon Amastae & Lucia Elias-Olivares (eds.), Spanish in the United States: Sociolinguistic aspects, 93–120. Cambridge: Cambridge University Press. Silva-Corvalán, Carmen. 1994. Language contact and change. Oxford: Oxford University Press. Silva-Corvalán, Carmen. 2001. Sociolingüística y pragmática del español. Washington, D.C.: Georgetown University Press. Sorace, Antonella. 2004. Native language attrition and developmental instability at the syntaxdiscourse interface: Data, interpretations and methods. Bilingualism: Language and Cognition 7. 143–145. Sorace, Antonella. 2011. Pinning down the concept of “interface” in bilingualism. Linguistic Approaches to Bilingualism 1. 1–33. Sorace, Antonella & Francesca Filiaci. 2006. Anaphora resolution in near-native speakers of Italian. Second Language Research 22. 339–368. Sorace, Antonella, Ludovica Serratrice, Francesca Filiaci & Michela Baldo. 2009. Discourse conditions on subject pronoun realization: Testing the linguistic intuitions of older bilingual children. Lingua 119. 460–477. Tarone, Elaine. 1982. Systematicity and attention in interlanguage. Language Learning 32. 69–82. Tarone, Elaine. 1983. On the variability of interlanguage systems. Applied Linguistics 4. 142–163. Tarone, Elaine. 1985a. Variability in interlanguage use: A study of style-shifting in morphology and syntax. Language Learning 35. 373–404. Tarone, Elaine. 1985b. The interlanguage continuum. In Barbara Wheatly, Ashley Hastings, Fred Eckman, Lawrence Bell, Gary Krukar & Rita Rutkowski (eds.), Current approaches to second language acquisition: Proceedings of the 1984 University of Wisconsin-Milwaukee Linguistics Symposium, 31–40. Bloomington: Indiana Linguistics Club. Tarone, Elaine. 1988. Variation in interlanguage. London: Edward Arnold. Tarone, Elaine. 1990. On variation in interlanguage: A response to Gregg. Applied Linguistics 11. 392–400. Thomas, Margaret. 2004. Universal grammar in second language acquisition: A history. London & New York: Routledge. Torres Cacoullos, Rena & Catherine E. Travis. 2011. Testing convergence via code-switching: Priming and the structure of variable subject expression. International Journal of Bilingualism 15. 241–267. Travis, Catherine E. 2005. The yo-yo effect: Priming in subject expression in Colombian Spanish. In Randall Scott Gess & Edward J. Rubin (eds.), Theoretical and experimental approaches to Romance linguistics: Selected papers from the 34th Linguistic Symposium on Romance Languages, 329–349. Amsterdam: John Benjamins. Travis, Catherine E. 2007. Genre effects on subject expression in Spanish: Priming in narrative and conversation. Language Variation and Change 19. 101–135.

282 | References Travis, Catherine E. & Rena Torres Cacoullos. 2014. Discourse syntax. In José Ignacio Hualde, Antxon Olarrea & Erin O’Rourke (eds.), The handbook of Hispanic linguistics, 653–672. Malden, Oxford & West Sussex: Wiley Blackwell. Tsimpli, Ianthi-Maria & Anna Roussou. 1991. Parameter-resetting in L2? UCL Working Papers in Linguistics 3. 149–169. Valenzuela, Elena. 2006. L2 end state grammars and incomplete acquisition of Spanish CLLD constructions. In Roumyana Slabakova, Silvina A. Montrul & Philippe Prévost (eds.), Inquiries in linguistic development, 283–304. Amsterdam: John Benjamins. VanPatten, Bill. 2007. Input processing in adult second language acquisition. In Bill VanPatten & Jessica Williams (eds.), Theories in second language acquisition: An introduction, 115–135. Mahwah, NJ & London: Lawrence Erlbaum. VanPatten, Bill & Teresa Cadierno. 1993. Input processing and second language acquisition: A role for instruction. The Modern Language Journal 77. 45–57. VanPatten, Bill & Jill Jegerski. 2010. Second language processing and parsing: The issues. In Bill VanPatten & Jill Jegerski (eds.), Research in second language processing and parsing, 3–23. Amsterdam/Philadelphia: John Benjamins. White, Lydia. 1985. The pro-drop parameter in adult second language acquisition. Language Learning 35. 47–62. White, Lydia. 1986. Implications of parametric variation for adult second language acquisition: An investigation of the pro-drop parameter. In Vivian Cook (ed.), Experimental approaches to second language learning, 55–72. Oxford: Pergamon Institute of English. White, Lydia. 1989. Universal grammar and second language acquisition. Amsterdam & Philadelphia: John Benjamins. Zagona, Karen. 2002. The syntax of Spanish. Cambridge: Cambridge University Press. Zubizarreta, María Luisa. 1998. Prosody, focus, and word order. Cambridge, MA: The MIT Press.

Index Abreu, Laurel 44, 211, 213, 215, 217–218, 220, 222, 224–226, 229–240, 242–244, 268 abstract properties 3, 13, 263 Al-Kasey, Tamara 48, 57–59, 61, 75, 248–249 Alonso-Ovalle, Luis 124–131, 133–134, 143, 148 ambiguity 57, 83, 87, 124, 129, 143, 150, 216, 231, 241–243, 267, 269 – contextual 208, 229–230, 266 – morphological 177, 184, 208, 213, 225, 230 – referent/referential 43, 146, 231 – verbal/of verb forms 43–44, 171, 208, 213–216, 229–231 anaphora – backwards 93–96, 136–138 – intrasentential 29, 122, 124, 139, 144 – resolution 122, 130, 139, 145, 147 anaphoric – expressions 122, 180 – functions/use 115 – hierarchy 13–14, 37, 152, 154–158, 161–168, 203, 263–264 – interpretation 11, 30, 151, 260–261 – resolution 29, 122–139, 143, 148–151, 258–260, 271 animacy 40, 45, 115, 168 antecedent – discourse-prominent 11, 140 – prominence 29, 122, 124 – resolution 12, 120, 127, 130–131, 150 Ariel, Mira 11, 14, 37, 124, 192, 203, 262 Asher, Nicholas 144 attention state 14, 18, 37, 153, 192, 203 – attrition see language loss Ávila-Shah, Bárbara I. 206, 218, 220, Ayoun, Dalila 7, 23 Baldo, Michela 10, 77–78, 81, 138, 253 Barron, Anne 12 Bayesian multinomial probit model 206, 227, 231, 235, 237, 241 Bayley, Robert 16, 206, 208, 216, 218, 235, 240

Bel, Aurora 78, 93, 100–102, 106–107, 253 Bentivoglio, Paola 43, 206, 214 Bialystok, Ellen 15 bilingual – acquisition 5, 77–79, 135–139 – adults 254 – children 78–84, 86, 117–118, 121, 133 biological endowment 3 Blackwell, Sarah E. 13, 37–39, 86, 152–153, 180, 182–187, 189–194, 196, 198–205, 215, 232–233, 263–265 Blake, Robert 187 Bock, J. Kathryn 217 bound variable interpretation 32, 62, 68, 110, 124–125 Brazilian Portuguese 218 Bley-Vroman, Robert 5 Burzio, Luigi 89 Cadierno, Teresa 17 Camacho, José A. 22, 25, 31 Cameron, Richard 168, 180, 216–217, 234, 237 Cape York Creole 24 Carminati, Maria N. 29–30, 120–124, 128, 130–134, 139, 142–144, 257 Carreiras, Manuel 10 child/L1 language acquisition see primary language acquisition Chinese 22, 28, 56–57, 61, 196 Chomsky, Noam 3–4, 6, 18, 22–23, 26, 47, 55 classroom setting/context 47, 49, 69–70, 73–74, 101, 111, 114 Clifton, Charles Jr. 124 cluster/clustering of properties 7–8, 18, 23–24, 27, 30, 47–49, 51, 53, 57, 59, 70, 73–75, 247–250 cognition 3, 6, 9, 35–36, 45, 232, 251, 272 cognitive – conditions 20, – domain 3, 11–12, 120, 135, 256, 258–259, 271 – focus 13, 152

284 | Index – processing 11, 256 – properties 42, 271 – state 2, 14, 205, 207, 262, 264, 266 – status 13–14, 17, 36–37, 152–153, 178, 192–202, 262 coherence 13–14, 152–153, 205, 261–262, 264, 271 cohesion 13–14, 36, 45, 153–155, 157, 159, 161–164, 168–169, 176, 262, 264 see also cohesiveness cohesion theory 154–167, 204 cohesiveness 156, 209, 218–222, 228, 237–242, 244–245, 268, 270–271 see also cohesion competence – pragmatic or communicative 12–13, 15–16, 261 – linguistic 15–16, 205, 261 – syntactic 96, 107, 205 – variable 207 conditioning factors 222–241, 272 – external/extralinguistic 1–2, 15, 17, 196, 265–266 – internal/linguistic 15, 207, 208–209, 222, 244–245, 268 – social/sociolinguistic 14–16, 206–209, 265 constraints – interpretative 27–30, 48 – discourse/discourse-pragmatic 10, 13–14, 23, 66–67, 74, 78, 96–99, 105, 108, 111, 204–205, 254–257 – pragmatic 64, 79, 81, 83–84, 152, 178–192, 261–263, – semantic/lexical 20, 23, 25, 67, 255, 257 – syntactic 78, 90, 108 – universal 96, 104–105, 269 contrast 44, 152, 168–178, 204 – emphatic 38–39, 85, 154, 171, 173, 177–178, 264, 270 contrastive focus 13, 28, 34, 36, 38, 42, 48, 61–62, 67–68, 71–72, 85, 105, 108, 110–113, 127, 130, 182–184, 188–192, 205, 263 see also focus – contrastive focus constraint (CFC) 67–68 coordinating discourse structure see discourse coordination

cross-linguistic – effects 81 – influence 79–88, 117, 135, 138, 144 – interference 78 – pragmatic differences 133–134, 144 Danish 54 Davidson, Brad 38–39, 168, 171, 184–185 deficits – pragmatic 9–10, 114, 252, 254–255, 257 – selective 117 deictic – functions/uses 40, 177, 212 – pronouns 38, 115, 175, 212 Díaz, Lourdes 53–57, 75 directionality 84, 118 see also unidirectional or unidirectionality disambiguation 13, 38, 152, 168–177, 265 discourse – connectedness 208–209, 218–219, 221–222, 238–241, 244–245, 266, 268–269 – coordination 122, 144–147, 259 – markers 13, 176 – subordination 122, 144–147, 149, 259 disjoint interpretation 110, 125 domain – cognitive see cognitive domain – linguistic 9, 251 – pragmatic 100, 102, 270 – syntactic 9, 84, 86, 100, 102, 251, 271 Dowens, Margaret Gillon 10 Ellis, Nick C. 64 Ellis, Rod 15, 207, 275 end result 5, 9, 10, 77, 251–252, 255–257, 260 English 7–8, 22, 24, 26–29, 31–33, 40, 48, 67–68, 135–146, 156, 163, 169, 193, 196, Enríquez, Emilia V. 38–39, 42, 168, 232 epistemic parenthetical 185–186, 188–189, 191–192 Erker, Daniel 227 Escobar, Anna María 42 expletive – pronouns 59–61

Index |

– subjects 7, 24–25, 31–32, 43–50, 52–53, 57–61, 69–71, 73–77, 108–111 ExpS see expletive pronouns Extended Projection Principle (EPP) 23, 31–32, 91–93, 96, 111 Feature Geometry Analysis 115 feature(s) – agreement 56, 93, 96 – checking 90, 108 – interpretable 91, 103, 120, 163 – specification 4, 6, 67 – [+/−strong] 56–57, 93, 96, – uninterpretable 91 Fernández-Solera, Susana 124–128 Filiaci, Francesca 9–10, 81, 100, 120, 131–138, 142–143, 148, 163, 253 final steady state see end state Finnish 24 Flores-Ferrán, Nydia 206, 217, 240 Flynn, Suzanne 6, 8 focus 13–14, 26, 32–36, 61–68, 79–81, 88–93, 96–99, 102, 118–119, 126, 152–153, 178–179 – broad 126, 128, – cognitive 13, 152 – narrow 125–126, 128 – neutral 25, 89–90, 94, 96–98, 102–106 – non-contrastive 36, 105, 108, 112 – presentational 94–95 see also contrastive focus; switch focus of attention formal-generative approach see generative approach or model Fox, Barbara 168 Frazier, Lyn 124–128 French 24, 49, 51, 54, 56–57, 248 Full Transfer/Full Access Model 5, 53–54, 100 functional – categories 4, 30–31, 55–56, 69 – features 9, 251 Functional Compensation Hypothesis (FCH) 43–44 functional-discourse perspective 2, 8, 168, 206, 246 Fundamental Difference Hypothesis 5

285

García Miguel, José María 39 García Salido, Marcos 206, 214, 229 Gass, Susan M. 17 Geeslin, Kimberly L. 2, 44, 206, 208–209, 211, 213, 215–217, 222, 224–228, 230–231, 234–235, 237–238, 240–242, 244 generative – approach or model 2–10, 17, 30, 47, 69, 77, 100, 246–256 – grammar see Universal Grammar (UG) German 56–57 Gili Gaya, Samuel 169 Givenness Hierarchy 14, 37, 153, 192, 192–203 Givón, Talmy 14, 37, 168, 192, 203, 262 Glass, William R. 48, 62–68, 70, 73, 75–77, 248, 250 Greek 67–68, 75, 87, 100–105, 248, 254 Gregg, Kevin R. 207 Griffin, Zenzi M. 217 Gudmestad, Aarnes 44, 206–208, 211, 213, 215–217, 222, 224–228, 230–231, 234–235, 237–238, 240–242, 244 Gundel, Jeanette K. 14, 36–37, 86, 152–153, 178, 192–196, 202–203, 263–264 Guy, Gregory 227

Harris, Randy Allen 2 Hedberg, Nancy 14, 36–37, 86, 152–153, 178, 192–196, 202–203, 263–264 heritage speakers 84–87, 118, 121, 139–143, 148–149, 224, 253, 259–260 Hertel, Tammy Jandrey 96–97, 106, Hildalgo-Downing, Raquel 182 Hochberg, Judith G. 43, 171, 214, 229 House, Leanna 206, 208, 213, 215, 217, 222, 227–228, 231, 235, 237, 241, 244 Hualde, José Ignacio 2 Hudson-D’Zmura, Susan 11, 122 Hulstijn, Jan H. 15 Hurtado, Luz Marcela 43, 206, 214, 232 Hyams, Nina 7–8, 55, 69 hybrid strategy 121, 151, 259–260 Hymes, Dell H. 12

286 | Index implicational hierarchy 49 indeterminateness see underdetermined input or poverty of the stimulus information structure 14, 18, 91, 97, 117, 119, 151–152, 261–263, 266 informative status 49, 255 informativeness 263, 271 – features 80 – of referents 14, 37, 153, 192, 203, 262–264 initial state 4–6, 8, 47, 54, 69, 247–248, 250 innate – capacity 5–6 – knowledge 4 – linguistic faculty 4 – principles 4, 260 – restrictions 10, 12, 121, 256, 258, 260–261 input 7, 61, 114, 120, 272 – linguistic 2, 4–5, 51 – L2, 64 – environmental 4 – naturalistic 48, 53, 69–74, 77–78, 143, 249 – in processing 11–12, 258–261 – quantity and quality of 138–139 – the role of 8, 47, 49, 51, 75–76, 247, 249–251 – underdetermined 5, 47, 65, 75, 111, 247 input processing model 17 interaction hypothesis 17 interface – hypothesis 10, 120, 163 – lexico-syntactic 26, 87, 93, 96, 105 – model 9–10, 117–119, 251–257 – syntactic/syntax-pragmatic(s)/discourse 1, 10, 26, 35, 76, 77–119, 135, 137, 139, 143, 258–259, 261 see also syntax-pragmatics interface model interlanguage 5–6, 18, 53–54, 120 – errors 18, 77 – variation in 15, 207 inversion see subject-verb inversion Isabelli, Casilde A. 25, 27, 48, 51–53, 69–70, 75, 248 Italian 54, 77, 117, – as a pro-drop/null subject language 9, 11, 22–23, 29, 87–88, 117,

– and the PAH 29, 121–122, 124, 128, 130–135, 138–140, 148–150, 258–259 – and the unidirectional hypothesis 79–84, Iverson, Michael 48, 53, 69–77, 108, 111, 114, 248–249 Jaeggli, Osvaldo 18, 22–23, 47, 69 Japanese 54–57, 72, 108, 196 Jegerski, Jill 11–12, 120–121, 139–147, 257, 259–260 Jordan, Geoff 207 Judy, Tiffany 48 Kanno, Kazue 70, 72, 108 Kasper, Gabriele 12 Keating, Gregory D. 11, 121, 139–147, 259–260 L1 – role of 49, 78–79 – transfer 64, 69, 79, 95, 97, 102, 104, 106, 163, 165, 167, 177 Labov, William 15, 206, 217 Lambrecht, Knud 182 language loss 84, 86, 118, 135 Lantolf, James P. 17 Lasnik, Howard 26, learnability problem 4, 47, 111, 247 Levin, Beth 89 Levinson, Stephen 13–14, 37, 153, 180, 192, 205, 262, 265 lexical – category/class/verb class/type 31, 78, 89, 91, 93, 96, 98, 100, 253 – noun phrase 20, 36–37, 40, 168, 174–175, 209, 221–224, 234–237, 241, 244, 251 – principles/properties 78, 87, 105, 106–107, 271 – semantics 16, – subjects 21, 55, 79, 99, 169, 171–173, 204, 237, 255 lexico-semantic properties 87, 117, 255 lexico-syntactic – factors – interface 26, 87, 96, – knowledge 78 – properties 79, 255

Index |

Lightfoot, David 55 Linford, Bret 208, 213, 215, 217, 222, 226–227, 231–235, 243, 268 Lipski, John 42–43, 84 logical problem see learnability problem Long, Michael H. 17 López-Ortega, Nuria Rocío 13, 38, 152, 154, 168–178, 203–204, 215, 232, 263–264 Lozano, Cristóbal 24, 32, 48, 59–61, 67–69, 73, 75, 91–92, 102–105, 107, 114–117, 242, 248, 254–255 Lubbers Quesada, Margaret 13, 39, 152–153, 178–180, 182–187, 189–192, 196, 198–205, 215, 232–233, 263–264 Luján, Marta 183 Mackey, Alison 17 MacWhinney, Brian 80, 129 Margaza, Panagiota 77, 100–102, 107 markedness theory 7–8 Mayer, Mercer 130 mental representation 59, 61, 75, 120, 249 Minimalist model or program 4, 22, 31, 108 see also Universal Grammar (UG) monolingual – acquisition 5 – adults 81, 118, 120–121, 139–144 – children 79, 81, 83, 93–94, 117–118, 120–121, 138, 150, 254, 260, 271 – speakers 11, 83–85, 133–137, 148–150, 208, 224–226, 229–232, 238–239, 242–244 Montalbetti, Mario 27, 48, 61, 78, 107, 247 Montrul, Silvina A. 5, 21–22, 31–32, 34–35, 42, 84–88, 98, 100, 106, 116, 118, 140 Morales, Amparo 42 morphological typology 49, 53–55 morphosyntactic properties 10, 78, 99–100, 117, 150, 254 naturalistic setting see study abroad context/setting Navarro, Samuel 79 non-focus 26, 32, 34, 36, 64–65, 77, 80, 88, 102, 251, 255 see also same referent/reference; topic continuity

287

non-pro-drop – anaphoric resolution in 135, 138, 145, 271 – language 5, 9, 11, 28–32, 61, 67, 77–80, 87–88, 118–122, 134, 248, 252, 254–257 – setting 49 non-uniformity of L2 acquisition 255–256 null subject language see pro-drop language Null Subject Parameter (NSP) 7, 22–27, 29–32, 47–53, 57–58, 61, 69–70, 73–78, 111, 113–114, 117, 120, 163

O’Rourke, Erin 2 Olarrea, Antxon 2 optionality 9, 79, 96, 104, 106, 120, 139, 144, 251, 260 Otheguy, Ricardo 206 Overt Pronoun Constraint (OPC) 27–30, 32, 45, 48–49, 61–78, 107–114, 247

Paoli, Sandra 10, 77–80, 86, 111, 253 Paradis, Johanne 79 parameter – (re)setting 5, 8, 18, 47–49, 51–53, 55–57, 59–60, 69–70, 74–75, 114, 119, 135, 247–249 see also triggering of parameters – unmarked option/setting/value 8, 49, 51, 260 parametric – approach or model 6–8, 47–48, 69, 74–76 – variation 30 Paredes Silva, Vera Lúcia 218–219 Partial Access Model see Fundamental Difference Hypothesis Pascual y Cabo, Diego 4, 9, 251, 271 Pease-Álvarez, Lucinda 206, 216, 218, 235, 240 Perales, Susana 124, 128–131, 133–134, 143, 148 Pérez-Leroux, Ana Teresa 48, 62–68, 70, 73, 75–77, 248, 250 Perlmutter, David M. 26, 89 perseveration or priming 208–209, 217, 222, 228, 235–238, 241, 244–245, 266, 268–269

288 | Index person 38–40, 114–117 – agreement 21 – role of 1 Pienemann, Manfred 17 Pinker, Steven 55 Pladevall Ballester, Elisabet 77, 93–96, 106, 118, 253 polarity 208–209 Portillo, Rosario 124, 128–131, 133–134, 143, 148 Position of Antecedent Hypothesis (PAH) 29–32, 45, 120–134, 136–137, 139–140, 144, 148–151 poverty of stimulus 7, 65, 70, 74 – hypothesis 4, 47, 61 pragmatic weight 13, 38–40, 152, 178, 184–185, 189–192, 205, 262–263 Preston, Dennis 15, 207–208 presupposition 13–14, 17, 35–36, 45, 152, 182, 203, 206, 262–264, 266 primary language acquisition 4–5, 27, 93, 269 principles – abstract 4, 6–7 – language specific 262, 264, 266 principles and parameters model see parametric approach or model pro-drop – anaphoric resolution in 122–135, 139, 142–145, 150, 259–260 – language 9, 11, 20, 22–32, 44, 48, 56, 61, 67, 79–81, 87–88, 117, 121–122, 247–248 – parameter 7–9, 49, 51, 55, 76, 93, 142, 150, 153 see also null subject parameter probabilistic weight 206, 208, 266

quantifier – determiner phrase (QDP)/quantifier subject 21, 70, 74, 107, 110, 113, 125–128 – raising 27 Quechua 24 raising – subject 91 – verb 58, 96 see also quantifier raising Ranson, Diana L. 38, 43, 207, 214 Rappaport-Hovav, Malka 89 re-setting of parameters see triggering of parameters Real Academia Española 169 redundancy 83, 87, 143, 150, 173, 241 referent – familiarity 13, 37, 152, 166, 204, 262–263 – role 13, 37, 152, 158, 160–161, 166–169, 203–204, 262–264 – saliency/salient 13–14, 36–37, 152–153, 178, 180, 187, 189–192, 203, 205 referential – pronouns 59–61, 75, 248, 250 – subjects 20, 23–24, 32, 44, 47, 60, 71, 73, 99, 111, 128, 192, 246, 248–249 regional variation see varieties of Spanish residual optionality 9, 106, 251 rich verbal morphology 7, 22, 150, 270–271 Rizzi, Luigi 23–24, 26, 47–48, 55, 75, 246–248 Rodríguez Louro, Celeste 34, 77, 98, 100, 106, 116, 140 Rose, Kenneth R. 12 Rosengren, Per 38 Rothman, Jason 4, 9, 31, 48, 53, 69–77, 108–114, 116, 248–249, 251, 255 Roussou, Anna 8, 30–31, 55

probability 208, 266 processability theory 17 processing models/studies 10–12, 17, 120–122, 256–261, 271 projection problem 4 see also learnability problem ProS see referential pronouns prosodic stress 34–35, 64, 79, 89, 255

/s/-weakening 43 Safir, Kenneth J. 18, 22–23 same reference/referent 32–33, 36, 45, 85, 106–107, 116, 143, 156, 185, 195, 205, 208–209, 215–219, 234–236, 239, 243–245 see also non-focus; topic continuity

Index |

Saunders, Joy Kathleen 13, 37, 152–168, 203–204, 263–264 Schachter, Jacquelyn 5 Schwartz, Bonnie D. 100 Segmented Discourse Representation Theory (SDRT) 144–145 semantics – lexical 16, 267–269 – verbal 38–41, 268, 270 sequential organization 13, 205, 261, 271 Serratrice, Ludovica 10, 77–81, 86, 88, 111, 138, 253 Shin, Naomi Lapidus 208, 213, 215, 217, 222, 226–227, 231–235, 243, 268 Silva-Corvalán, Carmen 43, 84, 140, 207, 214–216, 229, 231, 243 sociocultural theory 17 sociolinguistic model see variationist approach or model Sorace, Antonella 9–10, 77–82, 86, 88, 100, 111, 114, 117, 119- 120, 135–138, 142–143, 149, 163, 253, 260 speaker – attitude 13, 17, 45 – commitment 13, 38, 152–153, 168, 184, 204, 263, 265 – evaluation 13, 152, 174, 186, 265 – identity 13, 17, 152, 171, 175 – intentions 2, 13, 17, 206, 265 – presupposition 14, 17, 36, 155, 203, 206, 262, 264, 266 specification of features see feature specification specificity 40, 45, 168, 208–209, 211, 213, 222–228, 235, 242, 245, 266–267, 269, 271 speech acts 12–13 Sprouse, Rex A. 5, 100 strategies – competing 140, 149, 258 see also hybrid strategy – interpretation 11–12, 18, 120, 144 study abroad context/setting 48, 51, 69–70, 249 subject assignment strategy 29, 32, 121, 139–140, 257–258

289

subject-verb inversion 7, 25–27, 32, 47–53, 69–70, 73–75, 98, 101–102, 120, 150, 246, 248–250 subjects – postverbal 25, 39, 79, 86, 91, 93–96, 99, 126, 128, 251, 255 – preverbal 25, 29, 39, 86, 93–94, 99, 121, 126, 149, 251 – prominence 29, 130 – thematic 20, 57–59 see also referential subjects subordinating discourse structure see discourse subordination subsequent vs. simultaneous acquisition 9–10, 105, 107, 254 Swedish 54 switch focus of attention 180–182, 188–192, 205, 263 switch reference/of referent 32–33, 36, 45, 87, 99, 106, 113, 116, 133, 143, 168, 170, 173, 180, 182, 209, 216–219, 228, 230–231, 234–235, 238–239, 241, 243–245 see also focus; topic shift syntax-pragmatics interface model 9–10, 117–119, 251–258

Tanenhaus, Michael K. 11, 122 Tarone, Elaine 15–16, 206–207 tense, aspect, mood (TAM) 228–232, 239, 242–245, 267, 269, 271 that-trace 49, 70–71, 249–250 – effects 47, 50–53, 69, 73, 75–76, 246, 248 – filter 7, 26–27, 32, 48, 246, 248 Thomas, Margaret 4 Thorne, Steven L. 17 topic – continuity 32–36, 77, 80, 116, 138, 144, 151, 251, 255, 266 see also non-focus see also same referent/reference – involvement 13, 17, 152 – new 170–171 – shift 32–36, 55, 79–85, 100, 116–117, 120, 138–139, 143–144, 150–151, 163, 251, 255

290 | Index see also focus; switch reference/ of referent Torres Cacoullos, Rena 39, 43, 207, 214–215, 217, 232–233 transitivity and word order 39, 251 Travis, Catherine E. 38, 39, 43, 207, 214–215, 217, 232–233 triggering of parameters 8, 47, 51, 55–56, 69, 75, 247, 249 see also parameter (re)setting Tsimpli, Ianthi-Maria 8, 30, 55

Unaccusative Hypothesis 26, 87–89, 102–107 underspecification of – interpretable features 100, 120, 163 – knowledge representations 135 unidirectional or unidirectionality 81, 84–87, 117–118 uniform attainment/non-uniform attainment 9–10, 77, 251, 255–256 universal – cognitive abilities 14, 262, 264–266 – syntactic principles 20 – pragmatic principles 32, 124 Universal Grammar (UG) 3–6, 10–12, 47, 115, 121, 163, 246 – learners’ access to 61, 248

Valenzuela, Elena 78 VanPatten, Bill 11–12, 17, 120–121, 139–147, 257, 259–260 variable – competence model 207 – expression 107 see also quantifier determiner phrase (QDP)/quantifier subject; wh-matrix – structure 206, 208

variationist approach or model 1–2, 14–17, 206–209, 215–218, 242–245, 265–269, 271 varieties of Spanish 18, 42–44 – Andalusian 43 – Caribbean 21, 42–45, 65, 218, – Dominican 42–43 – non-Caribbean 42 – Panamanian 42 – Puerto Rican 44, 168, 218, 224 – Venezuelan 42 verbs – of mental activity 209, 215, 233 – of opinion 39 – psychological 39 – of perception 39, 178, 185, 188, 232 – transitive 26, 50, 88–89, 93–95 – unaccusative see word order and unaccusative verbs – unergative see word order and unergative verbs see also lexical, category/class/ verb class/type Vieu, Laure 144 wh– matrix 70, 73–74, 107, 110, 113 – subject extraction 7, 26–27, 71, 73, 246 White, Lydia 5, 163 word order and – unaccusative verbs 25–26, 39, 86, 89–99, 102–107, 251 – unergative verbs 25–26, 39, 89–91, 93–98, 102–107, 251 Zacharski, Ron 14, 36–37, 86, 152–153, 178, 192–196, 202–203, 263–264 Zagona, Karen 90 Zentella, Ana Celica 206 Zubizarreta, María Luisa 89, 182, 184