<%BANNER%>

Sources of Performance and Processing Errors in Near-Native L2 Spanish Speakers, L1 Farsi

MISSING IMAGE

Material Information

Title:
Sources of Performance and Processing Errors in Near-Native L2 Spanish Speakers, L1 Farsi
Physical Description:
1 online resource (312 p.)
Language:
english
Creator:
Judy, Tiffany E
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Romance Languages, Spanish and Portuguese Studies
Committee Chair:
Rothman, Jason Leonardo
Committee Members:
Lord-Ward, Gillian E
Wulff, Stefanie
Cowles, Heidi Wind
De Prada Perez, Ana
Haddad, Youssef A

Subjects

Subjects / Keywords:
farsi -- interface -- l2a -- processing -- representation -- spanish
Spanish and Portuguese Studies -- Dissertations, Academic -- UF
Genre:
Romance Languages thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
While normal child language acquisition results in complete productive and comprehension abilities at a relatively young age, adult language acquisition is more belabored and often results in linguistic abilities that differ from those of native speakers in terms of both productive and comprehension abilities. A major line of research in language acquisition studies seeks to determine and explain potential causes of these differences. One current and influential hypothesis, the Interface Hypothesis (Sorace 2011), claims that dissimilarities are a product of the increased cognitive load experienced by bilinguals. That is, since bilinguals possess two grammars, they must simultaneously inhibit one language while activating the other. As a consequence of the increased cognitive load, the Interface Hypothesis predicts that bilinguals will show differences on properties requiring integration of purely linguistic information (e.g. syntax) with extralinguistic information (e.g. discourse or pragmatics). I test this prediction in near-native adult second language acquirers of Spanish who are native speakers of Farsi using offline tasks measuring grammatical accuracy and interpretation of referential and expletive subject pronouns as well as online tasks measuring language processing. The particular participant group, linguistic properties and combination of offline and online experimental methodology are especially relevant to the debate surrounding near-native speaker ultimate attainment potential and bilingual processing. That is, the experimental group participants meet the requirements of the Interface Hypothesis in that they are highly fluent second language speakers that have lived in a Spanish-speaking environment for over 30 years. While the language pairing is quite novel, the linguistic property examined is interesting in that Spanish and Farsi are both null subject languages with analogous discourse distributions of subject pronouns. However, the underlying syntax of each language differs therefore necessitating restructuring in order for the speakers to converge on the syntax of subjects in Spanish. Finally, employing both offline and online measures makes it possible to test the Interface Hypothesis’s claim that bilingual processing is different from monolingual processing. The results, which provide mixed evidence regarding the claims of the Interface Hypothesis in that differences were found, but they were not isolated to overt subject pronouns, inform the field of linguistics regarding the human capacity for language acquisition, but also have significant implications for psycholinguistics, cognitive science and language teaching.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by Tiffany E Judy.
Thesis:
Thesis (Ph.D.)--University of Florida, 2013.
Local:
Adviser: Rothman, Jason Leonardo.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2013
System ID:
UFE0045217:00001

MISSING IMAGE

Material Information

Title:
Sources of Performance and Processing Errors in Near-Native L2 Spanish Speakers, L1 Farsi
Physical Description:
1 online resource (312 p.)
Language:
english
Creator:
Judy, Tiffany E
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Romance Languages, Spanish and Portuguese Studies
Committee Chair:
Rothman, Jason Leonardo
Committee Members:
Lord-Ward, Gillian E
Wulff, Stefanie
Cowles, Heidi Wind
De Prada Perez, Ana
Haddad, Youssef A

Subjects

Subjects / Keywords:
farsi -- interface -- l2a -- processing -- representation -- spanish
Spanish and Portuguese Studies -- Dissertations, Academic -- UF
Genre:
Romance Languages thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
While normal child language acquisition results in complete productive and comprehension abilities at a relatively young age, adult language acquisition is more belabored and often results in linguistic abilities that differ from those of native speakers in terms of both productive and comprehension abilities. A major line of research in language acquisition studies seeks to determine and explain potential causes of these differences. One current and influential hypothesis, the Interface Hypothesis (Sorace 2011), claims that dissimilarities are a product of the increased cognitive load experienced by bilinguals. That is, since bilinguals possess two grammars, they must simultaneously inhibit one language while activating the other. As a consequence of the increased cognitive load, the Interface Hypothesis predicts that bilinguals will show differences on properties requiring integration of purely linguistic information (e.g. syntax) with extralinguistic information (e.g. discourse or pragmatics). I test this prediction in near-native adult second language acquirers of Spanish who are native speakers of Farsi using offline tasks measuring grammatical accuracy and interpretation of referential and expletive subject pronouns as well as online tasks measuring language processing. The particular participant group, linguistic properties and combination of offline and online experimental methodology are especially relevant to the debate surrounding near-native speaker ultimate attainment potential and bilingual processing. That is, the experimental group participants meet the requirements of the Interface Hypothesis in that they are highly fluent second language speakers that have lived in a Spanish-speaking environment for over 30 years. While the language pairing is quite novel, the linguistic property examined is interesting in that Spanish and Farsi are both null subject languages with analogous discourse distributions of subject pronouns. However, the underlying syntax of each language differs therefore necessitating restructuring in order for the speakers to converge on the syntax of subjects in Spanish. Finally, employing both offline and online measures makes it possible to test the Interface Hypothesis’s claim that bilingual processing is different from monolingual processing. The results, which provide mixed evidence regarding the claims of the Interface Hypothesis in that differences were found, but they were not isolated to overt subject pronouns, inform the field of linguistics regarding the human capacity for language acquisition, but also have significant implications for psycholinguistics, cognitive science and language teaching.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by Tiffany E Judy.
Thesis:
Thesis (Ph.D.)--University of Florida, 2013.
Local:
Adviser: Rothman, Jason Leonardo.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2013
System ID:
UFE0045217:00001


This item has the following downloads:


Full Text

PAGE 1

1 SOURCES OF PERFORMAN CE AND PROCESSING ER RORS IN NEAR NATIVE L2 SPANISH SPEAKERS, L1 FARSI By TIFFANY JUDY A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREM ENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2013

PAGE 2

2 2013 Tiffany Judy

PAGE 3

3 To my family, friends and academic mentors that have supported me during my graduate studies

PAGE 4

4 ACKNOWLEDGEMENTS I thank my dear parents and extended family, who from childhood set the foundation for the pursuit of fulfillment and continuous betterment, have encouraged me to explore my academic potential and be curious and have supported me every step of this journe y. Sacrifices and gestures of kindness made on their part include linguistic judgments, participation in experiments, visits to see me when I could not get away, travel to see conference presentations and graduation celebrations, acceptance of my absence a t various family functions due to research related travels around the world, and, also the countless cards, care packages, messages and phone calls to brighten my spirit. They continue to amaze me with generosity and love! I thank my friends for accepting my passion for scholarly pursuit and for understanding that my academic commitments often require much social sacrifice. Brenda, Lindsay, Monica, Emily and Chelsea, I cherish the times we spend together, although they are too few and far between, and I tha nk you for giving me a sense of normalcy and always bringing me back to my Midwest roots. You women are admirable in your own unique ways and keep me grounded. I thank my participants from the Universidad de Buenos Aires, Facultad de Filosofa y Letras (Se de Puan ) and dear friends in Buenos Aires Argentina, without whom I could not have completed this Yazdani, and Gugliotti families, for their unending assista nce with respect to this project as well as their spiritual guidance and friendship. The many academic challenges faced throughout graduate school were weathered with the support and friendship of a wonderful cohort of colleagues and friends from the Unive rsity of Iowa and the University of Florida. It has been a privilege to learn from and study alongside each of you, and I share my successes with you as graduate school is such a combined effort. Among these, special thanks are extended to Mike Iverson, a classmate, roommate, colleague, friend, traveling research partner and soccer buddy

PAGE 5

5 present and encouraging since day one; Jeff Renaud, a quietly brilliant and sincere classmate that always raised the expectations; Anne Lingwall, a steadfast, multifaceted and wise Iowan woman that helped me ease into Florida life; Aazamosadat Feizmohammadpour, an astutely perceptive and generous friend and colleague that knows just the right words to offer; my dear monos, a wildly intelligent troop that filled my Iowa days with novelty, inspiration and playfulness; and Fede, whose dignity during his dissertation year set the bar for mine and whose unwavering support during mine provided much respite. And finally, I thank the phenomenal group of professors in the Spanish and Portuguese and Linguistics departments at the University of Iowa and in the Spanish and Portuguese Studies and Linguistics departments at the University of Florida, from whom I have had the pleasure to learn for investing in my potential for providing su ch wonderful training and for imparting their wisdom with the next generation of linguists. You are truly changing the world! At the University of Iowa, Dr. Roumyana Slabakova and Dr. Paula Kempchinsky played particularly significant roles in my developmen t as a linguist and acquisitionist. Roumyana and Paula afforded me much patience and understanding, especially in the early years of my studies. Therefore, I am forever grateful to both women, who are fiercely intelligent, sharp witted and precise, for the ir commitment to me and for the admirable standards that they set as women and as researchers. Also at the University of Iowa, I began my tutelage under Dr. Jason Rothman, first as a MA student and then as a PhD candidate. Jason has held a very special men torship and friendship role since the start of my graduate career, investing countless hours, energy and patience in my development. His method of hands on mentorship has afforded me many academic opportunities, but more importantly a model of student trai ning, the effects of which will be passed on to the next generation of students in honor of his legacy. Beginning my graduate studies at the University of Iowa made it possible to take courses form

PAGE 6

6 scholars such as Dr. Jill Beckman, Dr. Alice Davison, Dr. Judith Liskin Gasparro, Dr. Carlos Eduardo Pieros, Dr. Catherine Ringen, Dr. Jason Rothman and Dr. Roumyana Slabakova; completing them at the University of Florida made it possible to take courses from scholars such as Dr. Jessi Aaron, Dr. David Pharies, Dr. H. Wind Cowles, Dr. Brent Henderson, Dr. Edith Kaan, Dr. Jason Rothman and Dr. Stefanie Wulff, as well as work in the labs of Dr. Ludovica Serratrice at the University of Manchester and Dr. Mara del Pilar Garca Mayo at the University of the Basque C ountry. All of these experiences contributed positively to my educational experience and I am grateful to them and to my classmates Of course, this dissertation would not have been possible without the guidance and input of my committee: Dr. Jason Rothma n, Dr. Gillian Lord, Dr. Ana de Prada Prez, Dr. Wind Cowles, Dr. Youssef Haddad and Dr. Stefanie Wulff. I thank Dr. Rothman for advising this project since its conception in 2008 and for the infinite hours spent designing, revising and discussing it with me. I thank Dr. Lord for her comments that ultimately led to a clearer, more reader friendly product. I thank Dr. de Prada Prez for her acute insights throughout the dissertation, but especially regarding its overall cohesiveness, data interpretation and the organization of the results sect ion Perhaps more significant are the time, energy and persistence she poured into the project in its final hour, substantially improving its quality. I thank Dr. Cowles for her methodological expertise, her invaluable g uidance regarding data coding and analysis, her warm and encouraging style of mentoring and for her ability to structure our meetings as productive conversations. I thank Dr. Youssef Haddad for his syntactic guidance, for helping me become open to ideas no t previously considered, for insightful comments regarding proficiency measures and near native speakers and for his warm and witty humor! I thank Dr. Stefanie Wulff for her guidance with respect to data coding and organization, unparalleled

PAGE 7

7 statistical ex pertise and for the most detailed reading of any project I have ever completed! Of course, this project also benefited from the assistance of researchers, graduate students, student assistants and conference presentations. Specifically, Dr. Bill VanPatten, Dr. Edith Kaan, Carli Overfelt, Shuang Lu and the students and researchers associated with the Linguistics Development Lab all contributed significantly to the methodological design. James Colee (Department of Statistics) and the staff of the EDT office w ere also very generous with their time in helping me with statistics and formatting I also had the pleasure of working with four wonderful undergraduate assistants during my dissertation, each of whom aided in data coding and helping me organize my though ts regarding the tasks and data. Thank you Charlotte Park, Olivia Davey, Pamela Suriyachai and Arlene Chen, my COPA girls! Comments from audience members, especially Dr. Ayse Grel and Dr. Antonella Sorace, at the European Second Language Association confe rence doctoral workshop at the Universit di Modena e Reggio Emilia in Reggio Emilia, Italy and at the 7 th Annual Interdisciplinary Colloquium on Spanish, Portuguese & Catalan Linguistics, Literatures & Cultures at the University of Florida, Gainesville, F lorida also played their part in shaping this project.

PAGE 8

8 TABLE OF CONTENTS page ACKNOWLEDGEMENTS ................................ ................................ ................................ ............. 4 LIST OF TABLES ................................ ................................ ................................ ......................... 11 LIST OF FIGURES ................................ ................................ ................................ ....................... 13 ABSTRACT ................................ ................................ ................................ ................................ ... 15 CHAPTER 1 OPENING TO THE DISSERTATION ................................ ................................ .................. 17 1.1 General Introduction ................................ ................................ ................................ ......... 17 1.2 Child L1 versus Adult L2 Acquisition ................................ ................................ .............. 17 1.3 The Research Problem ................................ ................................ ................................ ...... 18 1.4 The Research Proposal ................................ ................................ ................................ ..... 21 1.5 The Research Questions ................................ ................................ ................................ ... 23 2 BACKGROUND INFORMATION AND PREVIOUS STUDIES ................................ ....... 26 2.1 General Introduction ................................ ................................ ................................ ......... 26 2.2 Interface Hypothesis ................................ ................................ ................................ ......... 26 2.2.1 Genesis and Purpose of the Interface Hypothesis ................................ .................. 26 2.2.2 Conceptualization of Inter faces ................................ ................................ .............. 28 2.2.3 Evolution of the Interface Hypothesis ................................ ................................ .... 30 2.2.4 Predictions ................................ ................................ ................................ .............. 32 2.2.5 Locus of Vulnerability ................................ ................................ ............................ 32 2.3 Previous Behavioral Research ................................ ................................ .......................... 33 2.3.1 Divergence on Interface conditioned Prope rties Related to Subjects: Child Studies ................................ ................................ ................................ .......................... 33 2.3.2 Divergence on Interface conditioned Properties Related to Subjects: Adult Studies ................................ ................................ ................................ .......................... 41 2.3.2.1 Word order and null subjects ................................ ................................ ....... 41 2.3.2.2 Word order ................................ ................................ ................................ ... 44 2.3.2.3 Null subjects ................................ ................................ ................................ 47 2.3.2.4 Subject and object expression ................................ ................................ ...... 50 2.3.3 Summary of Divergence Data ................................ ................................ ................ 51 2.3.4 Convergence on Interface conditioned Properties Related to Subjects .................. 53 2.3.4.1 Subject distribution ................................ ................................ ...................... 53 2.3.4.2 Word order ................................ ................................ ................................ ... 58 2.3.5 Summary of Convergence Data ................................ ................................ .............. 62 2.3.6 Criticisms of Previous Behavioral Research ................................ .......................... 63 2.4 Language Processing ................................ ................................ ................................ ........ 70 2.4.1 Parsing Model ................................ ................................ ................................ ......... 70

PAGE 9

9 2.4.2 Language Processing Methodology ................................ ................................ ....... 73 2.4.3 Previous Research ................................ ................................ ................................ .. 73 2.4.3.1 Neurophysiological studies ................................ ................................ .......... 74 2.4.3.2 Behavioral studies ................................ ................................ ........................ 80 2.4.4 Limitations of Previous Processing Literature ................................ ....................... 84 2.5 Testing the Interface Hypothesis ................................ ................................ ...................... 85 3 SYNTAX OF NULL SUBJECTS ................................ ................................ .......................... 87 3.1 Syntax ................................ ................................ ................................ ............................... 87 3.2 Null Subject Parameter (NSP) ................................ ................................ .......................... 87 3.2.1 Syntactic Properties Related to the NSP ................................ ................................ 94 3.2.2 Interface conditioned Properties Related to the NSP ................................ ............. 96 3.2.2.1 The Overt Pronoun Constraint (OPC) ................................ .......................... 96 3.2.2.2 Discourse distribution of referential subject pronouns ................................ 99 3.2.3 Learning Task: NSP ................................ ................................ ............................. 106 4 METHODOLOGY ................................ ................................ ................................ ............... 108 4.1 General Introduction ................................ ................................ ................................ ....... 108 4.2 Motivation for the Study ................................ ................................ ................................ 108 4.3 Participants ................................ ................................ ................................ ..................... 112 4.4 Non experimental Tasks ................................ ................................ ................................ 112 4.4.1 Background Questionnaire ................................ ................................ ................... 113 4.4.2 Proficiency Measure ................................ ................................ ............................. 118 4.5 Experimental Tasks ................................ ................................ ................................ ........ 121 4.5.1 Offline Experimental Tasks ................................ ................................ .................. 121 4.5.1.1 Grammaticality Judgment/Correction Task ................................ ............... 122 4.5.1.2 Context Matching Felicitousness Task ................................ ...................... 125 4.5.2 Online Experimental Tasks ................................ ................................ .................. 128 4.5.2.1 Online Grammaticality Task ................................ ................................ ...... 130 4.5.2.2 Online version of Context Matching Felicitousness Task ......................... 132 4.6 Conclusion ................................ ................................ ................................ ...................... 135 5 RESULTS ................................ ................................ ................................ ............................. 136 5.1 General Introduction ................................ ................................ ................................ ....... 136 5.2 Offline Experimental Results ................................ ................................ ......................... 136 5.2.1 Grammaticality Judgment/Correction Task (GJCT) ................................ ............ 136 5.2.1.1 Referential subject tokens ................................ ................................ .......... 137 5.2.1.2 Expletive subject tokens ................................ ................................ ............. 138 5.2.1.3 Statistical analysis and interpretation of results ................................ ......... 139 5.2.1.4 Summary of Grammaticality Judgment/Correction Task results ............... 142 5.2.2 Context Matching Felicitousness Task (CMFT) ................................ .................. 143 5.2.2.1 Contrastive Focus tokens ................................ ................................ ........... 143 5.2.2.2 Topic Shift tokens ................................ ................................ ...................... 145 5.2.2.3 Topic Maintenance tokens ................................ ................................ ......... 147

PAGE 10

10 5.2.2.4 Statistical analysis and interpretation of results ................................ ......... 148 5.2.2.5 Summary of Context Matching Felicitousness Task results ...................... 151 5.2.3 Summary of Offline Experimental Results ................................ .......................... 153 5.3 Online Experimental Results ................................ ................................ .......................... 154 5.3.1 Online Grammaticality Task (GT) ................................ ................................ ....... 155 5.3.1.1 Macro analysis of the online GT ................................ ................................ 155 5.3.1.2 Referential subject tokens ................................ ................................ .......... 157 5.3.1.3 Expletive subject tokens ................................ ................................ ............. 163 5.3.1.4 Condition Type comparison (referential versus expletive) ........................ 170 5.3.1.5 Comparative analysis of RTs ................................ ................................ ..... 171 5.3.1.6 Summary of online GT results ................................ ................................ ... 175 5.3.2 Online Context Matching Felicitousness Task (CMFT) ................................ ...... 176 5.3.2.1 Macro analysis of the online CMFT ................................ .......................... 176 5.3.2.2 Contrastive F ocus tokens ................................ ................................ ........... 178 5.3.2.3 Topic Shift tokens ................................ ................................ ...................... 187 5.3.2.4 Topic Maintenance tokens ................................ ................................ ......... 195 5.3.2.5 Context Type comparison (Contrastive Focus versus Topic Shift versus Topic Maintenance) ................................ ................................ ............................ 203 5.3.2.6 Comparative analysis of RTs across token type ................................ ......... 205 5.3.2.7 Summary of online CMFT results ................................ .............................. 215 5.3.3 Summary of Online Experimental Results ................................ ........................... 216 6 CONCLUSION ................................ ................................ ................................ ..................... 219 6.1 General Introduction ................................ ................................ ................................ ....... 219 6.2 The Research Questions ................................ ................................ ................................ 219 6.3 Limitations of the Study ................................ ................................ ................................ 225 6.4 Implications for the Interface Hypothesis and SLA Theory ................................ ........... 230 6.5 Conclusion ................................ ................................ ................................ ...................... 233 APPENDIX A OFFLINE GRAMMATICALITY JUDGMENT/CORRECTION TASK ........................... 235 B OFFLINE CONTEXT MATCHING FELICITOUSNESS TASK ................................ ...... 239 C ONLINE GRAMMATICALITY TASK ................................ ................................ .............. 244 D ONLINE CONTEXT MATCHING FELICITOUSNESS TASK ................................ ........ 252 E SAMPLE SCREEN SHOTS FROM THE ONLINE VERSION OF THE CONTEXT MATCHING FELICITOUSNESS TASK ................................ ................................ ............ 259 F AVERAGE AND STANDARD DEVIATION TABLES ................................ .................... 273 LIST OF REFERENCES ................................ ................................ ................................ ............. 291 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ....... 312

PAGE 11

11 LIST OF TABLES Table page 3 1 Present tense verbal morphology ................................ ................................ ....................... 94 3 2 Subject distribution in Spanish and Farsi ................................ ................................ ......... 105 4 1 Demographical i nformation ................................ ................................ ............................. 114 4 2 Native speaker linguistic information: languages spoken other than Spanish ................. 115 4 3 Native speaker linguistic information regarding Spanish ................................ ................ 116 4 4 L2 speaker linguistic information: languages spoken other than Spanish ....................... 118 4 5 L2 speaker linguistic information regarding Spanish ................................ ...................... 118 4 6 Proficiency scores ................................ ................................ ................................ ............ 120 5 1 Distribution of ratings ................................ ................................ ................................ ...... 144 F 1 Group average acceptance and standard deviation ................................ .......................... 274 F 2 Individual average acceptance and standard deviation ................................ .................... 275 F 3 ................................ ... 276 F 4 ............................ 277 F 5 Group average RT (ms) and standard deviation: First region ( ella/l/ello/siempre ) ....... 278 F 6 Individual average RT (ms) and standard deviation: F irst region ( ella/l/ello/siempre ) 279 F 7 Group average RT (ms) and standard deviation: Second region (verb) .......................... 280 F 8 Individual average RT (ms) and standard deviation: Second region (verb) .................... 281 F 9 Group average RT (ms) and standard deviation: Third region (DO) ............................... 282 F 10 Individual average RT (ms) and standard deviation: Second region (DO) ..................... 283 F 11 Group average RT (ms) and standard deviation: First region ( ella ...................... 284 F 12 Individual average RT (ms) and standard deviation: First region ( ella ................ 285 F 13 Group average RT (ms) and standard deviation: Second region (verb) .......................... 286 F 14 Individual average RT (ms) and standard deviation: Second region (verb) .................... 287

PAGE 12

12 F 15 Group average RT (ms) and stan dard deviation: Third region (DO) ............................... 288 F 16 Individual average RT (ms) and standard deviation: Third region (DO) ........................ 289

PAGE 13

13 LIST OF FIGURES Figure page 2 1 ................................ ................................ ..................... 29 2 2 ................................ ................................ ........ 30 2 3 Grammar parsing model adapted from Phillips (1996, p. 16) ................................ ........... 72 5 1 GJCT group results: referential subject tokens. ................................ ............................... 138 5 2 GJCT group results: expletive subject tokens. ................................ ................................ 139 5 3 CMFT group results: Contrastive Focus. ................................ ................................ ......... 145 5 4 CM FT grou p results: Topic Shift. ................................ ................................ .................... 146 5 5 CMFT group results: Topic Maintenance. ................................ ................................ ....... 148 5 6 Online GT group results: referential subject toke ns for first region ( ella/l/siempre ). ... 159 5 7 Online GT group results: referential subject tokens for second region (verb). ............... 160 5 8 Online GT group results: referential subject tokens for third region ( muchos/muchas ). ................................ ................................ ................................ ............ 162 5 9 Online GT group results: expletive subject tokens for first region ( ello/siempre ). ......... 16 5 5 10 Online GT group results: expletive subject tokens for second region (verb). ................. 166 5 11 Online GT group results: expletive subject t okens for third region ( mucho ). ................. 168 5 12 Online GT comparative group results for first region (subject/adverb). ......................... 172 5 13 Online GT comparative group results for second region (verb). ................................ ..... 173 5 14 Online GT comparative group results for third region ( muchos/muchas mucho ................................ ................................ ................................ ................. 174 5 15 Online CMFT group results: Contrastive Focus tokens for first region ( ella ...... 180 5 16 Online CMFT group results: Contrastive Focus tokens for secon d region (verb). .......... 181 5 17 Online CMFT group results: Contrastive Focus tokens for third region (DO). ............... 183 5 18 Online CMFT compar ative group results for second region (verb). ............................... 186 5 19 Online CMFT comparative group results for third region (DO). ................................ .... 187

PAGE 14

14 5 20 Onl ine CMFT group results: Topic Shift tokens for first region ( ella ................ 189 5 21. Online CMFT group results: Topic Shift tokens for second region (verb). ....................... 190 5 22 Online CMFT group results: Topic Shift tokens for third region (DO). ......................... 191 5 23 Online CMFT comparative group results for second region (verb). ............................... 194 5 24 Online CMFT comparative group results for third region (DO). ................................ .... 195 5 25 Online CMFT group results: Topic Maintenance tokens for first region ( ella .... 197 5 26 Online CMFT group results: Topic Maintenance tokens for second region (verb). ........ 198 5 27 Online CMFT group results: Topic Maintenance tokens for third region (DO) .............. 199 5 28 Online CMFT comparative group results for second region (verb). ............................... 201 5 29 Online CMFT comparative group results for third region (DO). ................................ .... 202 5 30 Online CMFT comparative group results for first region ( ella ........................... 206 5 31 Online CMFT comparative group results for second region (verb). ............................... 208 5 32 Online CMFT comparative group results for second region (verb). ............................... 209 5 33 Online CMFT comparative group results for third region (DO). ................................ .... 210 5 34 Online CMFT comparative group results for third region (DO). ................................ .... 212 E 1 Online CMFT: Instruction page 1 screenshot ................................ ................................ .. 259 E 2 Online CMFT: Instruction page 2 screenshot ................................ ................................ .. 259 E 3 Online CMFT: Pre ................................ ................. 260 E 4 Online CMFT: Practice session screenshot ................................ ................................ ..... 264 E 5 Online CMFT: Practice session comprehension question screenshot ............................. 265 E 6 Online CMFT: End of practice session screenshot ................................ .......................... 265 E 7 Online CMFT: Pre ................................ ................................ .. 266 E 8 Online CMFT: Sample Contrastive Focus token screenshot ................................ ........... 271 E 9 Online CMFT: Contrastive Focus comprehension question screenshot .......................... 272 E 10 Online CMFT: Task completion and thanks screenshot ................................ .................. 272

PAGE 15

15 Abstract of Dissertatio n Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy SOURCES OF PERFORMANCE AND PROCESSING ERRORS IN NEAR NATIVE L2 SPANISH SPEAKERS, L1 FARSI By Tiffany Judy August 2013 Chair: Jason Rothman Major: Romance Languages While normal child language acquisition results in complete productive and comprehension abilities at a relatively young age, adult language acquisition is more belabored and often results in linguistic abilities that differ from those of native speakers in terms of both productive and comprehension abilities. A major line of research in language acquisition studies seeks to determine and explain potential causes of these difference s. One current and influential hypothesis, the Interface Hypothesis (Sorace 2011), claims that dissimilarities are a product of the increased cognitive load experienced by bilinguals. That is, since bilinguals possess two grammars, they must simultaneously inhibit one language while activating the other. As a consequence of the increased cognitive load, the Interface Hypothesis predicts that bilinguals will show differences on properties requiring integration of purely linguistic information (e.g. syntax) w ith extralinguistic information (e.g. discourse or pragmatics). I test this prediction in near native adult second language acquirers of Spanish who are native speakers of Farsi using offline tasks measuring grammatical accuracy and interpretation of refer ential and expletive subject pronouns as well as online tasks measuring language processing. The particular participant group, linguistic properties and combination of offline and online experimental methodology are especially relevant to the debate surrou nding near native speaker ultimate

PAGE 16

16 attainment potential and bilingual processing. That is, the experimental group participants meet the requirements of the Interface Hypothesis in that they are highly fluent second language speakers that have lived in a Sp anish speaking environment for over 30 years. While the language pairing is quite novel, the linguistic property examined is interesting in that Spanish and Farsi are both null subject languages with analogous discourse distributions of subject pronouns. H owever, the underlying syntax of each language differs therefore necessitating restructuring in order for the speakers to converge on the syntax of subjects in Spanish. Finally, employing both offline and online measures makes it possible to test the Inter claim that bilingual processing is different from monolingual processing. The results, which provide mixed evidence regarding the claims of the Interface Hypothesis in that differences were found, but they were not isolated to overt subje ct pronouns, inform the field of linguistics regarding the human capacity for language acquisition, but also have significant implications for psycholinguistics, cognitive science and language teaching.

PAGE 17

17 CHAPTER 1 OPENING TO THE DISSERTATION 1.1 General Introduction This dissertation examines the acquisition of syntactic and external interface conditioned properties related to the Null Subject Parameter (NSP) in near native naturalistic adult second language (L2) acquirers who are native (L1) spe akers of Farsi. Results from two offline and two online tasks provide judgments and real knowledge of subject distribution in Spanish. The participants and methodological design were specifically constructed to test the Interface Hypothesis (IH; Sorace 2011), which predicts that processing differences will obtain for near native speakers for external interface conditioned linguistic properties even if the speakers have converged upon the related syntactic pro perty. With this information in mind, the following sections of Chapter 1 outline the research problem, research proposal and research questions. Before detailing these points, a brief overview of the basics of L1 acquisition in childhood is provided as ad ult L2 acquisition studies are often introduced against this backdrop for comparative purposes. 1.2 Child L1 versus Adult L2 Acquisition A salient difference that obtains when comparing child L1 acquisition with adult L2 acquisition is ultimate attainment, or the final result of in the process of language acquisition. Research shows that first language acquisition, despite its complex nature, is the only cognitive based task for which all humans, regardless of intelligence, socio economic status and/or moti 2005; Sakai 2005; Lust 2006). Chomsky (1965, 1981) explained this feat by proposing that human beings are endowed with an innate language acquisition device enabling them to converge on the grammar to which they are exposed despite the fact that the available linguistic

PAGE 18

18 Adult L2 learners, on the other hand, experience widely va rying levels of ultimate attainment (see White 2003a). This is not to say that it is impossible for adult L2 learners to have native like knowledge as much research has shown convergence across different domains of grammar, and, in particular, across prope rties for which the L1 and L2 differ ruling out L1 transfer (Lozano 2002; Grel 2006; Iverson, Kempchinsky & Rothman 2008; Rothman 2008a, 2009; Hopp 2009; Ivanov 2009; Bohnacker 2010; Donaldson 2001a, 2011b; Slabakova, Kempchinsky & Rothman 2012 among many others). Rather, the lack of uniform success evidenced in child L1 acquisition must be principally explained. Within the generative paradigm, research intending to explain the lack of universal success has moved from broad research questions concerning a ccess to UG and the Critical Period Hypothesis (Lenneberg 1967; see Oyama 1976; Coppetiers 1987; Johnson & Newport 1989; Birdsong 1992 for review of early formulations and Long 2005; Rothman 2008b for an updated view), to more specific questions regarding the quantity and quality of input (e.g. Flege, Yeni Komshian & Liu 1999; Rothman & Iverson 2007; Rothman & Guijarro Fuentes 2010) and to particular points of vulnerability in adult L2 acquisition (e.g. Sorace 2011). Motivated by the claims of the IH (Sora ce 2011), this dissertation explores a specific instance of vulnerability in near native adult L2 acquisition: differences in the processing of external interface conditioned properties related to the Null Subject Parameter (NSP). 1.3 The Research Proble m This dissertation assumes the generative framework, including the construct of UG and continued access to UG in adulthood 1 (see White 1989, 2003a for evidence supporting this 1 This is not to ignore theories that assume global impairment (e.g. Bley Vroman 1990; Clahsen & Hong 1995; Neeleman & Weerman 1997) or local impairment (Hawkins & Chan 1997; Beck 1998; Hawkins & Hattori 2006;

PAGE 19

19 assumption). However, even with full access to UG in adulthood, native like con vergence is not a guaranteed outcome (see Schwartz & Sprouse 2000), as anecdotal evidence from many L2 classrooms shows. Thus, the undeniable differences in ultimate attainment between child L1 acquisition and adult L2 acquisition cannot be and, in fact, a re not ignored by theories advocating adult UG accessibility. Rather, researchers adopting these theories are charged with the responsibility of describing and explaining these differences (see Dekydtspotter, Schwartz & Sprouse 2006). This task is complica ted by the empirical evidence demonstrating the success of adult L2 acquirers (White 2003a, 2003b, 2008, 2011a), indicating that it is not the case that this learner group is globally impaired but rather that certain aspects of language appear to be more difficult to converge on than others. Various theoretical proposals over the past decade have attempted to account for specific instances of divergence. For example, the Prosodic Transfer Hypothesis (Goad, White & Steele 2003; Goad & White 2006, 2008) clai ms that while the underlying representation of morphosyntactic features may be native like, the prosodic system of the native language may prevent accurate surface feature to form production in the second language, crucially, in predictable environments. A nother example, the Feature Re Assembly Hypothesis (Choi & Lardiere 2006; Lardiere 2008, 2009) posits that new feature acquisition is possible, but maintains that some non convergence obtains as a result of inaccurate featural mappings on the one hand, and the task of reassembling existing L1 formal features, possibly in different feature matrices, to new language specific morphophonological forms in the L2 on the other hand. While this is certainly not an exhaustive list of recent hypotheses, what is commo n to them is purported accessibility to UG. That is, the source of observed non convergence is not Tsimpli & Dimitrakopoulou 2007). For reasons of space and due to the focus of the dissertation, I do not explain these theories.

PAGE 20

20 posited to be a result of inaccessibility to UG, but rather of other factors such as difficulty with morphology, featural mappings and/or prosodic L1 influen ce. A complementary approach that has received much empirical attention as of late (cf. Sorace 2011) is the IH 2 (see Sorace 2011, 2012 for both the latest version as well as the history of its development), which, as it relates to adult L2 acquisition, se eks to explain L1 L2 differences much like the aforementioned hypotheses. Borrowing from psycholinguistics, the IH offers a reasonable method of explaining why some domains of non native language acquisition seem more belabored than native language acquisi tion, while still acknowledging the ample evidence demonstrating convergence in adult L2 acquisition (White 2003a, 2003b, 2011a). This reconciliation of facts is possible because the IH does not seek to weigh in on all types of L2 divergence 3 but rather m akes predictions for certain properties that are more likely to demonstrate residual optionality at the level of near native ultimate attainment. Specifically, the IH targets properties that require the integration of linguistic and cognitive information ( i.e. external interface conditioned properties). Sorace states that vulnerability arises from the presence of two linguistic systems in a single mind. In particular, vulnerability is claimed to be a result of the stress placed on finite cognitive resources (e.g. see Wilson, Keller & Sorace 2009; Sorace 2011). That is, as compared to monolinguals that have but one language to manage, second language acquirers must divide cognitive resources to inhibit the non relevant language. This creates an extra burden f or working memory, executive function and attentional resource allocation. The IH allows for the possibility of convergence on properties related to the narrow syntax by near native speakers. However, external interface conditioned properties, due to their 2 Differently from the approaches outlined above, the IH maintains agnosticism regarding accessibility to UG. 3 This is not to say that the other divergence accounts listed above predict across the board divergence; more precisely, evidence in favor of the null hypothesis for the majority of these accounts has been found, weakening their empirical strength and explanatory power.

PAGE 21

21 combination of both linguistic and general cognitive information, are predicted to be more costly in terms of processing, and thus, points of vulnerability in bilingualism (Sorace 2011 and sources cited therein such as Pinango, Burkhardt, Brun, & Avrutin 2001; Burkhardt 2005; Pinango & Burkhardt 2005; see Section 2.2.5 for further details). To date there is no consensus on the tenability of the IH (see Montrul 2011; Rothman & Slabakova 2011; Slabakova 2011; White 2011b). Nevertheless, the IH has proven to be largely influential in the field of generative L2 acquisition over the past decade, spurring much empirical investigation. In addition to stimulating innovative research and collaboration across sub understanding of the nature of human language since it addresses the source and cause of specific asymmetrical performances between highly proficient (near native) bilinguals and monolinguals. The following section outlines how this dissertation tests the explanatory predictions of the IH. 1.4 The Research Proposal In an attempt to identify and isolate potential sources of non convergence in adult L2 acquisition, this dissertation tests the competence and performance of an immigrant community living in Ar gentina, specifically, L1 Farsi speakers who are near native naturalistic learners of L2 Spanish, on two properties related to the NSP. To this end, the theoretical and predictive power of the IH is tested. As the IH permits different predictions for purel y syntactic properties as compared to related external interface conditioned properties, namely syntax discourse constraints, both types of properties are tested. Farsi and Spanish are both syntactic discourse configurational null subject languages, meanin g that pro is licensed and available in each language. While a syntactic difference exists between the two languages (Spanish, but not Farsi, permits null expletives (Karimi 2005)), the semantic and discourse interface conditioned properties associated wit h the null subject setting have been shown to be analogous across both

PAGE 22

22 languages (Judy & Feizmohammadpour 2012), at least for the properties examined herein 4 While the IH does not preclude transfer of the L1 value from allowing for convergence on the synt actic properties related to the NSP, the same is not true for the related external interface conditioned properties. That is to say, the IH predicts vulnerability on external interface conditioned properties despite the identical nature of these properties in Farsi and Spanish. This is due to the fact that the source of vulnerability is purported to be the presence of two grammars itself and, importantly, not only language specific differences. Previous research, which employed offline tasks and varied with respect to proficiency level, examining convergence on subject distribution in two null subject languages produced conflicting results. For example, Bini (1993), Margaza & Bel (2006) and Garca Alcaraz & Bel (2011) reported an overuse of overt subject pro nouns in Spanish speaking learners of Italian and Greek speaking learners of Spanish and Dariya speaking learners of Spanish, respectively. Similarly, the older Spanish Italian bilinguals of Sorace et al. (2009) demonstrated overuse of overt subject pronou ns when compared to monolingual children and younger bilinguals. Neither Lozano (2002) nor Prada Prez (2009), on the other hand, reports overuse of overt subjects in L1 Greek learners of L2 Spanish for internal interface conditioned properties related to the Overt Pronoun Constraint or in subject distribution of Spanish Catalan and Catalan Spanish bilinguals. Finally, Prada Prez (2010) established that word order as a function of focus (an external interface conditioned property) was not problematic for t he Catalan Spanish bilinguals tested therein. While the present study also examines speakers of two null subject languages, it contributes to a gap in the 4 Differences have been reported for typologically related languages such as Spanish and Italian (Filiaci 2010a, 2010b; Filiaci, Sorace & Carreiras 2013) and Spanish and Catalan (Prada Prez 2009), as well as among the different dialects of Spanish (Cameron 1996; Lapidus Shin 2006; Otheguy, Zentella & Livert 2007). Thus, f uture studies may demonstrate that differences between Farsi and Spanish do exist with respect to semantic and discourse interface conditioned properties.

PAGE 23

23 literature by examining near native speakers of two null subject languages using both offline and onl ine methodologies. These points are especially important for the claims of the IH. 1.5 The Research Questions With the research problem and proposal in mind, it is important to consider the research questions guiding the methodological design and data a nalysis of this dissertation. In general terms, this dissertation seeks to determine if the L2 participants demonstrate native like knowledge of and processing of the syntactic and discourse distribution of subjects in Spanish. More specifically, and consi dering the predictions of the IH, the language pairing and the properties tested herein, this dissertation addresses the following research questions: 1. Do near native L2 learners demonstrate more divergence with respect to external interface conditioned pro perties as opposed to purely syntactic ones in the offline tasks? Regarding this research question, the IH predicts that convergence on syntactic properties is possible (but not guaranteed) in adult L2 acquisition; however, because external interface condi tioned properties require the integration of strictly linguistic information and discourse information, the IH predicts differences between native speakers and adult near native L2 learners. As described above in Section 1.4, pro is licensed and identifiab le in both Spanish and Farsi, making them syntactic discourse configurational null subject languages. Furthermore, their semantic and discourse distribution of overt and null subjects is analogous for the properties tested herein (Judy & Feizmohammadpour 2 012). In spite of these similarities, the IH predicts vulnerability on external interface conditioned properties, such as the discourse distribution of overt and null subjects, since it is claimed that the presence of two grammars, and not differences/simi larities between the two languages, is the source of divergence. Thus, the IH would not predict convergence on external interface conditioned properties but divergence on purely syntactic properties. Previous research employing offline tasks supports this prediction in language pairings that are non facilitative for transfer since studies have shown convergence on

PAGE 24

24 the syntax of null subjects for English speaking learners of L2 Turkish (Grel 2006), English speaking learners of L2 Italian (Sorace & Filiaci 2006) and English speaking learners of L2 Spanish (Rothman 2008a, 2009), yet mixed results obtained across these and other studies with respect to convergence on the discourse distribution of overt and null referential subjects. For example, Rothman (2009) found that advanced L2 learners converged on subject distribution in Contrastive Focus and Topic Maintenance tokens, while Sorace & Filiaci (2006) found divergence with respect to the interpretation of overt subject pronouns in forward and backward anapho ra tokens. Nonetheless, and as mentioned above, studies examining convergence on subject related properties in speakers of two null subject languages, for language pairings in which L1 transfer might benefit the learner, also produced varied results in tha t an overuse of overt subjects was reported in Bini (1993), Margaza & Bel (2006) and Garca Alcaraz & Bel (2011). Still, Lozano (2002) and Prada Prez (2009, 2010) reported convergence among speakers of two null subject languages. Taken together, this body of studies points to a more complex acquisition situation, one that is not solely predicated on language pairing or property, but rather is likely dependent on proficiency level and testing measures. The second research question considers the same topic i n online tasks examining processing. 2. Do near native L2 learners exhibit greater processing difficulties with external interface conditioned properties as compared to purely syntactic ones in online tasks? Here, processing difficulties for external interfac e conditioned properties are expected under the IH, but processing differences on purely syntactic properties are neither strictly expected nor precluded. Therefore, the IH predicts that adult near native L2 learners will demonstrate processing differences (e.g. slower reaction times (RTs)) with respect to external interface conditioned properties and that differences in RTs may also be evidenced for purely syntactic properties. It would be odd, under the IH, for processing differences to be evidenced

PAGE 25

25 with respect to purely syntactic properties, but not external interface conditioned properties. Finally, the third research question examines testing modality. 3. Do near native L2 learners perform more native like in offline tasks as compared to online tasks? S ince the locus of divergence predicted to obtain in adult near native L2 learners is taken to be language processing, more native like performance is expected for the offline tasks as compared to the online tasks, although it is possible that non native li ke performance is evidenced in both tasks. It seems that finding more native like performance in online tasks as compared to offline tasks would be unexpected by the IH, though, should this result obtain, metalinguistic awareness could be taken as a variab le. With these research questions in mind, the following chapter describes relevant background information regarding the IH and previous research pertinent to the current study. Chapter 3 describes the syntax of the linguistic property examined herein, wh ile Chapter 4 presents the experimental methodology employed in the study. Finally Chapter 5 discusses the results of each of the four experimental tasks and the significance of the results is discussed in light of research questions in Chapter 6.

PAGE 26

26 CHAPTER 2 BACKGROUND INFORMATION AND PREVIOUS STUDIES 2.1 General Introduction The goal of this chapter is to present necessary background information regarding the theoretical assumptions adopted in the study as well as to present previous empirical research spa nning offline and online methodologies. To this end, the chapter begins with a description of the Interface Hypothesis (IH; Sorace 2011). Next in Section 2.3, previous offline behavioral research providing evidence both for and against the claims made by t he IH is described. While offline research of this type is ultimately unable to provide evidence for or against the claims of the IH, reviewing this research is nonetheless necessary as it has informed previous versions of the IH as well as shaped the goal s and methodologies of the current study. Following Section 2.3, the general topic of language processing, including the parsing model assumed herein, language processing methodologies and previous neurophysiological and online behavioral studies, are pres ented in Section 2.4. Criticisms of both the offline and online research are provided after each respective subsection. Finally, a proposal for methodologies that more 2.2 Interface Hypothes is The following sections describe the IH, inclusive of a brief history of the hypothesis, a over the years as a result of empirical investigation. The prediction s made by the IH are also described as well as the purported source of vulnerability. 2.2.1 Genesis and Purpose of the Interface Hypothesis The IH was originally proposed by Sorace (2005, 2006) as a hypothetical way to explain the divergence and residual o ptionality observed in highly advanced second language learners.

PAGE 27

27 formal linguistic analysis with research and concepts from psychology (and cognitive science more generally), building on the noti Sorace (2011) claims that the IH is agnostic 1 about the debate concerning accessibility to Universal Grammar, although it will be shown in Section 2.1.2 that the concept of interfaces is ent irely consistent with Universal Grammar native of native speakers of French and English learning Italian as a second language, two different types of non convergent intuitional patterns regarding syntactic and semantic properties related to u naccusativity and clitic climbing were found between the two experimental participant groups. The native French speaking participants demonstrated divergence 2 while the native English speaking participants demonstrated incomplete or indeterminate knowledge Divergence was described as judgments that differed from those of native speakers while incomplete or indeterminate knowledge was viewed as the inability to produce determinate judgments. Perhaps nati implies some sort of failure or shortcoming with respect to convergence on the target language. However, Hyltenstam and Abrahamsson (2003) correctly point out that these 1 While the Interface Hypothesis claims that narrow syntactic properties, as opposed to interface con ditioned properties, should be unproblematic, no direct claims are made about the exact nature of the underlying representation of these properties. Thus, it is possible that the underlying representation is not native like or that domain general learning mechanisms are employed, ultimately producing target like knowledge at the surface level. However, if the latter scenario were true, it is unclear how semantic properties that fall out of syntax are accurately converged on, as the literature amply shows (s ee Rothman 2008; Slabakova 2008 and sources cited therein). 2 But see Sorace (2011, p. 10) and Sorace (2005, p. 18ff) for different interpretations of the results.

PAGE 28

28 failures/shortcomings might only be exposed in near native speakers via scrutinous and specific methodologies, a position that has influenced the methodological practices of researchers investigating the IH (see Sections 2.3 and 2.4 for details). 2.2.2 Conceptualization of Interfaces To understand the claims of the IH, it is first necessary to examine the mechanics of linguistic interfaces and differentiate among the uses of the term in linguistics. To begin, it must iest use in Principles and Parameters (Chomsky 1981, 1986) and its later use in the Minimalist Program (Chomsky 1995). In the Principles and Parameters program, it was claimed that syntax and that compliance between syntactic computations and interface conditions was necessary in order for the derivation to survive. Similarly, in the Minimalist Program (Chomsky 1995, 2000, 2001) syntactic operations must satisfy interface conditions relate d to the conceptual intentional interface (semantics) and the perceptual interface (phonological). In these programs, interfaces were levels of representation that permitted language to accomplish its function of bringing together linguistic form and meani ng for communicative purposes. Yet, Ramchand and Reiss (2007) and Rothman and Slabakova (2011), among others, call attention to the fact that the strictly linguistic domains that map form with meaning also interact with extralinguistic domains such as disc ourse and context. This suggests that other interfaces communicative function to be meaningful and anchored. Uncovering the nature of both language specific interfaces as we ll as those that interface with cognition and how they interact stands to provide key insights into our understanding of the nature of language. Conceivably, it is for this reason that a number of proposals attempting to describe interfaces and their inter actions

PAGE 29

29 have been set forth in recent years (Jackendoff 2002; Burkhardt 2005; Ramchand & Reiss 2007; possibly due to its syntactocentric design (as compared to Ja Figure 2 1 As is straightforwardly seen in Figure 2 1, the Computational System (syntax) forms the with all other areas, lending it a privileged position. Reinhart claims that difficulties with interfaces arise as the result of reference set computations saying that comparing two or more possible syntactic derivations is burdensome on the language lear ner. This can be easily extended to both child and adult (2006) design is that it does not clearly differentiate between internal and external interfaces, an imp ortant concept in studies investigating the IH (especially since Sorace & Filiaci 2006; Sorace 2) remedies this

PAGE 30

30 drawback by separating the purely linguistic modules (seen in the uppe r, light gray boxes) from extralinguistic information (seen in the lower, dark gray box). Figure 2 2 To recapitulate, assuming a syntactocentric nature of language, interfaces are s yntactic structures that must meet minimum conditions determined by other modules (either internal or external) in order for the structure to be grammatical/felicitous. Importantly, the purported interaction between language and extralinguistic cognition i s not incompatible with the assumed modularity of language proposed by Universal Grammar. 2.2.3 Evolution of the Interface Hypothesis Like most theories and hypotheses, the IH has evolved over the years as empirical investigation informs the hypothesis 3 White (2011a) differentiates between two versions of the IH, although a third and more recent version will also be described herein. The first version juxtaposed narrow syntax with interface properties, claiming that only the former was immune to vulnerabi lity in second language acquisition (Sorace 2005, 2006). The second version separated purely linguistic (internal) interfaces from those that interface with cognition (external 3 While empirical research informs the Interface Hypothesis, one criticism of it is that th e argument is circular: vulnerability is predicted where it is found (Duffield 2011).

PAGE 31

31 interfaces) while narrow syntax maintained its privileged status (Sorace & Fili aci 2006; Sorace & Serratrice 2009; Tsimpli & Sorace 2006). Specifically, the second version claimed that external interfaces posed greater difficulties than internal interfaces. There are two commonalities across both early versions: (i) narrow syntax is not predicted to be a source of inevitable vulnerability while, conversely, (ii) external interfaces are predicted to be a source of inevitable vulnerability. As such, the past decade of research has witnessed a marked increase in studies investigating whe ther adult second language learners converge on properties lying at the syntax semantics, syntax pragmatics and syntax semantics pragmatics interfaces (see Slabakova 2006, 2008; Sorace 2011 and sources cited therein). In a sense, the IH (and studies conduc ted under its banner) seeks to describe second language acquisition potential by simultaneously explaining observed divergence in non native language acquisition and isolating the upper limits of second language acquisition. Currently, syntax pragmatics is often seen as the benchmark. In its most recent form (Sorace 2011), the IH maintains that only properties conditioned upon external interfaces are subject to vulnerability in second language acquisition. Different from earlier versions, language pairing i s ruled out as constituting the only deterministic factor in Rather, the current version places the locus of vulnerability on the presence of two linguistic system s, which purportedly prevents second language learners from processing the target in the distribution of finite cognitive resources (e.g. the limitations in atten tional resources related to inhibitory control (e.g. Green 1986, 1998)). As compared to monolinguals who have but one language to manage, cognitive resources must be divided by second language learners to inhibit the non relevant language, among other thin gs, creating an extra burden for working

PAGE 32

32 memory, executive function and attentional resource allocation. A further clarification was made mistakenly applied to in termediate stages of second language development and reiterates that its predictions are valid for near native speakers only (but see Slabakova 2008; White 2009, 2011b; Lardiere 2011 for compelling counterarguments). 2.2.4 Predictions The most recent insta ntiation of the IH predicts that near native speakers of a second language will demonstrate vulnerability/residual optionality with respect to properties lying at external interfaces only. That is to say, syntactic properties and those properties lying at internal interfaces are not predicted to be a source of inevitable vulnerability for near native speakers. 4 integration of syntactic and pragmatic conditions remains less than optimally efficient and gives 2009). 2.2.5 Locus of Vulnerability The vulnerability predicted with respect to external interface conditioned properties is seen as the result of processing problems brought about by bilingualism. In effect, it is claimed that the presence of two grammars is burdensome for bilinguals since one grammar must be deactivated or inhibited while the other is employed (i.e. Inhibit ory Control, Green 1986, 1998). The cognitive resources expended for deactivation/inhibition renders memory systems, executive function and attentional resource allocation less efficient. Since evidence suggests that all languages are concurrently activate d (Green 1986, 1998; Kroll & Stewart 1994; Marian & 4 one does not acquire levels of representation, since these are part of the grammatical apparatus (given by UG ) that language acquirers bring to bear when dealing with

PAGE 33

33 Spivey 2003; Costa 2005; Dijkstra 2005; Kroll, Bobb, & Wodniecka 2006; Schwartz, Kroll, & Diaz 2007; Bialystok 2009; Kroll, Bogulski, & McClain 2012 among others), the question as to why only external int erface conditioned properties are affected is raised. As described in Sorace and Tsimpli (2006), purely syntactic computations purportedly require less cognitive processing resources and internal interface conditioned properties are entirely grammar intern al. Conversely, the incorporation of linguistic and extralinguistic, or cognitive, information is deemed to be more taxing on processing irrespective of language pairings. 2.3 Previous Behavioral Research While this dissertation examines properties relate d to the distribution of overt and null subjects only, the literature review describes interface conditioned properties related to subjects including overt and null subject distribution as well as word order as the claims of the IH apply to all interface c onditioned properties The following sections describe current empirical investigations relevant to the IH and the current study. First, studies that provide evidence supporting the claims made by the IH are presented for child bilinguals (Section 2.3.1) a nd adult bilinguals (Section 2.3.2). These studies are presented by linguistic domain. Section 2.3.3 offers a conclusion of these studies and finally, Section 2.3.4 describes previous studies that provide evidence against the IH. These studies represent ad ult second language acquisition and are also presented according to the linguistic domain. 2.3.1 Divergence on Interface conditioned Properties Related to Subjects: Child Studies While observable differences between native speakers and adult second langua ge speakers have been cited for certain linguistic properties, so too has evidence suggesting native like convergence. Thus, in an attempt to, first, isolate which areas prove most problematic and, second, explain the potential cause of the observed differ ences, recent empirical investigations

PAGE 34

34 have focused on linguistic interfaces. The following sections briefly describe some relevant studies that have found divergence on interface conditioned properties among child learners. Child monolingual and bilingua l acquisition studies demonstrating that syntax is rather straightforwardly acquired, but that convergence on properties requiring the incorporation of linguistic information and other cognitive domains is somewhat delayed, provide evidence suggesting that certain linguistic properties (i.e. those that require integration of more than one source of information) are delayed relative to others. If empirical investigation demonstrates that purely syntactic properties emerge in children before interface conditi oned properties, it would not be surprising to find a similar developmental sequence in adults, thus highlighting the importance of examining near properties that are more costly, 5 such as those that lie at the interface. That is, as is emphasized by the most recent version of the IH (Sorace 2011), its claims are not intended to apply to adult second language speakers of lower proficiency levels since it cannot be determined whether relevant differenc es found are due to insufficient exposure to the language at the time of testing, some intermediary process of interlanguage development or inevitable variability in ultimate attainment (see for discussion White 1989, 2003a). Additionally, if the claim tha t processing consequences of bilingualism are the source of differences between monolinguals and bilinguals, both child and adult bilinguals are predicted to diverge on interface conditioned properties The following section summarizes select child langua ge acquisition studies that examine the distribution of null and overt subjects in Romance languages. First, two studies examining 5 formally complex, which would need to be shown in linguis tic computational terms and/or properties requiring more processing resources. It is often difficult to tease these two variables apart and not immediately clear what individual authors mean when they call something (formally) complex; complexity being a t heory internal notion and not an intuitive one. Suffice it to say that interface since integration of multiple types of information is at play (either across linguistic systems or betwee n linguistic systems and external domains of cognition).

PAGE 35

35 monolingual Spanish acquisition are described (Lapidus Shin 2006; Austin, Blume, Parkinson, N ez del Prado & Lust 1997). Nex t, Bel (2003) and Grinstead (2004), which examined both monolingual Spanish and Catalan speaking children and showed similar trends in the Catalan data, are outline. Paradis and Navarro (2003) and Liceras, Fern ndez Fuertes and Alba de la Fuerte (2012) te sted English Spanish bilingual children, with mixed results reported as a product of English influence. Finally, Sorace, Serratrice, Filiaci and Baldo (2009) examined both English Italian and Spanish Italian bilingual children as compared to monolinguals. While all studies examined Romance languages, varied methodologies were employed across different language learners and age groups. Across those studies that examined subject distribution in light of discourse, however, the results indicate that accessing or integrating interface conditioned knowledge is belabored for children. Additionally, several studies point to the need to carefully consider the input to which participants are exposed. In a study on the distribution of overt and null subject pronouns i n Mexican Spanish, Lapidus Shin and Cairns (2012) 6 tested for sensitivity to Continuity of Reference (i.e. Topic Maintenance and Topic Shift) with third person singular subjects. Stories that contained either Topic Maintenance or Topic Shift were narrated to the 139 child participants and 30 adult control participants with a target sentence that contained either an overt or a null subject and the preferences of the participants were recorded. The results pointed to differences between the children and adult s, especially the younger participants. Specifically, while adults preferred overt subject pronouns in Topic Shift contexts and null subject pronouns in Topic Maintenance contexts, younger participants overextended null subject pronouns to Topic Shift cont exts. With respect to the Topic Maintenance tokens, only the oldest group showed adult like preferences. 6 This work has its base in Lapidus Shin (2006).

PAGE 36

36 Lapidus Shin & Cairns also reported that the preference for overt subject pronouns over null subject pronouns in Topic Shift contexts becomes significa nt around 8 9 years of age 7 Austin et al. (1998) also examined the distribution of overt and null subject pronouns 8 in their study on the spontaneous speech of 10 monolingual Spanish speaking children ages 1;2 to 3;4 years (Mean Length Utterance (MLU) of 1.29 4.77). Null subject pronouns appeared at the earliest MLU tested, while overt subject pronouns began to appear at MLU 2.0. In general, substantially fewer overt subject pronouns were evidenced across all MLUs and across all person/number combinations (approximately 18% of subject pronouns were overt with the highest percentages evidenced for first person singular). Similar to Lapidus Shin & Cairns (2012), Austin et al. found that syntactic licensing and identification of null subject pronouns precedes their adult like distribution. In addition to monolingual Spanish speaking children, Bel (2003) also investigated monolingual Catalan subject pronouns. Specifically, longitudinal reco rded oral data from three Spanish and three Catalan speaking children ages 1;7 2;8 years were examined. The earliest evidenced overt subject pronoun in both Spanish and Catalan appeared relatively early at ages 1;9 and 1;10, respectively. The percentage o f overt as compared to null subject pronouns reaches what are considered adult like levels when MLU reaches approximately 2. Bel interprets the findings as supporting the Continuity Hypothesis in that early convergence in the domain of syntax is found. S imilarly, Grinstead (2004) investigated the distribution of subject pronouns in four monolingual Catalan speaking children and in three monolingual Spanish speaking children. 7 An overuse of overt subject pronouns was also reported in Bayley and Pease lvarez (1996); however, this study represents a language contact scenar io where English may have influenced overt subject pronoun distribution. 8 Auxiliary verb omission is also examined, but will not be discussed herein as the main focus of this study involves subject distribution.

PAGE 37

37 Grinstead found that, as compared to children acquiring a non null subject langu age, children acquiring a null subject language did not use overt subjects as early. Nonetheless and complimentary to the findings of Lapidus Shin & Cairns (2012), Austin et al (1998) and Bel (2003), overt subjects were evidenced in one Catalan and two Sp anish speaking children as early as 1;10 and more consistently around 2;0 or later. In line with previous research claiming that subjects are in the Complimentizer Phrase (CP; Contreras 1991; Otero 1993; Dobrovie Sorin 1994; Ordez 1997; Alexiadou & Anagn ostopoulou 1998; Ordez & Trevio 1999; Holmberg 2005 among many others), Grinstead argued that subjects are in the left periphery (which is deemed inactive in child production from the onset of language use) and that there is an interface delay between s yntax and pragmatics (although the discourse pragmatics module is argued to be present since children distinguish between new and old information). While some claim that delayed discour se pragmatics competence, Grinstead argued that children do in fact have adult like discourse pragmatic competence (as well as relatively adult like syntax), but that they are unable to properly access it. In line with the IH, Grinstead claimed that the di fference between adult and child production is a result of an immature interface between syntax and discourse pragmatics. Similarities between the Spanish monolingual data presented in Lapidus Shin and Cairns (2012) and Austin et al. (1998) and the Spanish and Catalan monolingual data presented by Bel (2003) and Grinstead (2004) include early emergence of overt subject pronouns, but protracted convergence on their discourse distribution. Having formulated a base of monolingual Romance data, the following st udies examine bilingual convergence on similar properties. Paradis and Navarro (2003) examined the spontaneous speech data of two Spanish monolingual children (ages 1;8 2;7 and 1;8 1;11years), one Spanish English bilingual child

PAGE 38

38 (age 1;9 2;6 years) and th eir parental interlocutors in light of subject use. They predicted that a Spanish English bilingual child may produce more overt subjects than a typically developing monolingual child speaker of Spanish and that the bilingual child may not exhibit the same two stage developmental pattern as the monolingual. By looking at the proportion of overt subject use versus null subject use as well as the context in which they were used, Paradis and Navarro claimed that the bilingual child overused overt subjects (35% (31% were overt pronouns)) as compared to the monolingual children (at or below 20% (25% and 16% were overt pronouns)). While all three children used overt subjects according to the five discourse pragmatic functions identified in the study, the bilingual high as that of the monolingual children (26% versus 10% and ~0%). Paradis and Navarro later noted that the Spanish to which the bilingual child was exposed is different from that of the monolingual children in that the bilingual child was exposed to Cuban Spanish, which is a variety that contains more overt subjects than null as compared to the Penins ular variety the monolingual children were exposed to: 60% overt, 40% null versus 40% overt, 60% null, respectively. Liceras et al. (2012) compare data on two English Spanish bilingual twins that was first presented in Liceras, Fern ndez Fuertes and P rez Tattam (2008) with the bilingual data reported in Paradis and Navarro (2003). The spontaneous speech data presented ranges from age 2;04 years to 4;11 years and MLUs ranging from 1.43 4.28 and 1.48 3.88 for each twin. The twins, like the Spanish speaking children discussed in Bel (2003), produce both overt and null subject pronouns in the earliest stage reported. The fact that no significant increase in production of overt subject pronouns is evidenced between the first two testing stages indicates that ch ildren

PAGE 39

39 converge on the syntax of null subjects early on, as data from aforementioned studies has also suggested. Differently from Paradis and Navarro (2003), the bilingual twins did not demonstrate an overuse of overt subject pronouns. This finding points to input, rather than the influence of English, as the factor affecting the overproduction of overt subject pronouns by bilingual child studied in Paradis and Navarro (2003). Lastly, Sorace et al. (2009) compared two child bilingual groups, English Italia n bilinguals and Spanish Italian bilinguals, to child and adult monolingual speakers of English and Italian. Again, participants were tested for knowledge of subject distribution in Topic Shift and non Topic Shift contexts in both English and Italian (in t he case of the English Italian bilinguals) and Italian (in the case of the Spanish Italian bilinguals). The participants completed an arized the events of the videos. The participants were asked to monolingual and bilingual children in the younger age group (6 7 years), especially those living in an English speaking environment, accepted significantly more overt subject pronouns in contexts that required null subjects as compared to the older bilingual children and the adult control. However, the older Spanish Italian bilinguals (8 10 years) also a ccepted more overt subject pronouns than all the other child groups. These results suggest divergence regardless of the language pairing, which is predicted by the IH since processing limitations, and not language pairing, are the purported source of diver gence. Finally, as Sorace et al. pointed out, this study highlights the importance of isolating and testing for the various discourse contexts (Topic Shift, Topic Maintenance, Focus, etc.), as not all contexts may be equally difficult.

PAGE 40

40 Thus, results from the subject distribution child acquisition studies described above demonstrated early production of both overt and null subject pronouns, but that, in those studies that examined this distribution in light of context, children had difficulty integrating in terface conditioned knowledge, even when the related syntax was presumed to be in place. Hulk and Mlller (2000) also echoed these findings in that the Dutch Freneh and German Italian bilingual children examined had difficulties with object drop, which the y claim is an external interface conditioned property 9 Contrary to the predictions of the IH, L1 effects were also evidenced in some of the literature reviewed above, which shows that, in addition to interface difficulties, language pairing may have an ef fect on child language acquisition. This was also evidenced in Serratrice, Sorace & Paoli (2004) who examined the distribution of subjects and objects in an English Italian bilingual child) and Serratrice et al. (2009). In the former study, the speech of a n English Italian bilingual child was found to be influenced unidirectionally (English to Italian) in that overt subjects were used in Italian where null subjects were preferred. In the latter, bilingual English Italian children performed less accurately w ith respect to specific and generic plural noun phrases in Italian than did monolingual Italian speaking children and adults or bilingual Spanish Italian children. As described in Section 2.3.1, we might expect adult second language (L2) learners to demons trate divergence on properties with which children also experience difficulties, like those described in above. This highlights the necessity of testing near native learners as they can be assumed to have converged on the grammar. Section 2.3.2 describes s ome adult acquisition studies that found divergence on interface conditioned properties related to subjects. 9 Despite this divergence, the same child bilinguals did not show difficulties with respect to Root Infinitives, which are also argued by Hoekstra and Hyams (1998) to be an interface conditioned property.

PAGE 41

41 2.3.2 Divergence on Interface conditioned Properties Related to Subjects: Adult Studies The following sections describe recent empirical studies t hat examine the acquisition of interface conditioned properties by adult bilinguals. These studies, which examine properties such as word order, the distribution of subjects and objects and topicalization, found divergence on interface conditioned properti es and are thus taken as providing evidence in support of the IH. 2.3.2.1 Word order and null subjects The first set of studies reviewed dealt with word order as it relates to the subject and verb as well as the distribution of null and overt subject prono uns. Importantly, these properties are related to the NSP. While Belletti and Leonini (2004), Belletti, Bennati and Sorace (2007) and Tsimpli, Sorace, Heycock and Filiaci (2004) examined different language pairings in different bilingual instances (L2 acqu isition in the first two studies and L1 attrition in the later), common across all three studies is the finding that interface conditioned properties related to the NSP are vulnerable in these populations. Specifically, Belletti and Leonini (2004) investi gated Verb Subject (VS) word order and null subject use in a group of adult second language speakers of Italian with different native languages (Albanian, French, German, Russian, Polish, etc.) of varying lengths of stay in Italy and various amounts of Ita lian study in their native country. Null subject languages, like Italian and Spanish, typically allow for VS word order more freely than non null subject languages like English. Additionally, VS word order is often 10 the result of focus (new information com ing last in the sentence as compared to topic, or old information, being presented first). The second language learners of Italian were presented with a PowerPoint presentation in which they viewed 22 short videos. Participants were asked to answer two que stions after viewing each video: one 10 Predicate type also affects word order in Spanish (Contreras 1976; Suer 1982; Ordez 199 7; Zubizarreta 1998).

PAGE 42

42 asked by a character in the video and one asked about the video. In their answer, participants were instructed to use the verb contained in the scene in the most natural way possible. Group 1 was composed of the most fl uent speakers of Italian, and they performed the most native like; however, Belletti and Leonini claim that this is a result of first language transfer. The native speakers of French or German performed least native like (however, they seemed to have the l owest proficiency as well), again suggesting language transfer issues as neither French nor German is a null subject language. Regarding null subject use, the native speakers of French and German used fewer null subjects than the control group and the othe r second language learners of Italian. However, their performance regarding null subjects was more accurate than their use of VS word order, suggesting that purely syntactic properties are converged upon prior to interface conditioned properties. Similar ly, Belletti et al. (2007) investigated the production and interpretation of postverbal subjects, overt pronominal subjects and null subjects in near native speakers of Italian (native speakers of English). Participants completed three production tasks: a Verb Subject Video Task to elicit narrow focus new information postverbal subjects, a Story Telling Task to examine spontaneous production of subjects, and a Headlines Task to examine the production of preverbal and postverbal subjects according to their d efiniteness in focus sentences. They also completed an interpretation task, which consisted of a Picture Verification Task to test for the interpretation of overt and null subject pronouns. While all properties examined are related to the null subject sett ing of the NSP in Italian (English is a non null subject language), properties such as focus implicate the syntax pragmatics interface. Regarding subject use, Belletti et al. found that the near native speakers of Italian overused overt subjects in context s where null subjects are felicitous to a significantly greater extent than monolingual controls. This finding is consistent

PAGE 43

43 with the patterns of interpretation of subject pronouns in forward and backward anaphora contexts: overt subject pronouns were inte rpreted as co referential with the matrix subject at a significantly higher rate by the near native speakers than by the control group. Additionally, the near native speakers underused postverbal subjects: Subject Verb Object was the most frequently produc ed word order. Since the near native speakers used null subjects, but seemingly did not converge on the distribution of overt and null subject pronouns in Italian, it appears that the syntax pragmatics interface was at fault, not syntax proper (i.e. the pa rticipants had successfully reset the NSP to the null subject setting in Italian). These findings led Belletti et al. to claim that i nterface conditioned properties, such as the use of overt and postverbal subjects, are somewhat problematic for the near na tive speakers due to simultaneous accessibility of more than one grammatical system (which makes more options available to second language speakers than the monolinguals have at their disposal). Finally, Tsimpli et al. (2004) tested for attrition effects i n native speakers of Greek or Italian who had lived in Britain for a minimum of 6 years and were near native speakers native language on a regular basis. Speci fically, this study tested the production and interpretation of overt and null subjects and for preverbal and postverbal subjects. Participants completed one production task (Headlines Task) and one interpretation task (Picture Verification Task). The purp ose of the Headlines Task was to examine the production of preverbal and postverbal subjects in all focus contexts. Participants were asked to form a sentence consisting of a given verb, a noun phrase and an adverbial phrase that were presented to them ran domly on a computer screen. The phrase had to begin with The phrases were presented along with a picture depicting the action to be described by the participant. The second task, the

PAGE 44

44 nterpretations of overt and null pronominal subjects. Participants were presented with a bi clausal token sentence and three pictures and then asked to point to the picture that accurately described what was said in the token sentence. The second token typ e, which tested for knowledge of new versus old information, contained two token sentences with the second sentence including a singular indefinite subject in either preverbal or postverbal position. Following Sorace (2000), Tsimpli et al. predicted that syntactic attrition primarily affects morphosyntactic features that are interpretable at the LF (2004, p. 263). Results from the two tasks were interpreted as provi ding evidence for the claim that narrow syntax is unaffected, but interface properties are vulnerable to language attrition. 2.3.2.2 Word order Next, word order is examined in L2 learners. The first study, Lozano (2006) shares some crossover with the stu dies reviewed in Section 2.3.2.1 since it examines distribution of SV/VS word order. However, Lozano (2006) leaves aside the topic of null subjects and instead focuses solely on word order. The second, Dominguez (2007), also investigates word order as a pr oduct of two focus types. Similarly, Hertel (2003) examines SV/VS word order as it relates to broad and narrow focus. The authors conclude that syntax may be in place before and is less subject to vulnerability than interface conditioned properties. Lozano 's (2006) investigation tested advanced adult second language learners of Spanish (native speakers of either English or Greek) for the distribution of SV and VS word orders. These word order alternations are determined both by formal syntax and by pragmati cs (presentational focus). Lozano predicted that in unfocused contexts (i.e. constrained by syntax only), the second language speakers would show native like knowledge but that in focused t intuitions as these

PAGE 45

45 Acceptability Judgment Task with paired target sentences and a 5 point Likert scale. The results of the experimental task demonstrated that the second language speakers converged on the syntax of Spanish word order for the properties tested, yet diverged (by showing optionality) with discursive focus. That is, the second language learners and native speakers alike significantly preferred SV to VS word order with unergative verbs, while they significantly preferred VS word order to SV with unaccusatives verbs. However, in presentational focus contexts, the second language learners accepted both SV and VS word order regardless of the verb type, s howing optionality. Lozano thus concludes that syntax is in place before interface conditioned knowledge. Domnguez (2007) also tested L2 learners of Spanish (native speakers of English) for convergence on SV and VS word order, but in both information focu s and contrastive focus contexts. Domnguez predicted that if learners were able to restructure from their native English grammar, which does not allow the SV/VS word order alternation, they would produce and accept SV and VS word order in the correct cont exts. The participants, whose proficiency ranged from advanced to intermediate to low, completed two tasks. The first was a Contextualised Production Task in which participants orally responded to a question (context was provided) and the second, an Accept ability Judgement Task, in which participants rated both possible word orders on a 2 to +2 scale according to a context. The results of the production task showed that the SV word order was preferred by participants regardless of their proficiency level. Nonetheless, the advanced group employed a cleft construction to differentiate between contrastive and non contrastive focus. Similarly, the SVO word order was preferred in the acceptability task, but the advanced group accurately accepted VSO word order, rejecting the

PAGE 46

46 SVO word order in narrow focus contexts. As some participants produced and accepted VS word order, which is not transferable from their L1, the results of Domnguez (2007), like Lozano (2006), point to earlier convergence on syntax than on th e discourse distribution of word order with a positive correlation between proficiency and convergence. In a similar study, Hertel (2003) tested native speakers of English who were learning L2 Spanish for sensitivity to the SV/VS word order alternation as a function of discourse and also verb class as in Lozano (2006). Unlike accusative verbs, which appear postverbally in both broad and narrow focus contexts, unergative verbs are split, appearing preverbally in broad focus contexts and postverbally in narro w focus contexts (Contreras 1976; Su er 1982; Ord ez 1997; Zubizarreta 1998). Assuming that syntax is in place before related discourse knowledge, Hertel predicted that learners would show gradual knowledge of Spanish word order alternations, and that the y would demonstrate sensitivity to the word order alternations as a product of verb type before focus. Participants of beginner, low intermediate, high intermediate and advanced proficiency completed a contextualized production task in which they were aske d to respond in writing (which obscures potential intonational stress placed on subjects in English) to a broad or narrow focused question. As predicted, the results showed that the advanced learners demonstrated sensitivity to word order according to verb class. Additionally, the advanced and intermediate learners showed knowledge of the word order alternation in focus contexts. While not discussed by Hertel, the fact that the intermediate learners were shown to have converged upon the discourse use of the SV/VS word order alternation but did not display this same sensitivity in the syntactic condition, provides evidence against the prediction that sensitivity to syntax emerges before sensitivity to discourse.

PAGE 47

47 2.3.2.3 Null subjects The following two secti ons describe some empirical work investigating the distribution of overt and null referential subject pronouns as well as anaphora interpretation. Montrul and to restructure from a non null subject grammar to a null subject grammar (i.e. Spanish in the case of Montrul and Rodrguez Louro and Italian in Sorace and Filiaci (2006)). Across both studies, evidence suggesting that the participants demonstrated null subj ect grammars was found; however, divergence obtained with respect to interface conditioned properties related to the distribution and interpretation of referential subject pronouns. Montrul and Rodrguez Louro (2006) investigated knowledge of the morphosyn tax and discourse dependent distribution of subject pronouns in native English speaking L2 learners of Spanish. Participants were divided into three proficiency groups (intermediate, advanced and near native) based on a proficiency measure and an oral inte rview that was rated by two native Little Red Riding Hood While Montrul and Rodrguez Louro acknowledged the potential limitations of only using one product ion task, they justify its implementation since it allowed them to examine properties related to the syntax of null subjects (i.e. subject verb agreement, referential and expletive subjects, preverbal and postverbal subjects, etc.). Data related to the dis course distribution of subject pronouns was coded according to topic maintenance (null subjects are licit), topic shift and focus (overt subjects are licit). Both the intermediate and advanced groups produced more illicit overt subjects than the near nativ e and native control groups (no significant difference was found here). Interestingly, the advanced and near native groups produced more illicit null subjects as compared to the native control group (and interestingly, the intermediate group).

PAGE 48

48 In a diffe rent language pairing (near native adult Italian L2 learners, L1 English), Sorace and Filiaci (2006) investigated knowledge of overt and null subjects in. The near native speakers began learning Italian after puberty and had lived in Italy for a minimum of 1.5 years. proficiency test (1996). The near native speakers, along with a native Italian control group, completed an off line Picture Verification Interpretation Task designed to test the interpretation of null and overt subjects in the context of forward and backward anaphora resolution. Sorace and Filiaci predicted that the near native group would differ only in their interpretation of overt pronouns (interpreti ng them as being co referential with the matrix clause subject), but not null pronouns and that the near native group would differ from the control group more on backward anaphora tokens. The results showed that the near native speakers interpreted null su bject pronouns in both forward and backward anaphora tokens as the native speakers did. However, the near native speakers differed from the native speaker group regarding their interpretations of overt subject pronouns in both forward and backward anaphora tokens. Specifically, they interpreted the overt subject pronoun as co referential with the matrix clause subject significantly more often than the native speaker group. Based on these findings, Sorace and Filiaci claimed that the near native speakers dem onstrate having acquired the syntax of subjects in Italian, yet It is necessary to acknowledge three other studies that exa mined subject distribution, importantly in bilinguals of two null subject languages. Both Bini (1993) and Margaza and Bel (2006) examined native Spanish grammar (Italian and Greek, respectively). B ini examined spontaneous speech data from eleven

PAGE 49

49 beginning proficiency and seven low intermediate L1 Spanish speaking learners of Italian. Margaza and Bel (2006) employed two experimental tasks in their study examining 10 intermediate proficiency and 9 adv anced proficiency L1 Spanish speaking learners of Greek. The first, was a cloze task and the second a written production task. The authors reported that both participant groups overused overt subject pronouns, despite the fact that their L1 would provide f acilitative transfer. Similarly, Garca Alcaraz and Bel (2011), examined 10 L1 Dariya speaking learners of Spanish and 10 Dariya Spanish bilinguals. Dariya, like Spanish, is a null subject h respect to convergence on referential subject pronoun distribution. Two important findings obtained from this study. First, while null subject pronouns were found to accurately maintain a discourse referent, overt subject pronouns had a more varied funct ion than that of the native speakers, which was to reintroduce a referent. This result differs from that of Lozano (2002) and Prada P rez (2009), which are discussed below in Section 2.3.4.1. Second, evidence of L1 transfer was not found in the L2 learner group, whereas the bilingual group seemed to perform more native like than the L2 group (although the authors claim that this knowledge cannot be said to be equivalent to that of the control group). While these studies provide results on unique language pa irings and all report overuse of overt subjects, it is important to point out that the linguistic and educational system of Spain presents a potential caveat: the speakers of Bini (1993) and Margaza and Bel (2006) were more likely than not third language ( L3) learners of Italian and Greek, not L2 learners. This is so because of the requirement to study an L2. Similarly, Garca Alcaraz and Bel (2011) disclosed that, due to their geographic location, the participants of their study were also exposed to Catala n and are, thus, trilingual. As this dissertation focuses on L2 acquisition and bilingualism, these studies will not be reviewed at length.

PAGE 50

50 2.3.2.4 Subject and object expression The ensuing paragraphs review studies investigating subject and object expres sion, the availability of which is dependent upon narrow syntax while the distribution is dependent upon the syntactically and pragmatically defined contexts. While Montrul (2004) examined heritage speakers of Spanish and Tsimpli and Sorace (2006) examined adult naturalistic learners of Greek, both examined the oral production and ultimately found divergence at the interfaces. Montrul (2004) examined subject and object expression in 24 Mexican American adult heritage speakers of Spanish living in the Unite d States. Subject and object expression were examined since these properties are tied to narrow syntax (e.g. the mere availability of both overt and null subjects in Spanish; object clitics), but their distribution is dependent upon the syntax semantics an d the syntax pragmatics interface. It was hypothesized that the participants would demonstrate knowledge of null subjects and object clitics in Spanish, but that they would demonstrate variability on interface conditioned distribution of related properties Participants completed an Oral Production Task in which they recounted the story Little Red Riding Hood. The results indicated that the narrow syntax had not been negatively affected by English since, for example, the heritage speakers used both overt an d null subjects as well as preverbal and the interface, however, was provided in that the lowest proficiency group used more overt subjects than null subject s and that heritage speakers were not accurate in the use of redundant overt subjects and illicit null subjects. Divergence was found regarding the distribution of the differential object marker a and the inherent dative case possession structure; hence th e data indicated that interfaces are vulnerable to linguistic influence or loss. Montrul claimed that the bilinguals resorted to the least costly semantic and pragmatic option offered by the two languages. She also claimed that the semantic and pragmatic f eatures of Spanish were eroded

PAGE 51

51 and that the heritage speakers have a reduced grammar that converges on the morpho syntactic characteristics of English. Finally, Tsimpli and Sorace (2006) tested for knowledge of the syntax semantics and the syntax pragmati cs interfaces of subjects and object focusing in adult native Russian speakers who were naturalistic learners of Greek. Participants completed a 10 20 minute long oral interview in which they described their background, told a story based on a set of eight pictures, completed two instruction giving tasks and gave a general discussion of life topics. Tsimpli and Sorace differentiated between internal interface conditioned properties and external interface conditioned properties, predicting that the former wo uld cause fewer problems at the advanced proficiency level than the latter, due to the interpretability of the features involved in the study. Divergence was predicted at the external interface due to the interference of discourse factors that regulate the distribution of overt and null pronouns in L1 Russian (which are different from those responsible for the distribution in Greek). Specifically, Tsimpli and Sorace predicted an overuse of overt subject pronouns. However, the authors found that all three pr oficiency groups overused null subject pronouns, a fact that cannot be attributed to first language transfer since Russian is a non null subject language. With respect to the internal interface, Tsimpli and Sorace found that all three participant groups we re aware of and used obligatory verb raising in focus contexts only (i.e. not in topicalization contexts, which is correct). Still, clitics were problematic for all groups. Tsimpli and Sorace conclude that no developmental pattern is seen with either inter face conditioned property and that external interface conditioned properties are more problematic for second language learners. 2.3.3 Summary of Divergence Data The child language acquisition studies (both monolingual and bilingual) described in Section 2 .3.1 investigated the distribution and interpretation of overt and null subject pronouns

PAGE 52

52 and in languages/language pairings involving Romance languages (Catalan and Spanish monolinguals, English Spanish bilinguals, English Italian bilinguals, Spanish Itali an bilinguals,). The general consensus of these studies is that interface conditioned properties are more problematic than narrow syntax for child monolinguals and bilinguals alike, thus lending support, in the cases of child bilinguals, for the IH. Howeve r, language pairing effects were found in Serratrice et al. (2009) (and also in Serratrice et al. (2004), which examined an English Italian child), suggesting that other factors may affect convergence. The adult language acquisitions studies (covering adul t second language learners, heritages speakers and first language attriters) described in Section 2.3.2 investigated such properties as SV/VS word order (English Spanish and Greek Spanish L2 learners; Italian English and Greek English attriters), distribut ion of referential subject pronouns (English Italian and Russian Greek L2 learners; L3 Spanish Italian learners, L3 Spanish Greek learners, L3 Dariya Spanish learners; Italian English and Greek English attriters), and argument expression (Spanish English h eritage speakers) As with the child studies, the studies largely found evidence supporting the IH in that, generally speaking, properties involving only the narrow syntax were less problematic while interface conditioned properties were more problematic. Differently from the child language acquisition studies, many adult acquisition studies reported an overuse of overt subject pronouns. Still, some evidence suggesting convergence on interface conditioned properties was found. For example, Sorace and Filia ci (2006) found that the near native speakers interpreted null subject pronouns in both forward and backward anaphora tokens as the native speakers did; likewise, Montrul and Rodrguez Louro (2006) found no significant differences between the near native s peakers and the native control group with respect to

PAGE 53

53 pragmatically conditioned distribution of overt and null subject pronouns in their Oral Elicitation Task. Sections 2.3.1 and 2.3.2 described both child and adult second language acquisition studies whos e results showed divergence on interface conditioned properties related to subjects. second language studies showing contradictory evidence. That is, the fol lowing studies report convergence on interface conditioned properties. 2.3.4 Convergence on Interface conditioned Properties Related to Subjects Just as evidence has been found suggesting that interface conditioned properties are subject to inevitable di vergence in bilingual grammars, much empirical research in the generative paradigm has also provided convincing evidence that adult bilinguals demonstrate native like knowledge of interface conditioned properties related to overt and null subject pronouns. The following sections describe some of this research. 2.3.4.1 Subject distribution The subsequent sections summarize experimental investigations examining properties related to the NSP. Rothman (2008a) and (2009) both examined native English speaking L2 learners of Spanish regarding their convergence on the syntax and interface conditioned convergence on similar properties in L2 Turkish. Lozano (2002) compared k nowledge of two pronominal constraints related to subject distribution in both Greek and English speaking L2 learners of Spanish while, Prada Prez (2009) examined subject expression in Spanish in contact with Catalan. According to the authors, the result s indicated that convergence on interface conditioned properties related to the distribution of subjects is possible in adult L2 acquisition. Still, a developmental pattern was seen in Rothman (2008a) and (2009) in that only the advanced

PAGE 54

54 participants demon strated convergence on interface conditioned properties. These studies, along with Prada Prez (2009) point to proficiency as a factor affecting convergence. The first study, Rothman (2008a), examined an earlier version of the IH (Sorace 2006) in two expe rimental participant groups, intermediate and advanced second language learners of Spanish (native speakers of English). Specifically, knowledge of the distribution of overt and null subject pronouns in Spanish was tested. Participants completed a Pragmati c Felicitousness Judgment Task and a Pragmatic Context Sentence Translation Task. The tokens of the first task were presented via recordings of native speakers of Spanish (no visual presentation of the tokens was made available to the participants), thus j udging their ability to respond to real time speech. Upon listening to each token, the participants rated its felicitousness on a 5 point scale. In the second task, participants translated an English sentence to Spanish. The results showed a developmental path: while both the intermediate and advanced learner groups demonstrated convergence on the syntax of subjects in Spanish, only the advanced learner group demonstrated mastery of the pragmatic distribution of overt and null subjects. Since the more advan ced second language learners demonstrated knowledge of the pragmatically conditioned distribution of overt and null subjects, Rothman claimed that the results provided evidence against the IH. It was argued that interface conditioned properties, such as th e distribution of overt and null subject pronouns in null subject languages, are inherently more complex; thus convergence on such properties is delayed, but not impossible. distribution of overt and null subjects in intermediate and advanced second language learners of Spanish (native speakers of English). Participants completed a Co reference Interpretation Task which tested for knowledge of the Overt Pronoun Constraint (OP C; Montalbetti 1984), a

PAGE 55

55 Pragmatic Context Matching Felicitousness Judgment Task which tested for knowledge of the pragmatically conditioned distribution of overt and null subjects and a Pragmatic Context Translation Task which tested for production of over t and null subjects in specific contexts. On all tasks, the advanced participants performed in a native like fashion (no significant differences were found). Regarding the intermediate group, 28 of the 38 participants showed sensitivity to the OPC, which R othman takes as sufficient evidence of a null subject grammar. However, in the Pragmatic Context Matching Felicitousness Judgment Task, the intermediate participants rated infelicitous overt subjects as felicitous as well as infelicitous null subjects as f elicitous significantly more often than the native control group and the advanced group. Finally, on the Pragmatic Context Translation Task, the intermediate participants used more null pronominal subjects in contexts requiring overt subjects (i.e. Contras tive Focus and answers to topic wh questions). They also used more overt pronominal subjects in contexts requiring null pronominal subjects (non focus contexts and yes/no questions). The intermediate participants differed significantly from the advanced an d control groups. Their use of overt versus null subjects was deemed to be insensitive to the discourse. Like Rothman (2008a), the results of the three tasks employed in Rothman (2009) indicated a developmental path and demonstrate that vulnerability at th e interfaces is neither inevitable nor permanent. The advanced participant group and 28 of the 38 intermediate participants demonstrated syntactic knowledge of the null subject setting of Spanish. The advanced participant group also demonstrated knowledge of the pragmatically conditioned distribution of overt and null pronominal subjects in Spanish, thus demonstrating convergence on interface conditioned properties. As research on first language acquisition has shown (e.g. Carroll 1983; Brownell, Carroll, R ehak & Wingfield 1992; Austin, Blume, Parkinson, Nez del Prado & Lust 1996 among others), Rothman concluded that the syntax of

PAGE 56

56 subjects is acquired before pragmatics and that, while external interface conditioned properties were perhaps more difficult, t hey were not subject to inevitable vulnerability. Grel (2006) investigated resetting of the NSP in adult second language learners of Turkish (native speakers of English that had lived in Turkey for a minimum of 10 years). Turkish is a null subject langua ge, however the Grel claims that the constraint on overt pronoun binding is not a result of the OPC (Montalbetti 1984, p. 279) as in Romance languages, but rather is a ning to that of Romance languages in that overt subjects are used to introduce new topics or a change while null subject pronouns are used when there is no topic change or contrastive focus. Twenty eight second language speakers of Turkish and a native control group completed four experimental tasks: a Picture Selection Task, a Written Interpretation Task, a Truth Value Judgment Task and a Picture Identification Listening Task. The results of the picture selection task, which tested for pragmatically conditioned uses of overt and null subjects, suggested that the second language participants converged on th e distribution of null and overt subjects. While the participants converged on the syntax and pragmatics of null and overt subjects, they failed to acquire the binding properties of the overt pronoun o, presumably due to native language transfer interferen ce. Thus, evidence against the IH was found for a unique language pairing: English Turkish. Next, Lozano (2002) compared two advanced L2 groups learning Spanish for knowledge of the OPC and a related property, Contrastive Focus Constraint (CFC). Unlike the L1 English

PAGE 57

57 group learned English prior to Spanish (2006, p. 59) Participants completed an Acceptability Judgement Test (AJT) in which they determined the acceptability of two target sentences, one containing an overt subject pronoun and one containing a null subject pronoun, in light of a context. The results from th e AJT showed that both the L1 English and L1 Greek speakers differentially judged overt and null subject pronouns in light of OPC tokens to a statistically significant degree and matched the native speakers. The same distinction was made with the CFC token s, but the L1 English group differed significantly from the native speakers with respect to the target sentences containing null subjects. Thus, these results show convergence on an internal interface conditioned property (OPC) and an external interface co nditioned property (CFC) by the L1 Greek speakers, but only the former in the case of the L1 English speakers, indicating that language pairing may play a role in convergence. Finally, Prada Prez (2009) explores the oral production of 11 L1 Spanish and 1 2 L1 Catalan bilingual speakers of Catalan Spanish in Minorca, Spain specifically with respect to subject distribution. Catalan and Spanish, like Greek and Spanish in Lozano (2002), are both null subject languages and are typologically very similar. Speech samples were also collected from 12 monolingual Spanish speakers and 12 Catalan dominant bilinguals to provide a basis of comparison. The data obtain from oral interviews showed, first, that production of overt subject pronouns in monolingual Spanish does not differ significantly as compared to Catalan. Nor did monolingual Spanish speakers. Prada P rez did find, however, that L1 Catalan speaking bilinguals demonst rated overt subject production in rates more similar to Catalan dominant bilinguals than the L1 Spanish speaking bilinguals. Proficiency effects, like those found in Rothman (2008a, 2009), were also reported in that high proficiency bilinguals from both gr oups

PAGE 58

58 produced overt subject pronouns in rates more similar to the native speakers than those with lower proficiency in the language tested. Thus, this set of studies examining subject distribution in L2 learners and bilinguals for whom the language pairin gs was either facilitative or non facilitative demonstrated that convergence on interface conditioned properties is not impossible (albeit in offline methodologies). In some cases, a developmental path was noted since proficiency was shown to positively co rrelate with native like knowledge. The following section reviews research conducted on a property related to subjects in language pairings not discussed yet herein. 2.3.4.2 Word order Three recent experimental studies that assessed knowledge of word orde r alterations in adult L2 learners are discussed below. Bohnacker (2010) examined German convergence on the prefield in Swedish. Prada Prez (2010) investigated word order as a function of focus and predicate type in native Catalan speak ing L2 learners of Spanish. Both used offline tasks. Differently, Hopp (2009) employed both offline and online tasks in his study of L2 acquisition of scrambling in German by native speakers of English, Dutch or Russian. While these studies differ with res pect to the bilingual populations, the properties studied and the experimental tasks employed therein, they are similar in that the results of these studies provided evidence that interface conditioned properties are not subject to inevitable divergence in adult L2 learners. Bohnacker (2010) examined properties related to word order at the syntax discourse interface in adult second language learners of Swedish (native speakers of German). Specifically, the prefield (what precedes the verb) was examined sin ce it is taken to be the element that situates the utterance in the overall discourse. Participants were classified as (largely naturalistic) adult second language learners of Swedish (although all participants had 7 9 years of English

PAGE 59

59 instruction and some also had knowledge of French or Latin). Bohnacker also placed the learners at the advanced proficiency level based on the fact that they had all passed a language proficiency test in order to enter Swedish university prior to the experimental tasks. The d ata used in the analysis consisted of naturalistic written and spoken data which were collected at 3 year intervals. The results showed that the second language learners initially demonstrated more German style prefield patterns using fewer expletives and more rhematic (new) information in the preverbal position. However, at the second and third testing interval, the second language participants demonstrated development towards the target grammar in that a substantial increase in expletive subjects, clefts and thematic information (or old information) was witnessed in the prefield. Since the participants demonstrated syntactic knowledge of related properties (namely the verb second (V2) nature of Swedish), yet demonstrated early difficulties with the associa ted interface conditioned properties, Bohnacker claimed that the results are in line with previous research showing convergence on syntax but divergence on interface properties. Still, Bohnacker interpreted the overall findings as suggesting that, while in terface conditioned properties such as word order may be more difficult, convergence is possible as proficiency increases. Building on work on English Spanish bilinguals (e.g. Hertel 2003; Zapata, Snchez & Toribio 2005; Lozano 2006; Domnguez & Arche 200 8), Prada Prez (2010) examined the effect that broad and narrow focus and verb type (unergatives and unaccusatives) have on word order in Spanish and Catalan. Word order as a function of focus, which is an external interface conditioned property, behaves similarly in both Catalan and Spanish. However, as a function of verb type, which is an internal interface conditioned property, differences obtain between the two languages. Specifically, preverbal subjects are preferred significantly more than postverbal subjects with unergative verbs in broad focus contexts in Catalan, yet this preference does not

PAGE 60

60 obtain in Spanish. With respect to the IH, an interesting prediction falls out of these facts. Since the IH predicts vulnerability at external interfaces, diff erences are predicted with respect to word order as a function of focus but not verb type even though Spanish and Catalan behave similarly with focus, but differ with respect to verb type which is an internal interface conditioned property. Eighteen monoli ngual Spanish speakers and 22 Catalan Spanish bilinguals (16 L1 Catalan, 6 L1 Spanish) completed an Oral Contextualized Preference Task 11 The results of this task showed that the bilingual participants demonstrated more vulnerability with subject verb word order as a function of verb type as compared to focus. This finding provides evidence against the IH in that the bilinguals were found to converge on the external interface conditioned property tested by Prada Prez but showed more differences with respe ct to the internal interface conditioned property. Prada Prez suggested that previous findings (e.g. Tsimpli & Sorace 2006) may be a product of language pairing, and not necessarily vulnerability at the external interface. Hopp (2009) explored convergence on word order alternations as a function of the syntax discourse interface in adult L2 learners of German. Specifically, the study examined convergence on scrambling in German by advanced to near native proficiency level native speakers of English, Dutch or Russian. The language pairings were chosen in order to tease apart native language transfer effects from interface vulnerability effects. Participants completed an offline Acceptability Judgment Task (AJT) and an online Self Paced Reading Task (SPRT). T he first task tested for knowledge of the syntax of scrambling in German and its interaction with the syntax discourse interface; the second task assessed online processing of scrambling. Regarding 11 Prada Prez (2010) also reports on data from a Sociolinguistic Interview completed by Spanish monolinguals, Spanish dominant Spanish Catalan bilinguals, Catalan dominant Catalan Spanish bilinguals and Catalan monolinguals. While Cat alan significantly prefers preverbal subjects, the Spanish preference did not reach significance.

PAGE 61

61 the online task, Hopp was particularly interested in deter mining whether or not the second language learners demonstrated slower reading times on infelicitous tokens as compared to felicitous tokens (which, according to Hopp, would indicate that the learners simultaneously map discourse information onto syntax as they build phrase structure in real time comprehension). For the AJT, Hopp found that the Russian German learners (regardless of proficiency) made native like distinctions regarding the syntax of scrambling in German. Additionally, the Russian German lear ners were sensitive to the discourse conditioned context with which the tokens were presented, indicating knowledge of the interface conditioned use of word order in German. The advanced proficiency native speakers of English and Dutch, on the other hand, demonstrated divergence on the syntax of scrambling as well as the discourse conditioned use of scrambling. Both the near native proficiency speakers of English and Dutch demonstrated knowledge of the syntax of scrambling in German, but only the native Eng lish speakers were sensitive to the discourse conditioned contexts. Regarding the SPRT, both proficiency levels of native speakers of English and Russian demonstrated significantly different reading times for the word order alternation according to the dis course conditioned context than the native control group did. Neither proficiency level of native Dutch speakers showed this contrast. Hopp concluded that the overall findings indicated that properties requiring the integration of syntactic and discourse i nformation do not represent an insurmountable difficulty for adult second language learners. This was true for offline comprehension as well as online processing. Thus, the common finding obtained from the three studies described above that examined word order alternations is that external interface conditioned properties were not shown to be points of inevitable divergence in adult L2 speakers. In Hopp (2009), this was shown in both offline and online tasks, and important methodological point for the clai ms of the IH.

PAGE 62

62 interface conditioned properties may be more difficult (i.e. converged upon at higher proficiency levels). 2.3.5 Summary of Convergence Data The adu lt language acquisitions studies described in Section 2.3.4 examined such properties as overt and null subject distribution (English Spanish, English Turkish and Greek Spanish L2 learners; Spanish Catalan bilinguals), and word order alternations (German Sw edish and English, Catalan Spanish bilinguals, Dutch or Russian German second language learners). The results of these studies demonstrate that while not all L2 learners, specifically those at lower proficiency levels, converge on properties at the externa l interface, these properties do not represent an inevitable source of divergence. That is, the studies outlined above highlight advanced and near perhaps more formally comple x and require integration of syntactic and pragmatic/discourse information. Similar results were found in other IH studies investigating convergence on dislocations and modality alternations. Specifically, Donaldson (2011a, 2011b) examined English speaking near native L2 learners of French for knowledge of left and right dislocation and found that these learners converged upon the interface conditioned properties examined. Ivanov (2009), which examined English speaking intermediate and advanced L2 learners of Bulgarian, found that these learners converged on CLLD, an external interface conditioned property. The English speaking L2 learners of Spanish in Slabakova et al. (2012) were tested for knowledge of CLLD and Fronted Focus. Like the previous studies, th e authors reported evidence demonstrating convergence on the external interface conditioned properties tested therein. Likewise, in their examination of mood alternations in English speaking L2 learners of Spanish, Iverson et al. (2008) found that, while t he syntax discourse interface was more difficult, the data

PAGE 63

63 demonstrated that L2 learners could acquire the distinction between the subjunctive and indicative mood with negated epistemics in advanced stages of acquisition. Taken together, these studies prov ide evidence against the IH (Sorace 2011) in that they demonstrate convergence on external interface conditioned properties. 2.3.6 Criticisms of Previous Behavioral Research The empirical studies examined in Sections 2.3.1 and 2.3.2 provided evidence for the IH in that data were found suggesting that external interface conditioned properties are subject to divergence in child and adult second language acquirers. However, Section 2.3.4 points to the opposite conclusion in that the empirical studies examined therein show convergence on external interface conditioned properties. Still, both bodies of results indicate that related syntactic properties are not the source of any delays or divergence on external interface conditioned properties. This finding is in line with the claims of the most recent version of the IH (Sorace 2011) since it claims that convergence on syntax is possible. Equally, these empirical studies seem to indicate that as proficiency increases, learners improve (if not converge) on external interface conditioned properties. Without a doubt, the empirical studies outlined in Section 2.3 have made significant contributions to the field of child and adult language acquisition. While not all studies explicitly test the predictions of the IH, th eir results can be, and have been, analyzed in light of the IH, especially the earlier versions. To recap, the most recent instantiation of the IH differs from earlier versions in that it claims that divergence on external interface conditioned properties is due to the presence of two linguistic systems, which consequently results in limitations in attentional resources related to inhibitory control (Green 1986, 1998). That is, since bilinguals must divide cognitive resources to inhibit the non relevant lan guage, working memory, executive function and attentional resource allocation are burdened, causing processing difficulties where

PAGE 64

64 syntactic information must integrate with pragmatic/discourse information. Sorace (2011) has also made explicit that the claim s made by the IH are valid for near native speakers only. With these claims in mind, some shortcomings of the research described in Section 2.3 in terms of their applicability to the current version of the IH, are described. Specifically, three shortcomin gs will be pointed out: lack of (standardized) proficiency tests to ensure near native proficiency level in some of the studies, relative lack of variety of language pairings and use of largely offline experimental tasks which cannot measure the processing ability of participants. Regarding the proficiency level of participants that took part in the studies described in Section 2.3, only the adult studies will be considered since the participants of the child monolingual and bilingual studies described in Section 2.3.1 should not be expected to have adult like proficiency yet. Thus, of the adult studies that found divergence at the external interface described in Section 2.3.2, some (Belletti & Leonini 2004; Tsimpli & Sorace 2006) either did not employ a p roficiency test or do not report on it. Of the studies that employed a proficiency test, it is not clear that all the participants of these studies have truly obtained near native proficiency level. For example, Belletti et al. (2007), Sorace and Filiaci ( 2006) and 12 This proficiency to evaluated by two native speaker judges. The native sp eaker evaluation, which is based on an 18 overall impression of near one were judged in t he near native range by both native speaker judges in the Belletti et al. (2007) study. However, the participants of Sorace and Filiaci (2006) and Tsimpli et al. (2004) 12 In the case of Belletti et al. (2007) and Sorace and Filiaci (2006), the measure was adapted for Italian.

PAGE 65

65 were not held to the same standard. That is, a lower cutoff point for near native profi ciency was and Genesee, we did not collect the data in a bilingual country and it is therefore more difficult to find speakers that satisfy the strictest cri teria for L2 near et al. (2004) offer a similar argument, claiming that near native proficiency level participants are harder to find in monolingual countries. Differently from these studies, Prada P rez (2009, 2010) emp loyed a self rating proficiency measure on a scale of 1 7. As might be expected from childhood bilinguals, the participants self rated highly. Additionally, the type of proficiency measure employed in these studies, oral elicitation, must be carefully cons idered. Standard expectations of the field call for three components deemed to be key to language proficiency measures validity, reliability and efficiency (see Bachman & Palmer 1996; Alderson 2000; Bachman 2004; Weir 2005; Hulstijn 2010). A test is deemed valid if it measures what it seeks to measure, reliable if it is consistent and precise and, lastly, efficient if the time invested in creation, implementation, coding and analysis of the test is less than the benefit gained from its employment. Since Whi which can be impressionistic and vary from one rater to the next (although inter rater reliability tests can minimize this effect), it is reasonable to question the reliability of an ora l elicitation measure, 13 especially when the experiment examines syntax and discourse dependent properties. Additionally, while this measure results in ample oral data in a conversation setting, due to the time required to analyze oral data, this measure i s far more time consuming than other measures, and, thus, potentially less efficient than other measures. 13 as shown that this type of task may be an accurate measure for phonological proficiency.

PAGE 66

66 participants were classified as advanced and therefore we cannot determ ine if they meet the requirement of being near native speakers. Finally, the participants of Montrul (2004) are heritage speakers. Sorace (2011) specifically states that the claims and predictions of the IH have been overextended to heritage speakers since the type of speaker most typically examined are second generation heritage speakers exposed to qualitatively and quantitatively different linguistic input (but see Montrul & Polinsky 2011 for a response to this claim). Thus, non convergence on interface c onditioned properties, as compared to monolinguals, it is difficult to tease apart the effects of input differences and vulnerability at the interfaces. Montrul y The present study addresses this shortcoming by employing a proficiency measure that is rather extensively used in generative research (e.g. Montrul 2000, 2 002, 2005, 2009; White, Valenzuela, Kozlowska MacGregor, & Leung 2004; Cuza & Frank 2011; Leeser, Brandl, & Weissglass 2011; Prada Prez & Pascual y Cabo 2011, 2012; Pascual y Cabo, Lingwall, & Rothman 2012; Slabakova et al.2012). Employing standardized pr oficiency measures, which can only be labeled as such when the same measurement are employed using the same scoring criterion, is important for the field of acquisition research, but especially when testing a hypothesis such as the IH. This is so because a cross study comparison of convergence on interface conditioned properties is not fruitful if variables like proficiency are not controlled for. The second shortcoming of the studies described in Section 2.3.2 and Section 2.3.4 is the relative lack of vari ety regarding language pairings. Of the 21 adult acquisition studies described in these two sections, 71% (15/21) include a Germanic language (Dutch, English, German or Swedish) as

PAGE 67

67 either the first or second language, and of these 134 studies, 93% (14/15) include English. Another 71% (15/21) include a Romance language (Italian or Spanish) as either a first or second language. Only 42% (9/21) include a non Germanic, non Romance language (e.g. Greek, Russian or Turkish). While all of these studies are informa tive and relevant to the field of adult second language acquisition, as has been pointed out by Rothman and Slabakova (2011), a variety of language pairings must be tested in order for the results to be generalizable. If not, the results (whether demonstra ting divergence or convergence) could simply be a product of the particular language pairing as pointed out by Prada P rez (2010). Thus, researchers must test the predictions of the IH not only on the language pairings on which the claims were based and th at differ with respect to the property tested, but also on language combinations in which the property is similar (as in Bini 1993; Lozano 2002; Belletti & Leonini 2004; Margaza & Bel 2006; Hopp 2009; Prada P rez 2010; Garc a Alcaraz & Bel 2011) to disenta ngle the effect of L1 transfer. This dissertation addresses this issue in that it studies a language pairing that shares the same parameter value for the property tested (i.e. null subject languages). Differently from the studies listed above which have al so examined this type of language pairing, the present study is unique in that, while Farsi and Spanish are both syntactic discourse configurational null subject languages that license pro and also share analogous discourse distribution of subject pronouns for the contexts tested herein, the underlying syntax of the two languages differs according to Karimi (2005). Additionally, Farsi Spanish is a completely novel language pairing. The final shortcoming of the adult acquisition studies described in Section 2.3 is the near across the board implementation of offline 14 behavioral methodology, either in the form of offline judgment, interpretation or production tasks, the only exception being Hopp (2009). 14 Offline tasks, as opposed to online tasks, are untimed (i.e. they do not gather reaction time data).

PAGE 68

68 While the field of generative second language acquisition has traditionally employed offline behavioral tasks of the pen and paper or picture matching sort when examining adult comprehension, these are not necessarily appropriate tasks for all current research questions. Collaborative work between psycholinguist ics and generative linguistics has highlighted the superior elucidatory value of online tasks (such as self paced reading or eye tracking tasks among others) for current generative research questions, especially those examining language processing. As desc ribed in Marinis (2010), a distinguishing characteristic of offline behavioral and online tasks is that the former do not control for, nor do they measure, the time each participant requires to read a sentence, respond to or complete a certain task. Marini s claims that after they have listened to the entire token, which he claims taxes working memory in that three steps are required in order to complete the task. First, the participant must process the token in real time as it is presented to them. Second, the token must be maintained in working memory. Finally, the participant completes the item by answering a yes/no question, matching the target sentence with a picture, or whatever the task requires of the participant in order to measure their answer. A natural consequence of this type of task is that participants with higher working memory will likely perform better, although this result may not be indicative of their underlying linguistic competence, but rather of their higher working memory capacity. A further disadvantage to offline tasks is that participants are normally not limited by time constraints, which allows for the possibility of employing domain gen eral skills and metalinguistic awareness, a problem that can lead to inaccurate conclusions about the representation of language in the participant. Finally, behavioral offline tasks do not allow the researcher to examine the temporal processing of languag e, which means that it is not possible to determine

PAGE 69

69 when and/or where ungrammaticality or infelicity is detected. Thus, the questions that can be answered with offline behavioral tasks are limited. Many online tasks, in contrast, remedy these shortcomings unconscious and automatic responses to linguistic stimuli while the stimuli are presented to the participant. Since data is obtained during stimuli presentation, it is less affected by the strain on working memory described by Ma rinis above. Even so, no task is entirely free of weaknesses. Marinis acknowledges that online measures require considerable preparation because every aspect of the tokens (length of sentence, frequency of vocabulary, etc.) must be controlled for. Addition ally, more tokens are required than for offline tasks, which increase the length of the study and, consequently, the burden on the participant. Data analysis is often more complex since two variables may require analysis s/no) as well as the reaction time (RT). Nonetheless, as the IH claims that the locus of vulnerability in near native second language speakers is an artifact of the increased processing burden experienced by bilinguals, online tasks are in the best positio n to test the claims and the predictions of the hypothesis. In the case that differences are found between monolingual and bilingual speakers, offline tasks are not able to distinguish between the potential sources of these differences (i.e. whether the di fferences obtain as a result of deficiencies in competence or in processing 15 ). Thus, while the studies described in Sections 2.3.1 and 2.3.2 found some level of divergence, these studies are not in a position to provide evidence for or against the IH (Sora ce 2011) 16 as it specifically claims that processing differences obtain as a result of the presence of two linguistic systems, which leads to limitations in attentional resource allocation (Green 1986, 1998). Section 15 Of course, this does not necessarily mean that the opposite i assumptions. For example, demonstration of similar processing may not mean that the representation is the same. 16 Of course, the same could be said for the studies that found convergence on interface con ditioned properties examined in Section 2.3.4 since they also employed offline behavior tasks.

PAGE 70

70 2.4 provides information regarding langu age processing, the parsing model adopted herein, as previous studies, as well as their limitations, are outlined. 2.4 Language Processing Section 2.3 above de scribed recent behavioral research examining convergence on interface conditioned properties while Section 2.3.6 described some shortcomings of these studies as related to the IH. The goal of the current section is to give a succinct overview of language p rocessing including the parsing model adopted herein (Section 2.4.1) and a brief summary of various types of methodologies employed in processing studies (Section 2.4.2) and some relevant findings of recent processing research (Section 2.4.3). While much r esearch has been conducted on the processing of garden path sentences and interpretive preferences (e.g. high or low attachment of relative clauses), the studies presented herein are limited to those that ose that examine the processing of grammatical versus ungrammatical stimuli. Finally, Section 2.5.4 describes some limitations of this body of research. 2.4.1 Parsing Model As stated by Felser (2005, p. 95), a principal goal of research conducted on secon d underlie L2 processing, 17 often in light of native versus non native speaker comparisons. Phillips (1996) presents two parsing models that differ in their treatment of the parser and the grammar. The first model, which he labels the PIG model (Parser is Grammar), does not distinguish between the parser and the grammar, whereas the second model does. While Phillips ultimately 17 Sentence processing as it relates to production will not be considered herein as it is not the focus of this dissertation.

PAGE 71

71 argues for the PIG Model, many psycholingu istic models assume a distinction between the grammar and the parser, perhaps as a function of their distinct roles in language comprehension (i.e. the grammar being the competence system and the parser being the performance system). Returning to the IH, i t is claimed that the simultaneous activation of the relevant grammar and the inhibition of the irrelevant grammar in adult bilinguals taxes cognitive resources. Thus, working memory and executive function systems (i.e. cognitive resources and, importantly not the grammar or the parser) are taken as the loci of divergence. It is not clear to me whether any theoretical implications pertinent to the present study fall out of the fusion or separation of the ond model aligns well with my conceptualization of language in that it separates performance (the parser) and competence (the grammar) and because the model is not problematic for the questions examined herein, I assume that the parser and grammar are sepa rate. The assumption of this model is supported by recent work by Hopp (2007), who, in an exhaustive study covering seven experiments, examined questions similar to those presented herein. Specifically, knowledge and processing of interface conditioned pro perties related to scrambling in near native L2 German speakers (the L1s were English, Dutch or Russian) was studied. Under this view, language comprehension is dependent on the grammar, the parser and a finite set of cognitive resources including executi ve function, working memory and inhibitory control (Phillips 1996) 18 As depicted in Figure 2 3 below, the grammar includes linguistic knowledge while the parser is composed of both universal and language specific parsing 18 the relevancy of these items regarding the questions examined in this study is unclear, they have been replaced with two very relevant items taken to be outside the grammar and th e parser: inhibitory control and executive function.

PAGE 72

72 mechanisms and resolution strategie s. The former is associated with the competence system, the latter with the performance system. Language comprehension = + Figure 2 3. Grammar parsing model adapted from Phillips (1996, p. 16) The available cognitive resources delimi t language processing. Each of these three components of language processing is a potential source of divergence in adult language acquisition. As described in Section 2.2, the IH does not assume that the underlying representation of language (the grammar) is divergent, nor does it make specific claims about particular processing strategies. Instead, t he IH claims that performance and interpretive differences between monolinguals and bilinguals may derive as a result of the division of finite cognitive reso urces in the latter instance only. That is, as previously described, due to competition between multiple grammars in a single mind, attentional resource allocation related to inhibitory control (Green 1986, 1998) is taxed in bilinguals, creating an extra b urden for the working memory and executive function systems. With this model of language comprehension in mind, the following section reviews several methodological paradigms employed in the examination of language processing. Grammar Lexicon Language universals Language particular properties Economy considerations Parser Structure building procedures universal l anguage specific Ambiguity resolution strategies Cognitive Resources Executive function Working Memory Inhibitory control

PAGE 73

73 2.4.2 Language Processing Met hodology Neurophysiological and behavioral tasks alike have been employed in psycholinguistic studies examining language processing, both contributing different pieces of the puzzle. On the one hand, neurophysiological tasks such as Position Emission Tomog raphy (PET) and Functional Magnetic Resonance Imaging ( f MRI) provide information regarding the areas of the brain that are activated during language processing. However, since these methods rely on changes in hemodynamic states, neither provides reliable i nformation regarding the fine grained time sequencing of language processing. Event related potentials (ERPs) and behavioral methodologies such as eye tracking, cross modal priming, self paced reading and speeded grammaticality judgment tasks, on the other hand, provide more precise information regarding real time language processing. It should be noted, though, that the granularity of precision among these methodologies varies, with ERPs and eye tracking as providing perhaps the highest level of online res olution. The following section describes previous research that employs both neurophysiological and behavioral tasks. 2.4.3 Previous Research While some previous research investigating second language processing has found differences between the regions of the brain that are recruited in language processing and/or the timing of language processing, it is not clear whether these differences are a result of processing differences only, or if other factors are involved. Contributing variables, such as proficie ncy or amount of exposure, may have an effect on second language processing, yet these variables are not always controlled for. Of course, these contributing variables may be intertwined with processing and they can be difficult to define. Referencing the parsing model illustrated in Figure 2 3 and for the purposes of the present study, proficiency is taken to be a product of the grammar and the parser, as well as being influenced by Cognitive Functions. The same difficulty in

PAGE 74

74 defining proficiency is encoun tered with respect to amount of exposure as simple questions regarding daily exposure or length of residence in an L2 environment may obscure reality. Below, I briefly review some second language processing research conducted in the neurophysiological para digm (Section 2.4.3.1) and behavioral paradigm (2.4.3.2). Section 2.4.4 discusses some limitations of this general body of research. 2.4.3.1 Neurophysiological studies Abutalebi, Cappa and Perani (2001) review several early functional neuroimaging studies examining bilingual production and comprehension. For this paradigm and those that follow, I will focus on comprehension data only since the experimental methodology employed in the present study does not include production data. Perani et al. (1996) condu cted a PET study in which a homogeneous group of nine male late bilinguals 19 of low L2 proficiency listened to stories in their native language (Italian), their second language (English) and in a third language of which they had no knowledge (Japanese). Abu talebi, Cappa and Perani found more extensive activations for native language processing as compared to second language processing and also report that the activation pattern for the second and third unknown language is the same (see Table 1 and Figure 1, Perani et al., 1996, p. 2440 2441). A similar methodology was employed by Dehaene et al. (1997). Instead of PET scanning, f MRI was used to determine the brain activation patterns of a homogenous group of eight male bilinguals 20 of moderate proficiency in se cond language English while listening to stories in their native language (French) or their 19 None of the participants were exposed to English, the second language, before age 7, and all 9 participants had studied English for a minimum of 5 years. A word translation and a sentence comprehension measure tested for their proficiency (see Abutalebi, Cappa & Perani 2001, p. 185). 20 None of the bilinguals had been exposed to English b efore age 7, and based on their completion of the same word translation and sentence comprehension measure employed in Perani et al. (1996), these bilinguals were deemed to have moderate proficiency in English.

PAGE 75

75 second language. In line with previous research, Abutalebi, Cappa and Perani found a correlation between native language processing and activation in the left tempo ral lobe. However, when the and frontal areas, sometimes restricted only to right (Dehaene et al., 1997, p. 3809). Kotz ( 2009) provides an updated review of f MRI and ERP experiments examining L2 processing, many of which examine highly proficient bilinguals. Two ERP studies and an f MRI studies indicating differences in native and non native processing are described. First, H ahne (2001) examined syntactic and semantic processing via ERPs. A second language group of 16 native speakers of Russian that had learned German after the age of 10 21 years was compared to a group of 16 native speakers of German. Stimuli included grammatic al tokens, tokens with semantic violations, tokens with syntactic violations and fillers (see Table 1, Hahne 2001, p. 254). Kotz reports that both the native and second language German groups demonstrated an N400 effect for the semantic violation tokens, w hich is expected as N400 effects are associated with integration of semantic information. However, there were differences between the two groups in that the N400 effect of the second language group was less pronounced in terms of amplitude and showed a lon ger peak latency than that of the native speakers (550 versus 750ms, respectively) Regarding the syntactic violation tokens, the native speaker group demonstrated sensitivity to syntactic violations in that, as compared to the grammatical sentences, an ea rly anterior negativity (ELAN) was elicited by ungrammatical sentences. This too is expected as ELAN effects (often with latencies between 100 300ms) are associated with word category or phrase structure violations. This was followed by a centro parietal p ositivity (i.e. a P600 effect) 21 The background of the bilingual participant s is quite variable and the averages presented are misleading if the range is considered. For example, the average time learning in a formal setting was 6 years, but the range was 2 168 months. Additionally, all but one participant had knowledge of English which could be another confounding factor.

PAGE 76

76 that peaked around 800ms 22 P600 effects are also associated with syntactic processing and may be elicited by ungrammatical sentences, garden path sentences and complex sentences. The L2 group, however, did not demonstrate dif ferences for the grammatical and ungrammatical tokens as measured by early anterior negativity. Still, like the native speakers, the syntactic violation tokens elicited a positivity, albeit with a later peak latency ( approximately 950ms). This seems to poi nt to timing differences between the two groups. Hahne, Mueller and Clahsen (2006) examined the processing of irregular and regular participial inflection and two types of plural marking on nouns via ERPs. A bilingual group comprised of 18 native speaker s of Russian 23 who were late second language learners of German participated in the study. The average age of first exposure was 17 years, although there is considerable variability (8 29 years 24 ). A self rating task served as a proficiency measure; the aver age self rating was 5 out of 6 (=high proficiency). In the participle task, a total of 50 regular and 50 irregular participles were embedded in simple declarative sentences composed of six words each. The tokens were presented word by word on a computer sc reen with the participle in sentence final position. In the noun task, a total of 480 tokens were presented to the participants. Half of the nouns were marked with the correct plural marking and half were marked with the incorrect plural marking. The nouns were embedded in sentences and always appeared in direct object position followed by an adverbial or a prepositional phrase. Regarding participle inflection, the ERP data show a regular/irregular distinction. Regularizations (i.e. the regular t suffixed to an irregular verb) elicited an anterior negativity between 250 and 600ms as well as a small parietal positivity between 600 and 1000ms compared with their correct counterparts in the 22 Although usually occurring approximately 600ms after the presentation of the stimuli, Kaan and Swab (2003) claims the effect may appear as early as 400ms, while Friederici (2002) claims it may appear between 6 00 1000ms. 23 The Russian participle and noun marking systems are substantially different than that of German. 24 A wide range of length of residence in Germany (.5 12 years) is possibly a confounding factor.

PAGE 77

77 L2 participants. By contrast, irregularizations (i.e. irregular n suf fixed to a regular verb) yielded a centrally distributed negativity between 450 and 600ms relative to the correct participle forms. With respect to plural marking, regularizations (i.e. regular s incorrectly suffixed to a noun taking n ) elicited a P600. Irregularizations (i.e. irregular n incorrectly suffixed to a noun taking s ) elicited an N400. Differently from native German speakers, no anterior negativity was found with the L2 participants for overuse of the s plural. This last finding suggests tha t overuse of the s plural was not detected as a word category or phrase structure violation in the L2 participant group as it was in the native German speaker group. R schemeyer, Fiebach, Kempe and Friederici (2005) used f MRI to explore syntactic and sem antic processing in native speakers of both German (n=18) and Russian (n=7) and native speakers of Russian that are advanced learners of German (n=14). The second language learners of German are described by R schemeyer et al. highly proficient, late l ( R (271) 25 The same stimuli from Hahne (2001) were used for the German portion of the study; I will focus exclusively on the findings related t o the comparison between native German and non native German speaking participants (i.e. Experiment 2; the data from Experiment 1 focused on native processing only). R schemeyer et al. report that only for the semantically anomalous tokens were comparable activations found between the native and second language participants, as compared to the grammatical tokens. Unlike the native speaker participants, the second language participants did not demonstrate different activation patterns when processing the syn tactically ungrammatical tokens as compared to the grammatical tokens. Similar results 25 N o standardized proficiency measure is mention ed nor is the range of for age of onset of acquisition or length of residency provided. Thus, the groups may contain considerable variability with respect to these variables.

PAGE 78

78 obtained in Rschemeyer, Zysset and Friederici (2006), which also examined native Russian speaking late learners of German. While different and similar patterns between native and non native processing were discussed in Abutalebi et al. (2001) and Kotz (2009), both reviews highlight the role of proficiency and its positive correlation with similar activation patterns. In fact, when the methodology from Perani et al. (1996 ) was employed with highly proficient 26 Italian English bilinguals, Perani et al. (1998) found similar activation patterns between native and second language processing. This finding was interpreted as indicating that proficiency is a determining factor (se e also Weber Fox & Neville 1996; Perani et al. 2003; Wartenburger et al. 2003; Dodel, Golestani, Pallier, ElKouby, Bihan & Poline 2005; Rschemeyer et al. 2005; Rschemeyer et al. 2006 for age and proficiency effects) Finally, the three studies presented above are just a few of many neurophysiological studies and therefore should not be taken as representative of the larger body of research. However, across both Abutalebi et al. (2001) and Kotz (2009), the lack of standardized proficiency measures and agre ement concerning the definition of early versus late bilingual is obvious. These issues should be kept in mind to rectify in future studies. Despite the studies suggesting differences in non native language processing, the overall conclusion of Kotz (2009 ) is that the body of neural data indicates that the areas of the brain activated in second language processing are similar to those activated in native language processing. Results across different language pairings, modality of stimuli presentation, ling uistic domain of investigation and early versus late bilinguals offer support that this claim is not simply restricted to certain language pairings or linguistic domains. For example, Perani et al. (1998) and Luke, Liu, Wai, Wan and Tan (2002) found simila r native and non native neural 26 In the case of the Italian English late bilinguals, proficiency was assessed vi a the same word translation and sentence comprehension task used in Perani et al. (1996).

PAGE 79

79 activation during sentence comprehension (involving syntax and semantics) whether presented aurally or visually. These results further contribute to the overall picture in that the language pairings examined therein are diffe rent: Perani et al. (1998) examine bilinguals of typologically similar languages (Italian English), while Luke et al. (2002) examines typologically dissimilar languages (Chinese English). Several ERP studies showed that even with little exposure, non nativ e lexical semantic processing demonstrated native like N400 effects (e.g. McLaughlin, Osterhout & Kim 2004; Osterhout, McLaughlin, Pitknen, Frenck Mestre & Molinaro 2006). Winkler et al. (1999) demonstrated that native Hungarian speakers who were second l anguage learners of Finnish evinced native like Mismatch Negativity (MMN) 27 for a phonemic contrast in Finnish. Frenck Mestre, Osterhout, McLaughlin and Foucart (2008) and Frenck Mestre, Foucart, Carrasco and Herschensohn (2009) demonstrated that phonologic ally realized morphosyntactic errors elicited a P600 effect in French. For morphosyntactic properties related to subject verb agreement, number agreement and gender agreement violations Bond, Gabriele, Fiorentino and Alemn Ban (2011) found a P600 effect (for similar results in different language combinations see also Sabourin 2003; Rossi, Gugler, Friederici & Hahne 2006 for subject verb agreement; Tokowicz & MacWhinney 2005; Gillon Dowens, Vergara, Barber & Carreiras 2010 for gender agreement). Finally, Hernandez, Hofmann and Kotz (2006) found a basic overlap for native (early Spanish English bilinguals) and non native (native English, late Spanish learners) gender processing. Still, retrieval differences between the early and late groups were found 28 In general, then, the neural studies and results reviewed by Kotz (2009), across various language pairings, testing modality and linguistic domain, points to the conclusion that similar brain areas are 27 Mismatch Negativities (MMNs) are generally elicited by unexpected stimuli in a series of visual or aural stimuli. 28 This difference could be a product of the expos ure to and use of Spanish early bilinguals might have a higher necessity for use of Spanish (i.e. family members) than late learners.

PAGE 80

80 activated in first and second language processing. The f ollowing section describes some studies from the behavioral paradigm. 2.4.3.2 Behavioral studies As discussed by Foucart and Frenck Mestre (2012), an advantage 29 of the eye tracking paradigm (as opposed to the visual presentation of stimuli in ERP studies or non cumulative self paced reading studies) is that the complete sentence can be presented to participants allowing them to read at their own pace and also to reread regions. I will only briefly describe the findings of two recent eye tracking studies e xamining grammatical violations. These studies provide evidence that late second language processing can be similar to that of native language processing. Finally, two studies using the self paced reading paradigm will be described. While eye tracking and self paced reading studies are highlighted here, behavioral tasks examining processing are not limited to these two paradigms. Nevertheless, only these will be described in accepted experimental tasks for the inves tigation of Keating (2009) examines second language processing of gender agreement between nouns and adjectives in Spanish. Specifically, Keating is interested in determining w hether processing effects are found as a function of syntactic distance between the noun and adjective. Three groups of native English speaking, late learners of Spanish (beginner, intermediate and advanced 30 ) as well as a native Spanish speaking control gr oup 31 completed a sentence 29 However, one disadvantage of the eye tracking paradigm is that it does not provide information regarding the nature of t he processing (i.e. whether syntactic or semantic processing is implicated). 30 Proficiency levels were based on years of exposure to Spanish as well as university course placement. 31 While the native Spanish speakers were reported to have resided in and (m inimally) completed their high school education in a Spanish speaking country, these participants were tested in the United States and were described as it is likely that they are somewhat proficient in English.

PAGE 81

81 comprehension task while eye tracking data were recorded. Participants were presented with 48 fillers and 36 experimental stimuli evenly divided by grammaticality, gender and condition (agreement within the Determiner Phrase (DP), the Verbal Phrase (VP) or Complementizer Phrase (CP)). While the beginner and intermediate groups did not demonstrate evidence of having acquired grammatical gender in Spanish, the advanced group did. That is, the advanced group, like the native Spanish s peaking group, showed sensitivity to ungrammatical gender agreement between nouns and adjectives within the DP, via longer reading times as compared to the grammatical DP condition. However, this finding was absent for the VP and CP condition. These result s demonstrate that late second language acquirers are able to converge on properties not present in their native language, for example, grammatical gender and adjectival word order. It should also be pointed out that native like sensitivity to gender agree ment in the VP and CP conditions might be evinced in near native speakers or with participants living in a Spanish speaking environment as recency and higher daily use of the target language may affect convergence. Foucart and Frenck Mestre (2012) conducte d a series of ERP and eye tracking experiments within the same group of native French speakers and native English speaking late learners of French 32 While the ERP experiments elicited P600 effects for gender agreement violations with postnominal adjectives both in native and second language speakers, an N400 (rather than a P600 effect) effect was found for gender agreement violations with prenominal adjectives for second language speakers only 33 Finally, while no effect was found for predicative adjectives in the third ERP experiment for the late learners of French, similar patterns 32 18 years) while the average length of formal learning was eight years (range 5 11 years). Each had pas de Langue Franaise) and five participants had low to intermediate knowledge of another language. 33 A P600 effect was evinced for the native speakers.

PAGE 82

82 were found for the same stimuli under the eye tracking paradigm. Specifically, longer reading times were recorded for gender violations between nouns and predicative adjectives f or both participant groups. Importantly, longer reading times were recorded for first pass, gaze duration and total reading time in both groups. Foucart and Frenck Mestre note that the second language tracking patterns with non local gender a greement were virtually indistinguishable from those of the native speakers (contra Keating 2009). In a recent self paced reading study, Sagarra and Herschensohn (2011) investigate sensitivity to grammatical gender in English speaking late acquirers of Spa nish as a second language. In addition to several offline measures, beginning (n=69) and intermediate 34 learners of Spanish (n=64) and native Spanish speakers (n=63) completed a self paced moving window task with comprehension questions in E Prime software. Experimental tokens included grammatical gender and number agreement, gender agreement violation and adjective and number agreement violation between the noun and adjective. While both the beginner and intermediate second language groups were slower reade rs for the critical regions analyzed statistically, the intermediate group showed sensitivity to gender and number agreement violations. That is, the gender and number agreement violation conditions elicited slower reading times as compared to grammatical condition; this sensitivity is qualitatively similar to that of the native Spanish speakers. Sagarra and Herschensohn interpret this finding as evidence suggesting that native speakers of languages that do not have grammatical gender, such as English, may converge on grammatical representations that are similar to those of native speakers. This is especially revealing given the relatively low proficiency level and lack of naturalistic learning (no more than 3 months in a Spanish speaking country). 34 Proficiency was based on both class enrollment as well as a grammar sect ion from the DELE.

PAGE 83

83 VanPatt en, Keating and Leeser (2012) also examined sensitivity to grammatical structures in non advanced late second language learners of Spanish (native English speakers). Specifically, the reading times of second language learners (n=25) were compared to those of native Spanish speakers (n=18) for subject verb inversion in questions, adverb placement and verbal inflection. Participants completed a self paced reading task (using the non cumulative moving window setup and with follow up comprehension questions) in SuperLab. With respect to verbal inflection, VanPatten et al. report that only the native speakers demonstrated sensitivity to person number agreement violations via slower reading times as compared to the grammatical items 35 However, regarding both subje ct verb inversion and adverb placement, sensitivity to violations was found in both participant groups via slower reading times for ungrammatical tokens as compared to grammatical tokens. Although VanPatten et al. ultimately argue that the aforementioned r esults suggest a representational problem for verbal morphology for the non native Spanish speakers, one must take into account the proficiency level of the participants. It is possible that higher proficiency second language learners tested with the same methodology would demonstrate sensitivity to verbal morphology. Importantly, even non advanced learners showed native like sensitivity to two structural violations. In summary, across both eye tracking and self paced reading paradigms, evidence supporting native like processing in late second language learners at least for Romance languages has been provided. While the four studies described above examined grammatical gender, VanPatten et al. (2012) also provides evidence that late second language learners show sensitivity to structural violations related to word order. Since the properties investigated in these 35 When the 1 st person singular was compared against 3 rd was found (VanPatten et al. 2012, p. 132). For no other comparisons (i.e. 2 nd person singular versus 3 rd person plural) or individual subjects (1 st person singular, 2 nd person singular, etc.) was sensitivity to subject verb agreement violations found.

PAGE 84

84 studies are not instantiated in the native language of the second language participants (English), the results provide two important pieces of evid ence. First, late second language learners are able to converge upon the underlying representation of properties absent in their native language and, second, that their processing of these properties can also be native like. 2.4.4 Limitations of Previous Processing Literature Although a complete review of the second language processing literature falls outside the scope of this study, the studies described in Section 2.4.3 provide significant conclusions. It was shown across various processing paradigms, l anguages and linguistic properties that second language learners can demonstrate native like processing. Even so, a variety of results have been reported (see Abutalebi et al. 2001; Kotz 2009; Foucart & Frenck Mestre 2012) and limitations of this literatur e have been made apparent. Hopp (2007) points out several characteristics of previous processing research that are limiting. First, previous research has demonstrated that there exists a positive correlation between higher proficiency and native like proce ssing; however, many of the studies examine second language processing with participants of lower proficiency and standardized proficiency measures are largely absent. Therefore, it is impossible to determine if the variable affecting divergence is related to differences in underlying grammatical representation or language processing itself, both of which are assumed herein to contribute to proficiency. Second, differences in exposure and use of the target language (which, similar to proficiency, likely aff ects automaticity in processing and distribution of cognitive resources) and working memory capacity are not always controlled for, thus introducing other contributing variables. Third, systematic attention has not been given to cross linguistic difference s as a potentially confounding factor. Finally, Hopp points out that second language processing research often relies on comparisons across studies with different participant populations and different experimental methods. While this is expected and encour aged within

PAGE 85

85 the field, a more complete picture of second language processing could be obtained by examining a variety of linguistic properties within the same participant group (or, at least a maximally comparable one). 2.5 Testing the Interface Hypothesi s Based on the limitations pointed out in Sections 2.3.6 and 2.4.4, this section briefly describes the type of research that is needed in order to more adequately test the most recent na tive speakers only, standardized proficiency measures should be administered. Additionally, it is important to ensure that the second language learners, like monolinguals, have high levels of exposure to and use of the second language as this can affect a utomaticity. Second, since Sorace claims that the presence of two linguistic systems creates a burden on the processing abilities of bilinguals, online testing measures must be employed. These tasks provide the type of data necessary (RT, for example) to d etermine if there are differences between monolingual and bilingual processing. A related matter concerns working memory and inhibitory control capacity. As lower levels of working memory or inhibitory control may affect language processing, determining a baseline languages (i.e. are not specific to a particular language pairing or directionality), a variety of languages from various language families must be tested As Hopp (2007) points out, comprehensive methodologies employed within the same or highly comparable populations will reduce extraneous factors, allowing for greater precision of interpretation of the results within in the scope of the particular study a nd decreasing the need for comparisons across different methodologies, participants and paradigms. Accordingly, the aim of the dissertation study described herein is to advance previous research by examining an interface conditioned property, referential subject pronoun

PAGE 86

86 distribution, in a novel language pairing, Farsi Spanish, using online tasks that are able to tap Prada Prez (2010), this scenario is ideal for are both null subject languages that apparently share an analogous distribution of referential subject pronouns, as in Bini (1993), Margaza and Bel (2006) and Garca Alcaraz and Bel (2011). Neverthele ss, the underlying syntax of the two languages is different and, according to Karimi (2005), Farsi does not possess the null expletive subject as Spanish does. In examining this ulnerability is based in the presence of two grammars and not a result of the specific language pairing and/or directionality. The following chapter outlines the syntactic analysis of preverbal subjects in Spanish and Farsi.

PAGE 87

87 CHAPTER 3 SYNTAX OF NULL SUBJE CTS 3.1 Syntax This chapter outlines the syntactic analysis adopted for the linguistic property tested in this dissertation study. Specifically, the syntactic literature of preverbal subjects is presented for both Spanish and Farsi. Furthermore, related i nterface conditioned properties regarding the discourse distribution of referential subject pronouns is also examined. 3.2 Null Subject Parameter (NSP) The well formedness principle pertinent to referential subject pronouns is the Avoid Pronoun Principle (APP; Chomsky 1981) 1 which simply states that overt pronouns should only be employed when absolutely necessary (i.e. for interpretive or discourse/pragmatic reasons). Thus, if an overt subject pronoun is not absolutely necessary, it should be avoided in f avor of a null subject pronoun 2 This is illustrated in example (1) below in Spanish, which is stated in discourse neutral context (i.e. no contrastivity is meant to be conveyed): (1)a. #Marcos i dice que l i est ocup ado. Marcos say 3. SG PRES that he be 3. SG PRES busy MASC SG b. Marcos i dice que pro i est ocupado. Marcos say 3. SG PRES that be 3. SG PRES busy MASC SG Example (1a) violates the APP in that it is not necessary to use the subject pronoun l l Marcos ; therefore, the utterance is infelicitous 3 with coreferentiality between the m atrix and embedded clause 1 How this principle of economy is conceived of in more modern terms is of no consequence. I use the Avoid Pronoun Principle as shorthand following others (e.g. Rothman 2009). 2 For representational sake, pro is used to indicate null subject pronouns. pro is a pronoun without phonological form that exists only in the underlying representation. 3 Infelicitousness is indicated b y a pound sign (#) while ungrammaticality is indicated by an asterisk (*).

PAGE 88

88 subjects. In comparison, example (1b) is felicitous in that it obeys the APP: there is no unnecessary use of an overt subject pronoun in the embedded clause. Since nominal phi features in Spanish license and identify null subject pronouns, the discourse neutral environment of example (1) renders (1b) the only felicitous option. 4 Notice, however, that the English glosses for (2b) shows: (2)a. Mark says that he is busy. b. *Mark says that pro is busy. Based on examples (1) and (2), one might ask why English seemingly disobeys the APP if it is a universal principle of well formedness that applies to all natural languages. The overt pronoun reason for its use. The answer to this question is related to language specific parametric settings and how they delimit what is possible in that language. Thus, as it relates to the APP, only Spanish syntax licenses and identifies null subjects, a necessary precursor to see some of the refluxes of null subjects predicted by the APP. English syntax, on the other hand, does not. So, while all languages are governed by the same grammatical principles that delimit the options available to human languages, crosslinguistic variation is a product of the particular setting of parameters related to how these principles are manifested in language production. Under Minimalist a ssumptions, parametric differences lie in the functional lexicon of individual grammars. In other words, lexical items encode formal features of language from which functional categories emerge as needed to value these features. The extent to which these 4 While examples (1a) and (1b) are both grammatical, they have different interpretive biases. According to interpretation of example (1a) is overwhelmingly disjoint whereas that of (1b) is co referential with the matrix clause subject. In the latter case, coreference obtains precisely because pro strongly biases for coreference with the subject in Spec,TP of th e matrix clause. Since this is outside the scope of this dissertation, it will not be revisited.

PAGE 89

89 f eatures are different between languages gives rise to what is perceived as crosslinguistic differences. The NSP has traditionally been conceived of as having two settings: non null subject and null subject (see Perlmutter 1971; Rizzi 1982) 5 Non null subj ect languages (e.g. English, French and German) require overt subjects 6 whereas null subject languages (e.g. Farsi, Greek and Spanish) allow for both null and overt subjects. Rizzi (1982) claims that the availability of null subjects is predicated upon the fulfillment of two requirements: (i) pro must be syntactically licensed and (ii) pro must be identified 7 While Farsi and Spanish are both null subject languages d ifference related to word order is that in Farsi, as opposed to Spanish, all phrasal elements may remain in the vP (verb shell). Only for interpretive reasons (Topic, Focus, etc.) do elements leave the vP (Karimi 2005). The second difference is related to satisfaction of the Extended Projection Principle (EPP), a principle of grammar that requires all clauses to have a subject. Chomsky (1995) proposes that, in general, the EPP is a D(eterminer) feature of T(ense) that must be checked by an element in Spec,T P (Specifier position of Tense Phrase). Goodall (2001, 2002) argues that this is how the EPP is satisfied in Spanish. However, Karimi (2005) contends that the 5 While this binary distinction describes the languages under consideration in this study, researchers have further distinguished between types of null subjec t languages listing as many as four (Biberauer, Holmberg, Roberts & Sheehan, 2010, p. 6 13): (i) Consistent null subject languages (Greek, Italian, Spanish, etc.) (ii) Expletive null subject languages (German, Haitian, Jamaican, etc.) (iii) Discourse pro drop/Radical pro drop languages (Chinese, Japanese, Korean, etc.) (iv) Partial null subject languages (Brazilian Portuguese, Finnish, Hebrew, etc.) 6 null subject languages such as English for example (see Haegemann 1997); however, this does not mean that English is a null subject language as pro cannot be licensed or identified in English. 7 by Huang (1984), many East Asian languages may identify null subjects via the discourse; it was also claimed that overt subject pronouns and pro are in free variation in some South Asian languages.

PAGE 90

90 same does not hold for Farsi. She explores three possibilities for EPP satisfaction (seen in the list below), ultimately concluding that it is satisfied morphologically and without movement in Farsi: XP movement (or Merge) to Spec,TP (English) V(erb) to Tense movement (most null subject languages; Alexiadou & Anagnostopoulou 1998) Morphologically (wit hout overt movement; Farsi, Hungarian) The first option described above is ruled out since the subject may remain in the vP in Farsi (unless topicalized or focused). Evidence for the vP internal status of the subject is found in adverbial placement. Based on the division of adverbs proposed by Cinque (1999), two groups are pertinent to Persian: the so which correspond to the sentence level, precede the lower adverbs, which correspond to the VP level. Karimi (2005) claims that both higher and lower adverbs adjoin to vP 8 ; thus, given the position of the adverbs with respect to the rest of the sentential elements, the following examples provide evidence that Farsi subjects do not raise to Spec,TP to check the EPP, and instead may remain in the vP. yesterday Kimea with Rahjue fight did 3. SG ketbxune dars mi xun e usually Kimea in library lesson DUR read 3. SG (5) kmelan Kimea be hush umade 8 According to Karimi (2005, p. 124 125), some higher ad verbs in Farsi are amdan xoshbaxtne zheran badbaxtne ehtemalan modal adverbials such as shyad byad emruz diruz zirakne hamishe ghelne hanuz hargez kmelan ngahn taghriban adverb (Hadv) positi ons, are proposed (all preceding the lower adverb (Ladv) position): (i) [ CP (Hadv) [ TopP (Hadv) [ TP (Hadv) (Ladv) [ vP ]]]]

PAGE 91

91 completely Kimea to sense come 3. SG (6) xoshbakhtne kmelan Kimea be hush umade (Karimi 2005, p. 94) fortunately completely Kimea to sense come 3. SG Fortunately, Kimea has completely regained her Although Karimi proposes three positions (see Footnote 8 above) for higher adverbs such as kmelan raised out of the vP since lower adverbs adjoin to the vP 9 The same examples in Spanish (with adjustments for word order) show a different picture. With neutral intonation and no pause after the adverb, these examples are ungrammatical since the subject must raise out of the vP. Thus not all phrasal elements may remain in the vP. (7) (*ayer) Carmen se pele con Ruben (ayer) yesterday Carmen RFLX fight 3. SG PRET wit h Ruben yesterday (8) (*usualmente) Carmen (usualmente) estudia en la biblioteca usually Carmen usually study 3. SG PRES in the library (9) (*completamente) Carmen entr en razn (completamente) completely Carmen enter 3. SG PRET in reason completely Carmen has comple 9 The following examples show that the subject may be realized in a position higher than the adverb if topicalized or focalized (A. F eizmohammadpour, personal communication, March 30, 2013). Kimea yesterday with Rahjue fight did 3. SG u ketbxune dars mi xun e Kimea usually in library lesson DUR read 3. SG (iii) Kimea kmelan be hush umade Kimea completely to sense come 3. SG (iv) Kimea xoshbakhtne kmelan be hush umade Kimea fortunately completely to sense come 3. SG

PAGE 92

92 (10) *afortunadamente completamente Carmen entr en razn 10 fortunately completely Carmen enter 3. SG PRET in reason Fortunately, Carmen has c The second option is also discarded due to the fact that Farsi is a verb final language and that Tense is phrase initial in this language. As follows, if the verb raised to Tense, the canonical S(ubject) O(bject) V(erb) word order of Farsi could not be maintained. In a neutral context, VSO is ungrammatical as the V does not raise out of the vP unless for emphasis 11 (11) Ramin nn khord. Ramin bread eat 3. SG PRET (12) k hord Ramin nn eat 3. SG PRET Ramin bread Conversely, if Tense were in final position and the EPP were satisfied by overt Verb to Tense movement, the sentential argument of V would have to appear preverbally, contrary to fact: (13)a. Kimea goft ke Parviz hune nist Kimea said 3. SG PRET that Parviz home not be.3. SG PRES b. *Kimea ke Parviz hune nist goft Kimea that Parviz home not be.3 SG PRES said 3. SG PRET 10 In Spanish, the most natural version of example (10) above would be: (i) afortunadamente, Carmen entr en razn completamente fortunately Carmen enter 3. SG PRET in reason completely 11 While example (12) above is ungrammatical, the presence of the accusative/specificity marker r (labeled below as DOM) makes the sentence grammatical. This follows from the structure described by Karimi (2005, p. 7). Both specific and non specific direct objects are base generated as the complement of V; however, like non specific subjects, non specific objects also remain inside PredP (Predicate Phase). Only specific subjects and specific objects (i.e. those m arked with r ) raise to Spec,vP. Thus, the sentence final position of the specific direct object nn r demonstrates that both the specific subject Ramin and the verb khord have raised out of the vP for emphasis. (i) khord Ramin nn r eat 3. SG PRET Ramin bread DOM

PAGE 93

93 Considering the canonical SOV word order of Farsi, a preverbal sentential argumen t is actually expected. However, the fact that SOV word order becomes ungrammatical, as in (13b) above, provides evidence that V does not raise to a sentence final Tense. Spanish also exhibits the same ban on preverbal sentential arguments as seen in examp le (14b) below. Differently from Farsi, Tense is sentence initial in Spanish. Thus, when V raises to Tense in Spanish, the canonical SVO word order of Spanish is maintained. (14)a. Carmen dijo que Paula no est en la casa. Carmen said 3. SG PRET that Paula no be.3. SG PRES in the house b. *Carmen que Paula no est en la casa dijo. Carmen that Paula no be.3. SG PRES in the house said 3. SG PRET 12 the EPP dividing languages according to the way in which the EPP is satisfied: either by XP mo vement to Spec,TP (as in (14) above) or via a property of the verbal morphology (as in (13) above). Karimi concludes that the EPP is satisfied morphologically in Farsi and that it is not a D (nominal) feature of T(ense) that triggers V to T movement as is in Spanish. Rather, Karimi deems it to be a property of the verbal morphology (overt agreement). Nominative Case is checked and valued morphologically in Spec,vP in Farsi while it is checked and valued in Spec,TP in Spanish 13 In both languages, the referen t of pro is identified via the verbal agreement morphology due to the fact that it is unique for each person. See Table 3 1 below for the verbal agreement morphology for the present tense in Farsi and Spanish. An additional difference between Spanish and F arsi, although both null subject languages, regards expletive subjects. This is further discussed in Section 3.2.1. 12 But, see Alexiadou and Anagnostop o ulou (1998) for a different analysis of the parameterization of the EPP. 13 Or V to T to C or even V to C depending on the analysis ta ken (e.g. whether or not one assumes a Spec,TP projection in Spanish). Since this is inconsequential to the discussion at hand, I note in passing this active debate (see for example Suer 1994; Ordez & Trevio 1999; Goodall 2001).

PAGE 94

94 Table 3 1. Present tense verbal morphology Farsi Spanish ( ar verbs 14 15 ) Singular Plural Singular Plural 1 st person am im o amo s 2 nd person i id as is 3 rd person ad/ast and a an 3.2.1 Syntactic Properties Related to the NSP In addition to the availability of null subject pronouns, Rizzi (1982, 1986), among others, has argued that certain syntactic properties, hypothes ized to be directly associated with the null subject setting of the NSP, cluster together. However, which properties actually cluster with the NSP is a controversial topic (Jaeggli & Hyams 1988; Rothman & Iverson 2007; Rothman 2009). In the original propos al by Rizzi (1982), it was claimed that four properties cluster with the NSP: (i) null referential subjects, (ii) obligatory null expletives, (iii) free word order (subject verb inversion) and (iv) that trace effects. As the debate concerning this cluster of properties is not fundamental for the current dissertation study, I assume, along with Rothman (2009), that null referential subjects (see example (15)) and obligatory null expletives (see example (16)) cluster with the null subject setting of the NSP ( but see Biberauer, Holmberg, Roberts & Sheehan 2010). (15)a. En mi opinin, pro habla bien el italiano. in my opinion speak 3. SG PRES well the Italian b. be qide mn pro italya'i xub sohbt mikon e in opinion my Italian well speak 3. SG PRES Examples (15a) and (15b) above show that in Spanish and Farsi, respe ctively, null referential subject pronouns are grammatical. The following examples examine expletive subjects. 14 Spanish has three type s of verbs, classified in Spanish language textbooks by their infinitive form: ar er and ir verbs. The present tense verbal paradigm for er and ir verbs is slightly different from that of ar verbs; however, it is regular. 15 Tense, verb class and person/number morphology are conflated in the examples above for ease of exposition.

PAGE 95

95 (16)a. *Dicen que ello es beneficioso relajarse cada da say 3. SG PRES that it be 3. S G PRES beneficial relax. INF each day para mantener buena salud. in order maintain. INF good health. b. Dicen que pro es beneficioso relajarse cada da say 3. SG PRES that be 3. SG PRES beneficial relax. INF each day para mantener buena salud. in order maintain. INF good health. c. mign mofid e ke in braye dashtn e zehn e salem say 3. PL PRES beneficial EZ that this for having EZ mind EZ healthy hr ruz esteraht kon i each day relax do 2. SG PRES d. mign mofid e ke braye dashtn e zehn e salem say 3. PL PRES beneficial EZ that for having EZ mind EZ healthy hr ruz esteraht kon i each day relax do 2. SG PRES Example (16a) shows that overt su bject pronouns in expletive contexts are ungrammatical in standard Spanish and that a null subject is required as in (16b). In Farsi, however, we see that both (16c), which seemingly contains an expletive in and (16d) are grammatical. Karimi ultimately ar gues that in 93). that have null expletives, Karimi claims that there is no evidence suggesting that they exist in Farsi. This claim is based largely on the fact that languages with expletives are subject to the Definiteness Effect: (17) *There is John in the room. (Karimi 2005, p. 93) (18) There is a boy in the room. Upon comparing examples (17) and (18), it is cle ar that only an indefinite subject may remain in the vP whereas definite subjects may not. Considering again the adverbial examples seen above

PAGE 96

96 in (3) through (6), Karimi demonstrates that Farsi is not subject to the Definiteness Effect. Recall that adverbs adjoin to vP in Farsi; thus the definite subject of (6), repeated as (19) below, is internal to the vP. (19) xoshbakhtne kmelan Kimea be hush umade (Karimi 2005, p. 94) fortunately completely Kimea to sense co me 3. SG PRET Farsi (Karimi 2005, p. 94). While the surface instantiation of the NSP appears t o be very similar in Spanish and Farsi two differences exist between the languages. The first, described in Section 3.2.1, relates to the way the EPP is satisfied in each language. Spanish satisfies the EPP with V to T movement while Farsi satisfies it mo rphologically within the vP. The second, described above, relates to the existence of expletive subjects in each language. These differences are important as they bear on the learning task of the native Farsi speaking second language learner of Spanish, wh ich is discussed in detail in Section 3.2.3. 3.2.2 Interface conditioned Properties Related to the NSP Two interface conditioned properties related to the null subject setting of the NSP in Spanish and Farsi are discussed in Sections 3.2.2.1 and 3.2.2.2. T he first property lies at the syntax semantics interface while the latter lies at the syntax discourse interface. 3.2.2.1 The Overt Pronoun Constraint (OPC) of t he NSP, Montalbetti (1984) added the OPC. The OPC is a restriction on the interpretation of overt referential subject pronouns. Montalbetti (1984, p. 26) describes this restriction as follows: A bound variable interpretation of an overt pronoun is prohibi ted if pro is available in the same

PAGE 97

97 Based on this restriction, it is expected that overt subject pronouns in null subject languages such as Spanish and Farsi may never be interpreted as a co referential with a variable antecedent since pro is av ailable in almost all subject positions. Examples (20) through (22) below show that this holds true for Spanish. In example (20), either the overt (20a) or null (20b) embedded clause subject pronoun can be co referential with the matrix clause subject ese hombre variable Determiner Phrase. However, in the presence of a variable antecedent, for example cada persona quin overt embedded clause subject pronouns cannot be interpreted as co re ferential with the variable antecedent. The asterisk (*) associated with the subscript linking of cada persona and quin in (21a) and (22a), respectively, indicates this prohibition. (20)a. Ese hombre i cree que l i/j es el mejor. that man think 3. SG .PRES that he be 3. SG PRES the best b. Ese hombre i cree que pro i/j es el mejor. that man think 3. SG .PRES that pro be 3. SG PRES the best (21)a. Cada persona i cree que l *i/j es el mejor. each person think 3. SG PRES that he be 3. SG PRE S the best b. Cada persona i cree que pro i/j es el mejor. each person think 3. SG PRES that pro be 3. SG PRES the best pro (22)a. Quin i cree que l *i/j es el mejor? who SG think 3. SG PRES that he be 3. SG PRES the best b. Quin i cree que pro i/j es el mejor? who SG think 3. SG PRES that pro be 3. SG PRES the best Regarding Farsi, data from 39 native speakers who were independently tested (Judy & Feizmoh ammadpour 2012) demonstrate that the OPC obtains in Farsi as well. This dataset shows strong categorical preferences. For example, no variation was seen with respect to the interpretation of the null embedded clause subject in example (23b) below. In Farsi the bound

PAGE 98

98 interpretation was categorically preferred. The opposite scenario is seen in the presence of variable antecedents. Examples (24a) and (25a) show that the overt embedded clause subject u atrix subject whereas the null embedded subject was only interpreted as bound in the presence of a variable matrix clause subject (see examples (24b) and (25b)). (23)a. Un mard i fekr mikonad ke u i/j az hameh behtar a st. that man thought DUR make 3. SG that he from all better be 3. SG PRES b. Un mard i fekr mikonad ke pro i/*j az hameh behtar ast. that man thought DUR make 3. SG that pro fro m all better be 3. SG PRES pro (24) a. Har kasi i fekr mikonad ke u *i/j az hameh behtar ast. each person thought DUR make 3. SG that he from al l better be 3. SG PRES b. Har kasi i fekr mikonad ke pro i/j az hameh behtar ast. each person thought DUR make 3. SG that pro from all better be 3. SG PRES (25)a. Ki i fekr mikonad ke u *i/j az hameh behtar ast? who thought DUR make 3. SG that he from all better be 3. SG PRES b. Ki i fekr mikonad ke pro i/j az hameh behtar ast? who thought DUR make 3. SG that pro from all better be 3. SG PRES Summarizing Kanno (1997, 1998) and Prez Leroux and Glass 1999, Whit e (2003a) argues that the OPC constitutes a poverty of the stimulus property since it is attested in a variety that the interpretive constraints that fall out fr om it are not derivable from the input alone and are not taught in the classroom, nor can they be accounted for via domain general learning strategies (for more information see Section 1.1). For these reasons, I concur with White in assuming that

PAGE 99

99 the OPC i s a poverty of the stimulus property. Furthermore, following Rothman (2005, 2007, 2009), I take the OPC as a diagnostic of a null subject language. 16 In conclusion, examples (20) through (25) above demonstrate that null subject languages do not always requi re null subjects. More accurately, these languages permit both overt and null subjects. However, where optional, the interpretation of the subject pronoun may depend on such things as the type of matrix clause antecedent. 3.2.2.2 Discourse distribution of referential subject pronouns Across many 17 null subject languages (including the two languages relevant for this dissertation, Farsi and Spanish), the distribution of overt and null subjects is constrained by the discursive context and intended meaning. Th us, it is not the case that pro and overt subject pronouns are in free variation, occurring by chance, but rather that discursive constraints drive the use of one form of over the other. Of course, there is much variation across all null subject languages; for example, Brazilian Portuguese uses overt subject pronouns more frequently than Spanish (Duarte 1993, 1995). With respect to the languages under investigation in the present study, Spanish and Farsi, distribution of referential subject pronouns is indi stinguishable for the domains examined. As opposed to non null subject languages like English, in which overt subject pronouns are required in expletive contexts, the use of overt referential subject pronouns in Spanish and Farsi is rather restricted, as w ould be expected by the APP (Chomsky 1981) and the availability of pro 16 The native speaker and L2 speaker groups of this study completed a task examining the OPC. The L2 speakers showed the restrictive constraints on the interpretation of ov ert embedded clause subjects in the presence of quantified antecedents, as predicted by the OPC. However, the OPC obtains in Farsi (Judy & Feizmohammadpour 2012), rendering it difficult to rule out L1 transfer effects in the current population. 17 Kissock subjects seem to be present with about the same frequency as pro and do not appear to mark any added emphasis or

PAGE 100

100 Ample theoretical research has examined the distribution of overt and null subjects (Rigau 1986, 1988; Lujn 1987, 1999; Fernndez Soriano 1989, 1993; Picallo 1994, 1998; Rizzi 1997; Alonso Alonso Ovalle, Fernandez, Frazier, & Clifton 2002 and sources cited within). Building on this work, Rothman (2009) demonstrates the restrictions on the distribution of subjects in Spanish. In order to illu strate the parallel nature of the distribution of null and overt referential subject pronouns in Spanish and Farsi, the same examples provided by Rothman (2009) for Spanish are considered in Farsi. Starting with topics, which is discourse given information examples (26) and (27) demonstrate that null subjects are felicitously used in Topic Maintenance contexts while examples (28) and (29) demonstrate that overt subjects are preferred in Topic Shift contexts to remove ambiguity. (26) Mara e Hilda no almorzaron hoy. Mary and Hilda not eat 3. PL PRET today a. # 18 Ellas tendrn mucha hambre. they have 3. PL FUT much hunger b. pro tendrn mucha hambre. pro have 3. PL FUT much hunger (27) Shabnam va Leila em ruz nahar nakhordand. Shabnam and Leila today lunch NEG eat 3. PL PRET a. #Unha bayad goshneh bashand. they must hungry be 3. PL PRES b. pro bayad goshneh bashand. pro must hungry be 3. PL PRES 18 In Rothman (2009), the judgm ent of this sentence in light of the preceding context is */?. Since there is no question regarding the grammaticality of this sentence, I have changed the judgment to # to represent the infelicitous, or odd, nature of the use of the overt subject given th e context.

PAGE 101

101 In examp les (26) and (27) and their respective follow up sentences seen in (26a) and (26b) and (27a) and (27b), there is no change in topic; that is, the subject of the context and follow up sentence alike refers to the same two women: Mara and Hilda in (26) and Shabnam and Leila in (27). Thus, the topic of the sentence has been maintained (hence the term Topic Maintenance) rendering the use of the overt subject pronoun ellas or unha as it is unnecessary whereas the use of pr o in (27b) and (27b) is felicitous. In contrast to the examples above, examples (28) and (29) show a Topic Shift context, or a context in which the topic changes. (28) No almorc hoy. pro not eat 1. SG PR ET today a. Ellas piensan que tengo hambre ahora. they think 3. PL PRES that pro have 1. SG PRES hunger now b. # 19 pro piensan que tengo hambre ahora. pro think 3. PL PRES that pro have 1. SG PRES hunger now (29) Man em ruz nahar nakhordam. I today lunch NEG eat 1. SG PRET a. Unha fekr mikonand ke goshneh basham. they thought DUR do.3. PL that pro hungry be 1. SG PRES b. # pro fekr mikonand ke goshneh basham. pro thought DUR do.3. PL that pro hungry be 1. SG PRES In examples (28) and (29) and the corresponding foll ow up sentences, a change in topic occurs as the subject of (28) and (29) are yo and man follow up sentences is ellas and unha 19 In Rothman (2009), the judgment of this sentence in light of the preceding context is ungrammatical (*). Since there is no question regarding the grammaticality of this sentence, I have changed the judgment to # to represent the infelicitous, or odd, nature of the use of the overt subject given the context.

PAGE 102

102 overt subject pronou n ellas / unha contrary, the use of pro is infelicitous. Examples (30) and (31) show a focus (discourse new information) question while examples (32) and (33) show a yes/no question. (30) Quines vieron la pelcula? who see 3 PL PRET the movie? a. Nosotros la vimos. we it CL FEM see 1 PL PRET b. # 20 pro la vimos. pro it CL FEM see 1 PL PRET (31) Ki ha film r didand? who PL movie DOM see 3. PL PRET a. Ma didim esh. we see 1. PL PRET it CL b. # pro didim esh. pro see 1. PL PRET it CL The focus questions in (30) and (31) ask for new information, specifically, for the subject of the onoun as in the (a) examples; the use of pro in this context is infelicitous. Examples (32) and (33) show a yes/no question. Answers to yes/no questions that contain overt subject pronouns are infelicitous as seen in the (a) examples, whereas those contain ing null subject pronouns, as in the (b) examples, are felicitous. 20 As in previous examples, the judgment of this sentence in light of the preceding context has been changed from ungrammatical (*) as in Rothman (2009) to infelicitous (#).

PAGE 103

103 (32) Vieron Uds. la pelcula? see 3. PL PRET you. PL the movie a. # 21 S nosotros la vimos. yes we it CL see 1 PL PRET b. S pro la vimos. yes pro it CL see 1 PL PRET (33) Shoma ha film r didid? you PL movie DOM see 2. PL PRET a. #Are ma didim esh. yea we see 1. PL PRET it b. Are pro didim esh. yea pro see 1. PL PRET it Examples (34) and (35) show subject use in Contrast ive Focus contexts, which are contexts that create a contrast between two entities of the sentence (e.g. the subjects). (34) O te lo digo yo o te lo dice ella. or you DAT it ACC tell.1. SG PRES I or you DAT it ACC tell 3 SG PRES she a. Quiero que me lo digas t. b. #Quiero que me lo digas. pro want 1 SG PRES that me DAT it ACC tell 2. SG PRES SUBJ you (35) Man be to beguyam ya Shabnam be to beguyad. I to you tell 1. SG PRES or Shabnam to you tell 3. SG PRES a. Man mixam ke to be man beguyi. I DUR wa nt.1. SG that you to me tell 2. SG PRES b. #Man mixam ke pro be man beguyi. I DUR want.1. SG that pro to me tell 2. SG PRES 21 This grammatic ality of this sentence was questioned in Rothman (2009), as indicated by (?); however, it has been changed to infelicitous (#) here since it is a licit sentence in Spanish.

PAGE 104

104 We see in the follow up sentences in (34a) and (35a) that the use of the overt subject pronoun t and to pro, in the (b) examples, is infelicitous (but see Matos Amaral & Schwenter 2005). Finally, consider the Phonological Focus examples seen in (36) and (37). (36) Nunca pens que tuvieras que ir a Never pro think 1. SG PRET that pro have 2. SG IMP SUBJ that go INF to buscar el paquete. get INF the package a. Juan me dijo que l lo recogera. John me DAT tell 3. SG PRET that he it ACC get 3. SG COND b. #Juan me dijo que pro lo recogera. John me DAT tell 3. SG PRET that pro it ACC get 3. SG COND (37) Man hich vaqt fekr nemikardam ke to majboor I no time thought NEG IND do 1. SG that you force beshi beravi baste r begiri. SUBJ bec ome. PRES 2. SG SUBJ go. PRES 2. SG package DOM SUBJ grab. PRES 2. SG a. Shamim be man goft ke u miravad va migirad esh. Shamim to me say 3. SG PRET that he DUR go.3. SG and DUR grab.3. SG it b. #Shamim be man goft ke pro miravad va migirad esh. Shamim to me say 3. SG PRET that pro DUR go.3. SG and DUR grab.3. SG it Farsi, like Spanish and English, can assign phonological stress to pronouns in order to establish focus. Clearly, the pronoun must be overt in order to do so; thus, phonological focus is properly established in example (36a) a nd (37a) but not (36b) (37b) due to the lack of the overt subject pronoun in the latter example. If the subject pronoun is null, no phonological stress can be assigned to it. To summarize, Table 3 2 below shows the discourse constrained distribution of re ferential subject pronouns for Spanish and Farsi.

PAGE 105

105 Table 3 2. Subject distribution in Spanish and Farsi Spanish Farsi Topic Maintenance Null Null Topic Shift Overt Overt Contrastive Focus Overt Overt Yes/No question Null Null Subject Focus questi on Overt Overt Phonological Focus Overt Overt It is not surprising that two null subject languages such as Spanish and Farsi should behave similarly with respect to the distribution of overt and null subjects. In fact, data from Judy and Feizmohammadpou r (2012) noted above demonstrates that for Contrastive Focus contexts, the speakers overwhelmingly accept overt subject pronouns and reject null subject pronouns. However, for Topic Maintenance contexts, the speakers overwhelmingly reject overt subject pro nouns and accept null subject pronouns. Finally, there is a clear preference for overt subject pronouns in Topic Shift contexts. These results are expected. Even so, the parallel behavior between the two languages is important for the aims of the current s tudy. The most recent version of the Interface Hypothesis (IH; Sorace 2011) asserts that interface vulnerability obtains as a result of bilingualism (i.e. the presence of two grammars) and, crucially, not because of language specific differences (parametri c, semantic, discourse or otherwise). The testable prediction that derives from this claim is that underlying syntactic and/or discourse similarities between two languages do not give the language learner an advantage with respect to convergence on interfa ce conditioned properties. Thus, near native second language learners of Spanish are predicted to show vulnerability regarding referential subject pronoun distribution in Spanish whether their native language has the opposite parameter setting or the same parameter value and discourse distribution of subjects as in the current study.

PAGE 106

106 3.2.3 Learning Task: NSP While Spanish and Farsi share the same setting for the NSP, important differences exist between the two languages that second language learners must o vercome if they are to converge on related properties. The first difference, as explained in Section 3.2.1 above, is related to word order and movement. Spanish, as opposed to Farsi, does not allow all phrasal elements to remain in the vP. Additionally, wh ile Spanish and Farsi are both head first, right branching languages, Farsi is a verb final language. The second difference pertains to satisfaction of the EPP, with Spanish satisfying this principle via V to Tense movement and Farsi doing so morphological ly. A related effect is that nominative Case is checked and valued in Spec,TP in Spanish, but in learners must also acquire expletives in Spanish. Thus, while the superficial representation of the two grammars appears analogous, the underlying representations are such that grammatical restructuring is necessary in second language acquisition of Spanish by native Farsi speakers. The question that arises then is: how will grammatical restructuring come about? Sprouse, development is failure driven, in the sense that the grammar undergoes revision in response to the inability to must obtain in order for restructuring to occur. In this case, native Farsi speakers learning Spanish as a second language must recognize that word order in Spanish is Subje ct Verb Object. There is ample evidence for this in the input. Furthermore, these same learners must recognize that Spec,vP in Spanish does not check/value nominative Case, but rather that this is accomplished in Spec,TP thus requiring subjects to leave th e vP. Restrictions on adverbial placement in Spanish (see example (38) below) may provide the necessary impetus for

PAGE 107

107 restructuring in this case, ultimately demonstrating to the Farsi speaker that the subject cannot remain within the vP shell: (38)a. Yo s iempre voy a la biblioteca. I always go .1. SG PRES to the library b. *Siempre yo voy a la biblioteca. always I go .1. SG PRES to the librar y

PAGE 108

108 CHAPTER 4 METHODOLOGY 4.1 General Introduction The goals of this chapter are to describe the methodology of the present study as well as to justify the specific tasks employed in light of the Interface Hypothesis (IH). The chapter opens with a brief summary of the IH (see Section 2.2 for details) and a review of the research questions, which together motivate the particular methodological design. The remainder of the chapter describes the participants and methodological des ign in detail. 4.2 Motivation for the Study In an effort to explain the observable differences between ultimate attainment in L1 acquisition and adult second language (L2) acquisition, various global and local impairment hypotheses have been put forward by generative researchers in recent decades, but to date the results are inconsistent. For example, while the Critical Period Hypothesis (Lenneberg 1967; Coppetiers 1987; Johnson & Newport 1989) and its extension to adult L2 acquisition spurred much research within the generative paradigm, the preponderance of the evidence produced by these empirical studies (see White 2003a, 2008, 2011a) indicates that global impairment of this type (i.e. inaccessibility to UG) is improbable. Importantly, adult L2 acquirers have demonstrated convergence on poverty of the stimulus properties for which the information necessary for convergence is (a) not present in the linguistic input, (b) is not explicitly taught nor (c) can be accounted for via domain general learning strate gies (Schwartz 1998; Schwartz & Sprouse 2000; Rothman 2008b; Rothman & Iverson 2008). For instance, Rothman and Iverson (2008) investigate English aspect as well as a related POS semantic prope rty in Brazilian Portuguese. Specifically, in sentences with adverbial quantification of universal force (e.g. sempre

PAGE 109

109 Phrase (DP) subjects such as os romanos or group denoting reading with imperfective as pect, whereas they may only take the group denoting reading in sentences with perfective aspect (Menndez Benito, p. 2001). While the imperfect and preterit aspect is quite frequent in the input the learners are exposed to, the restriction on the interpret ation of the DP subject in adverbially quantified sentences is not available from the input alone nor is this distinction explicitly taught. Since the linguistic input that the learners are exposed to does not provide them with these restricted interpretat ions, nor would it be apparent to the learner that their interpretation was non target like, no domain general learning strategies can be applied. Even so, researchers that assume continued access to UG in adult L2 acquirers are charged with describing and explaining the undeniable differences between L1 and adult L2 attainment. While numerous theoretical proposals have attempted this task over the past decade (e.g. Prosodic Transfer Hypothesis, Goad et al. 2003; Goad & White 2006, 2008; Feature Re Assembly Hypothesis, Lardiere 2008, 2009; Choi & Lardiere 2006, among others), the proposal motivating the present study is the most recent version of the IH (Sorace 2011). The IH, while remaining agnostic with respect to accessibility to UG, attempts to reconcile the observed convergence demonstrated by adult L2 acquirers (White 2003a, 2011a) with the observable ultimate attainment differences between L1 and adult L2 acquisition. By differentiating between properties that require the integration of linguistic and cognitive information (i.e. external interface conditioned properties) and those that only concern linguistic information (i.e. syntactic properties), the IH endeavors to describe and explain convergence in some areas (syntax) and non convergence in others (external interface conditioned properties). Vulnerability in bilinguals is purported to be a result of the increased stress placed on finite cognitive resources (see Wilson et al. 2009; Sorace 2011). That is, unlike monolinguals, who

PAGE 110

110 must manage only one linguistic system, bilinguals must divide their cognitive resources in order to inhibit (but not completely turn off; see Marian & Spivey 2003; Costa, 2005; Dijkstra 2005; Kroll, Bobb, & Wodniecka 2006; Schwartz, Kroll, & Diaz 2007), the non relevant ling uistic system while simultaneously activating the other (Green 1986, 1998; Kroll & Stewart 1994; Bialystok 2009; Kroll, Bogulski, & McClain 2012). Recapping the research questions first outline in Section 1.4, the present study tests the explanatory power of the IH by examining the following research questions. First, the question of whether near native second language learners demonstrate more divergence with respect to external interface conditioned properties as opposed to purely syntactic ones in offli ne tasks is examined. With respect to this question, the IH predicts that near native L2 speakers will show divergence on external interface conditioned properties, such as the discourse distribution of subjects. Convergence on syntactic properties is pred icted to be possible, but not guaranteed, as research has shown that it is less costly to process pure syntax as compared to an interface conditioned property (see Sorace 2011, p. 15 and sources cited therein). Secondly, this study asks whether near native second language learners exhibit greater processing difficulties with external interface conditioned properties as compared to purely syntactic ones in online tasks. Here again, the IH predicts processing difficulties for external interface conditioned pr operties, and possibly, but not necessarily, for syntactic properties. Finally, the question of whether near native second language learners perform more native like in offline tasks as compared to online tasks is explored. Since the IH claims that the sou rce of divergence in near native L2 learners is language processing, it is expected that differences will obtain on the online tasks. However, it is also possible that differences will obtain on the offline and online tasks alike. While a number of studies

PAGE 111

111 language pairings must be tested in order for the findings to be taken as generalizable. Equally as important is the testing method employed in studies testing the pred ictions of the IH. That is, since the IH claims that the presence of two linguistic systems results in an increased processing burden and that this is the locus of differences in ultimate attainment, online measures must be employed in order to determine i f differences in processing are indeed observable. Sorace does however, in the case that convergence on external interface conditioned properties may be is fou nd, one cannot be certain that this is not a product of metalinguistic knowledge. By employing online methodology, the potential effect of metalinguistic knowledge is lessened, and more importantly, the processing of external interface conditioned properti es may be tested via reaction time (RT) data (see Sorace 2011, p. 20). Accordingly, the language pairing, the property tested and the type of experimental tasks of the current study were purposefully selected in order to advance previous research and addr ess the most recent claims of the IH. Specifically, the distribution of referential subject pronouns, although extensively studied before, is examined in a new language pairing: Farsi Spanish bilinguals. While Spanish and Farsi are both null subject langua ges, a combination of the likes that was studied in Bini (1993), Lozano (2002), Margaza and Bel (2006), Prada P rez (2010) and Garc a Alcaraz and Bel (2011), this pairing innovates in a way that the studies cited above do not: the underlying syntax of Fars i and Spanish is different (see Section 3.2), yet they share analogous distributions of referential subject pronouns. Thus, simple transfer of the L1 values would not result in full convergence on all properties tested herein. Rather, the participants are faced with the learning task described in Section 3.2.3. Finally, both offline tasks (which tap

PAGE 112

112 ocessing to be tested. 4.3 Participants Two participant groups completed all non experimental and experimental tasks described in Sections 4.4 and 4.5 below. The native Spanish speaking group consisted of 24 participants while the second language group (L 2) consisted of eight native Farsi speaking adults that were first exposed to Spanish after puberty. All participants had normal or corrected to normal vision and no linguistic or reading impairment. Completion of all tasks was voluntary; however participa nts were monetarily compensated for their time. Native Spanish speaking participants were largely recruited through direct contact with the Facultad de Filosofa y Letras (Sede Puan), Universidad de Buenos Aires, word of mouth among recruited participants and personal contacts. The L2 speakers were contacted through personal contacts. With respect to the native speakers, participation in the study was contingent upon (i) having been exposed to River Plate Spanish since birth and, while most participants had basic knowledge of a second language (in most cases English), (ii) they had little fluency and used the second language minimally (see Section 4.4.1). For the L2 speakers, participation was contingent upon (i) having been born in Iran (and thus exposed to Farsi since birth), (ii) having been first exposed to Spanish naturalistically and post puberty (see Section 4.4.1) and (iii) having scored a minimum of 45/50 on the Proficiency Measure (see Section 4.4.2). Section 4.4 describes the demographic and lingui stic characteristics of both participant groups as well as the proficiency measure, the tasks measuring working 4.4 Non experimental Tasks In addition to the online and offline expe rimental tasks (see Section 4.5), both participant groups completed two non experimental tasks: a Background Questionnaire and a proficiency

PAGE 113

113 measure. The motivation and significance of each task are described below in Sections 4.4.1 through 4.4.4. 4.4.1 B ackground Questionnaire A short Background Questionnaire was completed by all participants in order to obtain demographical information regarding age, gender and country of origin as well as past and present linguistic experience. Specifically, participant s were asked to provide information regarding items such as the age at which they were first exposed to Spanish, the language(s) spoken in the their childhood homes as well as their current homes, the type and amount of formal instruction they have had wit h Spanish, other languages spoken and their current proficiency level. Table 4 1 below summarizes the demographical data for both participant groups while Tables 4 2 through 4 3 and 4 4 through 4 5, respectively, summarize the linguistic information obtain ed from the Background Questionnaire for the native speakers and the L2 speakers. Table 4 1 shows that all native speaker participants were born in Argentina and are native speakers of Argentina Spanish. This is important since the majority of the L2 spea to Spanish is to this distinct dialect; thus, the linguistic group to which the L2 speakers are compared is homogenous in terms of dialect and representative of the dialect to which the L2 speakers are exposed. Of the native speaker particip ants, 14 were female and 10 were male. The average age of the native speaker participants at the time of testing was 29.3 years (range 20 62 years, median 25 years). Regarding the L2 speakers, all were born in Iran. Of the L2 speakers, three were female an d five were male. Their average age was 51.8 years (range 43 58 years, median 53.5 years).

PAGE 114

114 Table 4 1 Demographical information Participant Age Gender Birth Country Native 3 1 21 Female Argentina 4 25 Male Argentina 5 21 Male Argentina 6 27 Female Argentina 7 23 Male Argentina 8 30 Male Argentina 9 24 Female Argentina 10 25 Female Argentina 11 27 Female Argentina 12 38 Male Argentina 13 25 Male Argentina 14 29 Male Argentina 15 48 Male Argentina 16 20 Female Arg entina 17 24 Female Argentina 18 23 Male Argentina 19 25 Female Argentina 20 62 Female Argentina 21 61 Female Argentina 22 25 Female Argentina 23 24 Female Argentina 24 24 Female Argentina 25 30 Female Argentina 26 22 Male Argentina Average 29.3 St. Dev. 11.6 L2 1 56 Female Iran 2 48 Male Iran 3 43 Female Iran 4 44 Male Iran 5 53 Male Iran 6 54 Male Iran 7 58 Female Iran 8 58 Male Iran Average 51.8 St. Dev. 6.02 1 Participants 1 and 2 data were lost. To maintain the faithfulness of the coding s chema for the remaining data, all tables begin with native speaker participant 3.

PAGE 115

115 Table 4 2 provides information regarding the nat background. From Table 4 2, it is evident that while some of the native Spanish speaking participants have knowledge of another language (English in most cases), their dominant language is Spanish. Of the 15 participant s that have knowledge of a language besides Spanish, 13 had studied English (ranging from mostly basic proficiency levels to intermediate and advanced), two had studied French (reaching basic proficiency levels), one had studied Italian and one had studied Portuguese (each reaching basic proficiency levels). Table 4 2 Native speaker linguistic information: languages spoken other than Spanish Native Other languages spoken Proficiency level of other language(s) Dominant lan guage Language spoken as a child at home Languages spoken at home currently 3 English advanced Spanish --------4 --------Spanish --------5 English intermediate Spanish --------6 --------Spanish --------7 English basic Spanish -------8 --------Spanish --------9 English/Italian basic (both) Spanish --------10 English intermediate Spanish Guaran ----11 --------Spanish --------12 --------Spanish --------13 English basic Spanish --------14 English basic Spanish --------15 --------Spanish --------16 English/French basic (both) Spanish --------17 English basic Spanish --------18 --------Spanish --------19 English intermediate Spanish --------20 French b asic Spanish --------21 English basic Spanish ----Daughter learning English 22 --------Spanish --------23 --------Spanish ----Sister learning English 24 English basic Spanish --------25 English basic Spanish --------26 Por tuguese basic Spanish Farsi ----

PAGE 116

116 Only two participants were exposed to a language other than Spanish as a child, but both participants reported having no knowledge (other than select lexical items) of the language. Similarly, only two participants repo rted being exposed to a language other than Spanish in their current home: each had a family member learning English and were therefore exposed to occasional conversations or English media sources. Finally, nine participants claimed to have no knowledge of a language other than Spanish. Table 4 3 below provides further linguistic information for the native speaker participants. All native speaker participants were exposed to Spanish from birth via family members and their surrounding environment. Twenty tw o participants reported having received formal instruction regarding the Spanish language, mostly in primary and secondary language arts courses, but also at the university in some cases. The approximate number of years of formal instruction, assuming prim ary through secondary instruction, is 12 years. Finally, 21 native speaker participants reported using Spanish 2 100% of the time (range 80 3 100%, median 100%). Table 4 3 Native speaker linguistic information regarding Spanish Native Age of first exposure Formal instruction 4 Duration of instruction % daily use of Spanish 3 birth Yes secondary school 100 4 birth No ----100 5 birth Yes primary secondary school 100 6 birth No ----100 7 birth Yes secondary school/uni versity 100 8 birth Yes primary secondary school 100 9 birth Yes university 100 2 This item specifically asked how much time the participants read/write/speak/listen to Spanish (and not another language) on a regular day. 3 Participant 14 reported rea ding the New York Times in English on a daily basis, which is what he counted as 20% use of another language. 4 All participants attended primary and secondary school in Buenos Aires, Argentina and thus likely received formal instruction regarding the Spa nish language.

PAGE 117

117 Table 4 3. Continued Native Age of first exposure Formal instruction 5 Duration of instruction % daily use of Spanish 10 birth Yes primary school university 100 11 birth Yes 12 years 100 12 birth Yes primary school 100 13 birth Yes 12 years 100 14 birth Yes 13 years 80 15 birth Yes secondary school university 100 16 birth Yes primary school university 100 17 birth Yes primary school university 100 18 birth Yes primary school university 100 19 birth Yes primary school university 100 20 birth Yes primary secondary school 95 21 birth Yes primary secondary school 100 22 birth Yes primary secondary school 100 23 birth Yes primary school/unive rsity 100 24 birth Yes primary secondary school 95 25 birth Yes primary school university 100 26 birth Yes 11 years 100 Average 98.75 St. Dev. 4.23 Table 4 4, we see that all L2 speakers were raised in Farsi speaking homes and that in all but one case (participant 6), Farsi is currently spoken at home. Regarding other languages spoken, Farsi is spoken natively by all L2 speakers and four of the participants have basic knowledge of one other language (Arabic in the case of participant 1, Portuguese in the cases of participants 3 and 4 and English in the case of participant 6). However, none of these third languages is currently spoken by the L2 participants. Inter estingly, three of the native Farsi speakers consider their dominant language to be Spanish, while five consider their dominant language to be Farsi. 5 All participants attended primary and secondary school in Buenos Aires, Argentina and thus likely received formal instruction regarding the Spanish language.

PAGE 118

118 Table 4 4. L2 speaker linguistic information: languages spoken other than Spanish L2 Other languag es spoken Proficiency level of other language(s) Dominant language Language spoken as a child at home Languages spoken at home currently 1 Farsi, Arabic native, basic Farsi Farsi Farsi 2 Farsi Native Spanish Farsi Spanish, Farsi 3 Farsi, Portuguese nat ive, basic Farsi Farsi Farsi 4 Farsi, Portuguese native, basic Farsi Farsi Farsi 5 Farsi, Portuguese native, basic Spanish Farsi Farsi 6 English Basic Farsi Farsi Spanish 7 Farsi Native Spanish Farsi Farsi 8 Farsi Native Farsi Farsi Farsi Table 4 5. L2 speaker linguistic information regarding Spanish L2 Age of first exposure Formal instruction 6 Duration of instruction % daily use of Spanish 1 22 Yes One class per week, less than one year 95 2 17 Yes 2 years of secondary school 100 3 18 No 90 4 21 No 90 5 19 No 100 6 17 Yes Less than 5 months 95 7 26 No 100 8 24 No 80 Average 20.5 93.75 St. Dev. 3.34 6.94 Table 4 5 above provides further linguistic information for the L2 speaker participants. All L2 speaker participants we re exposed to Spanish for the first time post puberty at an average age of 20.5 years (range 17 26 years, median 20). Only three participants reported having received formal instruction regarding the Spanish language, while the other five participants repo rted having learned the language naturalistically. Those that attended formal classes reported 6 All participants attended primary and secondary school in Buenos Aires, Argenti na and thus likely received formal instruction regarding the Spanish language.

PAGE 119

119 learning more from their environment than from the classroom. Finally, the self reported average percent daily use of Spanish was 93.75% (range 80 100%, median 9 7.5%). 7 4.4.2 Proficiency Measure Proficiency measures are now standard practice in L2 acquisition studies. However, since the claims made by the IH are only valid for near native speakers (non convergence at lower proficiency levels may simply be a resul t of insufficient exposure to input), the proficiency scores obtained herein are crucial for the goals of the experiment. Hulstijn (2010) reviewed several proficiency tests, two of which are relevant for the goals of this study: the vocabulary and cloze te sts. Vocabulary and cloze tests were rated as highly efficient in terms of test construction, administration and analysis (Hulstijn 2010). Additionally, Hulstijn claims that the vocabulary test is very reliable while the cloze test ostensibly measures ever ything from productive vocabulary knowledge, productive and receptive grammar knowledge, orthographic knowledge and knowledge of semantics, pragmatics and discourse. For these reasons, all participants (in the case of the L2 speakers, to ensure near native proficiency) completed an abbreviated version of the Diploma del Espaol como Lengua Extranjera (DELE 2002), which is widely used in generative second language acquisition studies of Spanish in North America (e.g. Montrul 2000, 2002, 2005, 2009; White et al. 2004; Cuza & Frank 2011; Leeser et al. 2011; Prada Prez & Pascual y Cabo 2011, 2012; Pascual y Cabo et al. 2012; Slabakova et al. 2012). The abbreviated version of the DELE is comprised of 50 items divided across two sections: a vocabulary section (n= 30) and a cloze section (n=20). 7 As with any self reported data, the data regarding how participants felt they learned Spanish and how much they use Spanish on a daily basis, should be taken with a certain le vel of reservation. However, it is clear that great differences exist regarding the amount of formal instruction about Spanish the native speaker participants and the L2 speakers received. Regarding the percent daily use of Spanish (and not another languag e), it should be noted that each of the L2 speakers forms part of the workforce, actively participates in religious events in the community, seven of eight have Spanish speaking children and the majority of their media sources are in Spanish.

PAGE 120

120 Table 4 6. Proficiency scores Participant Vocabulary Cloze Composite Native 3 29 19 48 4 30 19 49 5 29 20 49 6 30 20 50 7 30 17 47 8 30 20 50 9 28 19 47 10 30 20 50 11 30 19 49 12 30 20 50 13 30 20 50 14 30 19 49 15 30 19 49 16 30 20 50 17 30 18 48 18 29 20 49 19 30 20 50 20 30 20 50 21 29 17 46 22 30 20 50 23 28 19 47 24 29 19 48 25 30 16 46 26 30 18 48 Average 29.63 19.08 48.71 St. Dev. 0.65 1.14 1.33 L2 1 29 18 47 2 30 1 8 48 3 29 16 45 4 29 16 45 5 28 17 45 6 30 18 48 7 28 17 45 8 29 18 47 Average 29.00 17.25 46.25 St. Dev. 0.76 0.89 1.39

PAGE 121

121 In Table 4 6 above, the scores of each section as well as the composite score for both participant groups are shown The average vocabulary and cloze component scores as well as the composite proficiency scores for the L2 speaker participants (29, 17.25 and 46.25, respectively) are similar to those of the native speaker participants (29.63, 19.08 and 48.71, respectively). A dditionally, as the native speaker participants (28 30 and 16 18, 28 30 and 16 20, respectively). The standard cutoff for near native proficiency is 45, which all L2 speaker participants meet. 4.5 Experimental Tasks In this section, the experimental tasks are described in detail and examples of all token types are provided. In total, there are five experimental tasks. The three offline tasks were completed in Micr osoft Word documents while the two online tasks were completed via E prime software (Schneider, Eschman & Zuccolotto 2002). The specific tasks, the inclusion of both online and offline versions and the particular properties examined were expressly employed to processing differences for certain linguistic properties. 4.5.1 Offline Experimental Tasks In order to capture behavioral data, two offline experimental task s were completed by both participant groups. Obtaining behavioral (offline) and processing (online) data from the same participant groups is an important aspect of this study in that it makes it possible to test the ay be different. That is, while bilinguals are predicted to differ from monolinguals on online measures according to the IH, convergence on offline measures is not precluded. By using the same experimental stimuli in the offline and online tasks, this clai m can be tested.

PAGE 122

122 In total, two offline tasks were completed by both participant groups. Each task was completed in a Microsoft Word document on the same laptop computer on which the online tasks were completed. The offline version of the Grammaticality Ju dgment/Correction Task is described below in Section 4.5.1.1, followed by the offline version of the Context Matching Felicitousness Task in Section 4.5.1.2. Examples of all experimental stimuli are provided in each section. 4.5.1.1 Grammaticality Judgment /Correction Task The GJCT was designed to test for knowledge of the syntactic distribution of overt and null subjects in Spanish. It contained 106 tokens (n=24 subject tokens, n=72 Differential Object Marking tokens 8 n=10 filler tokens; see Appendix A). I n this task, participants were informed that they would read short sentences written in a Microsoft Word document. Upon reading each sentence, participants were asked to correct any sentences they deemed as having errors 9 Tokens related to subjects are de scribed in Sections 4.5.2.1.1 and 4.5.2.1.2 below while the filler agreement tokens are described in Section 4.5.2.1.3. 4.5.1.1.1 Referential subject tokens. Of the 24 tokens related to subjects, half were related to referential subjects (n=12) and half we re related to expletive subjects (n=12). Of the referential subject tokens, half were designed to test for acceptance of null referential subjects (n=6) and half for overt referential subjects (n=6). As discussed in Section 3.2.1, both null and overt refer ential subjects are possible in both Farsi and Spanish. Examples (39a) and (39b) below show a null and an overt referential subject token, respectively. 8 The Differe ntial Object Marking (DOM) tokens will not be discussed in the present study. 9 Participants were instructed to do nothing to sentences they deemed as error free. Additionally, they were instructed to circle a box containing the phrase Agramatical, no s corregirla

PAGE 123

123 (39)a. En mi opinin pro siempre 10 toma muchas vitaminas por la maana. in my opinin pro always take 3. SG PRES many vitamins for the morning b. En mi opinin ella toma muchas vitaminas por la maana. in my opinin she take 3. SG PRES many vitamins for the morning Half (n=3) of the overt referential subject tokens contained the third person singular masculine referential subject pronoun (i.e. l 3) contained the third person singular feminine referential subject pronoun (i.e. ella equally distributed across the null and overt referential subject tokens: toma hace come is, the verb toma Grammaticality Judgment/Correction Task: once with the third person masculine overt referential subject l ella 11 and twice with n ull subjects. The verbs were transitive activities, all conjugated for third person singular present tense to control for person and tense variables as per research in the variationist tradition. Barrenechea and Alonso (1977) and Silva Corval n (1994) repo rted that overt subject pronouns were more likely to be used in singular than in plural and a comparison across Quesada and Blackwell (2009) and Blackwell and Quesada (2010) reveals that first person singular was used more frequently in the Mexican speaker s studied therein (this was also found in Lapidus Shin (2012) for Mexican children. Silva Corval n (1982) reported less likelihood of overt subject pronouns with preterite and present tense than with imperfect or 10 Unlike the overt referential subject token seen in example (39b), the null subject token seen in example (39a) contains the adverbial siempre Grammaticality Task: the overt subject pronoun, verb and direct object serve as the regions of interest examined in the overt tokens, whil e the adverbial, verb and direct object serve as the regions of interest examined in null tokens. 11 Note that neither I nor the IH make predictions with respect to differences between masculine and feminine referential subject pronouns as I am not aware o f research that has shown an effect of gender here.

PAGE 124

124 conditional tense and continuity of tense a spect mood (TAM) has been shown to favor null subjects (Silva Corval n 1982, 1994; Otheguy et al. 2007). 4.5.1.1.2 Expletive subject tokens. Of the 12 expletive subject tokens, half were designed to test for acceptance of pro (n=6) and half for rejection of overt subjects (n=6) since, with the exception of some Caribbean dialects (Toribio 2000; Martnez Sanz 2011), Spanish does not allow for overt expletive subjects (Jaeggli 1982; Rizzi 1982; Picallo 1998). Example (40a) shows a null expletive token while (40b) shows an overt expletive token 12 (40)a. Los vecinos dicen que pro siempre llueve mucho en esa regin. the neighbors say 3. PL PRES that pro always rain 3. SG PRES mucho in this region b. *Los vecinos dicen que ello llueve mucho en esa regin. the neighbors say 3. PL PRES that it rain 3. SG PRES mucho in this region Only three weather verbs were employed in this token type ( llueve nieva llovizna tokens. Therefore, participants saw each verb four times, twice in null subject tokens and twice in overt subject tokens. The ungrammatical subject pronoun ello tokens. 4.5.1.1.3 Fillers. attention from the actual experimental tokens described abo ve as well as to ensure that they were paying attention to the stimuli. Six of the filler tokens contained subject verb number agreement errors of the type seen in example (41) below while four contained noun adjective number agreement errors such as those seen in example (42) below. 12 Section 3.2.1 discussed one difference between Spanish and Farsi: Farsi optionally permits in position. As Karimi (2005) claims that this is a demonstrative, the Spanish equivalent dem onstrative este could be considered a viable candidate for the ungrammatical overt expletive subject. However, este is commonly t ello was used.

PAGE 125

125 (41). *Las princesas ricas compr muchas decoraciones de la the princesses rich bought 3. PL PRET many decorations from the mueblera. furniture store (42). *Los polticos corrupto aceptaron muchos sobornos la semana the politicians corrupt accept 3. PL PRET many bribes the week pasada. last Importantly, the offline version of the Grammaticality Judgment/Correction Task was taken after the online version in order to avoid priming effects. The next section describes the second offline tas k. 4.5.1.2 Context Matching Felicitousness Task The second offline task was a CMFT, the goal of which was to test for knowledge of the distribution of overt and null subjects in three discourse constrained contexts. This task contained 36 tokens (n=12 Cont rastive Focus tokens, n=12 Topic Shift tokens, n=12 Topic Maintenance tokens; see Appendix B). Participants were instructed to read short contexts followed by a short sentence, which they were to judge based on the context. Specifically, participants were asked to judge the target sentence based on how well it was expressed on a scale of 1 (100% bien mal 13 Contrastive Focus tokens are described in Section 4.5.1.2.1, Topic Shift tokens are described in Section 4.5.1.2.2 and Topic Ma intenance tokens are described in Section 4.5.1.2.3. 4.5.1.2.1 Contrastive Focus. All Contrastive Focus contexts employed in this study introduced two subjects such that a contrast between them could be made. Each context began with an adverbial time phra se such as Cuando salimos a cenar 13 The use of a reverse scale (where the better linguistic option is scored lower than the inferior options) was an oversight. Future research of the same kind will employ standard scales.

PAGE 126

126 the two subjects employed were always the third person singular feminine DP, such as mi novia yo s tokens contained overt subject pronouns as in (43a) while half (n=6) contained null subject pronouns as in example (43b) below. Recall from Section 3.2.2.2 that overt subject pronouns are felicitous in Contrastive Focus contexts, whereas null subject pro nouns are infelicitous (but see Amaral Matos & Schwenter 2005). (43). Cuando salimos a cenar, mi novia prefiere comer platos livianos, pero yo prefiero comer algo sustancioso. fer to eat a. As que ella come ensaladas y yo como milanesas en so that she eat 3. SG PRES salads and I eat 1. SG PRES breaded meats in los restaurantes the restaurants. b. #As que pro come ensaladas y pro como milanesas en so that pro eat 3. SG PRES salads and pro eat 1. SG PRES breaded meats in los restaurantes. th e restaurants. 4.5.1.2.2 Topic Shift. All Topic Shift contexts also introduced two subjects such that a shift in reference can be made. The third person singular feminine subject was alw ays introduced in the first sentence of the context. Then, the first person singular subject pronoun was introduced in the second sentence and used in any subsequent sentences. As variationist literature has found a positive correlation for distance since last mention (Cameron 1995; Flores Ferr n 2002; Cameron & Flores Ferr n 2004; Travis 2005, 2007), a minimum of three verbs conjugated for first person singular separated the first sentence of the context in which the third person referent was introduced an d the target sentence in which it was reintroduced, reinforcing the Topic Shift context. Half (n=6) of the Topic Shift tokens contained overt subject pronouns as in

PAGE 127

127 (44a) while half (n=6) contained null subject pronouns as in example (44b) below. Like Cont rastive Focus contexts, overt subject pronouns are preferred in Topic Shift contexts (see Section 3.2.2.2; Bentivolgio 1987; Silva Corval n 1994; Cameron 1994, 1995; Flores Ferr n 2002; Cameron & Flores 2003; Travis 2005, 2007). (44). Mi hija quiere ser a utora y no tiene otros intereses. Yo creo que es mejor tener varios intereses y sugiero otras actividades, pero no importa lo que diga yo. have various interests a. Finalmente ella escribe cuentos y pro pasa todo el da finally she write 3. SG PRES stories and pro spend 3. SG PRES all the day en su cuarto. in her bedroom b. #Finalmente pro escribe cuentos y pro pasa todo el da finally pro write 3. SG PRES stories and pro spend 3. SG PRES all the day en su cuarto. in her bedroom 4.5.1.2.3 Topic Maintenance. Differently from the two contexts described above, the Topic Maintenance con texts employed in this task only introduced one third person singular feminine subject. This subject was initially introduced in the first sentence of the context. From there, the context contained a minimum of three verbs conjugated for third person singu lar with null subjects. Half (n=6) of the Topic Shift tokens contained overt subject pronouns as in (45a) while half (n=6) contained null subject pronouns as in example (45b) below. Recall from Section 3.2.2.2 that overt subject pronouns are less preferred in Topic Maintenance contexts; instead, the null subject pronoun is preferred (Bentivolgio 1987; Silva Corval n 1994; Cameron 1994, 1995; Flores Ferr n 2002; Cameron & Flores 2003; Travis 2005, 2007). (45). Mi cuada es muy sociable. Tiene muchos amigos y por eso va a muchas cenas a la canasta donde tiene que contribuir con algo.

PAGE 128

128 in law is very social. She has a lot of friends and for that reasons, she goes a. #As que ella lleva postres y pro comparte todo con so that she take 3. SG PRES desserts and pro share 3. SG PRES all with sus amigos. her friends b. As que pro lleva postres y pro comparte todo con so that she take 3. SG PRES desserts and pro share 3. SG PRES all with sus amigos. her friends Importantly, all subjects, matrix clause verbs and DOs were counterbalanced across token type (Contrastive Focus, Topic Shift and Topic Maintenance) and subject type (null or overt) in order for the tokens to be maximally comparable. Twelve matrix clause verbs ( come hace prepara toma ofrece lleva vende escribe lee compra pinta dibuja the three token types and sub ject types. All were transitive verbs (and most were activities) and as described above in Section 4.5.2.1.1, tense and person were controlled for. The same is true for the 12 matrix clause DOs ( ensaladas alfajores empanadas agua caf postres revistas cuentos poemas esculturas paisajes figuras was taken after the online version to avoid priming. 4.5.2 Online Experimental Tasks In addition to the offline tasks, complementary online tasks were employed in E prime software (Schneider, Eschman & Zuccolotto 2002) in order to capture RT data for the regions of interest and corresponding spillover regions. This was done to determine if these regions of interest are processed differently (i.e. more slowly) in the L2 speaker group than in the native

PAGE 129

129 speaker group. Specifically, a Self Paced Reading Task (SPRT; Aaronson & Scarborough 1976; Mitchell & Green 1978) was adopted for both online tasks. In order to mimic normal reading and to measure RT for each designated region of interest, a non cumulative moving window technique (Just, Woolley & Carpenter 19 82) was employed. In this type of display, participants are presented with a blank screen that contained several underscored sentence items whose content only became visible as the participant pressed a button to advance along in the sentence in a region b y region fashion. This way, participants saw the sentence in a non cumulative manner (only one region was visible at a time) and in linear fashion (as opposed to the centered technique in which each region is presented in the center of the screen). Example (46) provides a short English example of this technique. Each line below (e.g. (46a), (46b), etc.) represents what the participant would see as they progressed through the sentence, which starts with all regions of the sentence being hidden (as in (46a)) and ends with the final word (as in (46f)), but never seeing more than one region at a time. (46)a. ___ _____ ___ __ _______ b. The _____ ___ __ _______ c. ___ black ___ __ _______ d. ___ _____ cat __ _______ e. ___ _____ ___ is _______ f. ___ _____ ___ __ hungry. Upon completion of the moving window (i.e. having read the entire sentence in the region by region fashion described above), participants were asked to answer a short meaning based yes/no comprehension question designed to focus their attention on the content of the experimental stimuli and not the grammaticality/felicitousness of the tokens themselves. Sections 4.5.2.1 and 4.5.2.2 below describe the online version of the Grammaticality Task and the online versio n of

PAGE 130

130 the Context Matching Felicitousness Task, respectively, in detail and provide examples of all experimental stimuli. 4.5.2.1 Online Grammaticality Task Like the offline version, the online GT was designed to test for knowledge of the syntactic distrib ution of overt and null subjects via RT data. It also contained 106 tokens, 24 of which were subject tokens, 72 of which were DOM tokens 14 and 10 of which were filler agreement tokens (see Appendix C for examples). In this task, participants were asked to r ead short sentences via the moving window technique described above in Section 4.5.2 and then answer brief yes/no comprehension questions about each sentence. Comprehension questions were employed to check for understanding of the sentences and mask the pu rpose of the task. Participants were instructed to read the sentences at a normal rate and respond to the comprehension questions as quickly and as accurately as possible. Prior to the experimental task, participants completed a training session consisting of eight practice tokens, after which they were instructed to clarify any questions they had with the researcher. The actual experiment began with two additional non experimental tokens, after which the 106 experimental tokens were presented in random fas hion in E prime. Sections 4.5.2.1.1 and 4.5.2.1.2 below describe the subject tokens while Section 4.5.2.1.3 describes the filler agreement tokens. 4.5.2.1.1 Referential subject tokens. The same token distribution described in 4.5.1.1 was employed in the on line version. Yet, a s this task was designed to measure the RT to specific regions of interest within the sentence, sentences of the same token type were purposefully and consistently divided to allow for statistical comparison of the RTs to each region of interest. 14 The 72 experimental tokens examining the distribution of DOM are not discussed herein.

PAGE 131

131 With respect to the regions of interest, referential subject tokens were divided as follows, where the forward slash (/) indicates a division in the presentation of the token in E prime. (47)a. En mi opinin/siempre/toma/muchas/vitaminas/por l a maana. b. En mi opinin/ella/toma/muchas/vitaminas/por la maana. That is, in example (47a), the first region seen by participants is en mi opinin then siempre tokens such as example (47a) above are siempre toma muchos/muchas ce they surround the syntactic position of the verb (between the adverb siempre muchas vitaminas regions of interest analyzed for RT in thes e tokens above are ella toma spillover region muchos/muchas region neither began nor ended the token as those regions may have disproportionate RTs (Just & Carpenter 19 80). Additionally, the spillover region was the masculine or feminine plural form of the quantifier mucho muchos or muchas respectively). 4.5.2.1.2 Expletive subject tokens. As with the previous token type, all null and overt expletive subje ct tokens were purposefully and consistently divided to allow for statistical comparison of the RTs to each region of interest. With respect to the regions of interest, tokens such as (48a) and (48b) above were divided as follows (again, the forward slash (/) indicates a division in the presentation of the token in E prime). (48)a. Los vecinos/dicen/que/siempre/llueve/mucho/en esa regin. b. *Los vecinos/dicen/que/ello/llueve/mucho/en esa regin.

PAGE 132

132 The regions of interest in expletive subject tokens such as example (48a) above are siempre llueve mucho analysis since they surround the syntactic p osition of the weather verb (between the adverb siempre mucho the regions of interest analyzed for RT of this token type are ello llueve spillover region mucho 4.5.2.1.3 Fillers. Finally, the same 10 filler tokens from the offline version were above. Six contained subject verb number agreement errors o f the type seen in example (49) below, while 4 contained noun adjective number agreement errors such as those seen in example (50) below. (49). *Las princesas ricas compr muchas decoraciones de la mueblera. (50). *Los polticos corrupto aceptaron muchos sobornos la semana pasada. While the filler tokens will not be examined with respect to RT, they were presented with the same moving window technique and the regions were divided in the same manner. Experimental tokens and filler tokens alike were randomly distributed by E prime. 4.5.2.2 Online version of Context Matching Felicitousness Task The final onli ne task was an online version of the CMFT that was designed to test for knowledge of the distribution of overt and null subjects via RT data in three discourse constrained contexts. This task contained the same 36 contexts employed in the offline version ( n=12 Contrastive Focus tokens, n=12 Topic Shift tokens, n=12 Topic Maintenance tokens; see Appendix D). Recall that each context had two possible target sentences. In the case that a

PAGE 133

133 participant saw the overt subject version of the target sentence in the o ffline task, the same participant saw the null subject version of the target sentence in the online task. In this task, participants were informed that they would read short contexts presented in their entirety followed by a sentence presented via a moving window task (see Section 4.5.2) and then answer a brief yes/no comprehension question about each sentence 15 They were instructed to read the sentences at a normal rate and respond to the comprehension questions as quickly and as accurately as possible. Pa rticipants completed a training session consisting of three practice tokens and then were instructed to clarify any questions they had with the researcher. The actual experiment began with two additional non experimental tokens, after which all 36 experime ntal tokens were presented in random fashion in E prime. Sections 4.5.2.2.1, 4.5.2.2.2 and 4.5.2.2.3 below describe the Contrastive Focus, Topic Shift and Topic Maintenance tokens, respectively. 4.5.2.2.1 Contrastive Focus. The same token distribution desc ribed in Section 4.5.2.1 was employed in the online CMFT. The context was seen and read prior to the target sentence. All Contrastive Focus target sentences were purposefully and consistently divided for presentation in E prime to allow for statistical com parison of the RTs for each region of interest. With respect to the regions of interest, the target sentences were divided as shown in example (51), where the forward slash (/) indicates a division in the presentation of the token in E prime. (51) a. As que/ella/come/ensaladas/y/yo/como/milanesas/en los restaurantes. b. #As que/come/ensaladas/y/como/milanesas/en los restaurantes. The regions of interest to be examined in tok ens like (51a) which contain overt subjects are the subject, verb and the DO (the spillover region) of the matrix clause, in this case: ella come 15 For example, Mi novia come ensaladas?

PAGE 134

134 ensaladas regio ns of interest are come ensaladas 4.5.2.2.2 Topic Shift. The Topic Shift tokens were divided in the same manner such that statistical comparisons of the RTs for each region of interest could be performed. Specifically, the target sent ences were divided as seen in example (52), where the forward slash (/) indicates a division in the presentation of the token in E prime. Recall that the context was seen and read prior to seeing the target sentence. (52) a. Finalmente/ella/escribe/cuent os/y/ella/pasa/todo/el da/en su cuarto. b. #Finalmente/escribe/cuentos/y/pasa/todo el da/en su cuarto. The regions of interest to be examined in tokens like (52 a) which contain overt subjects are the subject, verb and the DO (the spillover region) of the matrix clause, in this case: ella escribe cuentos subjects, the regions of interest are escribe cuentos 4.5.2.2.3 Topic Maintenance. Finally, and as with the other context types, all Topic Maintenance tokens were divided as seen in example (53) such that statistical comparisons of the RTs for each region of interest could be conducted. (53)a. #As que/ella/lleva/postres/y/comparte/todo/con sus amigos. b. As que/lleva/postres/y/comparte/todo/con sus amigos. The regions subject, verb and the DO (the spillover region) of the matrix clause, in this case: ella lleva postres ain infelicitous null subjects, the regions of interest are lleva postres

PAGE 135

135 As described at the close of Section 4.5.2.2.3, all subjects, matrix clause verbs and DOs were counterbalanced across token type (Contrastive Focus, Topic Shi ft and Topic Maintenance) and subject type (null or overt) and the online tasks were taken before the offline tasks to avoid priming. 4.6 Conclusion This chapter presented the methodological design of the study as well as the theoretical claims motivating it. Since the IH maintains that differences in processing are a likely source of non convergence, especially for external interface conditioned properties in near native speakers, both online and offline tasks examining the distribution of overt and null subjects were described. Collecting online and offline data makes it possible to determine if participants perform differently according to the task, which in turn gives insight into their processing abilities as compared to the associated behavioral data collected.

PAGE 136

136 CHAPTER 5 RESULTS 5.1 General Introduction The goals of this chapter are to present the results of the offline and online experimental tasks described in Chapter 4 and to interpret their respective results. Results for both the native speaker an d second language (L2) speaker participant groups are provided, first for the offline experimental tasks and then for the online experimental tasks. For each task, the descriptive results are presented first along with the statistical analysis and, finally the interpretation of the results. An analysis of how these findings relate to the research questions of this dissertation will be offered in Chapter 6. 5.2 Offline Experimental Results e of the distribution of subjects in Spanish were described in Section 4.5.1: the Grammaticality Judgment/Correction Task and the Context Matching Felicitousness Task. A brief recap of the purpose of each task, the types of tokens examined and the empirica l findings of token type are presented below. 5.2.1 Grammaticality Judgment/Correction Task (GJCT) As described in Section 4.5.1.1, the goal of the GJCT was to test for knowledge of the syntactic distribution of overt and null subjects across 106 tokens, 24 of which examined the distribution of subjects, 72 of which examined the distribution of Differential Object Marking (DOM) and 10 of which were fillers. Recall that in this task, participants were simply asked to read short sentences written in a Micros oft Word document. Since some of the sentences were

PAGE 137

137 ungrammatical, participants were asked to correct any sentences they deemed as having errors 1 The following sections first discuss the referential and expletive subject tokens. 5.2.1.1 Referential subjec t tokens The 12 tokens related to referential subjects were designed to test for acceptance of overt and null referential subject tokens (RSO and RSN, respectively) both of which are grammatical in Spanish (see Section 3.2.1). An equal number of tokens fo r each subject type was employed. Example (39), which is reproduced below as (54), shows an RSN and a RSO token, respectively. Recall that the overt referential subjects were always third person singular (divided equally between masculine l ine ella toma hace come subject tokens (see Appendix A). (54)a. En mi opinin pro siempre toma muchas vitaminas por la maana. b. En mi opinin ella toma muchas vitaminas por la maana. Figure 5 1 below shows the group average number of items accepted for both the RSO and RSN tokens. An ungrammati cal item was only deemed to be accurately rejected (i.e. not accepted) when the participant made the appropriate correction. All participants accepted 6 of 6 RSO tokens. With the exception of one native speaker participant (19), all native speaker and L2 s peaker participants accepted a minimum of 4 of 6 RSN tokens. The average number of RSN tokens accepted by the native speaker group is 5.50 (standard deviation 1.31) while the average number of tokens accepted by the L2 speaker group is 5.75 (standard devia tion 0.46). These results are analyzed along with the results for the expletive subject tokens in Section 5.2.1.4. 1 Recall that when the participa nts deemed the sentence to be error free, they were instructed to simply continue on with the task. Additionally, participants were instructed to circle a box containing the phrase Agramatical, no s corregirla hey felt the sentence was ungrammatical, yet were not sure how to correct it.

PAGE 138

138 Figure 5 1. GJCT group results: referential subject tokens. NS=native speaker; L2=second language speaker; RSO=overt referential subject ; RSN=null referential subject 5.2.1.2 Expletive subject tokens The 12 tokens related to expletive subjects were designed to test for acceptance of pro (n=6) and for rejection of ello (n=6) in expletive subject contexts. Recall from Section 3.2.1 that in Argentine Spanish, pro is required in expletive subject contexts. Example (55), reproduced below as (40), shows a null expletive subject token (ESN) and an overt expletive subject token (ESO). Recall that three weather verbs were equally dispersed across t he null and overt expletive subject tokens: llueve nieva llovizna llueve the ungrammatical overt subject el lo (55)a. Los vecinos dicen que pro siempre llueve mucho en esa regin. b. *Los vecinos dicen que ello llueve mucho en esa regin. Figure 5 2 below provides the grou p averages for both the overt and null expletive subject tokens. With the exception of one L2 speaker participant (7), all native speaker and L2

PAGE 139

139 speaker participants accurately rejected all ESO tokens. The average number of overt expletive subject tokens a ccepted by the native speaker group is 0.00 (standard deviation 0.00) while the exception of one L2 speaker participant (2), all native speaker and L2 speaker part icipants accepted all ESN tokens. The average number of ESN tokens accepted by the native speaker group is 6.00 (standard deviation 0.00) while the average number of tokens accepted by the L2 speaker participant group is 5.88 (standard deviation 0.35) 2 Figure 5 2. GJCT group results: expletive subject tokens. NS=native speaker; L2=second language speaker; ESO=overt expletive subject; ESN=null expletive subject 5.2.1.3 Statistical analysis and interpretation of results A Generalized Mixed Models ANOVA with Subject Type (overt or null), Condition Type (referential or expletive) and Group (native speaker or L2 speaker) was conducted. Finding 2 L2 speaker participant P02 changed a grammatical ESN token that read Mis amigas dicen que ello llueve mucho en la costa Mis amigos dicen que hay mucha lluvia en la costa llover hay ticipant P07 failed to correct an ungrammatical token that read La doctora dice que ello llueve mucho en el verano

PAGE 140

140 a main effect for Subject Type would indicate that overt and null subjects were treated differently. Similarly, fi nding a main effect for Condition Type would indicate that referential and expletive subjects were treated differently. Finally, finding a main effect for Group would indicate that the native speaker and L2 speaker participants performed differently. Findi ng an interaction between Subject Type and Condition Type would indicate that overt and null subjects are treated differently in referential and expletive tokens. Finding interactions between Subject Type and Group would indicate that overt and null subjec ts were treated differently by the two participant groups. Finding interactions between Condition Type and Group would indicate that expletive and referential tokens were treated differently by the participant groups. Finally, finding an interaction betwee n Subject Type, Condition Type and Group would indicate that the native speaker and L2 speaker participants treated overt and null subjects in referential and expletive conditions differently from each other. Bonferroni post hoc two way and three way analy ses were also conducted to examine the sources of any interactions that obtained. The alpha level was set at 0.05; thus, p values less than 0.05 are reported as statistically significant. 3 The sequential Bonferroni adjusted significance level is 0.05. The analysis revealed a significant main effect for Subject Type (F(1,120) 71.210, p = 0.000) and Condition Type (F(1,120) 707.017, p = 0.000) but not for Group (F(1,120) 0.097, p = 0.766). There was an interaction between Subject Type and Condition Type (F(1 ,120) 3,120.097, p = 0.000) and an interaction between Subject Type and Group (F(1,120) 969.897, p = 0.000), but not between Condition Type and Group (F(1,120) 0.131, p = 0.718). Finally, a three way interaction between Subject Type, Condition Type and Gro up was found (F(1,120) 484.122, p = 0.000). 3 This is true of all analyses conducted herein and will, therefore, not be repeated.

PAGE 141

1 41 In order to compare the differential treatment between overt and null subjects, a post hoc pairwise t tests for independent samples (with Bonferroni adjustment) were conducted across conditions and groups. Nati ve speakers accepted overt and null subjects in referential tokens to the same degree ( p = 0.083), which is an expected finding since both are grammatical. With respect to expletive tokens, native speakers accepted null subjects significantly more than ove rt subjects ( p = 0.000), which is expected since only null expletives are permitted in Argentine Spanish. Like the native speaker group, the L2 speaker group accepted overt and null subjects in referential tokens to the same degree ( p = 0.139), an expected result as both are grammatical. This is true for both groups despite the difference in categorical acceptance of the RSO tokens and non categorical acceptance of the RSN tokens (four native speakers participants changed the verbal morphology of the verb a nd two added a subject; one L2 speaker participant added a subject). This is not problematic, however, as both overt and null referential subjects are permissible in Spanish. The L2 speaker group also accepted null subjects significantly more than overt su bjects ( p = 0.000), which is expected if their grammar does not permit overt expletive subjects. Regarding the comparison between referential and expletive tokens, the post hoc pairwise t tests for independent samples (with Bonferroni adjustment) showed t hat both the native speaker and L2 speaker groups accepted significantly more RSO tokens than ESO tokens ( p = 0.000 and p = 0.000), which is also expected since overt subjects are grammatical in referential contexts only. In contrast but as expected, both groups accepted referential and expletive tokens with null subjects to the same degree ( p = 0.083 and p = 0.559, respectively). Finally, with respect to the comparison between the native speaker and L2 speaker groups, the post hoc pairwise t tests for ind ependent samples (with Bonferroni adjustment)

PAGE 142

142 showed that the two groups accepted the referential tokens with overt subjects to the same degree ( p = 0.426). The same is true of the referential tokens with null subjects ( p = 0.588). Regarding the expletive tokens with overt subjects, both groups accepted the expletive tokens with overt subjects to the same degree ( p = 0.286). Lastly, both the native speaker and L2 speaker groups accepted the expletive tokens with null subjects to the same degree ( p = 0.329). All of these results are expected if both groups have converged on the syntactic distribution of subjects in Spanish. 5.2.1.4 Summary of Grammaticality Judgment/Correction Task results In conclusion, both the native speaker and L2 speaker group showed dif ferential treatment of overt and null subjects in expletive subject contexts, while accepting both subject types in referential subject contexts. The near rejection of overt expletive subjects and simultaneous near categorical acceptance of null expletive subjects in Spanish provides evidence of convergence on the Spanish syntax of subjects. Additionally, both groups differentiated between context type in the presence of an overt subject, but not null subjects. Finally, the results described in the precedi ng paragraph indicate that the L2 speaker group does not differ from the native speaker group with respect to their acceptance of the four token types described in Sections 5.2.1.2 (RSO and RSN) and 5.2.1.3 (ESO and ESN). Perhaps more importantly, both the native speaker and L2 speaker groups made a statistically significant distinction (p = 0.000 for both groups) between the ESO and RSO tokens, showing that they reject overt subjects in ungrammatical expletive tokens but accept them in grammatical referent ial tokens. Taken together, the findings outlined above are in line with theoretical syntactic descriptions (Rizzi 1982; Jaeggli & Hyams 1988) in that overt expletive subjects are not permissible in Spanish, but overt and null referential subjects are.

PAGE 143

143 5.2 .2 Context Matching Felicitousness Task (CMFT) As described in Section 4.5.2.1, the objective of the second offline task, the CMFT, was to test for knowledge of the discourse distribution of overt and null subjects in three specific contexts: Contrastive F ocus, Topic Shift and Topic Maintenance. The 36 tokens were divided evenly across these contexts and an equal number of tokens of each type contained overt subjects (always ella s read the context and following target sentence, which they judged in light of the context. Specifically, participants were asked to judge the target sentence based on how well it was expressed on a scale of 1 (100% bien mal all from Section 4.5.2.1 that all subjects, matrix clause verbs and direct objects were counterbalanced across the experimental items. Contrastive Focus tokens are discussed first below, then Topic Shift tokens and finally Topic Maintenance tokens. 5.2.2.1 Contrastive Focus tokens The following paragraphs provide the descriptive results only for Contrastive Focus tokens like example (56); a comprehensive statistical analysis and interpretation of the CMFT is provided in Section 5.2.2.4. (56). Cuando salim os a cenar, mi novia prefiere comer platos livianos, pero yo prefiero comer algo sustancioso. a. As que ella come ensaladas y yo como milanes as en los restaurantes. b. #As que pro come ensaladas y pro como milanesas en los restaurantes. While participants rated the target sentences on a scale of 1 through 4, the data were ultimate ly collapsed. As a four point scale was provided, gradiency was expected. Yet, as the

PAGE 144

144 ood) or 4 (=100% bad) as opposed to the intermediary ratings of 2 (= more or less good) and 3 (=more or less bad). Table 5 1. Distribution of ratings Rating 1 = 100% good 2 = more or less good 3 = more or less bad 4 = 100% bad NS 60.4% 9.4% 9.6% 2 .6% L2 70.5% 6.3% 10.1% 13.2% analyses performed below will be based on this binary distinction. Figure 5 null subject Contrastive Focus tokens (CFO and CFN, respectively) using the binary distinction described above. The average number of CFO tok native speaker and L2 speaker groups (5.92 and 5.75 out of 6, respectively). Additionally, there but two participan ts in the native speaker group and two in the L2 speaker group rated all six The standard deviation for the native speaker group was 0.28 and that of the L2 spe aker group native speaker group and the L2 speaker group: 0.71 and 0.25, respectively. The L2 speaker group showed very little variation (only participants 2 and 6 and the standard deviation was 0.46); however, the native speaker group demonstrated more

PAGE 145

145 was 1.27. Despite the variation seen in the native speaker group, all participants rated more C FO analyzed and interpreted in Section 5.2.2.4. Figure 5 3. CMFT group results: Contrastive Focus. CFO=Contrastive Focus token with overt subject; CFN=Contr astive Focus token with null subject 5.2.2.2 Topic Shift tokens Descriptive results for Topic Shift tokens like (57) are provided below, while a comprehensive statistical analysis and interpretation of the results is found in Section 5.2.2.4. (57). Mi h ija quiere ser autora y no tiene otros intereses. Yo creo que es mejor tener varios intereses y sugiero otras actividades, pero no importa lo que diga yo. have va a. Finalmente ella escribe cuentos y ella pasa todo el da en su cuarto. b. #Finalmente escribe cuentos y pasa todo el da en su cuarto. Figure 5 native speaker and L2 speaker groups (5.5 4 and 5.75 out of 6, respectively). For the L2 speaker

PAGE 146

146 group, there was very minimal variation; with the exception of one participant who rated 4 of 6 0.71. Whil p was 0.78. A higher than was evidenced for the Contrastive Focus contexts. Specifically, the native speaker group 2.75 and the L2 speaker group average was 6 and the L2 speaker 6. The standard deviations of each group are 1.70 and 1.51, respectively. In spite of the variation, all but one native speaker participant (8) and two L2 Figure 5 4. CMFT group results: Topic Shift. TSO=Topic Shift token with overt subject; TSN=Topic Shift token with null subject

PAGE 147

147 5.2.2.3 Topic Maintenance tokens The descriptive results for Topic Maintenance tokens like example (58) are presented below. The statistical anal ysis and interpretation of these tokens, along with Contrastive Focus and Topic Shift tokens, is provided in Section 5.2.2.4. (58). Mi cuada es muy sociable. Tiene muchos amigos y por eso va a muchas cenas a la canasta donde tiene que contribuir con alg o. in law is very social. She has a lot of friends and for that reasons, she goes a. #As que ella lleva postres y comparte todo con sus amigos. b. As que lleva postres y comp arte todo con sus amigos. Table 5 5 below provides the group average number of good ratings for the Topic was rather high for both the native speaker and L2 speaker groups (4.58 and 5.50 out of 6, respectively). For the L2 speaker group, there was little variation; with the exception of one ants who rated 5 out of 6 0.76. The native speaker participants showed slightly more variation with a range of 0 6 TMO tion of 1.61. Regarding the TMN tokens, both the Less variation, as compared to the TMO tokens, was evidenced for each group; the native speaker group range o 6 (standard deviation 0.88) while no variation was evidenced for the L2 speaker group. Based on this information, it is not surprising that fewer participants demonstrated the expected different ial rating between TMO and TMN tokens. In fact, only 11 of 24 native speaker participants and 3 of 8 L2 speakers did so. Of the 13 native speaker participants that did not make the expected

PAGE 148

148 five L2 speakers that did not make the expected distinction (3, 4, 5, 7 and 8) all rated an equal Figure 5 5. CMFT group results: Topic Maintenance. TMO=Topic Maintenance token with overt subject; TMN=Topic Maintenance token with null subject 5.2.2.4 Statistical analysis and interpretation of results As with the GJCT, a Generalized Mixed Models ANOV A was conducted for the CMFT. Subject Type (overt or null), Context Type (Contrastive Focus, Topic Shift and Topic Maintenance) and Group (native speaker or L2 speaker) constituted the factors. Finding a main effect for Subject Type would indicate that ove rt and null subjects were treated differently. Similarly, finding a main effect for Context Type would indicate that Contrastive Focus, Topic Shift and Topic Maintenance contexts were treated differently. Finally, finding a main effect for Group would indi cate that the native speaker and L2 speaker participants performed differently. Finding an interaction between Subject Type and Context Type would indicate that overt and null subjects are treated differently in the different contexts. Finding interactions between

PAGE 149

149 Subject Type and Group would indicate that overt and null subjects were treated differently by the two participant groups. Finding interactions between Context Type and Group would indicate that Contrastive Focus, Topic Shift and Topic Maintenance tokens were treated differently by the participant groups. Finally, finding an interaction between Subject Type, Context Type and Group would indicate that the native speaker and L2 speaker participants treated overt and null subjects in Contrastive Focus Topic Shift and Topic Maintenance contexts differently from each other. Bonferroni post hoc two way and three way analyses were also conducted to examine the sources of any interactions found. The analysis revealed a significant main effect for Subject T ype (F(1, 1,140) 5.397, p = 0.020), Context Type (F(2, 1,140) 89.487, p = 0.000) and Group (F(1,25) 37.641, p = 0.000). Additionally, interactions were found between Subject Type and Context Type (F(2, 1,140) 181.232, p = 0.000), Subject Type and Group (F( 1, 1,140) 35.774, p = 0.000) and Context Type and Group (F(2, 1,140) 58.946, p = 0.000). Lastly, a three way interaction between Subject Type, Context Type and Group was found (F(2, 1,140) 32.776, p = 0.000). With respect to the Subject Type and Context Type interaction, the post hoc pairwise t tests for independent samples (with Bonferroni adjustment) showed that the native speaker group p = 0.000). The native speaker also group rat p = ( p = 0.012). Regarding the L2 speaker group, they too rated significantly more CFO than CFN tokens as p = 0.000). The same result obtained for the Topic Shift contexts in that p = 0.000). p = 0.0 44). All

PAGE 150

150 of these results are expected since overt subjects are preferred in Contrastive Focus and Topic Shift contexts, while null subjects are preferred in Topic Maintenance contexts. For the Subject Type and Group interaction, the post hoc pairwise t te sts for independent samples (with Bonferroni adjustment) showed that the native speaker group rated significantly p = 0.031). The same is true of when comparing CFO tokens and TMO tokens ( p = 0.001). The native sp eaker group also rated significantly more TSO tokens than TMO tokens ( p = 0.014). These results are not surprising as the preference for overt subjects is preferred in both Contrastive Focus and Topic Shift contexts, although less categorical in Topic Shif t (Silva Corvaln 1982, 1994; Enrquez 1984; Bentivolgio 1987; Bayley & Pease lvarez 1996, 1997; Otheguy et al. 2007; Prada Prez 2009; Lapidus & Shin 2012), hence the significant difference between the CFO and TSO tokens and the CFN and TSN tokens. Regar ding the null subject tokens, the native speaker group rated significantly less CFN tokens p = 0.000). Similarly, significantly less CFN tokens were ( p = 0.0000). Lastly, signif icantly less TSN tokens p = 0.000). Again, these results are expected as only in Topic Maintenance contexts are null subjects preferred. The L2 speaker group, however, rated CFO and TSO tokens similarly ( p = 1.000), which is does not indicate divergence as both Contrastive Focus and Topic Shift contexts prefer overt subjects. A difference was expected, yet did not obtain, for the CFO versus TMO tokens ( p = 1.000) and the TSO versus TMO tokens ( p = 1.000). Th ese results are unexpected in that no difference in rating was seen between the Contrastive Focus or Topic Shift tokens as compared to the Topic Maintenance tokens. With respect to the null subject tokens, the L2 group rated significantly less CFN tokens a p = 0.000). The same result obtained for the CFN versus

PAGE 151

151 TMN comparison ( p = 0.000). Finally, the L2 speaker group rated significantly less TSN tokens p = 0.002). Here, the expected re sults obtained, matching the native speaker group. Regarding the Context Type and Group interaction, the post hoc pairwise t tests for independent samples (with Bonferroni adjustment) showed that the native speaker and L2 speaker groups rated CFO tokens t o the same degree ( p = 0.312). The same is true of the CFN tokens ( p = 0.148). These findings are expected if both participant groups have converged on the distribution of subject pronouns in Contrastive Focus contexts in Spanish. With respect to the Topic Shift tokens, the native speaker and L2 speaker groups rated the TSO tokens to the same degree ( p = 0.492); however the L2 group rated the TSN tokens significantly better than the native speaker group ( p = 0.011). Here, the L2 speaker group is showing more optionality, accepting more null subject pronouns in Topic Shift contexts. Previous child and adult research supports this finding ( Sorace 2000, 2003; Tsimpli et al. 2004; Prada Prez 2009; Rothman 2009; Lapidus Shin & Cairns 2012) Finally, the L2 speaker group rated significantly more TMO and p = 0.037 and p = 0.009, respectively), which is unexpected if the participant groups have converged on the distribution of subject pronouns in Topic Maintenance co ntexts. 5.2.2.5 Summary of Context Matching Felicitousness Task results To summarize, the results described above point to several similarities between the native speaker and L2 speaker groups. First, both the native speaker and the L2 speaker group showed differential treatment of overt versus null subjects within each context. Second, the native speaker and the L2 speaker group showed the same distinctions in ratings for the null subject tokens in that less CFN tokens than TSN and TMN tokens were rated as

PAGE 152

152 not obtain for the L2 speaker group as they rated these three token types similarly. Finally, recall that the three way interaction described above showed no statistically significant differences between the native spea ker and L2 speaker groups for the CFO tokens nor for the CFN tokens, thus demonstrating that the L2 speaker group performs native like for the Contrastive Focus avera way interaction revealed no statistically significant difference between the native speaker and L2 speaker group for TSO tokens; nevertheless, a significant difference was found for the TSN tokens, suggesting that the L2 group does not make an entirely native like distinction for Topic Shift tokens in the CMFT. Finally, with respect to th than in previous contexts. Since null subjects are preferred in Topic Maintenance contexts, these numbers are higher than exp ected. Nevertheless, both the native speaker and L2 speaker group overt subject tokens. However, the three way interaction revealed statistically significant dif ferences between the native speaker and L2 speaker group for both TMO and TMN tokens, tokens. Although a significant difference obtained between the native speaker and L2 speaker group, this finding is not entirely problematic for the claims made herein since Topic Maintenance contexts prefer null subjects. That is, since the null subject is the default subject

PAGE 153

153 that is employed when an overt subject is not required for contrast or reference to a previously mentioned topic, an overt subject pronoun in a Topic Maintenance context is perhaps less infelicitous than the opposite scenario (i.e. when a null subject is used in lieu of an overt subject in Contrastive Focus or Topic Shift contexts). Specifically, since overt subjects are required to mark contrastive focus or to refer to a distinct yet known (within the common ground) antecedent, we should not expect to see similar rating of overt and null subjects in Contrastiv e Focus or Topic Shift contexts. This is in fact what the data show: the ratings of both the native speaker and L2 speaker group are more polarized with regard to Contrastive Focus and Topic Shift contexts as compared to Topic Maintenance contexts. Additio nally, in spite of the apparent mixed results (i.e. the L2 speakers showing native like convergence on only half of the tokens), the Subject Type and Context Type comparison demonstrated that both the native speaker and L2 speaker groups made a distinction between overt and null subjects in all context types, thus establishing their differential treatment of overt subjects as compared to null subjects according Contr astive Focus contexts were seen to be largely unproblematic even at the advanced proficiency level in Rothman (2009) while Sorace and colleagues found difficulties with Topic Shift contexts. 5.2.3 Summary of Offline Experimental Results Having described t he results of the GJCT and the CMFT, this section brings together the findings of these two tasks. The GJCT tested for knowledge of the syntactic distribution of overt and null subject pronouns in Spanish. Regarding both referential and expletive s ubjects, the results showed that the L2 speaker group performed as the native speaker group in that they rejected ungrammatical overt expletive subjects but accepted grammatical overt and null referential subjects and null expletive subjects. Importantly, both groups distinguished between

PAGE 154

154 ungrammatical ESO and grammatical RSO tokens. Thus, it seems that the L2 speakers have converged upon the syntactic distribution of subjects in Spanish, including the presence of the null expletive, which Karimi (2005) arg ues is not permitted in Farsi This is consistent with the CMFT, tested for knowledge of the discourse distribution of subjects in Spanish in three contexts: Contrastive Focus, Topic Shift and Topic Maintenan ce. While some quantitative differences were found between the native and L2 speaker groups, it is important to note that both groups distinguished between the counterbalanced tokens (e.g. CFO versus CFN) showing that their ratings were dependent upon both the context type (Contrastive Focus, Topic Shift or Topic Maintenance) and the subject type (overt versus null). Taken together, the results of the offline experimental tasks provide some evidence of the L2 speaker like knowledge of the dis tribution of overt and null subjects in Spanish. However, since these results come from offline methodology, they might be viewed as compatible with the claims of the Interface Hypothesis (IH) (whether converge nce or divergence is found). Additionally, the differences found between the native speaker and L2 speaker group for TSN, TMO and TMN tokens are also predicted by the IH. The following section examines the processing of the distribution of overt and null subjects in two online tasks, as this is pr ecisely where the IH predicts that measurable differences between native speakers and near native L2 speakers on external interface conditioned properties will obtain. 5.3 Online Experimental Results Recall from Section 4.5.2 that two complementary onl ine experimental tasks were completed by the participants in order to examine their processing of the syntactic and discourse distribution of subjects in Spanish: a Grammaticality Task and the online Context Matching Felicitousness Task. The experiments we re presented using E prime software (Schneider et al. 2002) in order to gather reaction time (RT) data for the regions of interest, where slower RTs are

PAGE 155

155 expected when participants detect ungrammaticality or infelicitousness. A brief recap of the purpose of each task, the token types examined and the descriptive results for each token type are presented below in Sections 5.3.1 and 5.3.2. 5.3.1 Online Grammaticality Task (GT) As outlined in Section 4.5.2.1, the purpose of the online GT was to test for knowle dge of the syntactic distribution of overt and null subjects via RT data. Like the offline version, the GT contained 106 tokens, 24 of which examined the distribution of subjects, 72 of which examined the distribution of DOM 4 and 10 of which were fillers. Recall that in the online version of this task, which was a Self Paced Reading Task (SPRT), participants were simply asked to read short sentences presented to them via a non cumulative moving window technique (Just, Woolley & Carpenter 1982) (see Section 4.5.2 for details) and then answer a yes/no comprehension question. The following section provides a macro analysis of the results of the online GT, while Sections 5.3.1.2 and 5.3.1.3 discuss the results specific to referential and expletive subject tokens 5.3.1.1 Macro analysis of the online GT Just as in the offline tasks, a Generalized Mixed Models ANOVA was conducted for each region of interest in the online tasks. For the online GT, Subject Type (overt or null), Condition Type (expletive or referenti al) and Group (native speaker or L2 speaker) constituted the factors. Finding a main effect for Subject Type would indicate that overt and null subjects were treated differently. Similarly, finding a main effect for Condition Type would indicate that refer ential and expletive subjects were treated differently. Finally, finding a main effect for Group would indicate that the native speaker and L2 speaker participants performed differently. Finding an 4 As stated in Footnote 1, the 72 tokens examining the distribution of DOM are not discussed herein.

PAGE 156

156 interaction between Subject Type and Condition Type would indicate that overt and null subjects are treated differently in referential and expletive tokens. Finding interactions between Subject Type and Group would indicate that overt and null subjects were treated differently by the two participant groups. Findi ng interactions between Condition Type and Group would indicate that expletive and referential tokens were treated differently by the participant groups. Finally, finding an interaction between Subject Type, Condition Type and Group would indicate that the native speaker and L2 speaker participants treated overt and null subjects in referential and expletive conditions differently from each other. Bonferroni post hoc two way and three way analyses were also conducted to examine the sources of any interactio ns that obtained. For the first region of interest, which was comprised of either the overt subject ella/l/ ello siempre 17) 4.484, p = 0.049) and Group (F(1, 12) 9.831, p = 0.008), but not for Condition Type (F(1, 23) 2.549, p = 0.124). No interactions were found between Subject Type and Condition Type (F(1, 54) 0.353, p = 0.555), Condition Type and Group (F(1, 23) 0.008, p = 0.928), Subject Type and Group (F(1, 17) 1.394, p = 0.254) or between Subject Type, Condition Type and Group (F(1, 54) 0.025, p = 0.874). For the second region of interest, which was the verb, main effects were found for Condition Type (F(1,15) 24.645, p = 0.000) and Group (F(1,17) 9.027, p = 0.008), but not for Subject Type (F(1, 20) 3.981 p = 0.060). No interactions were found between Subject Type and Condition Type (F(1, 40) 0.867, p = 0.357), Condition Type and Group (F(1, 15) 3.040, p = 0.102), Subject Type and Group (F(1, 20) 0.413, p = 0.528) or between Su bject Type, Condition Type and Group (F(1, 40) 2.238, p = 0.142). Finally, for the last region of interest, which was comprised of either the quantifier muchos/muchas mucho p = 0.004), but not for Subject Type

PAGE 157

157 (F(1, 13) 2.015, p = 0.179) nor for Condition Type (F(1, 9) 0.764, p = 0.404). An interaction was found between Subject Type and Condition Type only (F(1, 13) 14.422, p = 0.002). No interactions were found between Conditio n Type and Group (F(1, 9) 0.322, p = 0.584), Subject Type and Group (F(1, 13) 1.438, p = 0.252) or between Subject Type, Condition Type and Group (F(1, 13) 0.593, p = 0.456). The corresponding post hoc analyses are presented below in the statistical analys is of referential and expletive subject tokens after each region of interest, respectively. 5.3.1.2 Referential subject tokens The 12 tokens related to referential subjects were of two types: overt referential subject tokens ( RSO ; n=6) and null referential subject tokens ( RSN ; n=6). As reviewed in Section 3. 2.1, both null and overt referential subjects are permissible in Spanish. Example (39), reproduced below as (59), shows an RSN and RSO token, respectively, with the divisions between regions of interest indicated by forward slashes (/). (59)a. En mi opini n/siempre/toma/muchas/vitaminas/por la maana. b. En mi opinin/ella/toma/muchas/vitaminas/por la maana. To allow for statistical comparison of the RTs to each regio n of interest, the regions of interest within sentences of the same token type were consistently divided in the manner seen in (59). The regions of interest analyzed for RT in null referential subject tokens such as example (59a) above are the adverb siemp re toma muchas surround the syntactic position of the verb (between the adverb siempre quantifier muchas regions of interest analyzed for RT in these tokens are the subject ella toma

PAGE 158

158 muchas ns for the online GT. The following subsections provide the descriptive results, statistical analysis and interpretation for all three regions of interest. 5.3.1.2.1 Descriptive and statistical analyses. This section provides descriptive and statistical an alyses of the RT data for th e RSO an d RSN tokens. Before beginning, the manner in which the RT data was treated must be explained. First, all RTs corresponding to missed comprehension questions (i. e. incorrect responses to the yes/no comprehension questions that followed the tokens) were excluded from the analysis. Secondly, the RT data was trimmed by excluding all RTs slower than 2 seconds 5 Additionally, the mean RT and standard deviation for each region of interest were calculated across the task. All RTs higher than 2 standard deviations from the mean RT were replaced with the cutoff value (x = mean RT + (standard deviation X 2)) 6 Figures and statistical analyses are provided according to the re gion of interest. To begin, Figure 5 6 provides a visual representation of the group average RT for the native speaker and L2 speaker groups for the first region of interest, which for the RSO tokens was the subject ella/ l s it was the adverb siempre tokens, the average RT for the native speaker group for the first region of interest, ella/l was 507.12ms while that of the L2 speaker group was 577.39ms. The native speaker range was 381.17 679.50m s with a standard deviation of 96.72ms. The L2 speaker range was a bit higher at 471.60 749.80ms with a standard deviation of 88.68ms. Regarding the RSN tokens, t he average RT for the native speaker group for the first region of interest, the adverb siempr e 516.52 while that of the L2 speaker group was 622.14ms. The native speaker range was 326.80 5 These data were excluded on the grounds that it is likely that such high RTs represent external distractions (e.g. asking the researcher a question, pausing to look up, a loud noise). 6 See Hopp (2007, p 224) and Leeser et al. (2011) for similar treatment of data.

PAGE 159

159 692.17ms with a standard deviation of 100.51ms. The L2 speaker range was higher at 523.40 772.67ms with a standard deviation of 102.10ms. Individual RTs and standard deviations for a ll GT tokens are found in Appendix F. F rom Figure 5 7, it is clear that the native speaker group average RT for the RSO and RSN tokens is faster than that of the L2 speaker group Figure 5 6. Online GT group results: referential subject tokens for first region ( ella/l/siempre ). NS=native speaker; L2=second language speaker; RSO=overt referential subject; RSN=null referential subject To compare the treatment of overt and null subjects, a post hoc pairwise t tests for independent samples (with Bonferroni adjustment) were conducted across conditions and groups. differ for the first region of interest ( p = 0.668). The same finding obtained for the L2 group in that the ir average RT to RSO and RSN tokens did not differ significantly ( p = 0.331). These results are expected since both token types are grammatical. With respect to the comparison between the native speaker and L2 speaker groups, the post hoc pairwise t tests for independent samples (with Bonferroni adjustment) showed that the

PAGE 160

160 p = 0.065). However, for the RSN tokens, the native speaker group RT was significantly faster than that of the L2 speaker group ( p = 0.0 12). 7 Next, Figure 5 7 below provides a visual representation of the group average RT for the native speaker and L2 speaker groups for the second region of interest, the verb. Regarding the RSO tokens, the average RT for the native speaker group was 464.4 8ms while that of the L2 speaker group was 575.98ms. The native speaker group range spanned from 307.67 796.00ms with a standard deviation of 114.18ms. Here again, the L2 speaker group range was a bit higher at 379.40 888.60ms with a standard deviation of 167.97ms. With respect to the RSN tokens, the average RT for the native speaker group was 463.52ms while that of the L2 speaker group was 497.74. Figure 5 7. Online GT group results: referential subject tokens for second region (verb). NS=native speak er; L2=second language speaker; RSO=overt referential subject; RSN=null referential subject 7 Recall that three statistical comparisons were conducted and described in the offline tasks: Subject Type (overt versus null), Condition Type (referential versus expletive) and Group (native speaker versus L2 speaker). This was the offline task. Three sets of comparisons are necessary for the online task since three regions of i nterest were examined. Therefore, the second comparison (referential versus expletive) will be presented for all three regions of interest in Section 5.3.1.4.

PAGE 161

161 The native speaker group range was 212.75 754.20 ms with a standard deviation of 117.45ms The L2 speaker range (364.00 628.83ms) fell within that of the native spea ker range and the standard deviation was lower (96.18ms). As was seen for the first region of interest, the native speaker group average RT to the RSO tokens and RSN tokens alike (although to a lesser extent in the latter case) is faster than that of the L 2 speaker group. To compare the treatment of overt and null subjects, a post hoc pairwise t tests for independent samples (with Bonferroni adjustment) were conducted across conditions and groups. RT to RSO and RSN tokens did not differ for the second of interest ( p = 0.951). The same result obtained for the L2 group; their average RT to RSO and RSN tokens did not differ significantly ( p = 0.158). Again, these results are expected since both are tok en types grammatical. With respect to the comparison between the native speaker and L2 speaker groups, the post hoc pairwise t tests for independent samples (with Bonferroni adjustment) showed that the p = 0.069) or for the RSN tokens ( p = 0.411). Finally, Figure 5 8 provides a visual representation of the group average RT for the native speaker and L2 speaker groups for the third region of interest, the quantifier muchos/muchas The native speaker group average for the RSO tokens was 514.50ms while that of the L2 speaker group was 575.98ms. The range of the native speaker group was 335.33 819.00ms with a standard deviation of 102.32ms. Here, the range of the L2 speaker group (456.00 699.00ms; 78.55m s range) fell within that of the native speaker group. Regarding the RSN tokens, the native speaker group average was 481.69ms while that of the L2 speaker group was 632.63ms. The native speaker group range was 314.33 643.60ms with a standard deviation of 94.96ms. The range of the L2 speaker group was higher (437.33 864.00ms) with a

PAGE 162

162 standard deviation of 148.04ms. Here again, the native speaker group average RT for the RSO and RSN tokens is faster than that of the L2 speaker group. Figure 5 8. Online GT group results: referential subject tokens for third region ( muchos/muchas ). NS=native speaker; L2=second language speaker; RSO=overt referential subject; RSN=null referential subject The same comparison of the treatment of overt and null subjects was con ducted via a post hoc pairwise t tests for independent samples (with Bonferroni adjustment) analysis across and RSN tokens nor that of the L2 speaker group differ ed for the third of interest ( p = 0.396 and p = 0.179, respectively).These results are expected since both are token types grammatical. For the comparison between groups, the post hoc pairwise t tests for independent samples (with Bonferroni adjustment) s both RSO tokens ( p = 0.009) and for RSN tokens ( p = 0.011). The following section provides and interpretation of the results found for all three regions of interest. 5.3.1.2.2 Interpretation of resu lts The first comparison made for the three regions of

PAGE 163

163 RSO versus RSN tokens differed significantly. This is the expected result as both tokens are grammatical and indicates that both groups have converged on the syntactic distribution of overt and null referential subject pronouns in Spanish. Lastly, when the native speake r RT was compared to the L2 speaker RT, statistically significant differences obtained between the two groups for RSO tokens in the third region of interest and RSN tokens in the first and third regions of interest. However no statistically significant dif ference in RT was found between the two groups for the second region of interest. These results are attributable to the L2 speaker tokens (sometimes by as much as 150ms as in the spillover region for RSN tokens). Importantly, these differences do not indicate lack of convergence on the target grammar, but rather slower overall RTs. The following section reviews the expletive subject tokens. 5.3.1.3 Expletive subject tokens The 12 expletive sub jects tokens consisted of two types: overt expletive subjects ( ESO; n=6) and null expletive subjects ( ESN; n=6). As discussed in Section 3.2.1, Argentine Spanish requires null expletive subjects; overt expletive subjects are ungrammatical. Example (60) below shows t his distinction. As with the referential subject tokens, divisions between regions of interest are indicated by forward slashes (/). (60)a. Los vecinos/dicen/que/siempre/llueve/mucho/en esa regin. b. *Los vecinos/dicen/que/ello/llueve/mucho/en esa regin. The regions of interest analyzed for RT in null expletive subject tokens such as example (60a) above are the adverb siempre llueve ver region mucho

PAGE 164

164 the verb (between the adverb siempre mucho (60b) contains an overt expletive subject, th e regions of interest analyzed for RT of this token type are the expletive subject ello llueve mucho interpretation 5.3.1.3.1 Descriptive and statistical analyses. Like Section 5.3.1.2.1, this section provides a description and a statistical analysis of the RT data f or the ESO and the ESN tokens. The data were treate d as described in Section 5.3.1.2.1. Figures and statistical analyses are provided according to the region of interest. Thus, Figure 5 9 provides a visual representation of the group average RT for the native speaker and L2 speaker groups for the first reg ion of interest, which for the ESO tokens was the subject ello siempre was 469.88ms while that of the L2 speaker group was 53 8.69ms. The native speaker range was 318.83 777.75ms with a standard deviation of 104.79ms. The L2 speaker range fell within the native speaker range at 416.60 662.60ms with a standard deviation of 89.17. For the ESN tokens, the average RT for the native s peaker group was 492.31ms while that of the L2 speaker group was 606.02ms. The native speaker range was 230.00 730.00ms with a standard deviation of 130.56ms. The L2 speaker range was a bit higher at 434.33 790.50ms, but their standard deviation (131.04ms) was almost identical to that of the native speaker group. As is seen in Figure 5 9, the native speaker group average RT for the ESO and ESN tokens is faster than that of the L2 speaker group.

PAGE 165

165 Figure 5 9. Online GT group results: expletive subject t okens for first region ( ello/siempre ). NS=native speaker; L2=second language speaker; ESO=overt expletive subject; ESN=null expletive subject In order to compare the treatment of overt and null subjects, a post hoc pairwise t test for independent samples (with Bonferroni adjustment) was conducted across conditions and groups. differ for the first region of interest ( p = 0.165), which is expected since both are gr ammatical. no significant difference was found between them in the first region of interest ( p = 0.084). These results are not unexpected as the ungrammaticality o f the ESO token may not yet be detected by either group. Regarding the comparison between the native speaker and L2 speaker groups in the first region of interest, the post hoc pairwise t tests for independent samples (with Bonferroni adjustment) showed th p = 0.075). However, the native speaker group RT to the ESN tokens was significantly faster than that of the L2 speaker group ( p = 0.029).

PAGE 166

166 Figure 5 10 below provides a visual representation of the group a verage RT for the native speaker and L2 speaker groups. For the ESO tokens, the native speaker group average RT was 592.67ms while that of the L2 speaker group was 762.26ms. The native speaker range spanned from 278.67 1000.60ms with a standard deviation of 189.08ms. The L2 speaker group range (486.33 956.33ms) fell within that of the native speaker group and their standard deviation was quite similar (182.82ms). Regarding the ESN tokens, the native speaker group average RT was 521.87ms while that of the L 2 speaker group was 700.18ms. The native speaker range spanned from 304.00 811.33ms with a standard deviation of 139.72ms. The L2 speaker group range (466.33 985.60ms) was higher than that of the native speaker group but their standard deviation was simil ar (157.58ms). From Figure 5 10, it is clear that the native speaker group average RT for the ESO and ESN tokens alike is faster than that of the L2 speaker group. Figure 5 10. Online GT group results: expletive subject tokens for second region (verb) NS=native speaker; L2=second language speaker; ESO=overt expletive subject; ESN=null expletive subject To compare the treatment of overt and null subjects, a post hoc pairwise t test for independent samples (with Bonferroni adjustment) was conducted acr oss conditions and groups.

PAGE 167

167 to ESO and ESN tokens differed ( p = 0.031), which is expected since only ESN tokens grammatical. However, this result did not obt ain for the L2 speaker group. Their average RT to ESO and ESN tokens showed no significant difference ( p = 0.288), which is unexpected since ESO tokens are ungrammatical and should therefore elicit slower RTs. For the comparison between groups, the post h oc pairwise t tests for independent and ESN tokens differed ( p = 0.025 and p = 0.004, respectively). Lastly, Figure 5 11 provides a visual representation of the gr oup average RT for the native speaker and L2 speaker groups for the final region of interest. For the ESO tokens, the native speaker group average RT was 540.54ms while that of the L2 speaker group was 641.28ms. The range of the native speaker group was 36 1.00 932.75ms with a standard deviation of 151.03ms. Here again, the L2 speaker group range (499.20 704.33ms) fell within that of the native speaker group and their standard deviation was substantially lower at 61.16ms. Finally, the native speaker group av erage for the ESN tokens was 467.39ms while that of the L2 speaker group was 573.82ms. The range of the native speaker group for the spillover region was 320.17 690.25ms with a standard deviation of 90.29ms. Differently from the previous two regions, the L 2 speaker group range (447.50 690.17ms) fell within that of the native speaker group and their standard deviation was lower at 68.46ms. As seen in all previous regions, the native speaker group average RT for the ESO and ESN tokens is faster than that of t he L2 speaker group.

PAGE 168

168 Figure 5 11. Online GT group results: expletive subject tokens for third region ( mucho ). NS=native speaker; L2=second language speaker; ESO=overt expletive subject; ESN=null expletive subject The same post hoc pairwise t test for independent samples (with Bonferroni adjustment) was conducted across conditions and groups for the third region of interest. The results showed p = 0.007), which is expected since only ESN tokens grammatical. The same result obtained for the L2 speaker group in that their average RT to ESO and ESN tokens was significantly different as well ( p = 0.000). Thus, in the third region, both participant groups demonstrate d the expected diff erentiation between overt and null subjects in expletive tokens. As in the previous region of interest, the comparison between groups, the post hoc pairwise t tests for independent samples (with Bonferroni adjustment) showed that the two th the ESO tokens and ESN tokens differed ( p = 0.026 and p = 0.016, respectively) in the third region of interest. 5.3.1.3.2 Interpretation of results. The first of two comparisons made for the three T to ESO tokens as compared to ESN

PAGE 169

169 ungrammatical ESO tokens differed significantly from their RTs to the grammatical E SN tokens indicates that they d id not detect the ungr ammaticality of the former token type in the first region of interest. This is not necessarily unexpected since, at this point in the sentence, they have not yet seen the verb. Without knowing what follows (a weather verb), the overt ello d etected as ungrammatical yet 8 With respect to the second region of interest, the native speaker group, but not the L2 speaker group, showed a significantly slower RT to the ungrammatical ESO tokens as compared to the ESN tokens. In the case of the native speaker group, this is expected as only the ESN tokens are grammatical; thus, this indicates that the native speakers detect the ungrammaticality in the second region. In the final region, both participant groups showed significantly slower RTs to the ungr ammatical ESO tokens, indicating that the ungrammaticality is detected by both groups. Lastly, when the native speaker RT was compared to the L2 speaker RT across all three regions of interest, statistically significant differences obtained in five of the six comparisons: ESN tokens in the first region of interest, ESO and ESN tokens in the second and third regions of interest. No statistically significant difference in RT was found between the two groups for the ESO tokens in the first region of interest. These d ifferences lack of convergence on the target grammar. After all, they notice the ungrammaticality of the ESO tokens. The following section examines the second compariso n made in the offline tasks, that of Condition Type. 8 At this point in the sentence, participants have seen a matrix clause subject ( los vecinos ), a matrix clause verb ( dicen ), the complementizer que and ello. The sentence could be felicitously completed as follows: (i) Los vecinos dicen que ello implica muchos cambios en el vecindario. The neighbors s ay 3. PL PRES that that/this imply 3. SG PRES many changes in the neighborhood

PAGE 170

170 5.3.1.4 Condition Type comparison (referential versus expletive) This section presents the second comparison between referential and expletive tokens described in Footnote 7 above. The results for each region of interest are presented, beginning with the first. T he post hoc pairwise t test for independent samples (with Bonferroni adjustment) significantly from that of the RT to the ungrammatical ESO tokens ( p = 0.071). The same result obtained for the L2 speaker group ( p = 0.427). These results are expected since it was shown in Section 5.3.1.3 above that neither group detected the ungrammaticality of the ESO token in the f irst region of interest. Regarding the RSN versus ESN comparison, neither the native speaker p =0.202 and p = 0.668, respectively). This is expected, as no difference between these two grammatic al tokens is necessarily expected. For the second region of interest, the post hoc pairwise t test for independent samples tokens was significantly faster than that of their RT to the ungrammatical ESO tokens ( p = 0.000). The same is true of the L2 speaker group ( p = 0.010). These findings are expected if the participant groups detect the ungrammaticality of the ESO tokens in the second region. With respect to the RS to the RSN tokens was significantly faster than their RTs to the ESN tokens ( p = 0.007 and p = 0.002, respectively). Since both tokens are grammatical, this finding is not necessari ly expected. Finally, for the third region of interest, the post hoc pairwise t test for independent samples (with Bonferroni adjustment) revealed that both the native speaker group RT and the L2 speaker group RT to the RSO tokens is significantly faster t han their respective RTs to the ESO tokens ( p = 0.005 and p = 0.015, respectively). This is expected since only the RSO tokens are

PAGE 171

171 grammatical. For the RSN versus ESN comparison, neither group was shown to have significantly different RTs ( p = 0.269 and p = 0.343, respectively). This is expected as both token types are grammatical. 5.3.1.5 Comparative analysis of RTs The analyses described in Sections 5.3.1.2.1 and 5.3.1.3.1 did not consider the differential RTs between the tokens; therefore, they do no t inform us if the two participant groups are different with respect to the ungrammaticality of the ESO tokens (as evidenced by comparatively slower RTs). As such, a comparison between the RTs to grammatical tokens and those of ungrammatical tokens must be performed. Recall that out of the four token types examined in both the offline and online versions of the grammaticality task, three are grammatical (RSO, RSN and ESN) while one is ungrammatical (ESO) in Spanish. As a consequence, two necessary compariso ns will be made, first between overt and null expletive subjects and then between overt expletive subjects and overt referential subjects. Since two comparisons are made from within the same data set, a Bonferroni correction was applied, effectively loweri ng the original p value of 0.05 to 0.025. In both comparisons made below, slower RTs are expected for the overt expletive subjects since this is the ungrammatical token type. As in Sections 5.3.1.2.1 and 5.3.1.3.1, the comparison will be made for all three regions of interest, starting with the subject region (either ella/ l ello (either muchos/muchas mucho Figure 5 12 below shows the average differential RT for the f irst region of interest for the two comparisons described above: overt expletive subjects (ESO) versus null expletive subjects (ESN) and overt expletive subjects (ESO) versus overt referential subjects (RSO). The average differential RT was calculated by s ubtracting the group average RT to the grammatical token

PAGE 172

172 type (in this case, ESN and RSO) from the group average RT to the ungrammatical token type (in both cases ESO). Figure 5 12. Online GT comparative group results for first region (subject/adverb). NS=native speaker; L2=second language speaker; ESO ESN=group average RT to ESO tokens minus group average RT to ESN tokens; ESO RSO=group average RT to ESO tokens minus group average RT to RSO tokens From Figure 5 12 above it is clear that the equations outli ned above resulted in a negative differential RT for both the native speaker group and the L2 speaker group for the first region of interest. The native speaker group difference in average RT for the ESO ESN tokens was less than that of the L2 speaker grou p: 22.43ms versus 67.33ms, respectively. For the ESO RSO comparison, the differential RT of each group was nearly identical: 37.24ms for the native speaker group and 38.69ms for the L2 speaker group. The fact that the results are negative indicates tha t, on average, the effect of detecting ungrammaticality (i.e. slower RTs) is not evidenced in the first region of interest. If it were, we would expect the result to be positive since in both equations the first element, ESO, would demonstrate a comparativ ely slower RT than that of the second element, either ESN or RSO. This finding is not unexpected, however,

PAGE 173

173 since without knowing what the following element of the sentence is, no syntactic crash occurs. That is, until the participants see the verb that fol lows the ungrammatical overt expletive subject ello tests showed no statistically significant difference in average differential RT between the two groups for either the E SO ESN counterbalance (t(30) = 1.254, p = 0.220) or for the ESO RSO counterbalance (t(30) = 0.032, p matched that of the native speaker group. Figure 5 13 examines the differential RT to the sec ond region of interest, the verb. As shown, the native speaker group average differential RT for the ESO ESN tokens was slightly greater than that of the L2 speaker group: 70.80ms versus 62.09ms, respectively. However, with respect to the group average dif ferential in RT for the ESO RSO tokens, the native speaker group average differential RT was less than that of the L2 speaker group: 128.19ms versus 186.28ms, respectively. Figure 5 13. Online GT comparative group results for second region (verb). NS= native speaker; L2=second language speaker; ESO ESN=group average RT to ESO tokens minus group average RT to ESN tokens; ESO RSO=group average RT to ESO tokens minus group average RT to RSO tokens

PAGE 174

174 Different from the first region of interest, the results fo r the second region were positive which indicates that the participants have detected the ungrammaticality (i.e. they have slower RTs for the ungrammatical tokens than they do for the grammatical tokens). The independent samples t tests revealed no statist ically significant difference in average differential RT between the two groups for either the ESO ESN counterbalance (t(30) = 0.132, p = 0.896) or for the ESO RSO counterbalance (t(30) = 0.901, p = 0.375), thus suggesting that the distinction the L2 spea ker group makes between the ungrammatical and grammatical tokens matched that of the native speaker group. Figure 5 14 below shows the average differential RT to the third and final region of interest, the spillover region ( muchos/muchos mucho a seen the native speaker and L2 speaker group average differential RT for the ESO ESN and ESO RSO tokens was remarkably similar. The native speaker group difference in average RT for the ESO ESN tokens was 73.15ms while that o f the L2 speaker group was 67.46ms. Figure 5 14. Online GT comparative group results for third region ( muchos/muchas mucho ESN=group average RT to ESO tokens minus group average RT to ESN tokens; ESO RSO=group average RT to ESO tokens minus group average RT to RSO tokens

PAGE 175

175 Regarding the group average differential RT for the ESO RSO tokens, the native speaker group average RT was 71.76ms while that of the L2 speaker group is 71.04ms. A s with the second region, the results for the third region are positive. This indicates that the participants have detected the ungrammaticality. The independent samples t tests revealed no statistically significant difference for either the ESO ESN counte rbalance (t(30) = 0.458, p = 0.650) or for the ESO RSO counterbalance (t(30) = 0.314, p = 0.756). This finding suggests that the distinctions the two participant groups make between the ungrammatical and grammatical tokens are equal. 5.3.1.6 Summary of on line GT results Sections 5.3.1.2 through 5.3.1.5 above examined the RT data to three regions of interest (the subject/adverb, the verb and the spillover region) for overt referential subject (RSO) tokens, null referential subject (RSN) tokens, overt explet ive subject (ESO) tokens and null expletive subject (ESN) tokens. Regarding the referential tokens, both participant groups demonstrated knowledge of the grammatical nature of both RSO and RSN tokens in all regions of interest in that their RTs to the RSO and RSN tokens did not differ statistically. With respect to the expletive tokens, both the native speaker and L2 speaker groups showed significantly different RTs to the ungrammatical ESO tokens as compared to the grammatical ESN tokens in the third regio n, indicating their knowledge of the ungrammaticality of the overt subject in an expletive context. However, the native speaker group also showed this distinction in the second region of interest thus revealing that their detection of the ungrammaticality is more spread out. Regarding the comparison between Condition Type, the L2 speaker group demonstrated significantly different RTs to the counterbalances in the same regions as the native speaker group: ESO versus RSO and ESN versus RSN tokens in the secon d region of interest and ESO versus RSO tokens in the third region of interest. This highlights the native like performance of the L2 speaker group.

PAGE 176

176 The third comparison made examined the RTs of the native speaker group versus those of the L2 speaker grou p. While the L2 speaker group was found to have significantly slower RTs for the RSN and ESN tokens in the first region of interest, the ESO and ESN tokens in the second region of interest and the RSO, RSN, ESO and ESN tokens in the third region of interes t, recall that the comparative analysis of average differential RTs conducted in Section 5.3.1.5 showed no statistically significant differences between the two participant groups. This ultimately establishes the native like distinction made between ungram matical ESO tokens and grammatical ESN and RSO tokens (via comparison of RTs) on the part of the L2 speaker group. 5.3.2 Online Context Matching Felicitousness Task (CMFT) As outlined in Section 4.5.2., the purpose of the online CMFT was to test for knowl edge of the discourse distribution of overt and null subjects via RT data. Like the offline version, the online CMFT contained 36 tokens, 12 of which pertained to Contrastive Focus contexts, 12 to Topic Shift contexts and 12 Topic Maintenance contexts. Rec all that the online version of the CMFT was a Self Paced Reading Task (SPRT) in which participants read short contexts (presented in their entirety) followed by a target sentence presented via a moving window task (see Section 4.5.2). The target sentence w as followed by yes/no comprehension question. The subsequent sections discuss the descriptive results, statistical analyses and interpretation of the results for all three token types. 5.3.2.1 Macro analysis of the online CMFT As with the GT, a Generalize d Mixed Models ANOVA was conducted for the online CMFT. Subject Type 9 (overt or null), Context Type (Contrastive Focus, Topic Shift and Topic Maintenance), and Group (native speaker or L2 speaker) were the factors. As in the offline 9 Subject Type is not a factor for the first region of interest since only overt subjects were analyzed. Therefore, the first region of interest has two factors, Token Type and Group.

PAGE 177

177 CMFT, finding a main effect fo r Subject Type would indicate that overt and null subjects were treated differently by the participants. Similarly, finding a main effect for Context Type would indicate that Contrastive Focus, Topic Shift and Topic Maintenance contexts were treated differ ently. Finally, finding a main effect for Group would indicate that the native speaker and L2 speaker participants performed differently across the task. Finding an interaction between Subject Type and Context Type would indicate that overt and null subjec ts are treated differently in the different contexts. Finding interactions between Subject Type and Group would indicate that overt and null subjects were treated differently by the two participant groups. Finding interactions between Context Type and Grou p would indicate that Contrastive Focus, Topic Shift and Topic Maintenance tokens were treated differently by the participant groups. Lastly, finding an interaction between Subject Type, Context Type and Group would indicate that the native speaker and L2 speaker participants treated overt and null subjects in Contrastive Focus, Topic Shift and Topic Maintenance contexts differently from each other. Bonferroni post hoc two way and three way analyses were also conducted to examine the sources of any interact ions found. As first mentioned in Footnote 7 the analysis for the first region of interest, which is the overt subject ella Context Type (F(2, 6) 10.257, p = 0.010) and Group (F(1,18) 63.805, p = 0.000). No interaction was found between Context Type and Group (F(2, 6) 1.029, p = 0.410). For the second region of interest, which was the verb, a main effect was found for Group (F(1,29) 67.546, p = 0.000), but none were found for S ubject Type (F(1, 7) 0.731, p = 0.422) or for Context Type (F(2, 36) 0.813, p = 0.451). No interactions were found between Subject Type and Context Type (F(2, 10) 0.444, p = 0.654), Subject Type and Group (F(1, 7) 0.128, p = 0.731), Context Type and Group (F(2, 36) 0.652, p = 0.527) or between Subject Type, Context Type and Group (F(2, 10) 0.003, p =

PAGE 178

178 0.997). With respect to the final region of interest, the DO, no main effects were found for Subject Type (F(1, 0) 0.964, p = 0.771), Context Type (F(2, 2) 5.9 09, p = 0.139) or Group (F(1, 1) 38.143, p = 0.192. Neither were interactions found between Subject Type and Context Type (F(2, 180) 1.040, p = 0.356), Subject Type and Group (F(1, 0) 3.942, p = 0.689), Context Type and Group (F(2, 2) 4.009, p = 0.193), or between Subject Type, Context Type and Group (F(2, 180) 2.149, p = 0.120). The corresponding post hoc analyses are presented below in the statistical analysis of Contrastive Focus, Topic Shift and Topic Maintenance tokens for each region of interest. 5.3. 2.2 Contrastive Focus tokens Of the 12 tokens related to Contrastive Focus contexts, half of the target sentences contained overt subject pronouns while half contained null subject pronouns. Recall from Section 3.2.2.2 that overt subjects are felicitous in Contrastive Focus contexts while null subjects are infelicitous. Example (43), reproduced below as (61), first provides a Contrastive Focus context which is then followed by an overt and null subject target sentence, respectively. Here again, the forward slashes (/) indicate the divisions between the regions of interest. (61). Cuando salimos a cenar, mi novia prefiere comer platos livianos, pero yo prefiero comer algo sustancioso. prefer to eat a. As que/ella/come/ensaladas/y/yo/como/milanesas/en los restaurantes. b. #As que/come/ensaladas/y/como/milanesas/en los restaurantes. To allow for s tatistical comparison of the RTs for each region of interest, all regions of interest were consistently divided as is seen in example (61) above. The regions of interest analyzed in the following sections for Contrastive Focus tokens with overt subjects su ch as example (61a) above are the subject ella the verb come

PAGE 179

179 DO ensaladas interest are the verb come ens aladas of these regions began or ended the target sentence. Additionally, the overt referential subjects were always third person singular feminine ( ella come hace prepara toma ofrece lleva vende escribe lee compra pinta dibuja offline CMFT were distributed evenly across the overt and null subject tokens. The same is true of the 12 spillover region DOs ( ensaladas alfajores empanadas agua caf postres revistas cuentos poemas esculturas p aisajes figuras (Contrastive Focus, Topic Shift and Topic Maintenance) and will not be repeated below. The descriptive and statistical analyses are provided below along wi th a description. 5.3.2.2.1 Descriptive and statistical analyses. This section provides a statistical analysis of the RT data for the Contrastive Focus with overt subject (CFO) tokens and the Contrastive Focus with null subject (CFN) tokens. The data were treated in the same way as described in Section 5.3.1.2.1. That is, all RTs corresponding to missed comprehension questions (i.e. incorrect responses to the yes/no comprehension questions that followed the target sentences) were excluded from the analysis. Next, the RT data was trimmed by excluding all RTs slower than 2 seconds. Finally, the mean RT and standard deviation for each region of interest were calculated across the task. All RTs higher than 2 standard deviations from the mean RT were replaced wit h the cutoff value (x = mean RT + (standard deviation X 2)). As in the online G T figures and statistical analyses are provided according to the region of interest. Thus, Figure 5 15

PAGE 180

180 below provides a visual representation of the group average RT for the native speaker and L2 speaker groups for the first region of interest, which for the CFO tokens was the subject ella 205.50 409.60ms with a standard deviation of 51.20ms. The L2 speake r range was higher at 327.67 564.17ms with a standard deviation of 80.30ms. Figure 5 15. Online CMFT group results: Contrastive Focus tokens for first region ( ella NS=native speaker; L2=second language speaker; CFO=Contrastive Focus with overt subject hoc pairwise t test for independent samples (with Bonferroni adjustment) was conducted across contexts and groups. The result of this analysis showed that the native spea p = 0.001). 10 10 In the offline version of the Context Matching Felicitousness Task, three statistical comparisons were conducted and described: Subject Type (overt ver sus null), Context Type (Contrastive Focus versus Topic Shift versus Topic Maintenance) and Group (native speaker versus L2 speaker). This was possible as just one set of comparisons sk. Three sets of comparisons are necessary for the online task since three regions of interest were examined. Therefore, the second comparison (Contrastive Focus versus Topic Shift versus Topic Maintenance) will be presented for all three regions of inter est in Section 5.3.2.5.

PAGE 181

181 Next, Figure 5 16 provides a shows the group average RT for the native speaker and L2 speaker groups for the second region of interest, the verb. For thi s region, the average RT for the CFO tokens for the native speaker group was 319.82ms while that of the L2 speaker group was 521.718ms. The native speaker range spanned from 190.60 502.50 ms with a standard deviation of 68.66ms. Here again, the L2 speaker r ange was higher at 389.67 618.17ms with a standard deviation of 92.88ms. Regarding the CFN tokens, the average RT for the native speaker group was 307.88ms while that of the L2 speaker group was substantially higher at 517.33ms. The native speaker range fo r the first region of interest was 190.50 398.40ms with a standard deviation of 51.70ms. The L2 speaker range was higher at 464.00 661.80ms with a standard deviation of 87.56ms. Figure 5 16. Online CMFT group results: Contrastive Focus tokens for secon d region (verb). NS=native speaker; L2=second language speaker; CFO=Contrastive Focus with overt subject; CFN=Contrastive Focus with null subject With respect to the Subject Type and Context Type interaction in the second region of interest, the post hoc p airwise t test for independent samples (with Bonferroni adjustment) showed that

PAGE 182

182 different ( p = 0.236). No significant difference was found between the RTs to the CFO t okens and the CFN tokens for the L2 speakers ( p = 0.914). Both of these findings are unexpected if the participants have detected the infelicitousness of the CFN tokens. Regarding the Group comparison, the native speaker group average RT was found to be si gnificantly faster than that of the L2 speaker group for CFO tokens in the second region ( p = 0.000). The same result was found for the CFN tokens ( p = 0.000). Lastly, Figure 5 17 provides a visual representation of the group average RT for the native spea ker and L2 speaker groups for the third region of interest, the direct object (DO). With respect to the spillover region, the average RT for the native speaker group was 354.39ms while that of the L2 speaker group was again higher at 792.78ms. The native s peaker group range was 196.67 493.50ms with a standard deviation of 85.99ms The L2 speaker range was again higher at 511.60 1171.00ms and the standard deviation was 219.62ms. For the TSN tokens, the na tive speaker group average was 374.63ms while that of the L2 speaker group was 726.68ms. The range of the native speaker group was 245.00 530.33ms with a standard deviation of 81.58ms. The range of the L2 speaker group (557.75 969.40ms) and their standard deviation was 158.17ms. For the Subject Type and Conte xt Type interaction in the third region of interest, the post hoc pairwise t test for independent samples (with Bonferroni adjustment) showed that the different ( p = 0.081 ). The same finding obtained for the L2 speaker group in that no significant difference was found between the RTs to the CFO versus CFN tokens ( p = 0.575). Again, these results are unexpected if the participants have detected the infelicitousness of the CF N tokens.

PAGE 183

183 Figure 5 17. Online CMFT group results: Contrastive Focus tokens for third region (DO). NS=native speaker; L2=second language speaker; CFO=Contrastive Focus with overt subject; CFN=Contrastive Focus with null subject Regarding the Group com parison, no significant difference in RT was found between the native speaker group average RT and that of the L2 speaker group for CFO tokens( p = 0.143). The same was found for the CFN tokens ( p = 0.299). Section 5.3.2.2.2 below provides an interpretation of the results found for Contrastive Focus tokens for all three regions of interest. 5.3.2.2.2 Interpretation of results. Regarding the comparison between overt and null subjects in Contrastive Focus contexts, neither the native speaker nor L2 speaker gro up showed significantly slower RTs to the infelicitous CFN tokens in either the second or third region of interest. This is an unexpected finding and may be interpreted as neither group distinguishing between CFO and CFN tokens. Additionally, it may indica te that the task was unsuccessful at speaker group performed as the native speaker group. With respect to the comparison between the two participant groups, and a s discussed in Section 5.3.2.2.1, the native speaker group average RT to the three regions of interest for both CFO and CFN tokens was faster than that of

PAGE 184

184 the L2 speaker group. For the first region of interest, the Bonferroni post hoc tests revealed a stat istically significant difference between the native speaker and L2 speaker groups with respect to their RTs to CFO tokens, which can only be a product of the slower overall RT of the L2 group since no other factors were present in the analysis of this regi on. Regarding the second region of interest, statistically significant differences were found between the two groups for the CFO and CFN tokens, which again are likely due to the overall slower RT of the L2 group as compared to the native speaker group. Di fferently from the two previous regions, no statistically significant differences were found between the native speaker and L2 speaker group for CFO and CFN tokens in the third region, suggesting a similar processing by both groups. Regardless, the signif icantly slower RTs in the first and second regions of interest do not indicate that the L2 speaker group has not converged on the discourse distribution of subject pronouns in Contrastive Focus contexts. As was cautioned with respect to the differences fou nd in average RT in the GJCT task, while it may initially seem as though the L2 speaker group differs from the native speaker group in terms of their processing of the discourse distribution of overt and null subjects in Spanish, a comparison between the R T to the infelicitous and felicitous tokens must be made before drawing this conclusion. As was described in Section 5.3, slower RTs are expected when participants detect ungrammaticality or infelicitousness. Therefore, the same examination of average diff erential RT between the felicitous and infelicitous tokens made in Section 5.3.1.5 for the referential and expletive subject tokens will be conducted for the Contrastive Focus tokens in the following section. Specifically, the average differential RT betwe en CFO and CFN tokens will be examined. Since two comparisons are made from within the same data set, a Bonferroni correction was applied, which lowered the original p value of 0.05 to 0.025.

PAGE 185

185 5.3.2.2.3 Comparative analysis of RTs for Contrastive Focus toke ns. As discussed above, a simple analysis of overall RT between the groups does not tell us if either group detects the infelicitousness (as evidenced by comparatively slower RTs) of the CFN tokens. Instead, a comparison between the RTs to felicitous and i nfelicitous tokens needs be performed. Recall that out of the three token types examined in the offline and online versions of the Context Matching Felicitousness Task, two prefer overt referential subjects (Contrastive Focus and Topic Shift) while one pre fers null subjects (Topic Maintenance). Thus, the average differential RT to the overt and null counterbalances within each token type will be compared across each group for the second and third region of interest 11 For Contrastive Focus tokens, slower RTs are expected for the null subject tokens since this is the infelicitous token type. Figure 5 19 below shows the average differential RT for the second region of interest (verb) for the CFN versus CFO tokens. The average differential RT was calculated by subtracting the group average RT to the felicitous CFO from the group average RT to the infelicitous CFN token type. From Figure 5 18, it is clear that the equation described above resulted in a negative average differential RT for both the native speaker group and the L2 speaker group. The native speaker group difference in average RT for the CFN CFO tokens was greater than that of the L2 speaker group: 11.94ms versus 4.37ms, respectively. The negative result indicates that, on average, the effect of det ecting infelicitousness (i.e. slower RTs) is not evidenced in the second region of interest. If it were, we would expect the result to be positive since the first element of the equation, CFN, would produce a comparatively slower RT than that of the second element, CFO. 11 No comparison can be made for the first region of interest, the subject ella the Contrastive Focus with overt subject tokens but necessarily absent in the Contrastive Focus with null subj ect tokens. This is true of the Topic Shift and Topic Maintenance tokens as well.

PAGE 186

186 Figure 5 18. Online CMFT comparative group results for second region (verb). NS=native speaker; L2=second language speaker; CFN CFO=group average RT to CFN tokens minus group average RT to CFO tokens An independent samples t test was co nducted to determine if the average differential between the groups was significantly different. The independent samples t test revealed no statistically significant difference in average differential RT for the CFN CFO counterbalance (t(30) = 0.506, p = 0 .617). This suggests that neither participant group distinguished between the infelicitous and felicitous Contrastive Focus tokens. Figure 5 19 below shows the average differential RT to the third region of interest (DO). As Figure 5 19 shows, an interesting di fference obtains for the third region of interest: the average differential RT for the native speaker group is negative ( 20.24ms), but for the L2 speaker group it is positive (66.10ms), as would be expected if the participants detected the infelicitous na ture of the null subject in the CFN tokens. The positive result evidenced in the L2 speaker group demonstrates that they have detected the infelicity of the null referential subject, but the lack thereof with respect to the native speaker participants indi cates that they have not.

PAGE 187

187 Figure 5 19. Online CMFT comparative group results for third region (DO). NS=native speaker; L2=second language speaker; CFN CFO=group average RT to CFN tokens minus group average RT to CFO tokens The independent samples t te st revealed a statistically significant difference in average differential RT for the CFN CFO counterbalance (t(30) = 2.513, p = 0.018). While this result indicates processing differences between the two groups, it should be pointed out that the L2 speake r group has performed as expected. 5.3.2.3 Topic Shift tokens Half of the 12 target sentences in Topic Shift contexts contained overt subject pronouns while half contained null subject pronouns. Recall from Section 3.2.2.2 that overt subjects are felicito us in Topic Shift contexts while null subjects are infelicitous. Example (44), which has been reproduced below as (62), provides a Topic Shift context as well as an overt and null subject target sentence, respectively. As with previous tokens, the forward slashes (/) indicate the divisions between the regions of interest. (62). Mi hija quiere ser autora y no tiene otros intereses. Yo creo que es mejor tener varios intereses y sugiero otras actividades, pero no importa lo que diga yo.

PAGE 188

188 be an author and she has no other interests. I think that it is best to a. Finalmente/ella/escribe/cuentos/y/ella/pasa/todo/el da/en su cuarto. b. #Finalmente/esc ribe/cuentos/y/pasa/todo el da/en su cuarto. In order to statistically compare the RTs for each region of interest, all regions of interest were consistently divided a s is seen in example (62) above. The regions of interest analyzed in the following sections for Topic Shift tokens with overt subjects such as example (62) above are the subject ella escribe cuentos examples like (62), which contain a null subject, the regions of interest are escribe cuentos and interpretation. 5.3.2.3.1 Descriptive and statistical analyses. This section provides descriptive and statistical analysis of the RT data for the Topic Shift with overt subject (TSO) tokens and the Topic Shift with null subject (TSN) tokens. The data were treated in the same manner described in Se ction 5.3.2.2.1. The figures and statistical analyses are provided according to the region of interest; accordingly, Figure 5 20 below provides a visual representation of the group average RT for the native speaker and L2 speaker groups for the first regio n of interest. Subsequently, Figures 5 21 and 5 22 provide information regarding the second and third regions of interest. Regarding Figure 5 20, it is shown that for the first region of interest the native speaker group average RT for the Topic Shift with overt subject tokens (TSO) was faster than that of the L2 speaker group: 307.64ms versus 440.96ms, respectively. The native speaker range was 197.50 378.50ms with a standard deviation of 51.98ms. The L2 speaker range was substantially higher at 382.67 487 .00ms with a standard deviation of 39.60ms.

PAGE 189

189 Figure 5 20. Online CMFT group results: Topic Shift tokens for first region ( ella NS=native speaker; L2=second language speaker; TSO=Topic Shift with overt subject he TSO tokens in the first region, a post hoc pairwise t test for independent samples (with Bonferroni adjustment) was conducted across contexts and than that of p = 0.000). Next, Figure 5 21 provides a visual representation of the group average RT for the native speaker and L2 speaker groups for the second region of interest, the verb. Here again, the native speaker group average RT fo r the TSO tokens was faster than that of the L2 speaker group: 325.66ms versus 509.35ms, respectively. The native speaker group range was 228.17 536.60ms with a standard deviation of 74.57ms The L2 speaker range was again higher at 356.80 673.80ms and the standard deviation was 114.04ms. The same is true with respect to the TSN tokens in that the native speaker group average RT was 347.05ms while that of the L2 speaker group was 536.93ms. Regarding the TSN tokens, the native speaker group average RT was 3 47.05ms while that of the L2 speaker group was higher at 536.39ms. The native speaker range

PAGE 190

190 was 237.00 518.75ms with a standard deviation of 70.71ms. The L2 speaker range was substantially higher at 434.25 666.83ms with a standard deviation of 98.82ms. Figure 5 21. Online CMFT group results: Topic Shift tokens for second region (verb). NS=native speaker; L2=second language speaker; TSO=Topic Shift with overt subject; TSN=Topic Shift with null subject With respect to the Subject Type and Context Type int eraction in the second region of interest, the post hoc pairwise t test for independent samples (with Bonferroni adjustment) revealed no significant difference in average RT for the TSO versus TSN tokens for the native speaker group ( p = 0.094). The same r esult obtained with respect to the L2 speaker group. Their average RT to the TSO versus TSN tokens did not differ for the second region of interest ( p= 0.600). These findings are unexpected if the participants detect the infelicitousness of the TSN tokens. For the Group comparison, the native speaker group average RT was found to be significantly faster than that of the L2 speaker group for TSO tokens in the second region ( p = 0.000). The same was found for the TSN tokens ( p = 0.000).

PAGE 191

191 Lastly, Figure 5 22 provides a visual representation of the group average RT for the native speaker and L2 speaker groups for the final region of interest, the DO. As with the previous regions of interest, the native speaker group average RT for the TSO tokens was faster than that of the L2 speaker group: 368.46ms versus 613.18ms, respectively. Finally, the native speaker group average for the spillover region, the direct object (DO), was 368.46ms while that of the L2 speaker group was 613.18ms. The range of the native speaker group was 229.17 573.00ms with a standard deviation of 84.97ms. The range of the L2 speaker group was again higher (462.50 842.25ms) and their standard deviation was 142.29ms. The same is true of the TSN tokens; the native speaker group average RT was 343 .87ms while that of the L2 speaker group was 718.75ms. Lastly, the average RT for TSN tokens in the native speaker group was 343.87ms while that of the L2 speaker group was quite a bit higher at 718.75ms. The native speaker group range was 249.00 444.33 ms with a standard deviation of 58.27ms The L2 speaker range was again higher at 562.60 995.33ms and the standard deviation was 177.49ms. Figure 5 22. Online CMFT group results: Topic Shift tokens for third region (DO). NS=native speaker; L2=second langu age speaker; TSO=Topic Shift with overt subject; TSN=Topic Shift with null subject

PAGE 192

192 Regarding the Subject Type and Context Type interaction in the third region of interest, the post hoc pairwise t test for independent samples (with Bonferroni adjustment) sh owed a significant difference in average RT for the TSO versus TSN tokens for the native speaker group ( p = 0.024). The same result obtained with respect to the L2 speaker group. Their average RT to the TSO versus TSN tokens differed for the third region o f interest ( p= 0.016). These findings are expected if the participants detect the infelicitousness of the TSN tokens. With respect to the Group comparison, the native speaker group average RT was found to be significantly faster than that of the L2 speake r group for TSO tokens in the second region ( p = 0.073). The same was found for the TSN tokens ( p = 0.090). The following section provides an interpretation of the results of the Topic Shift tokens for the three regions of interest. 5.3.2.3.2 Interpretatio n of results. For the comparison between overt and null subjects in Topic Shift contexts, neither the native speaker nor L2 speaker group showed significantly slower RTs to the infelicitous TSN tokens in the second region of interest. While this may be une xpected, both groups showed significantly slower RTs to the TSN tokens in the third region of interest, demonstrating their distinction between the use of overt and null subjects in Topic Shift contexts. Regarding the comparison made between the two partic ipant groups the native speaker group average RT to the three regions of interest for both TSO and TSN tokens was faster than that of the L2 speaker group. For the first region of interest, the Bonferroni post hoc tests revealed a statistically significant difference between the native speaker and L2 speaker groups with respect to their RTs to TSO tokens. Recall that this finding can only be due to the slower overall RT of the L2 group since no other factors were present in the analysis of this region. For the second region of interest, statistically significant differences were found between the two groups for the TSO and TSN tokens, which again are likely a product of the overall

PAGE 193

193 slower RT of the L2 group as compared to the native speaker group. Finally, a nd unlike the previous two regions, no statistically significant differences were found between the native speaker and L2 speaker group for TSO and TSN tokens in the third region, suggesting a similar processing by both groups. As discussed in Section 5.3. 2.2.2, the significantly slower RTs in the first and second regions of interest do not indicate that the L2 speaker group has not converged on the discourse distribution of subject pronouns in Contrastive Focus contexts. A comparison between the RT to the infelicitous and felicitous tokens must be made before drawing this conclusion. As such, the same examination of average differential RT between the felicitous and infelicitous tokens made in Section 5.3.1.5 for the Contrastive Focus tokens is shown for th e Topic Shift below. A Bonferroni correction was applied, which lowered the original p value of 0.05 to 0.025. 5.3.2.3.3 Comparative analysis of RTs for Topic Shift tokens This section provides a comparison of the RTs to felicitous and infelicitous Topic Shift tokens. As discussed in Section 5.3.2.1.3, the average differential RT to the overt and null Topic Shift tokens will be compared across each group for the second and third region of interest. In both comparisons, slower RTs are expected for TSN token s since this is the infelicitous token type. Figure 5 23 below shows the average differential RT for the second region of interest (the verb) for the TSN versus TSO tokens. Again, the average differential RT was calculated by subtracting the group average RT to the felicitous TSO from the group average RT to the infelicitous TSN token type. In Figure 5 23, we see that the equation outlined above resulted in a positive average differential RT for both the native speaker group and the L2 speaker group and tha t the results were quite similar: 21.39ms and 27.04ms, respectively. The positive result indicates that, on average, the effect of detecting infelicitousness (i.e. slower RTs) is in fact evidenced in the second region of interest.

PAGE 194

194 Figure 5 23. Online C MFT comparative group results for second region (verb). NS=native speaker; L2=second language speaker; TSN TSO=group average RT to TSN tokens minus group average RT to TSO tokens The independent samples t test revealed no statistically significant differe nce in average differential RT between the two participant groups for the TSN TSO counterbalance (t(30) = 1.205, p = 0.238), indicating that the distinction the L2 speaker group made between the infelicitous and felicitous Topic Shift tokens matched that o f the native speaker group. Finally, Figure 5 24 shows the differential RT to the third region of interest, which was the DO. Similarly to what was seen with the Contrastive Focus tokens in Section 5.3.2.2.3, the average differential RT for the native spe aker group is negative ( 24.59ms), but for the L2 speaker group it is positive (105.58ms), which is expected if the participants detect the infelicitous nature of the null subject in the TSN tokens. The positive result evidenced in the L2 speaker group dem onstrates that they have detected the infelicitousness of the null referential subject.

PAGE 195

195 Figure 5 24. Online CMFT comparative group results for third region (DO). NS=native speaker; L2=second language speaker; TSN TSO=group average RT to TSN tokens minu s group average RT to TSO tokens Nevertheless, no statistically significant difference was found by the independent samples t test with respect to the average differential RT for the TSN TSO counterbalance (t(30) = 1.575, p = 0.126). This indicates that the distinction between the infelicitous and felicitous Topic Shift tokens made by the L2 speaker group matched that of the native speaker group. 5.3.2.4 Topic Maintenance tokens Finally, the 12 Topic Maintenance tokens were divided evenly across target s entences with overt subject pronouns (n=6) and null subject pronouns (n=6). As discussed in Section 3.2.2.2, overt subjects are l ess preferred in Topic Maintenance contexts while null subjects are p referred Example (45), reproduced below as (63), shows a T opic Maintenance context with both an overt and null subject target sentence. As in previous token types, the forward slashes (/) indicate the divisions between the regions of interest. (63). Mi cuada es muy sociable. Tiene muchos amigos y por eso va a muchas cenas a la canasta donde tiene que contribuir con algo.

PAGE 196

196 in law is very social. She has a lot of friends and for that reasons, she goes a. #As que/ella/lleva/postres/y/com parte/todo/con sus amigos. b. As que/lleva/postres/y/comparte/todo/con sus amigos. All Topic Maintenance tokens were consistently divided to allow for statistical comparison of the RTs f or each region of interest. With respect to tokens like (63a) which contain an overt subject, the regions of interest are the subject ella lleva region DO postres null subject, the regions of interest are the verb lleva postres subsections provide the descriptive results, statistical analysis and interpretation. 5.3.2.4.1 Descriptive and statistical analyse s. Next, this section provides a statistical analysis of the RT data for the Topic Maintenance with overt subject (TMO) tokens and the Topic Shift with null subject (TMN) tokens. The data were treated as described for the Contrastive Focus and Topic Shift tokens. As with these two previous token types, figures and statistical analyses are provided according to the regions of interest (the subject ella verb and the direct object (DO)). Accordingly, Figure 5 25 provides a visual representation of t he group average RT to the first region of interest for the native speaker and L2 speaker groups. Figures 5 26 and 5 27 provide information regarding the second and third regions of interest, respectively. The average RT for the native speaker group for th e first region of interest, the subject ella The native speaker range for the first region of interest was 246.67 432.25ms with a standard deviation of 50.34. The L2 speaker ran ge was higher at 409.75 490.40ms with a very low standard deviation of 29.76.

PAGE 197

197 Figure 5 25. Online CMFT group results: Topic Maintenance tokens for first region ( ella NS=native speaker; L2=second language speaker; TMO=Topic Maintenance with ove rt subject the TMO tokens in the first region, a post hoc pairwise t test for independent samples (with Bonferroni adjustment) was conducted across contexts and groups. T he result showed that the for ( p = 0.000). Next, Figure 5 26 provides below a visual representation of the group average RT for the native speaker and L2 speaker group s for the second region of interest, the verb. For this region, the average RT for the TMO tokens for the native speaker group was 315.71ms while that of the L2 speaker group was again higher at 520.16ms. The native speaker group range was 231.83 483.50ms with a standard deviation of 66.65ms. The L2 speaker range was again higher at 417.80 635.60ms and the standard deviation was 88.80ms. Regarding the TMN tokens, the average RT for the native speaker group was 321.90ms while that of the L2 speaker group was higher at 535.73ms. The native speaker range for the first region of interest was 214.75

PAGE 198

198 455.67 ms with a standard deviation of 62.23ms. The L2 speaker range was substantially higher at 416.00 638.83ms with a standard deviation of 77.73ms. Figure 5 26. Online CMFT group results: Topic Maintenance tokens for second region (verb). NS=native speaker; L2=second language speaker; TMO=Topic Maintenance with overt subject; TMN=Topic Shift with null subject Regarding the Subject Type and Context Type interacti on in the second region of interest, the post hoc pairwise t test for independent samples (with Bonferroni adjustment) showed no significant difference in average RT was found for the TMO versus TMN tokens for the native speaker group ( p = 0.619). Likewise no significant difference in average RT was found for the TMO versus TMN tokens in the L2 speaker group ( p = 0.546). These results are unexpected in the participants detect the infelicitousness of the TMO tokens. For the Group comparison, the native spe aker group average RT was found to be significantly faster than that of the L2 speaker group for TMO tokens in the second region ( p = 0.000). The same was found for the TMN tokens ( p = 0.000). Finally, Figure 5 27 provides a visual representation of the gr oup average RT for the native speaker and L2 speaker groups for the TMO and TMN tokens in third region of interest,

PAGE 199

199 the DO. Here, the native speaker group average RT was 355.07ms while that of the L2 speaker group was 699.00ms. The range of the native spea ker group was 213.50 611.67ms with a standard deviation of 97.52ms. The range of the L2 speaker group was substantially higher at 527.40 891.60ms with a standard deviation of 135.31ms. With respect to the TMN tokens, the average RT for the native speaker g roup was 354.61ms while that of the L2 speaker group was substantially higher at 661.17ms. The native speaker group range was 240.00 539.00ms with a standard deviation of 86.63ms Here again, the L2 speaker range was higher at 396.00 1080.80ms and the stan dard deviation was 248.95ms. Figure 5 27. Online CMFT group results: Topic Maintenance tokens for third region (DO). NS=native speaker; L2=second language speaker; TMO=Topic Maintenance with overt subject; TMN=Topic Maintenance with null subject For the Subject Type and Context Type interaction in the last region of interest, the post hoc pairwise t test for independent samples (with Bonferroni adjustment) showed no significant difference in average RT was found for the TMO versus TMN tokens for the n ative speaker group ( p = 0.978). Likewise, no significant difference in average RT was found for the TMO

PAGE 200

200 versus TMN tokens in the L2 speaker group ( p = 0.804). These results are unexpected if the participants detect the infelicitousness of the TMO tokens. Regarding the Group comparison, no significant difference was found between the native speaker group average RT to TMO tokens as compared to that of the L2 speaker group ( p = 0.429). However, the native speaker group average RT was significantly faster fo r TMN tokens than that of the L2 speaker group ( p = 0.000).The next section provides an interpretation of the results found for all three regions of interest. 5.3.2.4.2 Interpretation of results. As with the previous two token types, a comparison between o vert and null subjects in Topic Maintenance contexts was made. In both the second and third region of interest, neither the native speaker nor L2 speaker group showed significantly slower RTs to the infelicitous TMO tokens as compared to the felicitous TMN tokens. This is an unexpected finding and may be interpreted as neither group distinguishing between TMO and TMN tokens. Nonetheless, the L2 speaker group performed as the native speaker group did. With respect to the comparison made between the two parti cipant groups and as in the previous two token types, the native speaker group average RT to the three regions of interest for TMO and TMN tokens was faster than that of the L2 speaker group. For the first region of interest, the post hoc tests revealed a statistically significant difference between the native speaker and L2 speaker groups with respect to their RTs to TMO tokens. This finding can only be due to the slower overall RT of the L2 group since no other factors were present in the analysis of this region. The same is true of both TMO and TMN tokens in the second region of interest and for the TMN tokens for the third region of interest. As previously discussed, the significantly slower RTs in the do not indicate that the L2 speaker group has not co nverged on the discourse distribution of subject pronouns in Contrastive Focus contexts. Rather, an examination of the

PAGE 201

201 conducted. The results are described b elow in Section 5.3.2.4.3 and the same Bonferroni correction was applied, effectively lowering the original p value of 0.05 to 0.025. 5.3.2.4.3 Comparative analysis of RTs for Topic Maintenance tokens. Finally, t his section presents a comparison of the RTs to felicitous and infelicitous Topic Maintenance tokens. The average differential RT to the overt and null Topic Maintenance tokens will be compared across each group for the second and third region of interest. Across both comparisons, slower RTs are exp ected for the overt subjects since these are less preferred than null subjects in Topic Maintenance contexts. Figure 5 28 below shows the average differential RT for the second region of interest (the verb) for the TMO versus TMN tokens. Here again, the av erage differential RT was calculated by subtracting the group average RT to the felicitous TMN from the group average RT to the infelicitous TMO token type. Figure 5 28. Online CMFT comparative group results for second region (verb). NS=native speaker ; L2=second language speaker; TMO TMN=group average RT to TMO tokens minus group average RT to TMN tokens

PAGE 202

202 In Figure 5 28, we see that the equation outlined above resulted in a negative average differential RT for both the native speaker and the L2 speaker group: 6.20ms and 15.57ms, respectively. The negative result indicates that, on average, the effect of detecting infelicitousness (i.e. slower RTs) was not evidenced in the second region of interest. The independent samples t test revealed no statistical ly significant difference in average differential RT for the TMO TMN counterbalance: t(30) = 0.141, p = 0.889. This finding suggests that the distinction the L2 speaker group made between the infelicitous and felicitous Topic Maintenance tokens equals tha t of the native speaker group. Figure 5 29. Online CMFT comparative group results for third region (DO). NS=native speaker; L2=second language speaker; TMO TMN=group average RT to TMO tokens minus group average RT to TMN tokens Lastly, Figure 5 29 ab ove shows the differential RT to the third region of interest, which was the direct object (DO). For both participant groups, the average differential RT was positive: 0.46ms for the native speaker group and 37.84ms for the L2 speaker group. This finding i s expected if the participants detect the infelicitous nature of the overt subject in the TMO tokens.

PAGE 203

203 Thus, the positive result evidenced in both participant groups demonstrates that they have detected this infelicity. The independent samples t test reveal ed no statistically significant difference with respect to the average differential RT for the TMO TMN counterbalance (t(30) = 0.247, p = 0.807). This indicates that the distinction the L2 speaker group made between the infelicitous and felicitous Topic Ma intenance tokens matched that of the native speaker group. 5.3.2.5 Context Type comparison (Contrastive Focus versus Topic Shift versus Topic Maintenance) This section explains the second comparison between Contrastive Focus, Topic Shift and Topic Mainten ance tokens described in Footnote 10 above. The results for the second and third regions of interest are presented. For the second region of interest, the post hoc pairwise t test for independent samples (with Bonferroni adjustment) showed that no signific ant difference between tokens ( p = 1.000). This finding is expected since both token types are felicitous. The same is tokens as compared to infelicitous TMO tokens ( p = 1.000). These findings are unexpected since only CFO tokens are felicitous. Finally, no TSO tokens as compared to infelicitous TMO tokens ( p = 1.000). The same results obtained for the L2 speaker group in that no differences between RTs were found between CFO and TSO tokens ( p = 1.000), CFO and TMO tokens ( p = 1.000) or TSO and TMO tokens ( p = 1.000). In the case o f the CFO and TSO comparison, this result is expected since both tokens are felicitous; however the latter two findings are unexpected. Regarding the null tokens, the native speaker their RT to the TSN tokens comparison ( p = 0.004). This is perhaps unexpected since both tokens are infelicitous. For

PAGE 204

204 the CFN versus TMN comparison, no significant difference was found ( p = 0.148), which is unexpected since only the TMN tokens are felicit RT to the TSN tokens did not differ significantly from their RT to the TMN tokens ( p = 0.064). This result is unexpected since only TMN tokens are felicitous. With respect to the L2 speaker group, no signifi cant difference was found in RT for the CFN versus TSN comparison ( p = 1.000), for the CFN versus TMN comparison ( p = 1.000) nor for the TSN versus TMN comparison ( p = 1.000). The first result is expected since both tokens are infelicitous; however a signi ficant difference was expected for the final two comparisons as only the TMN tokens are felicitous. Finally, for the third region of interest, the post hoc pairwise t test for independent samples (with Bonferroni adjustment) revealed no significant differe nce between the native p = 0.697). This finding is expected since both token types are felicitous. The same is true of the compared to infelicitous TMO tokens ( p = 0.482), which is unexpected since only CFO tokens are felicitous. Finally, no significant difference was infelicitous TMO t okens ( p = 0.592), which is also unexpected. For the L2 speaker group, the average RT to the felicitous CFO tokens was significantly slower than that of the felicitous TSO tokens ( p = 0.003), which is perhaps unexpected since both are felicitous. No signif icant difference was found in RT to the CFO and TMO tokens ( p = 0.673), which is unexpected since TSO did not differ significantly from that of the infelicitou s TMO tokens ( p = 0.401), which is

PAGE 205

205 tokens did not differ significantly from that of their RT to the TSN tokens comparison ( p = 1.000). This is not unexpected sinc e both tokens are infelicitous. For the CFN versus TMN comparison, no significant difference was found ( p average RT to the TSN tokens did not differ significantly from their RT to the TMN tokens ( p = 1.000). N either of the latter two results is expected since only TMN tokens are felicitous. For the L2 speaker group, no significant difference was found in RT for the CFN versus TSN comparison ( p = 0.542). However, their average RT to the CFN tokens as compared to the TMN tokens was significantly slower ( p = 0.017), which is expected since only TMN tokens are felicitous. Regarding the TSN versus TMN comparison, no significant difference in RT was found ( p = 0.542), which is unexpected since only TMN tokens are feli citous. 5.3.2.6 Comparative analysis of RTs across token type One final comparison will be made across the token types. Recall that Contrastive Focus and Topic Shift contexts prefer overt subjects while Topic Maintenance contexts prefer null subjects. Thu s, we can determine if the participant groups are slower with overt subjects in Topic Maintenance contexts than in Contrastive Focus and Topic Shift contexts, yet slower with null subjects in Contrastive Focus and Topic Shift contexts as compared to Topic Maintenance contexts. Equally, as the offline data showed more categorical responses to the Contrastive Focus tokens, we can compare them to Topic Shift tokens. Thus, the average differential RT between the CFO and TMO tokens, CFO and TSO tokens and TSO an d TMO is compared across each group for all three regions of interest. Similarly, the average differential RT between the CFN and TMN tokens, CFN and TSN tokens and TSN and TMN tokens is compared across each group for all three regions of interest. Slower RTs are expected for the overt subjects in Topic Maintenance contexts as compared to Contrastive Focus or Topic Shift contexts, since these are less preferred than null subjects in Topic Maintenance contexts. On the other hand, slower RTs

PAGE 206

206 are expected for the null subjects in Contrastive Focus and Topic Shift contexts as compared to Topic Maintenance contexts, since these are less preferred than overt subjects in Contrastive Focus and Topic Shift contexts. Independent samples t tests were performed to deter mine if the group average differential RTs were statistically significant. Since two comparisons are made from within the same data set, a Bonferroni correction was applied, which lowered the original p value of 0.05 to 0.025. Figure 5 30 below shows the a verage differential RT for the first region of interest ( ella TSO tokens. As before, the average differential RT was calculated by subtracting the group average RT to the felicitous CF O or TSO token type from the group average RT to the infelicitous TMO token type. Figure 5 30. Online CMFT comparative group results for first region ( ella speaker; L2=second language speaker; TMO CFO=group average RT to TMO tokens m inus group average RT to CFO tokens; TMO TSO=group average RT to TMO tokens minus group average RT to TSO tokens; CFO TSO=group average RT to CFO tokens minus group average RT to TSO tokens

PAGE 207

207 In Figure 5 30, we see that the equation outlined above resulted i n a positive average differential RT for both the native speaker group and the L2 speaker for the first two comparisons (39.15ms and 22.29ms, respectively, for the TMO CFO comparison and 33.80ms and 17.50ms, respectively, for the TMO TSO comparison), but a negative average for the third comparison (05.35ms and 04.79ms, respectively, for the CFO TSO comparison). The two positive results infelicitous TMO tok ens than for felicitous CFO or TSO tokens. The negative result found for region of interest for felicitous CFO tokens than for felicitous TSO tokens. An i ndependent samples t test was conducted to determine if the average differential between the groups was significantly different. The t test revealed no statistically significant difference in differential RT for the TMO CFO counterbalance (t(30) = 0.719, p = 0.478) or for the TMO TSO counterbalance ((t(30) = 1.110, p CFO TSO tokens was found (t(30) = 0.028, p = 0.978). These findings suggest that the distinction the L 2 speaker group made between the infelicitous TMO and felicitous CFO and TSO tokens equals that of the native speaker group and that the average differential RT to the CFO versus TSO tokens of the L2 speaker group matched that of the native speaker group. Next, Figure 5 31 shows the average differential RTs to the overt subject tokens for the second region of interest, which corresponds to the verb. With respect to the TMO CFO comparison, for both participant groups the average differential RT was negative : 4.12ms and 1.54ms, respectively. The native speaker group also demonstrated a negative average differential RT for the TMO TSO comparison ( 9.96ms) and the CFO TSO comparison while the L2

PAGE 208

208 d 12.36ms, respectively). With respect to the TMO CFO tokens, these findings are unexpected if the participant groups detect the infelicitous nature of the overt subject in TM contexts in the verb region. The same is true of the native speaker group for th e TMO TSO tokens. For the L2 speaker group, though, the positive result suggests that they have detected the infelicitous nature of the overt subject in TM contexts as compared to its felicitous nature in TSO contexts. Figure 5 31. Online CMFT comparat ive group results for second region (verb). NS=native speaker; L2=second language speaker; TMO CFO=group average RT to TMO tokens minus group average RT to CFO tokens; TMO TSO=group average RT to TMO tokens minus group average RT to TSO tokens; CFO TSO=gro up average RT to CFO tokens minus group average RT to TSO tokens The independent samples t tests revealed no statistically significant difference with respect to the average differential RT for the TMO CFO counterbalance (t(30) = 094, p = 0.926), for the TMO TSO counterbalance (t(30) = 0.756, p =0.456) or for the CFO TSO comparison (t(30) = 0.638, p = 0.528). These findings indicate that the distinction the L2 speaker group made between the infelicitous TMO tokens and the felicitous CFO and TSO

PAGE 209

209 tokens ma tched that of the native speaker group and that no differences were found between the two groups with respect to the CFO TSO comparison. The null subject token comparisons are shown below for the second region. Regarding the CFN TMN and CFN TSN comparison s, the average differential RT was negative for the native speaker and L2 speaker groups: 14.02ms and 18.40ms and 39.17ms and 19.05ms, respectively. However, for the TSN TMN comparison, both groups demonstrated a positive average differential RT: 25.15 ms for the native speaker group and 0.66ms for the L2 speaker group. If the participant groups detect the infelicitous nature of null subjects in Contrastive Focus contexts as compared to its felicitous nature in Topic Maintenance contexts, these findings are unexpected since we would expect a slower RT for the CFN tokens than for the TMN tokens. The same expectation is true of TSN tokens as compared to the TMN tokens, and here this expectation is evidenced for both groups. Figure 5 32. Online CMFT comp arative group results for second region (verb). NS=native speaker; L2=second language speaker; CFN TMN=group average RT to CFN tokens minus group average RT to TMN tokens; TSN TMN=group average RT to TSN tokens minus group average RT to TMN tokens; CFN TSN =group average RT to CFN tokens minus group average RT to TMN tokens

PAGE 210

210 The independent samples t tests revealed no statistically significant differences with respect to the average differential RT for the CFN TMN counterbalance (t(30) = 0.203, p= 0.841), for the CFN TMN counterbalance ((t(30) = 0.766, p =0.450) or for the CFN TSN comparison (t(30) = 0.550, p = 0.586). These findings indicate that the distinction the L2 speaker group made between the infelicitous CFN and TSN tokens and the felicitous TMN token s matched that of the native speaker group and that their average differential RT to the CFN TSN comparison also matched that of the native speaker group. Next, Figure 5 33 below provides the average differential RTs to the overt subject tokens for the final re gion of interest, the DO. For the TMO CFO comparison for the DO region of interest, the average differential RT of both the native speaker and the L2 speaker group was negative: 19.56ms and 27.68ms, respectively. Figure 5 33. Online CMFT comparative group results for third region (DO). NS=native speaker; L2=second language speaker; TMO CFO=group average RT to TMO tokens minus group average RT to CFO tokens; TMO TSO=group average RT to TMO tokens minus group average RT to TSO tokens; CFO TSO=group aver age RT to CFO tokens minus group average RT to TSO tokens

PAGE 211

211 The native speaker group also demonstrated a negative average differential RT for the TMO TSO comparison ( (85.83ms). Final ly, both participant groups evidenced a positive average differential RT for the CFO TSO comparison (6.17ms and 113.50ms, respectively). Considering the TMO CFO tokens, these findings are unexpected if the participant groups detect the infelicitous nature of the overt subject in TM contexts in the DO region 12 This also applies to the native speaker group for the TMO TSO tokens since slower RTs are expected for the infelicitous TMO tokens as compared to the felicitous TSO tokens. For the L2 speaker group, th ough, the positive result suggests that they have detected the infelicitousness of the TMO tokens as compared to the felicitousness of the TSO tokens. The independent samples t tests revealed no statistically significant differences with respect to the a verage differential RT for the TMO CFO counterbalance (t(30) = 0.207, p = 0.837). Nevertheless, one was found for the TMO TSO counterbalance (t(30) = 2.698, p =0.011). Thus, it appears that the L2 speaker group does not differ from the native speaker group for the TMO CFO comparison, but that they do for the TMO TSO comparison. Note positive, as expected. Finally, t he independent samples t test showed a significant difference between the two groups for the CFO TSO comparison: (t(30) = 3.528, p = 0.001). This indicates that the L2 speaker group average RT for the third region of interest is significantly slower in Contrastive Focus tokens than in Topic Shift tokens w hereas these same RTs are more similar in the native speaker group. 12 However, recall from Figure 5 31 that both the native speaker and L2 speaker performed as expected with respect to the TMO CFO distinction. This indicates that their detec tion of the infelicitous nature of the overt subject in TM contexts as compared to the felicitous nature of it in Contrastive Focus contexts occurs in the subject region.

PAGE 212

212 Finally, Figure 5 34 shows the average differential RTs to the null subject tokens for the DO. The average differential RT between the CFN and TMN tokens was negative for the native speak er group ( 0.22) but positive for the L2 speaker group (131.61ms). For the TSN TMN comparison, the same pattern is evidenced: the native speaker group average differential RT between the TSN and TMN tokens was negative ( 11.20ms) but positive for the L2 sp eaker group (19.75ms). Regarding the native speaker group, if they detect the infelicitous nature of null subjects in Contrastive Focus contexts as compared to their felicitous nature in Topic Maintenance contexts, these findings are unexpected since we wo uld expect a slower RT for the CFN tokens than for the TMN tokens. The same is true of TSN tokens as compared to the TMN tokens. Only in the case of the L2 speaker group does this expectation obtain. Finally, both participant groups showed positive average differential RTs for the CFN TSN comparison. Figure 5 34. Online CMFT comparative group results for third region (DO). NS=native speaker; L2=second language speaker; CFN TMN=group average RT to CFN tokens minus group average RT to TMN tokens; TSN TMN= group average RT to TSN tokens minus group average RT to TMN tokens; CFN TSN=group average RT to CFN tokens minus group average RT to TMN tokens

PAGE 213

213 The independent samples t tests revealed a statistically significant difference with respect to the average dif ferential RT for the CFN TMN counterbalance (t(30) = 3.937, p= 0.000) while no statistically significant difference was found for the TSN TMN counterbalance ((t(30) = 0.832, p =0.412) or for the CFN TSN comparison (t(30) = 1.148, p = 0.111). These finding s indicate that the distinction the L2 speaker group made between the infelicitous CFN tokens and the felicitous TMN tokens differed from that of the native speaker group. It is important to note RT was positive, they demonstrated distinctions between the infelicitous TSN tokens and the felicitous TMN tokens as well as the CFN TSN comparison matched that of the native speaker group. To conclude, of the 15 comparisons described in Section 5.3.2.6, 12 showed no statistically significant difference between the native speaker and L2 speaker group (first region of interest: TMO CFO, TMO TSO and CFO TSO; second regio n of interest: TMO CFO, TMO TSO, CFO TSO, CFN TMN, TSN TMN and CFN TSO; third region of interest: TMO CFO, TSN TMN and CFN TSN). All three statistically significant differences were found in the third region of interest. For two of the statistically signif icant differences found, CFN TMN and TMO TSO, the L2 speaker group performed as expected in that their average differential RT was positive while that of the native speaker group was not. The final statistically significant difference was found for the CFO TSO comparison for which the L2 speaker group, but not the native speaker group, showed a high average differential RT. Of the 10 comparisons for which there are clear predictions 13 (first region of interest: TMO CFO and TMO TSO; second region of interest: TMO CFO, TMO TSO, CFN TMN and TSN TMN; third region of interest: TMO 13 It is not clear that any distinction is expected between comparisons CFO TSO or CFN TSN, as both Contrastive Focus and Topic Shift contexts prefer overt subjects over null subjects.

PAGE 214

214 CFO, TMO TSO, CFN TMN and TSN TMN), the native speakers demonstrated the predicted distinction in three comparisons (first region of interest: TMO CFO and TMO TSO; second region of inter est: TSN TMN). The L2 speaker group demonstrated the predicted distinction in seven of the 10 comparisons (first region of interest: TMO CFO and TMO TSO; second region of interest: TMO TSO and TSN TMN; third region of interest TMO TSO, CFN TMN and TSN TMN) Thus, both the native speaker and L2 speaker groups distinguish between overt and null subject pronouns in the three contexts examined herein, but differences are seen regarding how widespread the distinction is. The native speaker group, for example, di stinguishes TMO tokens from CFO and TSO tokens in the first region of interest only, showing an early and brief distinction between the use of an overt subject pronoun in Topic Maintenance contexts as compared to Contrastive Focus or Topic Shift contexts. The L2 speaker group also shows this same early distinction (significant differences were not found between the two participant groups in the first region of interest); yet, as they also show the predicted distinction between TMO TSO in the second and thir d region of interest, the distinction is not as pinpointed, temporally CFO distinction, no widespread distinction is seen as compared to the native speaker group. Similarly, a more widespread distinction between TSN and TMN tokens is seen in the L2 speaker group as they show the predicted distinction in the second and third region of interest, while the native speaker group does so in the second region of interest only. Interestingly, the L2 speaker group but not the native speaker group, distinguished between CFN and TMN tokens in the third region of interest only. In general, these findings point to native like, but more widespread in terms of regions of interest, distinctions on the part of the L2 spea ker group.

PAGE 215

215 5.3.2.7 Summary of online CMFT results Sections 5.3.2.2, 5.3.2.3, and 5.3.2.4 above examined the RT data to three regions of interest (the subject ella CMFT: CFO, CFN, TSO, TSN TMO and TMN. No statistically significant distinction obtained between the overt and null subjects in Contrastive Focus contexts for either participant group for the second and third regions of interest, an unexpected finding that may indicate that the t ask was tokens, neither group distinguished between overt subjects and null subjects in Topic Shift contexts in the second region of interest, but both gro ups did so in third region of interest. Equally, the comparisons made between the average differential RTs described in Section 5.3.2.3.3 revealed no significant difference between the groups for average differential RT for the TSN TSO counterbalance in ne ither the second nor the third region of interest. This finding indicates that the distinction made by the L2 speaker group between the infelicitous and felicitous Topic Shift tokens matched that of the native speaker group. Finally, for the third token ty pe, no statistically significant differences were found between overt subjects and null subjects for either group in neither the second nor the third region of interest, which, although unexpected since null pronouns are preferred in Topic Maintenance cont exts, is supported by variationist literature showing use of overt referential subject pronouns in Topic Maintenance contexts (e.g. Silva Corval n 1994; Cameron 1995; Lapidus Shin & Cairns 2012). The comparison examined in Section 5.3.2.4.3 demonstrated th at the distinction the L2 speaker group made between the infelicitous and felicitous Topic Maintenance tokens matched that of the native speaker group for the second and third regions of interest, although only in the third region was the expected finding evidenced. Finally, the comparisons made in Section 5.3.2.6 demonstrated that, while more widespread, the distinctions made by the L2 speaker group largely matched those of the

PAGE 216

216 native speaker group, although the L2 speaker group did show more temporally wi despread distinctions. Together, these findings provide some evidence in favor of native like processing with interface conditioned properties like the distribution of subjects in specific discourse contexts. 5.3.3 Summary of Online Experimental Results Se ction 5.3 described two online experimental tasks, the purpose of which was to he syntactic distribution of overt and null referential and expletive subject pronouns in Spanish. The statistical analyses described in Section 5.3.1.2.1 showed that neither the L2 speaker group nor the native speaker group demonstrated significantly diff erent RTs to overt subjects versus null subjects in referential subject tokens in any of the regions of interest. This result is expected since both overt and null referential subjects are grammatical in Spanish. With respect to the expletive tokens descri bed in Section 5.3.1.3.1, neither group demonstrated statistically significant RTs to the ungrammatical ESO tokens as compared to the grammatical ESN tokens in the first region of interest (but see Section 5.3.1.3.2 for explanation). And while the L2 speak er group did not demonstrate a statistically significant difference between these counterbalances in the second region as the native speaker group did, both groups made this distinction for the third region of interest. Since overt subjects in expletive to kens are ungrammatical, differences with respect to the RT to overt and null subjects are expected. Finally, Section 5.3.1.5 provided further evidence for the claim that the processing of the tokens examined in the online version of the GT is native like. That is, no statistically significant differences were found between the native speaker and the L2 speaker group with respect to the average differential RT between the

PAGE 217

217 ESO versus ESN and the ESO versus RSO. This establishes that the distinction they made between the ungrammatical and grammatical tokens is native like. discourse distribution of overt and null subjects in Spanish in Contrastive Focus, Topic Shift and Topic Maint enance contexts. As regards the Contrastive Focus tokens, the statistical analyses described in Section 5.3.2.2.1 showed that no statistically significant difference was found r, neither group was shown to make the appropriate distinction between overt and null subject referential subject pronouns in the second region of interest and only the L2 speaker group did so in the third region nability to detect the participants sensitivity to the infelicitous nature of the null subject in this context. Nevertheless, evidence pointing to native like processing between the two participant groups was found for the Topic Shift tokens. Recall that n either group showed a statistically significant difference between their treatment of overt and null subjects in Topic Shift contexts for the second region of interest, but that both groups did so in the third region of interest. This latter finding sugges ts that the infelicitousness of the null subject is detected by both groups in this context in the same region. Additionally, the comparisons made between the average differential RTs described in Section 5.3.2.3.3 revealed no significant difference betwee n the groups for the TSN TSO counterbalance in the third region of interest where the distinction between overt and null subjects in Topic Shift contexts was made. This finding indicates that the distinction made by the L2 speaker group between the infelic itous and felicitous Topic Shift tokens matched that of the native speaker group. Next, neither group showed statistically significant differences with respect to their treatment of overt and null subjects in Topic Maintenance contexts in neither the secon d nor third region of interest,

PAGE 218

218 suggesting issues with the task or optionality between the overt and null referential subject average differential RT to the co unterbalanced tokens showed no statistically significant difference in the third region of interest, where both groups evidenced the expected distinction. Lastly, the comparisons made in Section 5.3.2.6 showed that distinctions made by the L2 speaker group final point is interesting in that the native speaker group was shown to demonstrate more widespread distinction in the online GJ, but it is the L2 learners that show t his in the online CMFT. In sum, evidence suggesting similarities between the two participant groups regarding the processing of the discourse distribution of overt and null referential subjects in Spanish was found in the online version of the CMFT. In s um, based on the results of the online GJ, the L2 speaker group largely demonstrated native like processing of the syntactic distribution of overt and null subjects in Spanish. The results regarding the discourse distribution are less straightforward in th at the distinctions made by the L2 speaker group described in Section 5.3.2.6 seem to be more widespread in terms of regions of interest than those of the native speaker group. Additionally, as the expected distinction was not made by the native speaker gr oup or the L2 speaker group for the Contrastive Focus or Topic Maintenance tokens, the task may not have been sensitive enough to capture distinctions made by the participants. Recall that the IH predicts differences between native speakers and near native speakers for the processing of interface conditioned properties like the discourse distribution of referential subjects. Some support of this prediction was found in the widespread distinctions made between token types. Chapter 6 brings together the offline an d online results and examines them in light of the predictions of the IH.

PAGE 219

219 CHAPTER 6 CONCLUSION 6.1 General Introduction This final chapter offers general conclusions concerning the data presented in Chapter 5, especially in light of the original research q uestions outlined in Chapter 1. I will argue that certain data provide evidence against the predictions made by the most recent version of the Interface Hypothesis (IH), although the results are not definitive and some results were predicted by the IH Fro m there, I will discuss the limitations of the study, some of which are methodological in nature Then, the general implications of the type of research presented in this study will be described with respect to the IH and general L2 acquisition theory. Fin ally, the chapter closes with a brief conclusion. 6.2 The Research Questions Chapter 1 briefly described the basics of normal child L1 acquisition, highlighting its ultimate attainment and contrasting this with a very distinct scenario witnessed in adult second language (L2) acquisition: many adult L2 learners do not achieve native like knowledge of their L2. However, native like knowledge is attainable and much research has shown that this population converges on a number of properties for which domain ge neral learning strategies nor frequency can account for (White 2003a, 2011a). These differences and the variable level of attainment reached by adult L2 learners constitute the research problem addressed in this study. A recent proposal that attempts to re concile some of the noted disparity, the IH, was described. The IH claims that external interface conditioned properties are more susceptible to divergence due to the presence of two linguistic systems (the first and the second language) in one mind and, c onsequently, the division of finite cognitive resources that managing these two systems necessitates (Wilson et al. 2009; Sorace 2011). Based on the claims of the latest version of the IH

PAGE 220

220 (Sorace 2011), three research questions (1) (3) were outlined in Sec tion 1.5 and examined in this dissertation. They are repeated below as (4) (6): 4. Do near native L2 learners demonstrate more divergence with respect to external interface conditioned properties as opposed to purely syntactic ones in offline tasks? With res and discourse distribution of subjects in Spanish was tested in two offline experimental tasks (see Section 4.5.1), the Grammaticality Judgment/Correction Task (GJCT) and the Context Matching Felicitousness Task (CMFT). Regarding the GJCT, two important findings point to native like convergence on the part of the L2 speaker group. First, no main effect was found for group and, second, both participant groups distinguished betw een ungrammatical overt subjects in ESO tokens as compared to grammatical overt subjects in RSO tokens. Thus, it seems that the L2 speaker group has converged upon the syntactic distribution of subjects in Spanish, including the null expletive subject whic grammar. It is worth noting that insofar as Spanish and Farsi differ syntactically in the domain of null subjects, this evidence demonstrates clear acquisition by the L2 speaker group. Si milarly, on the CMFT, no significant difference was found between the native speaker and L2 speaker groups with respect to Contrastive Focus with overt subject (CFO), Contrastive Focus with null subject (CFN) or Topic Shift with overt subject (TSO) tokens. Significant differences were found between the two groups, however, for Topic Shift with null subject (TSN), Topic Maintenance with overt subject (TMO) and Topic Maintenance with null subject (TMN) tokens. Together, these findings echo previous research t hat found divergence on Topic Shift (Sorace and colleagues) but not Contrastive Focus (Rothman 2009) and show that the L2 speaker group is rather accepting of overt referential subject pronouns in Topic Maintenance contexts despite a preference for null su bject pronouns in L1 Farsi and Spanish. In fact, and

PAGE 221

221 differently from the native speaker group, no significant difference was found between the CFO or TSO tokens versus TMO tokens. Despite the seemingly mixed results with respect to the CMFT, another three way comparison demonstrated that both participant groups made a distinction between overt versus null subjects in each context type. This finding confirms the text tested. Considering these results as a whole, it appears that the L2 speaker group converged on the syntactic distribution of subjects in that native like distinctions were made between the counterbalanced tokens. The L2 participant group accepted ov ert and null referential subject pronouns and null expletive subjects while rejecting overt expletive subjects in the GJCT in spite differences between their native language and the target language. Still, the L2 speaker group showed more mixed results wit h respect to the discourse distribution of subjects since convergence was found for Contrastive Focus (as in Rothman 2009), but differences were found for Topic Shift (as found by Sorace and colleagues) and Topic Maintenance (where the L2 learners rated bo between the use of overt subjects in CFO versus TMO and TSO versus TMO contexts, rating ential subject pronouns ( Bini 1993; Margaza & Bel 2006; Sorace & Filiaci 2006; Belletti et al. 2007; Sorace et al. 2009; Garca Alcaraz & Bel 2011). Without further investigation, one cannot be sure if these differences are a product of dialect (11 of 24 n ative speaker participants rated TMO or a product of the task itself. Importantly, the L2 speakers rated more overt subjects than null subjects ive Focus and Topic Shift contexts, and rated more null subjects

PAGE 222

222 corroborate data found in previous offline studies examining the distribution and interpretation of pron ominal subjects in Romance languages (Lozano 2002; Sorace & Filiaci 2006; Rothman 2008a, 2009; Prada P rez 2009, 2010; S orace et al. 2009, and those listed above). 5. Do near native L2 learners exhibit greater processing difficulties with external interface conditioned properties as compared to purely syntactic ones in online tasks? discourse distribution of subjects in Spanish was tested in two online experimental tasks (s ee Section 4.5.2). For the online Grammaticality Task (GT), no statistically significant differences compared to null subjects in referential tokens in either the f irst, second or third region of interest. In accord with the syntactic literature (Rizzi 1982; Jaeggli & Hyams 1988), no evidence (i.e. slower comparative RTs) was found suggesting that either of the participant groups perceive these tokens as ungrammatica l. With respect to the expletive tokens, both participant groups distinguished between ungrammatical overt and grammatical null subjects in the third region of interest. The native speaker group, but not the L2 speaker group, did so as well in the second r egion of interest. This suggests that the two participant groups detect the ungrammaticality of the ESO tokens, the only difference being that the native speaker group also does so in the second region showing a more widespread distinction between the over t and null subjects. This finding may be interpreted as evidence in favor of the IH as the detection of ungrammaticality is evidenced at the earliest possible point, the verb. This indicates that the native speaker group is faster to process the ungrammati average differential RT between the counterbalanced tokens was calculated and examined

PAGE 223

223 statistically. the ESO versus ESN and the ESO versus RSO tokens showed no statistically significant differences, providing evidence that the L2 speaker group made a native like distinction between the ungrammatical ESO tokens and the grammatical ESN and RSO tokens. Regarding the online CMFT, the results are not as clear cut; neither participant group showed a significant difference between overt and null subjects in either region of interes t for the Contrastive Focus tokens. As explained in Section 5.3.3, this likely indicates that the task was type. With respect to the Topic Shift tokens, no st atistically significant difference between overt and null subjects in Topic Shift contexts was made by either the native speaker or the L2 speaker group in the second region of interest, but both groups exhibited a statistically significant difference betw een the two subjects in the third region of interest. Additionally, the comparison significant difference for the second or third regions of interest. Finally, the same result seen for the Contrastive Focus tokens obtained for the Topic Maintenance tokens in that neither group displayed a statistically significant difference between overt and null subjects in either region of interest. As stated above in reference t o the Contrastive Focus tokens, this may be a product of the task. Finally, the comparisons made across token types described in Section 5.3.2.6 showed a more widespread distinction on the part of the L2 speaker group, yet the distinctions largely matched those of the native speaker groups. To conclude, then, the online syntactic data demonstrates more evidence of native like processing in that neither group distinguished between the RSO and RSN tokens as they are both grammatical, but both groups distingui shed between the grammatical ESN tokens and the

PAGE 224

224 ungrammatical ESO tokens in the third region of interest. Additionally, the comparison of each showed no statistically sensitivity to the Contrastive Focus and Topic Maintenance tokens, less convergence is seen with respect to the discourse data since the L2 speaker group showed more temporally w idespread detection of differences in the comparison made across context type (see Section 5.3.2.6). This less pinpointed detection differs from that seen in the online GT in that the native speakers showed earlier detection, while the L2 learners show lon ger sustained detection. Whether this is a product of processing differences between the native speaker group and the L2 participant group or the task cannot be determined from this data alone. 6. Do near native L2 learners perform more native like in offlin e tasks as compared to online tasks? Finally, and bearing in mind the data described in answering the two previous research questions, some differences between the two task modalities were found. With respect to the GJCT and the CMFT, both groups disting uished between the counterbalanced tokens, yet only the native speaker group rated TMO tokens significantly differently as compared to CFO and TSO tokens. Regarding the online tasks, one difference obtained with respect to the regions in which the detectio n of the infelicitous nature of the ESN tokens was found in the online GJ: the native speaker group demonstrated evidence of detection in both the second and third regions of interest, while the L2 speaker group showed evidence of detection in the third re gion of interest only. Furthermore, the L2 speaker group showed more widespread distinction between context type in the comparisons made in Section 5.3.2.6. While some differences were found in offline and online tasks, it seems that more differences were found in the online tasks, suggesting that

PAGE 225

225 experimental methodologies may affect the results and also that online methodologies may be more sensitive measures. In summary, the three research questions that were formed with the specific intention of testin g the claims of the IH (Sorace 2011) seem to be answered in the positive when the data provided by the two offline and two online experimental tasks in Chapter 5 are considered. The near native L2 speakers of Spanish demonstrated more divergence with respe ct to the discourse conditioned properties in the offline CMFT than the syntactic properties in the GJCT. Similarly, the data from these same learners provided evidence suggesting more processing differences with the discourse distribution of subjects in t he online CMFT as compared to the syntactic distribution in the GT, especially with respect to the widespread nature of some of the distinctions the L2 speaker group made. This is not to say, however, that no evidence of native like performance was found, as certain results from each task do show similarities. Finally, task modality was shown to have an effect on the findings in that more differences were found in the online tasks. While these findings are valuable, this study is not free of shortcomings. T he following section discusses some of the limitations of this study. 6.3 Limitations of the Study To my knowledge, the present study is the first to examine L2 Spanish in the context of Farsi. Notwithstanding, significant populations of Farsi Spanish bili nguals, especially those that are homogenous in terms of age of first exposure, exposure to the same dialect, maintenance of the L1and little to no formal instruction, do not exist. Recall, that it was necessary to test near native speakers as the predicti ons of the IH apply to this population. These factors have contributed to the relatively low number of L2 speaker participants. An advantage of testing this specific language pairing however, is that while they share an analogous discourse distribution of subjects, the underlying syntax differs. While this is not a necessary condition in order to test

PAGE 226

22 6 the IH, it is unique to this study and can provide new insights. Regardless of the language pairing, a larger L2 group is ideal in order to increase the powe r of the current study. Thus, potential future research includes identifying additional homogenous groups of near native L2 Spanish speakers that meet these or similar criteria (e.g. Hungarian Spanish speakers), speakers of language pairings that differ, a lbeit slightly, with respect to the distribution of referential subject pronouns (e.g. Italian Spanish) and speakers of language pairings that differ with respect to the Null Subject Parameter (e.g. English Spanish) It could be claimed that a further lim itation of the study is the examination of critical properties (i.e. the discourse distribution) for which the native and L2 of the bilingual participants do not differ. Recall from Chapter 3 that Spanish and Farsi are both syntax discourse configurational null subject languages that permit overt and null referential subjects (Rigau 1986, 1988; Lujn 1987, 1999; Fernndez Soriano 1989, 1993; Picallo 1994, 1998; Karimi 2005) and that the participants of Judy and Feizmohammadpour (2012) showed the same discou rse distribution of subjects in Farsi as shown in Rothman (2009) for Spanish. While this parallel behavior across Spanish and Farsi may at first seem to undermine the results described in Chapter 5 under the assumption that convergence on these properties in Spanish is possible through simple L1 transfer, it is in fact important for the goals of the study. Recall that the IH (Sorace 2011) alleges that interface vulnerability obtains as a byproduct of the managing of two grammars in one mind and, importantly not as a consequence of language specific differences (parametric, semantic, discourse or otherwise). With exceptions like Bini (1993), Lozano (2002, 2006), Belletti and Leonini (2004), Margaza and Bel (2006), Sorace et al. (2009), Serratrice et al. (200 9), Prada P rez (2009, 2010) and Garca Alcaraz & Bel (2011) many studies examining the IH (directly or indirectly) have done so in language pairings for which the two languages differ

PAGE 227

227 (for example Hertel 2003; Tsimpli et al. 2004; G rel 2006; Sorace & Fi liaci 2006; Valenzuela 2006; Belletti et al. 2007; Dom nguez 2007; Iverson et al. 2008; Rothman 2008a, 2009; Ivanov 2009; Donaldson 2011a, 2011b; Slabakova et al. 2012). Consequently, it is important to test the with respect to the properties tested, like those cited above. Sorace (2011) discusses some preliminary data showing that, despite similarities across Spanish and Italian, such bilinguals are not immune to the overuse of overt subjects. From these data, So race claims that it is precisely their status as bilinguals and its consequences on resource allocation that explains this finding. The present language pairing, thus, follows the same logic while innovating since Farsi and Spanish, unlike Spanish and Ital ian, are not typologically similar languages nor have they been studied in light of each other before. Nonetheless, a limitation of the present study is that the findings could have been bolstered by either examining an additional near native L2 Spanish sp eaking population whose L1 is a non null subject language, or by concurrently examining an additional linguistic property for which Farsi and Spanish differ. Doing so would allow us to tease apart the potential effect of the analogous nature of the discour se distribution of referential subject pronouns in Farsi in Spanish. First, if divergence were to be found for the same properties across the same tasks employed herein by non null subject language speakers (e.g. English), this would provide evidence that language pairing has an effect on convergence, which was also found in Lozano (2002), Serratrice et al. (2004) and Serratrice et al. (2009). Second, if divergence was found in the same population of native Farsi speaking, near native L2 Spanish speakers st udied herein for a property for which Farsi and Spanish differ, this result would also point to language pairing as a factor that affects convergence. To test this scenario, convergence on the distribution of Differential Object

PAGE 228

228 Marking (DOM) could be exam ined in the same population as DOM distribution is more restricted in Farsi than in Spanish. Finally, some methodological limitations will be discussed. First, Section 3.2.3 outlined the learning task of the native speakers of Farsi with respect to their convergence on the syntax of Spanish subjects. Native Farsi speakers must converge on the fact that Spanish does not allow all phrasal elements to remain in the vP and that Spanish is not a verb final language. Therefore, while nominative case on the subje ct can be checked and valued in Spec,vP in Farsi and the EPP is satisfied morphologically without movement, nominative case is checked and valued in Spec,TP and the EPP is satisfied via V to T movement in Spanish. Finally, Farsi speakers learning Spanish m ust acquire expletives according to Karimi (2005). A potential way of determining whether or not the L2 speaker group has converged on the syntax of Spanish subjects is by testing for knowledge of restrictions on adverbial placement (which demarcate the VP boundary). However, since these token types were not included in the methodology, no definitive claims can be made regarding their underlying syntax. I leave this to future research. Second, and with respect to the online CMFT, an unexpected result obtain ed for the Contrastive Focus and Topic Maintenance tokens in that neither the native speaker nor the L2 speaker group was found to distinguish to a significant degree between the use of overt or null subjects in these two contexts. However, they made the d istinction for Topic Shift tokens. Recall that a distinction between the two subject types was predicted if infelicitousness was detected. Therefore, the fact that the native speakers did not make the expected distinction could point to Maintenance tokens, or possibly the need to reevaluate what is to be considered the expected result. However, the offline task showed that both groups distinguished between CFO a nd CFN

PAGE 229

229 tokens, the TSO and TSN tokens and the TMO and TMN tokens, but that the L2 speaker group rated the TSN, the TMO and the TMN tokens as significantly better than did the native speaker group. This unexpectedly high rating of TSN tokens could result fr om the L2 speaker group using context and intuition in order to make meaning of the sentence, including the subject of the verb in question. While this is also possible with respect Contrastive Focus tokens, there seems to be less necessity for overt subje cts in Topic Shift contexts than in Contrastive Focus contexts (Silva Corvaln 1982, 1994; Enrquez 1984; Bentivolgio 1987; Bayley & Pease lvarez 1996, 1997; Otheguy et al. 2007; Prada Prez 2009; Lapidus & Shin 2012) Future research should not only atte mpt to remedy these limitations, but also explore the differences between these context types. Third, the online measure employed in this study was a self paced reading task (SPRT). A limitation of SPRTs is that, while this task gathers RT data for the regions of interest, this measurement is not as fine grained as other tasks used to examine processing, such as eye tracking or ERP methodology. Future research may expand on the present study by employing such methodologies and, importantly, by comparing the results found across methodologies to determine the effect of the modality of testing. Fourth, some syntactic differences between Farsi and Spanish were discussed in Section 3.2.1, one of those being that Farsi optionally allows in osition. Karimi (2005) claims that in is a demonstrative; thus, the Spanish equivalent, este as the ungrammatical overt expletive subject to more closely match the Farsi demonstrative However, a major drawback of this option i s that este is frequently used hesitation disfluency or hesitations affect language processing in spoken speech (Corley & Stewart 2008; Corley,

PAGE 230

230 MacGregor & Donald son 2007) and that they can also affect grammaticality ratings assigned to garden path sentences in offline tasks (Bailey & Ferreira 2003). To avoid these potential effects, ello which is not used as a hesitation was used instead. Finally, the proficienc y measure employed herein, the Diploma del Espaol como Lengua Extranjera (DELE 2002), is commonly used in Spanish L2 acquisition studies; however, this measure may not be able to capture fine grained differences in the proficiency levels of such highly fl uent L2 speakers as those examined herein. That is, while all L2 speaker participants tested at the near beyond what the DELE can measure. It is possible that a more sophistic ated proficiency test may reveal differences between the native and L2 speaker groups, and also between the L2 speaker participants themselves. If so, subtle differences may be found, based on the fine grained gence and processing of the properties tested herein, perhaps showing a gradual path of convergence. This hypothesis could be tested by employing more sophisticated proficiency measures and also by examining and comparing the L2 acquisition of speakers wit h varying numbers of years of exposure to the L2 (e.g. 5 years, 10 years, 15 years, etc.). 6.4 Implications for the Interface Hypothesis and SLA Theory In recent years, the IH has undergone two key clarifications (Sorace 2011, 2012). First, the original c laim that interface conditioned properties (both internal and external) were more susceptible to divergence in near native grammars than purely syntactic properties was modified to isolate only external interfaces after research showing convergence on synt ax semantic properties was provided (see work cited in Tsimpli & Sorace 2006; Slabakova 2006, 2008; White 2008; Rothman & Slabakova 2011 ao.). Second, this revised version was subsequently clarified in response to data demonstrating convergence on external interface conditioned

PAGE 231

231 properties examined via offline methodology, ultimately leading Sorace to claim that only properties dependent on the external interfaces are subject to vulnerability as regards their processing (i.e. this vulnerability should ideall y be tested using online methodology). As discussed in Section 2.3.6, the vast majority of the adult acquisition studies testing the claims of the IH combine a Germanic language (usually English) with a Romance language (usually French, Italian or Spanish) alternating as the first or second language of the participants. Very few studies combine a non Germanic, non Romance language as one of the languages (but see Tsimpli & Sorace 2006). Related to this point is the fact that many studies, as described in Se ction 6.3, examine convergence on a property for which the L1 and L2 differ. Thus, if divergence is found, this may very well be a question of grammatical restructuring, not of processing differences between native and non native speakers. Differences on i nterface conditioned properties could also be a product of different working memory or inhibitory control capacities. To avoid this conflation of variables, researchers can and should determine native like knowledge of the related syntax of the interface c onditioned property tested. Nonetheless, as Rothman and Slabakova (2011) encourage, a variety of language pairings must be examined while testing the predictions of the IH in order for the results to be generalizable. That is, if we are to take seriously t he claim that the presence of two grammars ultimately causes processing differences with external interface conditioned properties, evidence for this claim needs be provided from a variety of language pairings and linguistic properties. Specifically, an ar ray of linguistic properties for which the L1 and L2 do not differ, either on the surface or at the even more restricted set of predictions with respect to whe re divergence is expected.

PAGE 232

232 The current study contributes to this body of research in that the languages examined are both null subject languages with analogous discourse conditioned distributions of subject pronouns. Against the predictions of the IH, th e results showed convergence on certain aspects of the discourse distribution of subjects across offline (specifically with respect to the Contrastive Focus tokens, an external interface conditioned property) and online experimental tasks (specifically reg arding the Topic Shift tokens). If, as the IH claims, processing differences obtain expected on all discourse conditioned properties. This is not to minimize t he differences that were found, but rather to point out that across the board divergence was not found. Considering this result with previous research examining the predictions of the IH, further modification may be warranted. More precisely, these modific down and may specifically involve changes regarding the types of properties for which divergence is predicted, the particular methodological designs to be used in testing its claims and/or the language pair ings tested. For instance, it could the case that certain external interface conditioned properties perhaps Topic Shift as Tsimpli and Sorace (2006) and Sorace et al. (2009) seem to support, but not all are more vulnerable to divergence (see for example I verson et al. 2008; Hopp 2009; Ivanov 2009; Prada P rez 2009, 2010; Bohnacker 2010; Donaldson 2011a, 2011b; Slabakova et al. 2012) It could also be the case that the subtleties of processing differences between native speakers and near native speakers, as claimed by the IH, can only be captured via particular methodologies, such as eye tracking, thus explaining the convergence seen in the abovementioned studies that report convergence in offline tasks. Finally, language pairing may come to bear on converge nce on processing in that near native L2 learners whose L1 matches the L2 for the property tested may demonstrate more native like processing in the L2

PAGE 233

233 than those whose L1 differs from the L2 as discussed above. It would be particularly interesting to test the same group of near native speakers on one property for which the L1 and L2 match and on one property for which they differ, the distribution of referential subject pronouns and DOM being good candidates for this type of study. Of course, future resear ch testing various properties across various language pairings and using different online methodologies will guide these modifications and I certainly do not mean to suggest that the IH must be modified based on the findings of one study. 6.5 Conclusion This dissertation study examined L1 Farsi speakers who are near native L2 Spanish speakers regarding their knowledge and processing of two properties related to subjects in Spanish in light of the predictions of the IH (Sorace 2011). Namely, the distributi on of referential and expletive subjects as well as the discourse conditioned distribution of referential subject pronouns were tested across four experimental tasks. The offline GJCT tested for knowledge of the syntactic distribution of subjects and the C MFT tested for knowledge of the discourse conditioned distribution of subjects in Spanish. Evidence of native like knowledge was demonstrated on both offline tasks; however, as in previous work by Sorace and colleagues, the L2 speaker group differed from t he native speaker group with respect to their ratings of TSN, TMO and TMN tokens and they rated TMO tokens similarly to CFO and TSO tokens showing an overuse of overt subjects in Topic Maintenance contexts (in line with Bini 1993; Margaza & Bel 2006; Sorac e & Filiaci 2006; Belletti et al. 2007; Sorace et al. 2009; Garca Alcaraz & Bel 2011 ). Since the same participants did not differ with respect to their ratings of the CFO or CFN tokens (as in Rothman 2009), these findings may point to differential levels of difficulty in external interface conditioned properties (at least when examined in offline tasks) and call for reassessment of the properties on which near native L2 speakers are expected to demonstrate

PAGE 234

234 processing differences. Finally, a complementary o nline version of each offline experimental task was employed via a SPRT in E prime. While the L2 speaker group demonstrated slower overall RTs in all regions of interest across both tasks, some evidence showing native like processing of these linguistic pr operties was found. Importantly, no differences between the native speaker and L2 speaker group were found for the average differential RT to the counterbalanced Topic Shift tokens in either the second or third region of interest, demonstrating that even w ith a potentially more sensitive measure (i.e. online measure, see Section 6.2), the L2 speaker group detected the infelicity of the TSN tokens.

PAGE 235

235 APPENDIX A OFFLINE GRAMMATICALITY JUDGMENT/CORRECTION TASK A.1 Referential Subject Tokens A.1.1 Referential S ubject Overt (RSO) En mi opinin l toma muchas clases en esta academia. En mi opinin ella come muchos dulces durante el verano. En mi opinin l hace muchos comentarios durante el taller. Segn sus hermanos ella toma muchas pastillas por la noche. Segn sus enfermeras l come muchos vegetales por la tarde. Segn su maestra ella hace muchas crticas en esta asignatura. A.1.2 Referential Subject Null (RSN) En mi opinin siempre toma muchas vitaminas por la maana. En mi opinin siempre come mu chos postres durante el invierno. En mi opinin siempre hace muchas consultas en este negocio. Segn sus docentes siempre toma muchos cursos en este instituto. Segn sus hermanos siempre come muchas golosinas por la noche. Segn los alumnos siempre hace muchas preguntas durante la clase. A.2 Expletive Subject Tokens A.2.1 Expletive Subject Overt (ESO) Mi abuela dice que ello llueve mucho en la primavera. La locutora dice que ello nieva mucho en las montaas. El meteorlogo dice que ello llovizna mucho en el sur. La doctora dice que ello llueve mucho en el verano. Mi hermano dice que ello nieva mucho en ese estado. Mi pap dice que ello llovizna mucho en esa ciudad. A.2.2 Expletive Subject Null (ESN) Los vecinos dicen que siempre llueve mucho en esa re gin. Mis primas dicen que siempre nieva mucho en esa provincia. Mis abuelos dicen que siempre llovizna mucho en el otoo. Mis amigas dicen que siempre llueve mucho en la costa. Las profesoras dicen que siempre nieva mucho en las sierras. Los expertos di cen que siempre llovizna mucho en el invierno. A.3 Differential Object Marking (DOM) Tokens A.3.1 +specific, +animate, +indicative with DOM El docente busca a un chico que hace bien el trabajo.

PAGE 236

236 La chica prefiere a un chico que toca bien la guitarra. El director desea a un chico que canta bien la msica. La vecina busca a un chico que hace bien el laburo. El cantante desea a un chico que toca bien el bajo. El hombre prefiere a un chico que canta bien la letra. A.3.2 +specific, +animate, +indicative w ithout DOM La maestra busca un chico que hace bien la tarea. El compositor desea un chico que toca bien el acorde. El msico busca un chico que canta bien la estrofa. La jefa prefiere un chico que hace bien la investigacin. La clienta prefiere un chico que toca bien el piano. El guitarrista desea un chico que canta bien el coro. A.3.3 +specific, animate, +indicative with DOM La abogada busca a un libro que describe bien la legislacin. El intelectual desea a un libro que explica bien la teora. El e conomista busca a un libro que define bien el mercado. El viajero desea a un libro que describe bien la regin. El jugador prefiere a un libro que explica bien el objetivo. El instructor prefiere a un libro que define bien la tcnica. A.3.4 +specific, animate, +indicative without DOM El filsofo busca un libro que describe bien la cuestin. El ingeniero desea un libro que explica bien la industria. El tcnico prefiere un libro que define bien la estrategia. La chica prefiere un libro que describe bie n la materia. La mujer busca un libro que explica bien la historia. El director desea un libro que define bien el problema. A.3.5 specific, +animate, +subjunctive with DOM La maestra busca a un chico que haga bien la tarea. El compositor desea a un ch ico que toque bien el clarinete. La adolescente prefiere a un chico que cante bien la cancin. El hombre busca a un chico que haga bien el trabajo. El concertista prefiere a un chico que toque bien el oboe. El guitarrista desea a un chico que cante bi en el tema. A.3.6 specific, +animate, +subjunctive without DOM El docente busca un chico que haga bien el ensayo. La chica prefiere un chico que toque bien la batera. El director desea un chico que cante bien el himno.

PAGE 237

237 La abuela busca un chico que ha ga bien el mandado. El baterista desea un chico que toque bien la corneta. El maestro prefiere un chico que cante bien la balada. A.3.7 specific, animate, +subjunctive with DOM El profesor busca a un libro que describa bien el punto. La estudiante pref iere a un libro que explique bien la idea. El cientfico desea a un libro que defina bien la hiptesis. La mujer busca a un libro que describa bien la tradicin. El presidente prefiere a un libro que defina bien el asunto. El bajista desea a un libro que explique bien el instrumento. A.3.8 specific, animate, +subjunctive without DOM El filsofo busca un libro que describa bien el argumento. El chico prefiere un libro que explique bien la leyenda. El investigador desea un libro que defina bien la frmu la. El gerente busca un libro que describa bien el negocio. El empleado prefiere un libro que explique bien el sistema. El detective desea un libro que defina bien el mtodo. A.3.9 CLLD with +definite DO with DOM Los colegas vieron el tubo, pero a la bi loga no la vieron ayer en el laboratorio. Los nios vieron el globo, pero a la payasa no la vieron ayer en el carnaval. Los espectadores vieron el tut, pero a la bailarina no la vieron ayer en el teatro. Los pasajeros vieron el uniforme, pero a la azafa ta no la vieron ayer en el avin. Los muchachos vieron el vaso, pero a la florista no la vieron ayer en la florera. Los coleccionistas vieron el pincel, pero a la pintora no la vieron ayer en el estudio. A.3.10 CLLD with +definite DO without DOM Los pra cticantes vieron el rosario, pero la monja no la vieron ayer en la iglesia. Los admiradores vieron el lpiz, pero la autora no la vieron ayer en la librera. Los clientes vieron el espejo, pero la peluquera no la vieron ayer en la peluquera. Los disead ores vieron el vestido, pero la modelo no la vieron ayer en el probador. Los banqueros vieron el recibo, pero la cajera no la vieron ayer en el pasillo. Los actores vieron el guin, pero la actriz no la vieron ayer en el escenario. A.3.11 CLLD with +ind efinite DO with DOM Los colegas vieron un tubo, pero a una biloga no la vieron ayer en el laboratorio. Los nios vieron un globo, pero a una payasa no la vieron ayer en el carnaval. Los espectadores vieron un tut, pero a una bailarina no la vieron ayer en el teatro. Los pasajeros vieron un uniforme, pero a una azafata no la vieron ayer en el avin. Los muchachos vieron un vaso, pero a una florista no la vieron ayer en la florera.

PAGE 238

238 Los coleccionistas vieron un pincel, pero a una pintora no la vieron ayer en el estudio. A.3.12 CLLD with +indefinite DO without DOM Los practicantes vieron un rosario, pero una monja no la vieron ayer en la iglesia. Los admiradores vieron un lpiz, pero una autora no la vieron ayer en la librera. Los clientes vieron un esp ejo, pero una peluquera no la vieron ayer en la peluquera. Los diseadores vieron un vestido, pero una modelo no la vieron ayer en el probador. Los banqueros vieron un recibo, pero una cajera no la vieron ayer en el pasillo. Los actores vieron un guin, pero una actriz no la vieron ayer en el escenario. A.4 Agreement Filler Tokens A.4.1 Subject Verb Agreement Fillers Las princesas ricas compr muchas decoraciones de la mueblera. Las actrices maleducadas tom muchas cervezas en la celebracin. Las sec retarias perezosas ley muchas revistas en la oficina. Las mams embarazadas comi muchos chocolates por la tarde. Las artistas jvenes pint muchos cuadros para la exhibicin. Los financieros listos vendieron muchas propiedades el ao pasado. A.4.2 N oun Adjective Agreement Fillers Los polticos corrupto aceptaron muchos sobornos la semana pasada. Los diplomticos refinado leyeron muchos documentos para la reunin. Los bomberos nuevo recibieron muchas llamadas durante la noche. Los soldados fuerte lev antaron muchas pesas durante el entrenamiento.

PAGE 239

239 APPENDIX B OFFLINE CONTEXT MATCHING FELICITOUSNESS TASK B.1 Contrastive Focus with Overt Subjects (CFO) Cuando salimos a cenar, mi novia prefiere comer platos livianos pero yo prefiero comer algo sustancioso As que ella come ensaladas y yo como milanesas en los restaurantes. Cuando cenamos en casa, mi esposa prefiere preparar el postre, pero yo prefiero cocinar el plato principal. As que ella hace alfajores y yo hago milanesas para la cena. Cuand o preparamos la cena, mi hermana prefiere preparar los entremeses, pero yo prefiero preparar la carne. As que ella prepara empanadas y yo preparo milanesas para la cena. Cuando trabajamos en la librera, mi colega prefiere estar en la seccin de cu ltura actual, pero yo prefiero la de libros. As que ella vende revistas y yo vendo novelas en la librera. Cuando escribimos sobre la historia del pas, mi compaera prefiere contar relatos cortos, pero yo prefiero contar relatos largos. As que ell a escribe cuentos y yo escribo novelas para la editorial. Cuando estamos de vacaciones, mi prima prefiere leer algo corto, pero yo prefiero leer algo muy denso y largo. As que ella lee poemas y yo leo novelas durante las vacaciones. B.2 Contrastive Focus with Null Subjects (CFN) Cuando merendamos por la tarde, mi compaera prefiere tomar algo fro, pero yo prefiero tomar algo caliente. As que toma agua y tomo mate con la merienda.

PAGE 240

240 Cuando hablamos con los invitados, mi esposa prefiere su propia be bida, pero yo prefiero compartir una bebida con ellos. As que ofrece caf y ofrezco mate despus de la cena. Cuando hacemos picnic con nuestros amigos, mi novia prefiere llevar algo dulce, pero yo prefiero llevar una bebida. As que lleva postres y llevo mate a los picnics. Cuando vamos a exposiciones de arte, mi colega prefiere ver artes plsticas, pero yo prefiero ver cuadros. Al final compra esculturas y compro retratos en las exposiciones. Cuando vamos a la clase de pintura, mi sobrina prefiere pintar la naturaleza, pero yo prefiero retratar a personas. As que pinta paisajes y pinto retratos en la clase. Cuando vamos al estudio de arte, mi amiga prefiere dibujar otras cosas, pero yo prefiero dibujar a personas. As que dibuja figuras y dibujo retratos en el estudio. B.3 Topic Shift with Overt Subjects (TSO) Mi vecina es muy quisquillosa en cuanto a la comida y solo come ensaladas. Yo le preparo muchos platos ricos y se los llevo, pero no importa lo que haga yo. Finalmente ell a come ensaladas y no prueba otras comidas nunca. Mi suegra es buena cocinera, pero no acepta los gustos de los dems. Yo no como dulces ya que estoy a dieta y prefiero comer comidas sanas. No obstante ella hace alfajores y viene a mi casa para compa rtirlos. Mi nuera sabe cocinar muchas cosas ya que estudia la gastronoma. Yo no quiero comer ms platos argentinos sino que prefiero probar platos nuevos y se lo digo. Sin embargo ella prepara empanadas e ignora mis pedidos gastronmicos.

PAGE 241

241 Mi hermana tena un puesto de golosinas pero no ganaba mucho dinero. Yo s que sera mejor vender varios productos y suger revistas y peridicos. Hoy en da ella vende revistas y gana ms plata que antes. Mi hija quiere ser autora y no tiene otros intereses. Y o creo que es mejor tener varios intereses y sugiero otras actividades, pero no importa lo que diga yo. Finalmente ella escribe cuentos y pasa todo el da en su cuarto. Mi compaera de clase nunca estudia para los exmenes ya que prefiere relajarse. Yo soy una persona muy nerviosa as que siempre estudio y repaso toda la materia. Sin embargo ella lee poemas y deja el estudio para otro da. B.4 Topic Shift with Null Subjects (TSN) Mi amiga es adicta a la cafena y por eso no duerme bien por la noche. Yo no tengo ese problema ya que no tomo caf y evito totalmente la cafena. No obstante toma caf y est despierta toda la noche. Mi esposa est obsesionada con la salud y cambi totalmente el desayuno. Yo no entiendo por qu tantos cambios y solo q uiero comer medialunas con chocolate. Sin embargo ofrece agua y sirve fruta para el desayuno. Mi abuela es muy golosa y siempre hace postres para las reuniones familiares. Yo soy diabtico y no puedo comer los dulces. Igual, no importa lo que diga y o. Finalmente lleva postres e ignora mis restricciones dietticas. Mi vecina es coleccionista de arte y es fantica de las artes plsticas. Yo soy vendedor de arte y por eso s que es ms barato colectar pinturas y se lo digo. No obstante compra esc ulturas y gasta mucho dinero en sus compras. Mi nuera se dedica al arte y puede pintar de todo. Yo creo que sera ms lucrativo pintar retratos y sugiero eso mismo. Igual, no importa lo que diga yo. Finalmente pinta paisajes y vende sus obras por poco dinero.

PAGE 242

242 Mi hija no estudia bien en casa ya que se distrae fcilmente. Yo s que es difcil y por eso intento eliminar distracciones. Igual, no importa lo que haga yo. Finalmente dibuja figuras y mira los rboles por la ventana. B.5 Topic Maintenance wi th Overt Subjects (TMO) El ao pasado mi mam estaba muy estresada y coma de todo. Ahora quiere bajar de peso para sentirse mejor. As que ella come ensaladas y ella hace ejercicio por la maana. El semestre pasado, mi amiga hizo un curso de cocina y aprendi muchas cosas. Ahora quiere mostrar sus talentos. As que ella hace alfajores y ella reparte todo entre sus amigos. Ayer mi vecina encontr su receta de empanadas favorita que haba perdido hace tiempo. No trabaja esta semana. As que ella prepara empanadas y ella comparte todo con el vecindario. Mi nieta quiere ganar plata para un viaje escolar. Hace unos meses se postul para un puesto en un kiosco y ya empez a trabajar. As que ella vende revistas y ella ahorra sus ingresos para el viaje. Mi profesora hizo un curso de escritura. Estudi varios gneros y aprendi mucho. Ahora quiere mostrar sus talentos. As que ella escribe cuentos y ella comparte sus historias con los estudiantes. Mi jefa est muy estresada con el trabajo. No quiere sentirse tan estresada y por eso busc formas de relajarse. As que ella lee poemas y ella escucha msica relajante en su oficina. B.6 Topic Maintenance with Null Subjects (TMN) ltimamente, mi abuela no se siente muy bien. Tiene mucha sed, tiene dolor de cabeza y se cansa fcilmente.

PAGE 243

243 As que toma agua y descansa un rato antes de salir. Mi suegra es muy detallista. Cuando tiene cenas en casa, no se olvida de nada y hace todo a la perfeccin. Por ejemplo ofrece caf y pone msica despu s de cenar. Mi cuada es muy sociable. Tiene muchos amigos y por eso va a muchas cenas a la canasta donde tiene que contribuir algo. As que lleva postres y comparte todo con sus amigos. Mi hermana acaba de comprarse una casa. Quiere decorar la ca sa pero dice que no quiere colgar muchos cuadros. As que compra esculturas y decora la casa de forma moderna. Mi ta es artista profesional y pasa mucho tiempo en el campo. Se inspira mucho all y suele retratar el entorno en sus obras. As que p inta paisajes y vende sus obras en la ciudad. Mi prima es artista y tambin participa en la poltica. Quiere juntar las dos cosas e incorporar sus ideas en sus obras. AS QUE DIBUJA FIGURAS Y ESCRIBE MENSAJES SOBRE LAS OBRAS.

PAGE 244

244 APPENDIX C ONLINE GRAMMA TICALITY TASK C.1 Referential Subject Tokens C.1.1 Referential Subject with Overt (RSO) En mi opinin ella toma muchas vitaminas por la maana. Toma muchas vitaminas por la noche? En mi opinin l come muchos postres durante el invierno. Come muchos po stres durante el invierno? En mi opinin l hace muchas consultas en este negocio. Hace pocas consultas en este negocio? Segn sus docentes ella toma muchos cursos en este instituto. Toma muchos cursos en este instituto? Segn sus hermanos ella come m uchas golosinas por la noche. Come muchas golosinas por la tarde? Segn los alumnos l hace muchas preguntas durante la clase. Hace muchas preguntas durante la clase? C.1.2 Referential Subject with Null (RSN) En mi opinin siempre toma muchas clases e n esta academia. Toma muchas clases en esta academia? En mi opinin siempre come muchos dulces durante el verano. Come muchas frutas durante el verano? En mi opinin siempre hace muchos comentarios durante el taller. Hace muchos comentarios durant e el taller? Segn sus hermanos siempre toma muchas pastillas por la noche. Toma muchas pastillas por la noche? Segn sus enfermeras siempre come muchos vegetales por la tarde. Come muchos vegetales por la tarde? Segn su maestra siempre hace much os crticos en esta asignatura. Hace pocos crticos en esta asignatura?

PAGE 245

245 C.2 Expletive Subject Tokens C.2.1 Expletive Subject with Overt (ESO) Los vecinos dicen que ello llueve mucho en esa regin. Dicen que llueve mucho en esta regin? Mis primas di cen que ello nieva mucho en esa provincia. Dicen que llueve mucho en esa provincia? Mis abuelos dicen que ello llovizna mucho en el otoo. Dicen que llovizna mucho en el otoo? Mis amigas dicen que ello llueve mucho en la costa. Dicen que nunca llueve en la costa? Las profesoras dicen que ello nieva mucho en las sierras. Dicen que nieva mucho en las sierras? Los expertos dicen que ello llovizna mucho en el invierno. Dicen que nunca llovizna en el invierno? C.2.2 Expletive Subject Tokens with Nul l Subjects (ESN) Mi abuela dice que siempre llueve mucho en la primavera. Dice que nieva mucho en la primavera? La locutora dice que siempre nieva mucho en las montaas. Dice que siempre nieva en las montaas? El meteorlogo dice que siempre llovizna mucho en el sur. Dice que nunca llueve en el sur? La doctora dice que siempre llueve mucho en el verano. Dice que siempre llueve en el verano? Mi hermano dice que siempre nieva mucho en ese estado. Dice que nieva mucho en ese pas? Mi pap dice que siempre llovizna mucho en esa ciudad. Dice que llovizna mucho en esa ciudad? C.3 Differential Object Marking (DOM) Tokens C.3.1 +specific, +animate, +indicative with DOM La maestra busca a un chico que hace bien la tarea.

PAGE 246

246 Busca a un chico que hace bien la tarea? El compositor desea a un chico que toca bien el acorde. Desea a un chico que toca bien el acorde? El msico busca a un chico que canta bien la estrofa. Busca a un hombre que canta bien la estrofa? La jefa prefiere a un chico que hace bien la investigacin. Prefiere a un chico que hace bien la investigacin? La clienta prefiere a un chico que toca bien el piano. Prefiere a un chico que toca bien la guitarra? El guitarrista desea a un chico que canta bien el coro. Desea a un chico que c anta bien la balada? C.3.2 +specific, +animate, +indicative without DOM El docente busca un chico que hace bien el trabajo. Busca a un hombre que hace bien el trabajo? La chica prefiere un chico que toca bien la guitarra. Prefiere a un chico que toc a bien la guitarra? El director desea un chico que canta bien la msica. Desea a un chico que canta bien la msica? La vecina busca un chico que hace bien el laburo. Busca a un chico que hace bien el laburo? El cantante desea un chico que toca bien el bajo. Desea a un chico que toca bien la batera? El hombre prefiere un chico que canta bien la letra. Prefiere a un chico que toca bien la letra? C.3.3 +specific, animate, +indicative with DOM El filsofo busca a un libro que describe bien la cues tin. Busca un libro que describe bien la cuestin? El ingeniero desea a un libro que explica bien la industria. Desea un libro que explica bien la industria? El tcnico prefiere a un libro que define bien la estrategia. Prefiere un libro que define b ien la estrategia?

PAGE 247

247 La chica prefiere a un libro que describe bien la materia. Prefiere un libro que describe bien la tarea? La mujer busca a un libro que explica bien la historia. Busca un video que explica bien la historia? El director desea a un libro que define bien el problema. Desea un video que define bien el problema? C.3.4 +specific, animate, +indicative without DOM La abogada busca un libro que describe bien la legislacin. Prefiere un libro que define bien la tcnica? El intelectual d esea un libro que explica bien la teora. Desea un libro que explica bien la teora? El economista busca un libro que define bien el mercado. Busca un libro que define bien el mercado? El viajero desea un libro que describe bien la regin. Desea u n libro que describe bien el juego? El jugador prefiere un libro que explica bien el objetivo. Prefiere un libro que define bien la estrategia? El instructor prefiere un libro que define bien la tcnica. Prefiere un panfleto que describe bien la legi slacin? C.3.5 specific, +animate, +subjunctive with DOM El docente busca a un chico que haga bien el ensayo. Busca un chico que haga bien el ensayo? La chica prefiere a un chico que toque bien la batera. Prefiere un chico que toque bien la bater a? El director desea a un chico que cante bien el himno. Desea un chico que cante bien el himno? La abuela busca a un chico que haga bien el mandado. Busca un hombre que haga bien el mandado? El baterista desea a un chico que toque bien la corneta. Desea un chico que toque bien el saxofn?

PAGE 248

248 El maestro prefiere a un chico que cante bien la balada. Prefiere un chico que toque bien la balada? C.3.6 specific, +animate, +subjunctive without DOM La maestra busca un chico que haga bien la tarea. Busca un chico que haga bien la tarea? El compositor desea un chico que toque bien el clarinete. Desea un chico que toque bien el clarinete? La adolescente prefiere un chico que cante bien la cancin. Prefiere un chico que cante bien la cancin? El homb re busca un chico que haga bien el trabajo. Busca una chica que haga bien el trabajo? El concertista prefiere un chico que toque bien el oboe. Prefiere un chico que toque bien la flauta? El guitarrista desea un chico que cante bien el tema. Desea un chico que cante bien el coro? C.3.7 specific, animate, +subjunctive with DOM El filsofo busca a un libro que describa bien el argumento. Busca un libro que describa bien el argumento? El chico prefiere a un libro que explique bien la leyenda. Pref iere un libro que explique bien la leyenda? El investigador desea a un libro que defina bien la frmula. Desea un libro que defina bien la frmula? El gerente busca a un libro que describa bien el negocio. Busca un libro que describa bien el asunto? El empleado prefiere a un libro que explique bien el sistema. Prefiere un panfleto que explique bien el sistema? El detective desea a un libro que defina bien el mtodo. Desea un libro que muestre bien el mtodo? C.3.8 specific, animate, +subjuncti ve without DOM El profesor busca un libro que describa bien el punto. Busca un libro que describa bien el punto?

PAGE 249

249 La estudiante prefiere un libro que explique bien la idea. Prefiere un libro que explique bien la idea? El cientfico desea un libro que defina bien la hiptesis. Desea un libro que defina bien la hiptesis? La mujer busca un libro que describa bien la tradicin. Busca un libro que describa bien el negocio? El presidente prefiere un libro que defina bien el asunto. Prefiere un libro que describa bien el asunto? El bajista desea un libro que explique bien el instrumento. Desea un video que explique bien el instrumento? C.3.9 CLLD with +definite DO with DOM Los practicantes vieron el rosario, pero a la monja no la vieron ayer en la i glesia. El rosario estuvo en la calle? Los admiradores vieron el lpiz, pero a la autora no la vieron ayer en la librera. El lpiz estuvo en la biblioteca? Los clientes vieron el espejo, pero a la peluquera no la vieron ayer en la peluquera. El es pejo estuvo en la silla? Los diseadores vieron el vestido, pero a la modelo no la vieron ayer en el probador. Vieron el vestido? Los banqueros vieron el recibo, pero a la cajera no la vieron ayer en el pasillo. Vieron el recibo? Los actores vieron e l guin, pero a la actriz no la vieron ayer en el escenario. Vieron el guin? C.3.10 CLLD with +definite DO without DOM Los colegas vieron el tubo, pero la biloga no la vieron ayer en el laboratorio. El tubo estuvo en el laboratorio? Los nios vieron el globo, pero la payasa no la vieron ayer en el carnaval. El globo estuvo en el carnaval? Los espectadores vieron el tut, pero la bailarina no la vieron ayer en el teatro. El tut estuvo en el teatro?

PAGE 250

250 Los pasajeros vieron el uniforme, pero la azafa ta no la vieron ayer en el avin. Vieron la bufanda? Los muchachos vieron el vaso, pero la florista no la vieron ayer en la florera. Vieron las tijeras? Los coleccionistas vieron el pincel, pero la pintora no la vieron ayer en el estudio. Vieron e l cuadro? C.3.11 CLLD with +indefinite DO with DOM Los practicantes vieron un rosario, pero a una monja no la vieron ayer en la iglesia. Vieron un tubo? Los admiradores vieron un lpiz, pero a una autora no la vieron ayer en la librera. Vieron un gl obo? Los clientes vieron un espejo, pero a una peluquera no la vieron ayer en la peluquera. Vieron un tut? Los diseadores vieron un vestido, pero a una modelo no la vieron ayer en el probador. Un uniforme estuvo en el aeropuerto? Los banqueros vier on un recibo, pero a una cajera no la vieron ayer en el pasillo. Un vaso estuvo en la cocina? Los actores vieron un guin, pero a una actriz no la vieron ayer en el escenario. Un pincel estuvo en el escenario? C.3.12 CLLD with +indefinite DO without D OM Los colegas vieron un tubo, pero una biloga no la vieron ayer en el laboratorio. Vieron un rosario? Los nios vieron un globo, pero una payasa no la vieron ayer en el carnaval. Vieron un lpiz? Los espectadores vieron un tut, pero una bailarina no la vieron ayer en el teatro. Vieron un espejo? Los pasajeros vieron un uniforme, pero una azafata no la vieron ayer en el avin. El vestido estuvo en la pasarela? Los muchachos vieron un vaso, pero una florista no la vieron ayer en la florera. E l recibo estuvo en la caja? Los coleccionistas vieron un pincel, pero una pintora no la vieron ayer en el estudio.

PAGE 251

251 El guin estuvo probador? C.4 Agreement Filler Tokens C.4.1 Subject Verb Agreement Fillers Las princesas ricas compr muchas decoraciones de la mueblera. Las princesas gastaron mucho dinero? Las actrices maleducadas tom muchas cervezas en la celebracin. Las actrices son descorteses? Las secretarias perezosas ley muchas revistas en la oficina. Las secretarias hacen mucho trabajo? Las mams embarazadas comi muchos chocolates por la tarde. Las mams estn a dieta? Las artistas jvenes pint muchos cuadros para la exhibicin. Las artistas son trabajadoras? Los financieros listos vendieron muchas propiedades el ao pasado. Los financieros compraron muchas propiedades? C.4.2 Noun AdjecgtiveAgreement Fillers Los polticos corrupto aceptaron muchos sobornos la semana pasada. Aceptaron muchos sobornos la semana pasada? Los diplomticos refinado leyeron muchos documentos para la reunin. Leyeron muchos libros para la reunin? Los bomberos nuevo recibieron muchas llamadas durante la noche. Recibieron muchas llamadas durante el da? Los soldados fuerte levantaron muchas pesas durante el entrenamiento. Levantaron muchas pesas durante el entrenamiento?

PAGE 252

252 APPENDIX D ONLINE CONTEXT MATCHING FELICITOUSNESS TASK D.1 Contrastive Focus with Overt Subjects (CFO) Cuando merendamos por la tarde, mi compaera prefiere tomar algo fro, pero yo prefiero tomar algo caliente. As que ella t oma agua y yo tomo mate con la merienda. Mi compaera toma t con la merienda? Cuando hablamos con los invitados, mi esposa prefiere su propia bebida, pero yo prefiero compartir una bebida con ellos. As que ella ofrece caf y yo ofrezco mate despu s de la cena. Mi esposa ofrece limonada despus de la cena? Cuando hacemos picnic con nuestros amigos, mi novia prefiere llevar algo dulce, pero yo prefiero llevar una bebida. As que ella lleva postres y yo llevo mate a los picnics. Mi novia lle va vino a los picnics? Cuando vamos a exposiciones de arte, mi colega prefiere ver artes plsticas, pero yo prefiero ver cuadros. Al final ella compra esculturas y yo compro retratos en las exposiciones. Mi colega prefiere artes plsticas? Cuand o vamos a la clase de pintura, mi sobrina prefiere pintar la naturaleza, pero yo prefiero retratar a personas. As que ella pinta paisajes y yo pinto retratos en la clase. Mi sobrina prefiere pintar la naturaleza?

PAGE 253

253 Cuando vamos al estudio de arte, mi amiga prefiere dibujar otras cosas, pero yo prefiero dibujar a personas. As que ella dibuja figuras y yo dibujo retratos en el estudio. Mi amiga prefiere dibujar otras cosas? D.2 Contrastive Focus with Null Subjects (CFN) Cuando salimos a cena r, mi novia prefiere comer platos livianos pero yo prefiero comer algo sustancioso. As que come ensaladas y como milanesas en los restaurantes. Mi novia come ensaladas? Cuando cenamos en casa, mi esposa prefiere preparar el postre, pero yo prefier o cocinar el plato principal. As que hace alfajores y hago milanesas para la cena. Mi esposa hace alfajores? Cuando preparamos la cena, mi hermana prefiere preparar los entremeses, pero yo prefiero preparar la carne. As que prepara empanadas y prepara milanesas para la cena. Mi hermana prepara empanadas? Cuando trabajamos en la librera, mi colega prefiere la de cultura actual, pero yo prefiero estar en la seccin de libros. As que vende revistas y vendo novelas en la librera. Mi c olega prefiere la seccin de ciencia ficcin? Cuando escribimos sobre la historia del pas, mi compaera prefiere contar relatos cortos, pero yo prefiero contar relatos largos. As que escribe cuentos y escribo novelas para la editorial. Mi compaer a prefiere contar leyendas?

PAGE 254

254 Cuando estamos de vacaciones, mi prima prefiere leer algo corto, pero yo prefiero leer algo muy denso y largo. As que lee poemas y leo novelas durante las vacaciones. Mi prima prefiere leer algo largo? D.3 Topic Shift w ith Overt Subjects (TSO) Mi amiga es adicta a la cafena y por eso no duerme bien por la noche. Yo no tengo ese problema ya que no tomo caf y evito totalmente la cafena. No obstante ella toma caf y est despierta toda la noche. Mi amiga duerme bi en por la noche? Mi esposa est obsesionada con la salud y cambi totalmente el desayuno. Yo no entiendo por qu tantos cambios y solo quiero comer medialunas con chocolate. Sin embargo ella ofrece agua y sirve fruta para el desayuno. Mi esposa e st obsesionada con la ropa? Mi abuela es muy golosa y siempre hace postres para las reuniones familiares. Yo soy diabtico y no puedo comer los dulces. Igual, no importa lo que diga yo. Finalmente ella lleva postres e ignora mis restricciones diettica s. Mi abuela hace postres para sus amigos? Mi vecina es coleccionista de arte y es fantica de las artes plsticas. Yo soy vendedor de arte y por eso s que es ms barato colectar pinturas y se lo digo. No obstante ella compra esculturas y gasta m ucho dinero en sus compras. Mi vecina gasta mucho dinero en sus compras? Mi nuera se dedica al arte y puede pintar de todo. Yo creo que sera ms lucrativo pintar retratos y sugiero eso mismo. Igual, no importa lo que diga yo. Finalmente ella pinta paisajes y vende sus obras por poco dinero. Mi nuera vende sus obras por poco dinero?

PAGE 255

255 Mi hija no estudia bien en casa ya que se distrae fcilmente. Yo s que es difcil y por eso intento eliminar distracciones. Igual, no importa lo que haga yo. Finalme nte ella dibuja figuras y mira los rboles por la ventana. Mi hija mira los rboles por la ventana? D.4 Topic Shift with Null Subjects (TSN) Mi vecina es muy quisquillosa en cuanto a la comida y solo come ensaladas. Yo le preparo muchos platos ricos y se los llevo, pero no importa lo que haga yo. Finalmente come ensaladas y no prueba otras comidas nunca. Mi vecina es muy quisquillosa? Mi suegra es buena cocinera, pero no acepta los gustos de los dems. Yo no como dulces ya que estoy a dieta y pr efiero comer comidas sanas. No obstante hace alfajores y viene a mi casa para compartirlos. Mi suegra es muy buena cocinera? Mi nuera sabe cocinar muchas cosas ya que estudia la gastronoma. Yo no quiero comer ms platos argentinos sino que prefie ro probar platos nuevos y se lo digo. Sin embargo prepara empanadas e ignora mis pedidos gastronmicos. Mi nuera estudia la gastronoma? Mi hermana tena un puesto de golosinas pero no ganaba mucho dinero. Yo s que sera mejor vender varios product os y suger revistas y peridicos. Hoy en da vende revistas y gana ms plata que antes. Mi hermana gana menos plata que antes? Mi hija quiere ser autora y no tiene otros intereses. Yo creo que es mejor tener varios intereses y sugiero otras activ idades, pero no importa lo que diga yo. Finalmente escribe cuentos y pasa todo el da en su cuarto. Mi hija pasa todo el da en la biblioteca?

PAGE 256

256 Mi compaera de clase nunca estudia para los exmenes ya que prefiere relajarse. Yo soy una persona muy nerv iosa as que siempre estudio y repaso toda la materia. Sin embargo lee poemas y deja el estudio para otro da. Mi compaera estudia cada da? D.5 Topic Maintenance with Overt Subjects (TMO) ltimamente, mi abuela no se siente muy bien. Tiene mucha se d, tiene dolor de cabeza y se cansa fcilmente. As que ella toma agua y ella descansa un rato antes de salir. Mi abuela tiene dolor de cabeza ltimamente? Mi suegra es muy detallista. Cuando tiene cenas en casa, no se olvida de nada y hace todo a la perfeccin. Por ejemplo ella ofrece caf y ella pone msica despus de cenar. Mi suegra es muy detallista? Mi cuada es muy sociable. Tiene muchos amigos y por eso va a muchas cenas a la canasta donde tiene que contribuir algo. As que ella l leva postres y ella comparte todo con sus amigos. Mi cuada tiene muchos amigos? Mi hermana acaba de comprarse una casa. Quiere decorar la casa pero dice que no quiere colgar muchos cuadros. As que ella compra esculturas y ella decora la casa de forma moderna. Mi hermana decora la casa de forma anticuada? Mi ta es artista profesional y pasa mucho tiempo en el campo. Se inspira mucho all y suele retratar el entorno en sus obras. As que ella pinta paisajes y ella vende sus obras en la ci udad. Mi ta vende sus obras en el campo?

PAGE 257

257 Mi prima es artista y tambin participa en la poltica. Quiere juntar las dos cosas e incorporar sus ideas en sus obras. As que ella dibuja figuras y ella escribe mensajes sobre las obras. Mi prima escr ibe mensajes en las paredes? D.6 Topic Maintenance with Null Subjects (TMN) El ao pasado mi mam estaba muy estresada y coma de todo. Ahora quiere bajar de peso para sentirse mejor. As que come ensaladas y hace ejercicio por la maana. Mi mam es taba enferma el ao pasado? El semestre pasado, mi amiga hizo un curso de cocina y aprendi muchas cosas. Ahora quiere mostrar sus talentos. As que hace alfajores y reparte todo entre sus amigos. Mi amiga hizo un curso de arte el semestre pasado? Ayer mi vecina encontr su receta favorita de empanadas que haba perdido hace tiempo. No trabaja esta semana. As que prepara empanadas y comparte todo con el vecindario. Mi vecina encontr una receta de pastelitos ayer? Mi nieta quiere ganar plat a para un viaje escolar. Hace unos meses se postul para un puesto en un kiosco y ya empez a trabajar. As que vende revistas y ahorra sus ingresos para el viaje. Mi nieta ahorra sus ingresos? Mi profesora hizo un curso de escritura. Estudi vari os gneros y aprendi mucho. Ahora quiere mostrar sus talentos. As que escribe cuentos y comparte sus historias con los estudiantes. Mi profesora comparte sus historias?

PAGE 258

258 Mi jefa est muy estresada con el trabajo. No quiere sentirse tan estresada y p or eso busc formas de relajarse. As que lee poemas y escucha msica relajante en su oficina. Mi jefa escucha msica relajante?

PAGE 259

259 APPENDIX E SAMPLE SCREEN SHOTS FROM THE ONLINE VERSION OF THE CONTEXT MATCHING FELICITOUSNESS TASK Figure E 1. Onl ine CMFT: Instruction page 1 screenshot Figure E 2. Online CMFT: Instruction page 2 screenshot

PAGE 260

260 Figure E 3. Online CMFT: Pre

PAGE 261

261

PAGE 262

262

PAGE 263

263

PAGE 264

264 Figure E 4. Online CMFT: Practice session screenshot

PAGE 265

265 Figure E 5. Online CMFT: Practice session comprehension question screenshot Figure E 6. Online CMFT: End of practice session screenshot

PAGE 266

266 Figure E 7. Online CMFT: Pre

PAGE 267

267

PAGE 268

268

PAGE 269

269

PAGE 270

270

PAGE 271

271 Figure E 8. Online CM FT: Sample Contrastive Focus token screenshot

PAGE 272

272 Figure E 9. Online CMFT: Contrastive Focus comprehension question screenshot Figure E 10. Online CMFT: Task completion and thanks screenshot

PAGE 273

273 APPENDIX F AVERAGE AND STANDARD DEVIATION TABLES

PAGE 274

274 F.1 Gram maticality Judgment/Correct Task Table F 4 Group average acceptance and standard deviation RSO RSN ESO ESN Group Ave S.D. Ave S.D. Ave S.D. Ave S.D. NS 6.00 0.00 5.50 1.29 0.00 0.00 6.00 0.00 L2 6.00 0.00 5.75 0.46 0 .13 0.35 5.88 0.35

PAGE 275

275 Table F 2. Individual average acceptance and standard deviation RSO RSN ESO ESN Participant Ave S.D. Ave S.D. Ave S.D. Ave S.D. Native 3 6 0.00 6 0.00 0 0.00 6 0.00 4 6 0.00 6 0.00 0 0.00 6 0.00 5 6 0.00 6 0.00 0 0 .00 6 0.00 6 6 0.00 6 0.00 0 0.00 6 0.00 7 6 0.00 4 0.52 0 0.00 6 0.00 8 6 0.00 5 0.41 0 0.00 6 0.00 9 6 0.00 5 0.41 0 0.00 6 0.00 10 6 0.00 6 0.00 0 0.00 6 0.00 11 6 0.00 6 0.00 0 0.00 6 0.00 12 6 0.00 6 0.00 0 0.00 6 0.00 13 6 0.00 6 0.00 0 0.00 6 0.00 14 6 0.00 5 0.41 0 0.00 6 0.00 15 6 0.00 6 0.00 0 0.00 6 0.00 16 6 0.00 6 0.00 0 0.00 6 0.00 17 6 0.00 6 0.00 0 0.00 6 0.00 18 6 0.00 6 0.00 0 0.00 6 0.00 19 6 0.00 0 0.00 0 0.00 6 0.00 20 6 0.00 6 0.00 0 0.00 6 0.00 21 6 0.00 6 0.00 0 0.00 6 0.00 22 6 0.00 6 0.00 0 0.00 6 0.00 23 6 0.00 6 0.00 0 0.00 6 0.00 24 6 0.00 6 0.00 0 0.00 6 0.00 25 6 0.00 5 0.41 0 0.00 6 0.00 26 6 0.00 6 0.00 0 0.00 6 0.00 L2 1 6 0.00 5 0.41 0 0.00 6 0.00 2 6 0.00 6 0.00 0 0.00 5 0.41

PAGE 276

276 Tabl e F 2. Continued RSO RSN ESO ESN Participant Ave S.D. Ave S.D. Ave S.D. Ave S.D. L2 3 6 0.00 6 0.00 0 0.00 6 0.00 4 6 0.00 6 0.00 0 0.00 6 0.00 5 6 0.00 5 0.41 0 0.00 6 0.00 6 6 0.00 6 0.00 0 0.00 6 0.00 7 6 0.00 6 0.00 1 0.41 6 0.00 8 6 0.00 6 0.00 0 0.00 6 0.00 F.2 Context Matching Felicitousness Task Table F CFO CFN TSO TSN TMO TMN Group Ave S.D. Ave S.D. Ave S.D. Ave S.D. Ave S.D. Ave S.D. NS 5.92 0.28 0.71 1.27 5.54 0.78 2.75 1.70 4.58 1.61 5.54 0.88 L2 5.75 0.46 0.25 0.43 5.75 0.71 4.38 1.51 5.50 0.76 6.00 0.00

PAGE 277

277 Table F CFO CFN TSO TSN TMO TMN Group Ave S.D. Ave S.D. Ave S. D. Ave S.D. Ave S.D. Ave S.D. Native 3 6 0.00 0 0.00 3 0.55 1 0.41 3 0.55 3 0.55 4 6 0.00 1 0.41 6 0.00 2 0.52 3 0.55 6 0.00 5 6 0.00 0 0.00 6 0.00 0 0.00 4 0.52 6 0.00 6 6 0.00 0 0.00 6 0.00 2 0.52 0 0.00 6 0.00 7 6 0.00 0 0.00 5 0.41 3 0.55 5 0.41 6 0.00 8 6 0.00 3 0.55 4 0.52 6 0.00 1 0.41 6 0.00 9 6 0.00 4 0.52 5 0.41 4 0.52 6 0.00 6 0.00 10 5 0.41 3 0.55 6 0.00 3 0.55 5 0.41 5 0.41 11 6 0.00 0 0.00 6 0.00 1 0.41 6 0.00 6 0.00 12 6 0.00 2 0.52 6 0.00 2 0.52 4 0.52 5 0.41 13 6 0.0 0 3 0.55 5 0.41 4 0.52 6 0.00 6 0.00 14 6 0.00 0 0.00 6 0.00 5 0.41 6 0.00 6 0.00 15 6 0.00 0 0.00 5 0.41 3 0.55 6 0.00 5 0.41 16 6 0.00 1 0.41 5 0.41 2 0.52 6 0.00 6 0.00 17 6 0.00 0 0.00 6 0.00 5 0.41 4 0.52 6 0.00 18 6 0.00 0 0.00 6 0.00 2 0.52 5 0 .41 5 0.41 19 6 0.00 0 0.00 6 0.00 0 0.00 5 0.41 3 0.55 20 6 0.00 0 0.00 6 0.00 5 0.41 4 0.52 6 0.00 21 6 0.00 0 0.00 6 0.00 1 0.41 4 0.52 5 0.41 22 6 0.00 0 0.00 5 0.41 3 0.55 5 0.41 6 0.00 23 5 0.41 0 0.00 6 0.00 5 0.41 6 0.00 6 0.00 24 6 0.00 0 0. 00 6 0.00 4 0.52 6 0.00 6 0.00 25 6 0.00 0 0.00 6 0.00 2 0.52 6 0.00 6 0.00 26 6 0.00 0 0.00 6 0.00 1 0.41 4 0.52 6 0.00 L2 1 6 0.00 0 0.00 6 0.00 1 0.41 4 0.52 6 0.00 2 6 0.00 1 0.41 6 0.00 4 0.52 5 0.41 6 0.00

PAGE 278

278 Table F 4. Continued CFO CFN TSO TSN TMO TMN Group Ave S.D. Ave S.D. Ave S.D. Ave S.D. Ave S.D. Ave S.D. L2 3 6 0.00 0 0.00 6 0.00 5 0.41 6 0.00 6 0.00 4 6 0.00 0 0.00 6 0.00 6 0.52 6 0.00 6 0.00 5 6 0.00 0 0.00 6 0.00 5 0.41 6 0.00 6 0.00 6 6 0.00 1 0.41 6 0.00 4 0.52 5 0.41 6 0.00 7 5 0.41 0 0.00 6 0.00 5 0.41 6 0.00 6 0.00 8 5 0.41 0 0.00 4 0.50 5 0.41 6 0.00 6 0.00 F.3 Online Grammaticality Task Table F 5. Group average RT (ms) and standard deviation: First region ( ella/l/ello/s iempre ) RSO RSN ESO ESN Group Ave S.D. Ave S.D. Ave S.D. Ave S.D. NS 507.12 96.72 516.52 100.51 469.88 104.79 492.31 130.56 L2 577.39 88.68 622.14 102.10 538.69 89.17 606.02 131.04

PAGE 279

279 Table F 6. Individual average RT (ms) and standard deviation: First region ( ella/l/ello/siempre ) RSO RSN ESO ESN Participant Ave S.D. Ave S.D. Ave S.D. Ave S.D. Native 3 540.50 182.96 406.20 36.48 454.00 174.51 446.83 86.40 4 381.17 59.31 543.60 231.41 403.67 85.19 447.00 143.44 5 424.75 176.60 647.20 241.83 520.60 223.95 645.80 81.27 6 652.00 190.99 507.17 96.35 412.60 66.69 433.80 48.38 7 355.75 318.84 368.00 123.88 401.33 276.63 274.25 98.51 8 447.75 92.88 401.00 80.65 339.60 32.94 359.17 39.54 9 570.50 262.96 639.50 174.02 522.67 48.23 529.17 22 2.81 10 418.75 25.24 542.17 109.28 466.75 88.81 421.83 75.81 11 494.67 210.77 543.20 111.74 462.75 100.77 503.00 153.59 12 300.17 69.79 367.80 202.56 318.83 101.34 317.00 61.85 13 559.20 241.01 506.75 85.62 590.50 205.95 719.75 219.38 14 679.50 201.30 502.80 129.19 511.00 253.68 579.60 234.63 15 608.33 108.07 594.75 178.80 777.75 233.03 730.00 36.77 16 471.50 84.63 529.80 83.55 428.33 102.76 434.17 134.39 17 570.60 224.88 692.17 198.48 511.50 56.38 664.83 164.61 18 451.20 123.97 600.17 245.07 469.5 0 99.57 502.00 256.78 19 431.20 126.64 396.83 36.96 341.00 30.38 443.50 178.04 20 589.25 265.17 611.20 192.53 470.80 75.72 426.33 37.68 21 481.00 67.30 474.75 30.29 523.17 127.15 550.20 274.08 22 512.00 186.73 484.60 57.44 462.83 101.90 452.50 107.19 23 640.80 190.03 660.17 163.62 654.40 224.24 579.33 87.16 24 533.80 246.56 326.80 66.51 368.50 297.51 230.00 40.91 25 580.00 225.35 502.50 59.01 350.00 35.57 487.00 79.48 26 476.50 71.20 547.33 129.24 515.00 98.75 638.40 150.94 L2 1 495.25 105.30 741.50 225.75 662.60 102.36 790.50 124.06 2 598.00 184.31 599.40 203.25 416.60 65.06 434.33 67.17

PAGE 280

280 Table F 6. Continued RSO RSN ESO ESN Participant Ave S.D. Ave S.D. Ave S.D. Ave S.D. L2 3 585.17 224.61 557.20 110.32 483.0 0 41.05 473.80 117.11 4 596.00 192.81 549.80 219.47 610.50 315.28 646.33 149.45 5 506.60 177.55 772.67 288.79 469.20 154.62 596.33 195.31 6 471.60 84.79 526.33 166.92 637.17 256.74 514.33 230.58 7 749.80 203.85 706.83 150.69 542.80 122.16 776.25 153.57 8 616.67 206.02 523.40 178.67 487.67 150.66 616.25 292.40 Table F 7. Group average RT (ms) and standard deviation: Second region (verb) RSO RSN ESO ESN Group Ave S.D. Ave S.D. Ave S.D. Ave S.D. NS 464.48 114.18 463.52 117.45 592.67 189.08 521.87 1 39.72 L2 575.98 167.97 497,74 96.18 762.26 182.82 700.18 157.58

PAGE 281

281 Table F 8. Individual average RT (ms) and standard deviation: Second region (verb) RSO RSN ESO ESN Participant Ave S.D. Ave S.D. Ave S.D. Ave S.D. Native 3 446.67 177.3 1 376.20 52.63 399.67 72.20 332.83 22.33 4 369.83 20.41 379.40 50.58 430.17 72.53 482.50 134.28 5 479.40 78.21 588.20 288.76 901.40 390.38 535.40 105.35 6 497.25 28.10 484.33 74.06 518.20 151.97 443.40 89.16 7 314.50 96.65 212.75 39.74 602.00 284.25 31 9.00 95.31 8 338.00 55.29 384.17 71.30 332.40 49.59 370.67 28.87 9 394.00 61.22 599.00 248.53 504.67 101.04 489.40 90.25 10 424.00 64.95 453.17 80.75 475.50 80.01 435.83 87.64 11 439.17 103.68 455.20 83.30 503.50 78.34 524.00 134.79 12 307.67 92.61 27 2.80 25.20 278.67 60.67 351.25 87.30 13 712.40 277.68 525.00 234.49 910.25 330.07 497.50 166.12 14 539.50 254.84 535.80 254.20 602.80 369.30 550.00 202.28 15 796.00 205.98 754.20 269.73 782.00 321.52 781.00 308.74 16 382.00 88.16 484.00 115.47 636.67 3 39.48 516.67 64.26 17 560.40 160.16 528.33 176.92 597.67 148.03 811.33 148.43 18 447.80 131.02 546.83 306.25 546.50 314.88 533.33 203.40 19 405.60 77.22 333.83 80.35 435.50 152.03 454.00 175.40 20 428.25 56.33 535.60 70.96 623.75 293.53 508.50 108.30 21 462.17 78.67 430.75 72.15 888.40 266.83 617.60 144.39 22 466.50 112.96 440.20 78.91 547.17 101.17 601.83 268.73 23 582.00 87.10 584.83 150.20 1000.60 260.49 675.17 163.59 24 417.20 224.59 330.00 185.86 422.00 391.06 304.00 97.66 25 419.75 102.67 461 .83 132.26 590.60 279.81 703.83 277.15 26 517.50 124.58 428.00 694.00 260.49 685.80 325.01 L2 1 564.25 102.30 575.33 164.70 956.33 397.79 985.60 280.75 2 466.83 108.06 464.60 48.99 541.60 325.12 468.33 100.29

PAGE 282

282 Table F 8. Continued RSO RSN ESO ESN Participant Ave S.D. Ave S.D. Ave S.D. Ave S.D. L2 3 450.33 91.12 441.20 37.18 802.83 367.46 528.80 161.92 4 583.75 190.64 364.00 209.69 950.25 319.28 733.50 340.12 5 524.00 196.98 616.00 156.35 486.33 200.23 743.50 347.0 4 6 379.40 73.33 449.33 53.13 672.33 184.24 736.33 252.50 7 888.60 308.55 628.83 312.80 763.67 379.04 653.00 115.95 8 750.67 268.17 442.60 68.89 924.75 482.69 752.33 380.25 Table F 9. Group average RT (ms) and standard deviation: Third region (DO) RSO RSN ESO ESN Group Ave S.D. Ave S.D. Ave S.D. Ave S.D. NS 468.78 102.32 481.69 94.96 540.54 151.03 467.39 90.29 L2 570.24 78.55 632.63 148.04 641.28 61.16 573.82 68.46

PAGE 283

283 Table F 10. Individual average RT (ms) and standard deviation: Second region ( DO) RSO RSN ESO ESN Participant Ave S.D. Ave S.D. Ave S.D. Ave S.D. Native 3 467.50 253.23 358.40 92.45 412.00 183.05 359.33 29.96 4 351.00 38.91 381.00 38.54 472.00 265.99 455.50 150.70 5 573.00 155.39 471.60 85.94 682.20 329.84 508.40 88.26 6 396.20 90.63 459.00 80.20 414.60 52.63 414.40 54.83 7 380.00 141.32 348.00 102.97 374.00 29.46 389.75 146.99 8 338.25 44.36 314.33 55.04 411.80 95.50 367.67 58.78 9 415.25 33.06 605.67 248.29 516.00 159.90 448.67 94.82 10 440.50 86.43 480.83 152.37 571.50 117.26 440.00 105.39 11 439.17 85.53 499.60 81.52 674.00 131.09 405.75 47.65 12 335.33 74.54 391.60 136.67 407.83 137.28 361.75 55.95 13 561.40 150.81 653.50 261.36 577.75 166.97 539.75 111.95 14 448.25 124.21 562.00 285.29 641.60 176.49 614.40 242.39 15 819.00 119.38 643.60 154.62 932.75 162.50 690.25 244.58 16 510.17 208.50 484.00 84.15 474.83 230.43 452.00 78.23 17 520.80 279.18 546.20 270.58 544.00 47.29 601.17 190.84 18 432.20 105.09 482.17 172.92 648.50 213.70 483.83 164.23 19 4 16.00 87.44 426.17 165.54 365.83 73.55 417.83 150.56 20 480.25 94.68 466.60 33.71 598.60 235.55 489.17 31.37 21 454.33 90.55 459.50 76.22 666.67 211.42 530.60 82.83 22 517.83 257.25 495.40 134.23 582.33 225.19 428.00 48.12 23 562.80 255.23 587.67 132.9 9 830.00 175.22 576.50 229.96 24 376.00 113.86 339.40 77.46 432.40 201.62 320.17 80.83 25 501.00 179.67 548.67 157.30 380.80 115.11 428.67 83.55 26 514.50 96.83 555.67 139.03 361.00 16.52 493.80 125.45 L2 1 575.00 106.86 864.00 186.2 3 661.40 199.90 579.00 78.36 2 481.00 84.80 499.00 23.54 499.20 132.93 447.50 55.03

PAGE 284

284 Table F 10. Continued RSO RSN ESO ESN Participant Ave S.D. Ave S.D. Ave S.D. Ave S.D. L2 3 456.00 65.94 566.80 169.60 669.83 170.84 566.80 149.27 4 69 9.00 116.16 744.60 320.01 633.50 190.24 552.83 188.12 5 575.80 272.04 776.67 207.44 651.60 241.52 545.75 173.96 6 548.60 209.39 437.33 123.72 704.33 180.41 690.17 201.15 7 587.00 100.03 554.83 127.81 644.00 80.77 613.00 88.40 8 639.50 102.74 617.80 205 .26 666.33 135.72 595.50 177.39 F.4 Online Context Matching Felicitousness Task Table F 11. Group average RT (ms) and standard deviation: First region ( ella CFO TSO TMO Group Ave S.D. Ave S.D. Ave S.D. NS 302.29 51.20 307.64 51.98 341.44 50 .34 L2 436.18 80.30 440.96 39.60 458.47 29.76

PAGE 285

285 Table F 12. Individual average RT (ms) and standard deviation: First region ( ella CFO TSO TMO Group Ave S.D. Ave S.D. Ave S.D. Native 3 284.83 51.70 278.33 144.11 334.00 36.23 4 216.3 3 17.44 231.80 21.45 248.50 32.42 5 280.40 72.01 337.60 64.15 326.67 156.05 6 380.50 123.83 327.00 49.09 352.80 44.39 7 205.50 69.60 254.40 114.95 363.60 78.97 8 268.50 21.73 271.17 99.01 351.60 28.87 9 306.00 86.68 333.33 23.32 319.67 55.67 10 331.6 7 26.52 358.60 62.44 371.33 56.76 11 307.17 57.78 274.17 64.77 345.60 24.21 12 241.00 40.78 264.67 37.56 247.00 42.46 13 312.50 61.54 353.00 48.42 336.25 56.04 14 268.83 38.10 285.50 132.69 375.50 59.79 15 295.17 73.42 371.17 164.87 386.60 123.51 16 305.50 49.79 303.33 40.63 317.67 40.94 17 376.83 37.41 372.17 124.70 432.25 22.87 18 339.67 42.71 308.00 32.88 314.25 44.97 19 234.00 47.70 221.33 184.64 296.00 83.24 20 379.75 25.26 378.50 15.44 367.00 43.98 21 314.80 34.46 373.00 17.25 406.50 41.66 22 301.33 48.65 309.00 41.04 352.50 22.50 23 409.60 27.86 371.20 102.54 421.67 32.28 24 299.17 162.35 197.50 42.52 246.67 18.29 25 287.50 25.66 295.40 59.14 379.67 18.39 26 308.33 162.23 313.17 181.92 301.20 104.30 L2 1 564.17 50.57 48 7.00 49.09 490.40 103.16 2 484.60 60.73 482.00 45.49 489.83 51.45

PAGE 286

286 Table F 12. Continued CFO TSO TMO Group Ave S.D. Ave S.D. Ave S.D. L2 3 462.60 52.12 445.40 68.34 481.00 49.93 4 375.20 96.83 397.17 92.10 444.75 111.91 5 387.67 128.54 3 82.67 144.12 432.20 93.49 6 381.50 19.23 424.67 49.62 445.20 42.81 7 506.00 115.25 429.80 81.20 409.75 99.39 8 327.67 175.08 479.00 46.95 474.60 81.01 Table F 13. Group average RT (ms) and standard deviation: Second region (verb) CFO CFN TSO TSN TM O TMN Group Ave S.D. Ave S.D. Ave S.D. Ave S.D. Ave S.D. Ave S.D. NS 319.82 68.66 307.88 51.70 325.66 74.57 347.05 70.71 315.71 66.65 321.90 62.23 L2 521.71 92.88 517.33 87.56 509.35 114.04 536.39 98.82 520.16 88.80 535.73 77.73

PAGE 287

287 Table F 14. Individua l average RT (ms) and standard deviation: Second region (verb) CFO CFN TSO TSN TMO TMN Group Ave S.D. Ave S.D. Ave S.D. Ave S.D. Ave S.D. Ave S.D. Native 3 280.50 23.13 278.00 23.82 272.00 60.59 272.67 39.83 266.67 24.73 258.80 43.5 4 4 259.33 46.50 254.33 60.91 254.67 33.19 264.40 39.31 233.60 23.09 258.67 52.25 5 421.60 236.96 284.25 68.11 297.33 30.90 419.17 110.78 324.00 38.43 360.17 105.42 6 329.17 101.55 285.67 38.09 310.60 22.57 329.17 36.13 362.00 153.29 288.00 47.09 7 341 .25 61.89 322.80 183.44 269.40 79.24 379.40 147.06 259.40 111.10 237.20 95.35 8 351.83 133.46 279.67 49.39 317.00 55.07 302.33 19.38 270.33 17.00 275.67 29.36 9 288.00 41.55 349.83 103.94 285.00 31.55 303.00 35.09 322.00 43.96 388.50 222.87 10 341.00 34 .84 366.20 54.55 319.67 23.16 382.40 42.18 302.40 66.10 317.67 63.65 11 302.33 49.43 286.00 28.44 302.40 57.76 325.40 71.44 276.83 20.09 396.75 128.97 12 228.50 17.97 257.67 41.77 228.17 30.50 237.00 6.66 231.83 40.44 257.33 31.38 13 350.83 216.72 369.8 3 90.13 268.25 26.25 400.67 223.83 483.50 196.52 361.60 102.44 14 346.00 241.18 328.83 101.74 410.67 218.64 292.50 52.06 258.00 54.86 342.20 138.02 15 502.50 193.79 345.25 178.85 536.60 213.01 518.75 216.74 428.33 217.63 455.67 119.94 16 311.67 55.37 30 8.17 66.61 315.17 97.27 286.33 46.63 303.50 92.33 314.50 47.50 17 366.83 50.55 345.20 31.08 462.75 152.76 488.33 171.87 394.17 60.04 418.40 84.24 18 321.00 53.46 334.75 28.37 326.75 29.28 347.20 56.30 319.50 90.09 327.20 44.30 19 190.60 25.77 190.50 32. 43 273.80 67.75 272.83 59.65 280.50 70.69 241.83 65.13 20 374.25 26.27 353.75 33.04 376.80 10.66 367.33 15.98 377.67 27.53 339.60 23.18 21 311.60 38.33 350.00 42.02 364.67 36.89 367.25 16.68 325.75 31.43 340.17 49.47 22 306.00 37.79 323.17 29.71 394.83 82.29 353.60 49.96 338.83 47.25 343.83 45.88 23 394.60 59.24 398.40 52.60 405.33 108.54 409.80 86.44 377.40 26.12 394.67 48.83 24 234.17 61.52 239.17 104.11 231.17 44.48 379.40 107.64 208.00 17.47 214.75 24.88 25 306.50 58.24 323.00 65.24 269.50 21.39 3 69.60 89.15 265.40 14.91 300.17 32.29 26 215.67 57.55 214.80 39.93 323.40 128.63 260.67 50.42 367.33 187.15 292.33 92.83 L2 1 648.17 210.27 608.50 29.28 610.00 238.12 666.83 189.72 621.20 176.23 565.40 71.54 2 494.40 81.38 479 .75 151.75 537.33 108.14 526.20 80.00 471.00 49.69 540.60 126.34

PAGE 288

288 Table F 14. Continued CFO CFN TSO TSN TMO TMN Group Ave S.D. Ave S.D. Ave S.D. Ave S.D. Ave S.D. Ave S.D. L2 3 486.60 158.16 571.20 137.31 484.00 132.59 530.00 101.0 0 417.80 46.47 616.50 97.45 4 590.40 126.83 399.80 243.47 599.00 289.32 644.80 208.10 608.67 238.83 505.17 266.29 5 389.67 87.18 464.00 189.25 356.80 185.89 495.00 322.27 482.20 88.75 444.00 ------6 417.25 81.20 487.40 72.29 415.60 29.24 434.25 35.20 424.83 94.23 416.00 50.28 7 618.17 119.62 466.20 129.41 398.25 54.31 607.00 189.40 500.00 197.57 559.33 186.06 8 529.00 259.92 661.80 202.05 673.80 140.04 387.00 43.03 635.60 99.53 638.83 219.02 Table F 15. Group average RT (ms) and standard deviatio n: Third region (DO) CFO CFN TSO TSN TMO TMN Group Ave S.D. Ave S.D. Ave S.D. Ave S.D. Ave S.D. Ave S.D. NS 374.63 81.58 354.39 85.99 368.46 84.97 343.87 58.27 355.07 97.52 354.61 86.63 L2 726.68 158.17 792.78 219.62 613.18 142.29 718.75 177.49 699.00 135.31 661.17 248.95

PAGE 289

289 Table F 16. Individual average RT (ms) and standard deviation: Third region (DO) CFO CFN TSO TSN TMO TMN Group Ave S.D. Ave S.D. Ave S.D. Ave S.D. Ave S.D. Ave S.D. Native 3 284.67 48.84 270.60 37.55 281.50 3 6.65 267.83 38.47 269.33 43.30 248.40 20.59 4 346.60 60.80 255.50 57.04 291.83 54.53 332.80 102.76 304.60 41.10 324.67 88.48 5 479.80 138.89 367.00 58.86 352.50 83.17 317.17 59.07 282.00 42.01 451.00 170.78 6 440.83 156.73 309.83 41.19 353.80 99.68 350. 00 67.58 329.00 23.96 354.60 74.98 7 377.50 367.65 451.20 275.20 502.20 190.61 424.40 193.41 415.60 414.64 539.00 402.19 8 291.67 27.78 279.33 12.48 343.80 111.75 316.83 45.88 302.50 19.37 269.67 32.98 9 282.33 45.58 276.67 54.91 339.00 116.51 297.50 61 .56 322.00 41.89 286.17 42.35 10 326.17 29.32 351.20 13.86 354.00 14.98 389.00 60.89 412.80 169.30 334.50 80.92 11 431.33 260.09 304.40 52.72 337.60 62.15 310.20 40.98 304.17 33.02 371.00 113.50 12 284.25 112.09 264.00 64.52 229.17 30.78 249.00 23.19 22 9.17 22.82 240.00 24.31 13 530.33 377.20 457.17 290.18 311.25 109.90 350.83 142.15 519.00 381.05 358.80 143.37 14 342.83 148.91 347.83 122.84 367.67 136.18 272.83 32.25 273.50 63.88 303.20 53.79 15 500.67 106.51 493.50 92.34 573.00 219.08 438.50 160.30 611.67 126.57 459.17 155.41 16 345.17 77.30 386.00 104.38 318.50 54.89 312.67 31.77 332.33 37.51 324.00 40.26 17 404.33 192.27 393.20 55.93 515.75 124.87 444.33 203.42 491.17 136.96 481.80 109.91 18 373.17 29.01 346.00 43.41 338.50 43.34 340.20 38.07 40 2.50 70.08 331.80 38.17 19 252.80 29.01 273.33 75.01 251.00 54.71 278.17 34.41 219.67 49.34 256.50 73.73 20 372.50 31.55 379.25 18.57 387.00 34.11 371.67 38.92 359.00 32.89 360.80 16.72 21 465.40 145.52 564.20 352.29 434.83 103.73 335.75 25.22 446.00 13 6.04 307.00 28.37 22 359.83 41.56 350.67 37.11 391.17 49.79 422.60 154.58 345.83 91.14 346.17 48.55 23 460.00 43.78 448.60 106.56 394.00 78.52 414.40 111.19 388.20 30.28 541.17 182.47 24 245.00 80.77 196.67 21.19 278.83 100.93 269.60 32.81 213.50 73.92 283.50 130.91 25 332.33 71.25 342.80 124.70 419.67 143.62 397.80 182.74 327.80 105.28 306.00 74.76 26 461.50 60.54 396.40 295.92 476.40 209.22 348.83 77.19 420.33 115.84 431.83 148.77 L2 1 949.17 266.66 1065.33 129.40 828.25 2 35.42 995.33 162.32 891.60 217.34 1080.80 95.05 2 644.00 234.68 679.50 334.26 587.83 176.36 562.60 133.26 527.40 75.68 566.60 176.23

PAGE 290

290 Table F 16. Continued CFO CFN TSO TSN TMO TMN Group Ave S.D. Ave S.D. Ave S.D. Ave S.D. Ave S.D. Ave S.D. L2 3 783.60 291.42 804.40 308.12 578.67 1422.04 778.00 196.09 530.80 147.93 590.00 192.30 4 969.40 242.13 1171.00 0.00 842.25 380.53 851.75 253.02 776.17 363.69 966.67 263.53 5 588.67 504.42 651.40 434.21 462.50 155.77 446.40 199.24 712.60 429. 39 396.00 ------6 557.75 257.54 511.60 116.66 542.20 94.32 645.25 210.37 641.50 229.02 404.60 82.07 7 648.83 179.10 741.00 331.17 541.50 150.71 639.67 318.12 662.75 236.46 724.67 366.03 8 672.00 355.10 718.00 425.10 522.20 329.88 831.00 282.04 849.20 197.26 560.00 342.84

PAGE 291

291 LIST OF REFERENCES Aaronson, D., & Scarborough, H.S. (1976). Performance theories for sentence coding: Some quantitative evidence. Journal of Experimental Psychology: Human Perception and Performance, 2, 56 70. Abutalebi, J., Cappa, S.F., & Perani, D. (2001). The bilingual brain as revealed by functional neuroimaging. Bilingualism: Language and Cognition, 4 179 190. Alderson, J.C. (2000). Assessing Reading. Cambridge: Cambridge University Press. Alexiadou, A., & Anagnostopoulou, E. (1998). Parametrizing AGR: Word Order, V movement and EPP checking. Natural Language and Linguistic Science, 16 (3), 491 539. Alonso hypothesis. In H. Campos et al. (Eds .), Hispanic Linguistics at the Turn of the Millennium Somerville: Cascadilla Press. Alonso Ovalle, L., Fernandez, S., Frazier, L., & Clifton, C. (2002). Null vs. Overt pronouns and the Topic focus articulation in Spanish, Journal of Italian Linguistics, 14 151 169. Austin, J., Blume, M., Parkinson, D., Nnez del Prado, Z. & Lust, B. (1996). Interactions between pragmatic and syntactic knowledge in the first language acquisition of Spanish null and overt pronominals. In J. Lema, & E. Trevino (Eds.), Theor etical analyses on Romance languages (pp. 35 47). Amsterdam, John Benjamins. Bachman, L.F. (2004). Statistical Analyses for Language Assessment. Cambridge: Cambridge University Press. Bachman, L.F., & Palmer, A.S. (1996). Language Testing in Practice. Oxfo rd: Oxford University Press. ect the parsing of garden path sentences. Journal of Memory and Language, 49, 183 200. Barrenechea, A.M., & Alonso, A. (1977). Los pronombres personales sujetos en el espa ol hablado en Buenos Aires. In Juan M. Lope Blanch (Ed.), Estudios sobre el espaol hablado en las principales ciudades de Amrica (pp. 333 349). Mexico City: Universidad Nacional Autnoma de Mxico. Bayley, R., & Pease lvarez, L. (1996). Null and expre ssed pronoun variation in Mexican Solomon (Eds.) Sociolinguistic variation: Data, theory, and analysis (pp. 85 99). Stanford: Center for the Study of Language and Informati on. Beck, M.L. (1998). L2 acquisition and obligatory head movement: English speaking learners of German and the local impairment hypothesis. Studies in Second Language Acquisition, 20 311 348.

PAGE 292

292 Belletti, A., & Leonini, C. (2004). Subject Inversion in L2 It alian. EUROSLA Yearbook, 4, 95 118. Belletti, A., Bennati, E., & Sorace, A. (2007). Theoretical and developmental issues in the syntax of subjects: evidence from near native Italian. Natural Language and Linguistic Theory, 25 657 689. Bentivoglio, P. (198 7). Los sujetos pronominales de primera persona en el habla de Caracas. Caracas: Universidad Central de Venezuela. Bergelson, E., & Swingley, D. (2012). At 6 9 months, human infants know the meanings of many common nouns PNAS, 109 (9), 3253 3258. Bialystok, E. (2009). Bilingualism: The good, the bad, and the indifferent. Bilingualism: Language and Cognition, 12 (1), 3 11. Biberauer, T., Holmberg, A., Roberts, I., & Sheehan, M. (2010). Parametric Variation: Null Subjects in Minimalist Theory Cambridge, MA: C ambridge University Press. Bini, M. (1993). La adquisicin del italiano: ms all de las propiedades sintcticas del parmetro pro drop In J.M. Liceras (Ed.), La lingstica y el anlisis de los sistemas no nativos (pp. 126 139). Ottawa: Dovehouse. Bird song, D. (1992). Ultimate attainment in second language acquisition. Language, 68, 706 755. Blackwell, S., & Quesada, M.L. (2010). The L2 acquisition of discourse pragmatic constraints on Spanish third person subject use. Paper presented at the Hispanic Li nguistics Symposium 2010. Bloomington, Indiana, Oct. 14 17. Blake, R. (1983). Mood selection among Spanish speaking children, ages 4 to 12. The Bilingual Review, 10, 21 32. Bley Vroman, R. (1990). The logical problem of foreign language learning. Linguist ic Analysis, 20 (1 2), 3 49. Bley Vroman, R. (2009). The evolving context of the fundamental difference hypothesis. Studies in Second Language Acquisition, 31 175 198. Bohnacker, U. (2010). The clause initial position in L2 Swedish declaratives: Word order variation and discourse pragmatics. Nordic Journal of Linguistics, 33, 105 143. Bond, K., Gabriele, A., Fiorentino, R., & Alemn Ban, J. (2011). Individual differences and the role of the L1 in L2 processing: An ERP investigation. In D. Tanner, & J. He rschensohn (Eds.), Proceedings of the 11th Generative Approaches to Second Language Acquisition Conference (GASLA 2011) ( pp. 17 29). Somerville, MA: Cascadilla Proceedings Project.

PAGE 293

293 Brown, R. (1973). A First Language: The Early Stages Harvard University Pr ess, Cambridge, MA. Brownell, H.H, Carroll, J.J, Rehak, A, & Wingfield, A. (1992). The use of pronoun anaphora and speaker mood in the interpretation of conversational utterances by right hemisphere brain damaged patients. Brain and Language, 43 (1), 121 14 7. Burkhardt, P. (2005). The Syntax Discourse Interface: Representing and Interpreting Dependency Amsterdam: John Benjamins. Cameron, R. (1994). Switch reference, verb class and priming in a variable syntax. In K. Beals (Ed.), Papers from the 30th region al meeting of the Chicago Linguistics Society: Volume 2: The parasession on variation in linguistic theory (pp. 27 45). Chicago Linguistics Society. Cameron, R. (1995). The scope and limits of switch reference as a constraint on pronominal subject expressi on. Hispanic Linguistic,s 6/7, 1 27. Cameron, R. (1996). A community based test of a linguistic hypothesis. Language in Society, 25 61 111. Cameron, R., & Flores Ferrn, N. (2004). Perseveration of subject expression across regional dialects of Spanish. Spanish in Context, 1, 41 65. Carminati, M.N. (2002). The Processing of Italian Subject Pronouns Doctoral dissertation, University of Massachusetts Amherst. Carroll, J.J. 1983. Underlying Determinants of Complex Interpretations in Discourse. Doctoral diss ertation, Cornell University. Choi, M.H., & Lardiere, D. ( 2006). The interpretation of wh in situ in Korean second language acquisition. In A. Belletti, E. Bennati, C. Chesi, E. Di Domenico, & I. Ferrari (Eds.), Language acquisition and development : Procee dings of GALA 2005 (pp. 125 35). Cambridge, MA: Cambridge Scholars Press. Chomsky, N. (1965). Aspects of the Theory of Syntax. Cambridge, MA: MIT Press. Chomsky, N. (1981). Lectures on Government and Binding Dordrecht: Foris Publications. Chomsky, N. (198 6). Knowledge of Language: Its Nature, Origin and Use New York: Praeger. Chomsky, N. (1995). The Minimalist Program Cambridge, MA: MIT Press. Chomsky, N. (2000). On Nature and Language New York: Cambridge University Press. Chomsky, N. (2001). Derivation by phase. In M. Kenstowicz (Ed.), Ken Hale: A Life in Language (pp. 1 52). Cambridge, MA: MIT Press.

PAGE 294

294 Clahsen, H., & Hong, U. ( 1995). Agreement and null subjects in German L2 development: new evidence from reaction time experiments. Second Language Researc h, 11 57 87. Cinque, G. (1999). Adverbs and Functional heads: A Cross Linguistic Perspective. New York Oxford: Oxford University Press. Contreras, H. (1976). A theory of word order with special reference to Spanish. Amsterdam: North Holland Publishing Co mpany. Contreras, H. (1991). On the position of subjects. Syntax and Semantics, 25 63 79. Coppetiers, R. (1987).Competence differences between native and near native speakers. Language, 63, 544 573. Corley, M. & Stewart, O.W. ( 2008). Hesitation disfluencies in spontaneous speech: The meaning of um. Language and Linguistics Compass, 2 (4), 589 602. Hesitations in speech a ect lan guage comprehension. Cognition, 105 658 668. Costa, A. (2005). Lexical access in bilingual production. In J.F. Kroll, & A.M.B. De Groot (Eds.), Handbook of bilingualism: Psycholinguistic approaches (pp. 308 325). New York: Oxford University Press. Curtiss S. (1977). Genie: A Psycholinguistic Study of a Modern Day "Wild Child." New York: Academic Press. Curtiss, S. (1989a). The Case of Chelsea: A New Test Case of the Critical Period for Language Acquisition. Unpublished manuscript, University of California Los Angeles. Curtiss, S. (1989b). The independence and task specificity of language. In M. Bornstein, & J. Bruner (Eds.), Interaction and Human Development (pp. 105 137). Mawah, N J: Lawrence Erlbaum. Cuza, A., & Frank, J. (2011). Transfer Effects at the Syntax Semantics Interface: The Case of Double que Questions in Heritage Spanish. Heritage Language Journal, 8 (1), 66 89. Dehaene, S.D., Dupoux, E., Mehler, J., Cohen, L., Paulesu, E., Perani, D., van de Moortele, P. F., Lehericy, S., & Le Bihan, D. (199 7). Anatomical variability in the cortical representation of first and second languages. Neuroreport, 8, 3809 3815. Diessel, H., & Tomasello, M. (2000). The development of relative clauses in spontaneous child speech. Cognitive Linguistics, 11 (1 2), 131 15 1. Dijkstra, T. (2005). Bilingual word recognition and lexical access. In J.F. Kroll & A.M.B. De Groot (Eds.), Handbook of bilingualism: Psycholinguistic approaches (pp. 179 201). New York: Oxford University Press.

PAGE 295

295 Dobrovie Sorin, C. (1994). The Syntax of Romanian: Comparative Studies in Romance Dordrecht: Foris. Dodel, S., Golestani, N., Pallier, C., ElKouby, V., Le Bihan, D., & Poline, J.B. (2005). Condition dependent functional connectivity: Syntax networks in bilinguals. Philosophical transactions of t he Royal Society of London. Series B, Biological sciences, 360, 921 935. Domnguez, L. (2007). The L2 acquisition of Spanish focus: a case of incomplete and divergent grammars. In S. Baauw, J. van Kampen, & M. Pinto (Eds.) The Acquisition of Romance Langua ges: Selected papers from The Romance Turn II 2006 (pp. 45 57). Utrecht, The Netherlands: LOT. Domnguez, L., & Arche, M.J. (2008). Optionality in L2 Grammars: The Acquisition of SV/VS Contrast in Spanish. In H. Chan, H. Jacob, & E. Kapia (Eds.), Proceedin gs of Annual Boston Conference on Language Development 32 (pp. 96 107). Sommerville, MA: Cascadilla Press. Donaldson, B. (2011a). Left dislocation in near native French. Studies in Second Language Acquisition, 33 (3), 399 432. Donaldson, B. (2011b). Nativel ike right dislocations in near native French. Second Language Research, 27 (3), 361 390 Duarte, M.E. (1993). Do pronome nulo ao pronome pleno: a trajetria do sujeito no portugus do brasil. In I. Roberts, & M.A. Kato (Eds.), Do pronome nulo ao pronome ple no: a trajetria do sujeito no portugus do Brasil (pp. 107 128). Campinas: Editora da UNICAMP. Duarte, M.E. (1995). Doctoral dissertation, UNICAMP. Duffield, N. (2011). Loose Ends? Commentary on Sorace. Linguistic Approaches to Bilingualism 1 35 38. Dekydtspotter, L., Schwartz, B.D., & Sprouse, R.A. (2006). The Comparative Fallacy in L2 Proceedings of the 8th Gener ative Approaches to Second Language Acquisition Conference (GASLA 2006) (pp. 33 40). Somerville, MA: Cascadilla Proceedings Project. Felser, C. (2005). Experimental psycholinguistic approaches to second language acquisition. Second Language Research 21 9 5 97. Fernndez Soriano, O. (1989). Strong pronouns in null subject languages and the Avoid Pronoun Principle. MIT Working Papers in Linguistics, 11, 228 239.

PAGE 296

296 Fernndez Soriano, O. (1993). La visin paramtrica del lenguaje: ms sobre los sujetos y objetos nulos. In J.M. Liceras (Ed.), La lingstica y el anlisis de los sistemas no nativos (pp. 161 176).Ottawa: Dovehouse. Filiaci, F. (2010). Anaphoric Preferences of Null and Overt Subjects in Italian and Spanish: a Cross linguistic Comparison. Doctoral dis sertation, University of Edinburgh. Filiaci, F. (2010). Null and overt subject biases in Spanish and Italian: A crosslinguistic comparison. In C. Borgonovo, M. Espanol Echevaria, & Prevos (Eds.), Selected proceedings of the 12 th Hispanic Lingusitics Sympos ium (pp. 171 182). Somerville, MA: Cascadilla Proceedings Project. Filiaci, F., Sorace, A., & Carreiras, M. (2013). Anaphoric biases of null and overt subjects in Italian and Spanish: a cross linguistic comparison. Language and Cognitive Processes. Flege, J.E., Yeni Komshian, G.H., & Liu, S. (1999). Age constraints on second language acquisition. Journal of Memory and Language, 41 78 104. Fodor, J.A. (1983). Modularity of Mind: An Essay on Faculty Psychology Cambridge, MA: MIT Press. Frenck Mestre, C., Osterhout, L., McLaughlin, J., & Foucart, A. (2008). The effect of phonological realization of inflectional morphology on verbal agreement in French: Evidence from ERPs. Acta Psychologica, 128 (3), 528 536. Frenck Mestre, C., Foucart, A., Carrasco, H., & Herschensohn, J. (2009). Processing of grammatical gender in French as a first and second language. EUROSLA Yearbook, 9 76 106. Friederici, A. (2002). Towards a neural basis of auditory sentence processing. TRENDS in Cognitive Sciences, 6 (2), 78 84. Ga rca Alcaraz, E., & Bel, A. (2011). Seleccin y distribucin de los pronombres en el espaol L2 de los hablantes del rabe. Revista de Lingstica y Lenguas Aplicadas, 6, 165 179. Gillon Dowens, M., Vergara, M., Barber, H.A., & Carreiras, M. (2010). Morpho syntactic processing in late second language learners. Journal of Cognitive Neuroscience, 22 (8), 1870 1887. Goad, H., & White, L. (2006). Ultimate attainment in interlanguage grammars: A prosodic approach. Second Language Research, 22 243 268. Goad, H., & White, L. (2008). Prosodic structure and the representation of L2 functional morphology: A nativist approach. Lingua, 118 577 594. Goad, H., White, L., & Steele, J. (2003). Missing inflection in L2 acquisition: Defective syntax or L1 constrained prosod ic representations? Canadian Journal of Linguistics, 48 243 263.

PAGE 297

297 Goodall, G. (2001). The EPP in Spanish. In W. Davies, & S. Dubinsky (Eds.), Objects and Other Subjects: Grammatical Functions, Functional Categories and Configurationality (pp. 193 223). Do rdrecht: Kluwer Academic. Goodall, G. (2002). On Preverbal Subjects in Spanish. In T. Satterfield, C. Tortora, & D. Cresti (Eds.), Current Issues in Romance Languages: Selected Papers from the 29 th Linguistics Symposium on Romance Languages (LSRL) (pp. 95 109). Amsterdam: John Benjamins. Green, D.W. (1986). Control, activation and resource. Brain and Language, 27 210 223. Green, D.W. (1998). Mental control of the bilingual lexico semantic system. Bilingualism: Language and Cognition, 1 67 81. Grinstead, J. (2004). Subjects and Interface Delay in Child Spanish and Catalan. Language, 80 (1), 40 72. Guasti, M.T. (2002). Language Acquisition: The Growth of Grammar Cambridge: MIT Press. Grel, A. (2006). L2 acquisition of pragmatic and syntactic constraints in the use of overt and null subject pronouns. In R. Slabakova, S. Montrul, & P. Prvost (Eds.), Inquiries in linguistic development: in honor of Lydia White (pp. 259 282). Amsterdam: John Benjamins. Haegeman, L. (1997). Register variation, truncation and su bject omission in English and in French. English Language and Linguistics, 1 233 270. Hahne, A. (2001). What's different in second language processing? Evidence from event related brain potentials. Journal of Psycholinguistic Research, 30 (3), 251 266. Hah ne, A., Mueller, J.L., & Clahsen, H. (2006). Morphological processing in a second language: behavioral and event related brain potential evidence for storage and decomposition. Journal of Cognitive Neuroscience, 18, 121 134. Hawkins, R., & Chan, C. (1997). The partial availability of universal grammar in second Second Language Research, 13 187 226. Hawkins, R., & Hattori, H. (2006). Interpretation of English multiple wh questions by Japane se speakers: A missing uninterpretable feature account. Second Language Research, 22 (3), 269 301. Hernandez, A.E., Hofmann, J., & Kotz, S.A. (2007). Age of acquisition modulates neural activity for both regular and irregular syntactic functions. NeuroImag e, 36 (3), 912 923. Hertel, T. (2003). Lexical and discourse factors in the second language acquisition of Spanish word order. Second Language Research 19 273 304.

PAGE 298

298 Hoekstra, T., & Hyams, N. (1998). Agreement, the finiteness of V2: evidence from child lang uage. Proceedings of the 22nd BUCLD (pp. 360 370). Somerville, MA: Cascadilla. Holmberg, A. (2005). Is there a little pro? Evidence from Finnish. Linguistic Inquiry, 36 (4), 533 564. Hopp, H. (2004). Syntactic and interface knowledge in advanced and near n ative interlanguage grammars. EUROSLA Yearbook, 4, 67 94. Hopp, H. (2007). Ultimate Attainment at the Interfaces in Second Language Acquisition: Grammar and Processing. Groningen: Grodil Press. Hopp, H. (2009). The syntax discourse interface in near nativ e L2 acquisition: Off line and on line Performance Bilingualism: Language and Cognition, 12 (4), 463 483. Huang, C. T.J. (1984). On the distribution and reference of empty pronouns. Lin g uistic Inquiry, 15, 531 574. Hulk, A. & Mller, N. (2000). Bilingual f irst language acquisition at the interface between syntax and pragmatics. Bilingualism: Language and Cognition, 3 (3), 227 244. Hulstijn, J. (2010). Measuring second language proficiency. In E. Blom, & S. Unsworth (Eds.), Experimental Methods in Language Ac quisition Research (pp. 185 199). Amsterdam: John Benjamins. Hyams, N. (1986). Language acquisition and the theory of parameters Dordrecht: Reidel. Hyltenstam, K., & Abrahamsson, N. (2003). Maturational constraints in SLA. In C.J. Doughty, & M.H. Long (E ds.), The handbook of second language acquisition (pp. 539 88). Oxford: Blackwell. Ivanov, I. (2009). Topicality and clitic doubling in L2 Bulgarian: a test case for the interface hypothesis. In M. Bowles, T. Ionin, S. Montrul, & A. Tremblay (Eds.), Proce edings of the 10th Generative Approaches to Second Language Acquisition Conference (GASLA 2009) (pp. pp. 17 24). Somerville, MA: Cascadilla Proceedings Project. Iverson, M., Kempchinsky, P., & Rothman, J. (2008). Interface vulnerability and knowledge of th e subjunctive indicative distinction with negated epistemic predicates in L2 Spanish. EUROSLA Yearbook, 8 135 163. Jackendoff, R. (2002). Foundations of Language: Brain, Meaning, Grammar, Evolution. Oxford: Oxford University Press. Jaeggli, O. (1982). To pics in Romance Syntax. Dordrecht: Foris. Jaeggli, O., & Hyams, N. (1988). Morphological uniformity and the setting of the null subject parameter. In J. Blevins, & J. Carter (Eds.), Proceedings of NELS 18 GLSA (pp. 238 223). University of Massachusetts, Am herst.

PAGE 299

299 Johnson, J.S., & Newport, E.L. (1989). Critical period effects in second language learning: The influence of maturational state on the acquisition of English as a second language Cognitive Psychology, 21, 60 99. Judy, T., & Feizmohammadpour, A. (2 When Parsing Failures Fail: Referential Subjects Generative Approaches to Language Acquisition North America 5. October 11 12, 2012; Lawrence, KS. Just, M.A., & Carpente r, P.A. ( 1980). A theory of reading: from eye fixations to comprehension. Psychological Review, 87, 329 354. Just, M.A., Woolley, J.D., & Carpenter, P.A. (1982). Paradigms and processes in reading comprehension. Journal of Experimental Psychology: General, 111 (2), 228 238. Kaan, E., & Swaab, T. (2003). Repair, revision, and complexity in syntactic analysis: an electrophysiological differentiation. Journal of Cognitive Neuroscience, 15 (1), 98 110. Kanno, K. ( 1997). The acquisition of null and overt pronomin als in Japanese by English speakers. Second Language Research, 13, 265 287. Kanno, K. (1998). Consistency and variation in second language acquisition. Second Language Research, 14, 376 388. Karimi, S. (2005). A Minimalist Approach to Scrambling: Evidence from Persian Berlin: de Gruyter. Keating, G.D. (2009). Sensitivity to violations of gender agreement in native and nonnative Spanish: An eye movement investigation. Language Learning, 59 (3), 503 535. Kissock, M. (1995). Reflexive middle Constructions and Verb Raising in Telugu Doctoral Dissertation, Concordia University. Kotz, S. (2009). A critical review of ERP and fMRI evidence on L2 syntactic processing. Brain & Language, 109 68 74. Krashen, S. (1973). Lateralization, language learning, and the critic al period. Language Learning, 23, 63 74. Krashen, S. (1975). The critical period for language acquisition and its possible bases. In D. Aaronson, & R. Rieber (Eds.), Developmental Psycholinguistics and Communicative Disorders ( pp. 211 224). New York: New York Academy of Science. Krashen, S. (1976). Cerebral asymmetry. In H. Whitaker, & A. Whitaker (Eds.), Studies in Neurolinguistics, Vol. 2 (pp. 157 191). New York: Academic Press,

PAGE 300

300 Kroll, J.F., & Stewart, E. (1994). Category interference in translation and picture naming: Evidence for asymmetric connections between bilingual memory representations. Journal of Memory and Language 33, 149 174. Kroll, J., Bogulski, C., & McClain, R. (2012) Psycholinguistic perspectives on second language learning and bilingua lism: The course and consequence of cross language competition. Linguistic Approaches to Bilingualism, 2 (1), 1 24. Kroll, J.F., Bobb, S., & Wodniekca, Z. (2006). Language selectivity is the exception, not the rule: Arguments against a fixed locus of langu age selection in bilingual speech. Bilingualism: Language and Cognition, 9, 119 135. Lapidus Shin, N. (2006). The Development of Null vs. Overt Subject Pronoun Expression in Monolingual Spanish speaking Children: The Influence of Continuity of Reference. D octoral dissertation, The City University of New York. Lapidus Shin, N. (2012). Variable Use of Spanish Subject Pronouns by Monolingual Children in Mexico. In K. Geeslin, & M. Daz Campos (Eds.), Selected Proceedings of the 14th Hispanic Linguistics Sympo sium (pp. 130 141). Somerville, MA: Cascadilla Proceedings Project Lapidus Shin, N. & Cairns, H.S. (2011). The Development of NP Selection in School Age Children: Reference and Spanish Subject Pronouns. Language Acquisition 19 (1), 3 38 Lardiere, D. ( 2000 ). Mapping features to forms in second language acquisition. In J. Archibald (Ed.), Second language acquisition and linguistic theory (pp. 102 129). Oxford: Blackwell. Lardiere, D. ( 2007). Acquiring (or assembling) functional categories in second language acquisition. In A. Belikova, L. Meroni, & M. Umeda, (Eds.), Proceedings of the 2nd Conference on Generative Approaches to Language Acquisition North America (GALANA) (pp. 233 244). Somerville, MA: Cascadilla Proceedings Project. Lardiere, D. ( 2008). Featur e assembly in second language acquisition. In J. Liceras, H. Goodluck, & H. Zobl (Eds.), The role of formal features in second language acquisition (pp. 107 140). New York: Lawrence Erlbaum. Lardiere, D. (2009). Some thoughts on the contrastive analysis of features in second language acquisition. Second Language Research, 25 (2), 173 227. Lardiere, D. (2011). Who is the interface hypothesis about? Linguistic Approaches to Bilingualism 1 (1), 48 53. Leeser, M.J., Brandl, A., & Weissglass, C. (2011). Task effe cts in second language sentence processing research. In K. McDonough, & P. Trofimovich (Eds.), Insights from psycholinguistics: Applying priming research to L2 learning and teaching (pp. 179 198). Amsterdam: John Benjamins.

PAGE 301

301 Lenci, A., & Bertinetto, P.M. ( 2000). Aspect, Adverbs and Events: Habituality and Perfectivity. In J. Higginbotham, F. Pianesi, & A.C. Varzi (Eds.), Speaking of Events (pp. 65 287). Oxford/New York: Oxford University Press. Lenneberg, E. (1967). Biological foundations of language New York, NY: Wiley. Liceras, J.M., Fernndez Fuertes, R., & Alba de la Fuente, A. (2012). Overt subjects and copula omission in the Spanish and the English grammar of English Spanish bilinguals: On the locus and directionality of interlinguistic influence. Fi rst Language, 32 (1 2), 88 115. Liceras, J.M., Fernndez Fuertes, R., & Prez Tattam, R. (2008). Null and overt subjects in the developing grammars (L1 English/L1 Spanish) of two bilingual twins. In C. Prez Vidal, M. Juan Garau, & A. Bel (Eds.), A portrai t of the young in the new multilingual Spain (pp. 111 134). Clevendon: Multilingual Matters Long, M. (2005). Problems with Supposed Counter Evidence to the Critical Period Hypothesis. International Review of Applied Linguistics, 43 (4), 287 317. Lpez Orna t, S. (1994). La adquisicin de la lengua espaola Madrid, Siglo XXI. Lozano, C. ( 2002). The interpretation of overt and null pronouns in non native Spanish. In H. Marsden, S. Pourcel, & M. Whong Barr (Eds.). Durham Working Papers in Linguistics, 8 53 6 6. Lozano, C. (2006). Focus and Split intransitivity: the Acquisition of Word Order Alternations in Non Native Spanish. Second Language Research, 22 145 187. Lozano, C., & Mendikoetxea, A. (2010). Interface conditions on postverbal subjects: a corpus stu dy of L2 English. Bilingualism, Language and Cognition, 13 (4), 475 497. Lujn, M. (1987). Los pronombres implcitos y explcitos del espaol. Revista Argentina de Lingstica, 3 19 54. Lujn, M. (1999). Expresin y omisin del pronombre personal. In I. Bo sque, & V. Demonte, (Eds.), Gramtica descriptiva de la lengua espaola (pp. 1275 1315). Madrid: Espasa Calpe. Luke, K. K., Liu, H. L., Wai, Y. Y., Wan, Y. L., & Tan, L. H. (2002). Functional anatomy of syntactic and semantic processing in language compreh ension. Human Brain Mapping, 16, 133 145. Lust, B. (2006). Child Language: Acquisition and Growth. Cambridge: Cambridge University Press. Margaza, P., & Bel, A. (2006). Null Subjects at the Syntax Pragmatics Interface: Evidence from Spanish Interlanguage o Archibald, (Eds.), Proceedings of the 8th Generative Approaches to Second Language

PAGE 302

302 Acquisition Conference (GASLA 2006) (pp. 88 97). Somerville, MA: Cascadilla Proceedings Project. Marian, V., & Spivey M.J. (2003). Competing activation in bilingual language processing: Within and between language competition. Bilingualism: Language and Cognition 6 97 115. Marinis, T. (2010). Using on line processing methods in language acquisition research. In E. Blo m, & S. Unsworth (Eds.), Experimental Methods in Language Acquisition Research (pp. 139 162). Amsterdam: John Benjamins. Martnez Sanz, C. (2011). Null and overt subjects in a variable system: The case of Dominican Spanish. Doctoral Dissertation, Universi ty of Ottawa. Matos Amaral, P., & Schwenter, S.A. (2005). Contrast and the (Non ) Occurrence of Subject Pronouns. In D. Eddington, Selected Proceedings of the 7th Hispanic Linguistics Symposium (pp. 116 127). Somerville, MA: Cascadilla Proceedings Project. McLaughlin, J., Osterhout, L., & Kim, A. (2004). Neural correlates of second language word learning: Minimal instruction produces rapid change. Nature Neuroscience, 7 (7) 703 704. Menndez Benito, P. (2001).Aspect and Adverbial Quantification in Spanish Manuscript, University of Massachusetts, Amherst. Mitchell, D.C., & Green, D.W. (1978). The effects of context and content on immediate processing in reading. Quarterly Journal of Experimental Psychology, 30 (4), 609 636. Montalbetti, M. (1984). After Bin ding: On the interpretation of pronouns. Doctoral dissertation, MIT. Montrul. S. (2000). Causative psych verbs in Spanish L2 acquisition. In R.P. Leow, & C. Sanz (Eds.), Spanish Applied Linguistics at the Turn of the Millennium (pp. 97 118). Somerville, MA : Cascadilla Press. Montrul, S. (2002). Incomplete Acquisition and Attrition of Spanish Tense/aspect Distinctions in Adult Bilinguals. Bilingualism: Language and Cognition 5 39 68. Montrul, S. (2004). Subject and object expression in Spanish heritage spe akers: A case of morpho syntactic convergence. Bilingualism: Language and Cognition 7 125 142. Montrul, S. (2005). Second language acquisition and first language loss in adult early bilinguals: Exploring some differences and similarities. Second Language Research, 21 (3), 199 249. Montrul, S. (2009). Knowledge of Tense Aspect and Mood in Spanish Heritage Speakers. International Journal of Bilingualism, 13 (2), 239 269. Montrul, S. (2011). Interfaces and incomplete acquisition. Lingua, 121 (4), 591 604.

PAGE 303

303 Montr ul, S., & Louro Rodrguez, C. (2006). Beyond the syntax of the null subject parameter. In V. Torrens, & L. Escobar (Eds.), The acquisition of syntax in Romance languages (pp. 401 418). Amsterdam: John Benjamins. Montrul, S., & Polinsky, M. (2011). Why not heritage speakers? Linguistic Approaches to Bilingualism, 1, 58 62. Neeleman, A., & Weerman, F. (1997). L1 and L2 word order acquisition. Language Acquisition 6 (2), 125 170. How children learn language Cambridge, UK: Cambridge Univer sity Press. Oates, J., & Grayson, A. (2004). Cognitive and Language Development in Children. Malden, MA: Blackwell. Ordez, F. (1997). Word order and clause structure in Spanish and other romance languages. Doctoral dissertation, New York, CUNY. Ordez, F. & Trevio, E. (1999). Left dislocated subjects and the pro drop parameter: A case study of Spanish. Lingua, 107, 39 68. Osterhout, L., McLaughlin, J., Pitknen, I., Frenck Mestre, C., & Molinaro, N. (2006). Novice learners, longitudinal designs, and e vent related potentials: A means for exploring the neurocognition of second language processing. Language Learning, 56, 199 230. Otero, C. (1996). Head Movement, Cliticization, Precompilation, and Word Insertion. In R. Friedin (Ed.), Current Issues in Comp arative Grammar (pp. 296 337). Dordrecht: Kluwer Academic Publishers. Otheguy, R., & Zentella, A.C. (2012). Spanish in New York: Language contact, dialectal leveling, and structural continuity. Oxford: Oxford University Press. Otheguy, R., Zentella, A.C., & Livert, D. (2007). Language and Dialect Contact in Spanish in New York: Toward the Formation of a Speech Community. Language, 83, 770 802. Oyama, S.C. (1976). A Sensitive Period for the Acquisition of a Phonological System Journal of Psycholinguistic Re search, 5, 261 283. Paradis, J., & Navarro, S. (2003). Subject realization and crosslinguistic interference in the bilingual acquisition of Spanish and English: what is the role of input? Journal of Child Language, 30 371 393. Pascual y Cabo, D., Lingwall A., & Rothman, J., (2012). Applying the Interface Hypothesis to Heritage Speaker (HS) Acquisition: Evidence from Spanish Mood. In A.K. Biller, E.Y. Chung, & A.E. Kimball (Eds.) Proceedings of the 36th annual Boston University Conference on Language Devel opment (BUCLD 36) (pp. 437 448). Somerville, MA: Cascadilla Press.

PAGE 304

304 Perani, D., Abutalebi, J., Paulesu, E., Brambati, S., Scifo, P., Cappa, S., & Fazio, F. (2003). The role of age of acquisition and language usage in early, high proficient bilinguals: An f MRI study during verbal fluency. Human Brain Mapping, 19 170 182. Perani, D., Dehaene, S., Grassi, F., Cohen, L., Cappa, S.F., Dupoux, E., Fazio. F. & Mehler, J. (1996). Brain processing of native and foreign languages. NeuroReport, 7 2439 2444. Perani, D., Paulesu, E., Sebastian Galles, N., Dupoux, E., Dehaene, S., Bettinardi, V., Cappa, S. F., Fazio, F. & Mehler, J. (1998). The bilingual brain: proficiency and age of acquisition of the second language. Brain, 121, 1841 1852. Prez Leroux, A.T. (1998). T he acquisition of mood selection in Spanish relative clauses. The Journal of Child Language, 25, 585 604. Prez Leroux, A.T., & Glass, W. (1999). Null anaphora in Spanish second language acquisition: probabilistic versus generative approaches. Second Lan guage Research, 15 220 249. Perlmutter, D. (1971). Deep and surface constraints in syntax. New York: Hold, Rinehart and Winston. Phillips, C. (1996). Order and Structure Doctoral dissertation, MIT. Picallo, M.C. (1994). Catalan possessive pronouns: the Avoid Pronoun Principle revisited. Natural Language and Linguistic Theory, 12 259 299. Picallo, M.C. (1998). On the extended projection principle and null expletive subjects. Probus, 10 219 241. Pinango, M., & Burkhardt, P. (2005). Pronominal interpretat ion and the syntax discourse interface: Real time comprehension and neurological properties. In A. Branco, T. McEnery, & R. Mitkov (Eds.), Anaphora processing: Linguistic, cognitive and computational models (pp. 221 238). Amsterdam: John Benjamins. Pinango M., Burkhardt, P., Brun, D., & Avrutin, S. (2001). The architecture of the sentence processing system: The case of pronominal interpretation. Paper presented at the Cognitive Science Annual Meeting. Prada P rez, A. de (2009). Subject expression in Minorc an Spanish: Consequences of contact with Catalan Doctoral dissertation, Pennsylvania State University. Prada Prez, A. de (2010). Subject Position in Spanish in Contact with Catalan: Language Similarity vs. Interface Vulnerability. In M. Iverson, I. Ivano v, T. Judy, J. Rothman, & R. Slabakova (Eds.), Proceedings of the 2009 Mind/Context Divide Workshop (pp. 104 115). Somerville, MA: Cascadilla Proceedings Project. Prada Prez, A. de, & Pascual y Cabo, D. (2011). Invariable gusta in the Spanish of Heritage Speakers in the US. In J. Herschensohn, & D. Tanner (Eds.), Proceedings of the

PAGE 305

305 11thGenerative Approaches to Second Language Acquisition (GASLA 2011) (pp. 110 120). Somerville, MA: Cascadilla Proceedings Project. Prada Prez, A. de, & Pascual y Cabo, D. (20 12). Interface Heritage speech across proficiencies: Unaccusativity, focus, and subject position in Spanish. In K. Geeslin, & M. Daz Campos (Eds.), Proceedings of the Hispanic Linguistic Symposium (HLS) (pp. 308 318). Somerville, MA: Cascadilla Proceeding s Project. Quesada, M.L., & Blackwell, S. (2009). The L2 acquisition of null and overt Spanish subject pronouns: A pragmatic approach. In J. Collentine, M. Garca, B. Lafford, & F. Marcos Marn (Eds.), Selected Proceedings of the 11th Hispanic Linguistics Symposium (pp. 117 130). Somerville, MA: Cascadilla Proceedings Project. Ramchand, G., & Reiss, C. (2007). The Oxford handbook of linguistic interfaces. Oxford: Oxford University Press. Reinhart, T. (2006). Interface strategies: Reference set computation Cambridge, MA: MIT Press. Rigau, G. (1986). Some remarks on the nature of strong pronouns in null subject languages. In I. Bordelois, H. Contreras, & K. Zagona (Eds.), Generative Studies in Spanish Syntax (pp.143 163). Dordrecht: Foris. Rigau, G. (1988). Strong pronouns. Linguistic Inquiry 19 503 511. Rizzi, L. (1982). Issues in Italian Syntax Dordrecht: Reidel. Rizzi, L. (1986). Null Objects in Italian and the Theory of Pro. Linguistic Inquiry, 17 501 557. Rizzi, L. (1997). A parametric approach to c omparative syntax: properties of the pronominal system. In L. Haegeman (Ed.), The New Comparative Syntax (pp. 268 285) London: Longman. Rizzi, L. (2003). On the form of chains: Criterial positions and ECP effects. In L.L Cheng, & N. Corver (Eds.), Wh Move ment: Moving on (pp. 97 135). Cambridge, MA: MIT Press. Rossi, S., Gugler, M.F. Friederici, A.D., & Hahne, A. (2006). The Impact of Proficiency on Syntactic Second language Processing of German and Italian: Evidence from Event related Potentials. Journal of Cognitive Neuroscience, 18 (12), 2030 2 48. Rothman, J. (2005). On Adult Accessibility to Universal Grammar in Second Language (L2) Acquisition: Some Evidence from the L2 Acquisition of Spanish Doctoral dissertation, UCLA. Rothman, J. (2007). Sometimes Discussion of L2 Morphological Production and Its Use as a Competence Measurement. Applied Linguistics 28 (4), 609 615.

PAGE 306

306 Rothman, J. (2008a). How pragmatically odd! Interface delays and pronominal subje ct distribution in L2 Spanish. Studies in Hispanic and Lusophone Linguistics 1 (2), 317 339. Rothman, J. (2008b). Why All Counter Evidence to the Critical Period Hypothesis in Second Language Acquisition is not Equal or Problematic: Implications for SLA L anguage and Linguistics Compass, 2 (6), 1063 1088. Rothman, J. (2008c). Aspectual Morphology Use in Adult L2 Spanish and the Competing Systems Hypothesis: When Pedagogical and Linguistic Rules Conflict. Languages in Contrast, 8 (1), 74 106. Rothman J. (2009 ). Pragmatic Deficits with Syntactic Consequences?: L2 pronominal subjects and the Syntax Pragmatics Interface. Journal of Pragmatics 41 (5), 951 973. Rothman, J., & Guijarro Fuentes, P. (2010). Input Quality Matters: Some Comments on Input Type and Age Ef fects in Adult SLA. Applied Linguistics, 31 (2), 301 306. Rothman, J., & Iverson, M. (2007). On L2 Clustering and Resetting the Null Subject Parameter in L2 Spanish: Implications and Observations. Hispania, 90 (2), 329 342. Rothman, J. & Iverson, M. (2008). Poverty of the Stimulus and SLA Epistemology: Considering L2 Knowledge of Aspectual Phrasal Semantics. Language Acquisition, 15 270 314. Rothman, J., & Slabakova, R. (2011). The mind context divide: On acquisition at the linguistic interfaces. Lingua, 121 (4), 568 576. Rschemeyer, S.A., Zysset, S., & Friederici, A.D. (2006). Native and non native reading of sentences: An fMRI experiment. NeuroImage, 31, 354 365. R schemeyer, S.A., Fiebach, C.J., Kempe, V., & Friederici, A.D. (2005). Processing lexical sem antic and syntactic information in first and second language: fMRI evidence from German and Russian. Human Brain Mapping, 25 (2) 266 286. Sabourin, L. ( 2003) Grammatical gender and second language processing Groningen Dissertations in Linguistics, volume 42. Groningen: CLCG. Sagarra, N. & Herschensohn, J. (2011). Asymmetries in Gender and Number Agreement Processing in Late Bilinguals. In L.A. Ortiz Lpez (Ed.), Selected Proceedings of the 13th Hispanic Linguistics (pp. 169 177). Somerville, MA: Cascadilla Proceedings Project. Sakai, K.I. (2005). Language Acquisition and Brain Development. Science, 310 (5749), 815 819. Schwartz, B.D. (1998). On two hypotheses of 'Transfer' in L2A: Minimal Trees and Absolute L1 Influence." In S. Flynn, G. Martohardjono, & W. O'Neil (Eds.), The Generative Study of Second Language Acquisition (pp. 35 59). Mahwah, NJ: Lawrence Erlbaum.

PAGE 307

307 Schwartz, B.D. (2011). Parsing up the Interface Hypothesis. Linguistic Approaches to Bilingualism, 1 (1), 84 88. Schwartz, B.D., & Sprouse, R.A. (1994). Word order and nominative case in nonnative language acquisition: A longitudinal study of (L1 Turkish) German interlanguage. In T. Hoekstra, & B.D. Schwartz (Eds.), Language acquisition studies in generative grammar (pp. 317 368). Amsterdam: John B enjamins. Schwartz, B.D., & Sprouse, R.A. (1996). L2 cognitive states and the full transfer full access model. Second Language Research, 12 40 72. Schwartz, B.D., & Sprouse, R.A. (2000). When Syntactic Theories Evolve: Consequences for L2Acquisition Resea rch. In J. Archibald (Ed.), Second Language Acquisition and Linguistic Theory (pp. 156 186). Malden, MA: Blackwell. Schwartz, A.I., Kroll, J.F., & Diaz, M. (2007). Reading words in Spanish and English: Mapping orthography to phonology in two languages. La nguage and Cognitive Processes 22 ,106 129 Schneider, W., Eschmann, A., & Zuccolotto, A. (2002). E Prime [Computer software]. Pittsburgh, PA: Psychology Software Tools Inc. Serratrice, L., Sorace, A., & Paoli, S. (2004). Crosslinguistic influence at the s yntax pragmatics interface: subjects and objects in English Italian bilingual and monolingual acquisition. Bilingualism: Language and Cognition, 7, 183 205. sp ecificity and genericity: Evidence from metalinguistic awareness. Bilingualism: Language and Cognition, 12 1 19. Silva Corvaln, C. (1982). Subject expression and placement in Mexican American Spanish. In J. Amastae, & L. Elas Olivares (Eds.), Spanish i n the United States: Sociolinguistic aspects (pp. 93 120). Cambridge/New York: Cambridge University Press. Silva Corvaln, C. (1994). Language contact and change New York: Oxford University Press. Slabakova, R. (2006). Is there a critical period for sema ntics? Second Language Research, 22 302 338. Slabakova, R. (2008). Meaning in the Second Language. Berlin: Mouton de Gruyter. Slabakova, R. (2009). What is easy and what is hard to acquire in a second language? In H. Caunt Nulton, M. Bowles, T. Ionin, S. Montrul, & A. Tremblay (Eds.), Proceedings of the 10th Generative Approaches to Second Language Acquisition Conference (GASLA 2009) (pp. 280 294). Somerville, MA: Cascadilla Proceedings Project. Slabakova, R. (2011). Which features are at the syntax pragma tics interface? Linguistic Approaches to Bilingualism, 1 (1), 89 93.

PAGE 308

308 Slabakova, R., & Ivanov, I. (2011). A more careful look at the syntax discourse interface. Lingua, 121 637 651. Slabakova R., Kempchinsky, P. & Rothman, J. (2012). Clitic doubled Left D islocation and Focus Fronting in L2 Spanish: A case of successful acquisition at the syntax discourse interface. Second Language Research 28 (3), 319 343. Snyder, W. (2007). Child Language: The Parametric Approach Oxford, UK: Oxford University Press. Sorace, A. (1993). Incomplete vs. divergent representations of unaccusativity in nonnative grammars of Italia n. Second Language Research 9 22 47. Sorace, A. (2000). Differential effects of attrition in the L1 syntax of near native L2 speakers. In. S.C. Howell, S. A. Fish, & T, Keith Lucas (Eds.) Proceedings of Boston University Conference on Language Developmen t, BUCLD 24 (pp. 719 725) Somerville, MA: Cascadilla Press. Sorace, A. (2005). Selective optionality in language development. In L. Cornips, & K.P. Corrigan (Eds.), Syntax and variation: Reconciling the biological and the social (pp. 55 80). Amsterdam: Joh n Benjamins. Sorace, A. (2006). Gradedness and optionality in mature and developing grammars. In G. Fanselow, C. Fry, R. Vogel, & M. Schlesewsky (Eds.), Gradience in Grammar (pp. 106 123). Oxford: Oxford University Press. Sorace, A. (2011). Pinning down Linguistic Approaches to Bilingualism, 1 (1), 1 33. Sorace A ( 2012) Pinning down the concept of "interface" in bilingualism: a reply to peer commentaries. Linguistic Approaches to Bilingualism, 2, 209 216. Sora ce, A., & Filiaci, F. (2006). Anaphora resolution in near native speakers of Italian. Second Language Research, 22 339 368. Sorace, A., & Serratrice, L. (2009). Internal and external interfaces in bilingual language development: Beyond structural overlap. International Journal of Bilingualism, 13 195 210. Sorace, A., Serratrice, L. Filiaci, F., & Baldo, M. (2009). Discourse conditions on subject pronoun realization: Testing the linguistic intuitions of bilingual children. Lingua, 119 460 477. Sprouse, R. (2007). The bankruptcy of the stimulus. In K. Ud Deen, J. Nomura, B. Schulz, & B.D. Schwartz (Eds.), The Proceedings of the Inaugural Conference on Generative Approaches to Language Acquisition North America (pp. 51 63). University of Connecticut Occasion al Papers in Linguistics 4.

PAGE 309

309 Suer, M. (1982). Syntax and semantics of Spanish presentational sentence types. Washington, DC: Georgetown University Press. Suer, M. (1994). V movement and the licensing of argumental wh phrases in Spanish. Natural Language a nd Linguistic Theory 12, 335 372. Tokowicz, N., & MacWhinney, B. (2005). Implicit and explicit measures of sensitivity to violations in second language grammar: An event related potential investigation. Studies in Second Language Acquisition, 27, 173 204 Toribio, A.J. (2000). Setting parametric limits on dialectal variation in Spanish. Lingua, 110 (5), 315 341. Travis, C. (2005). The yo yo effect: Priming in subject expression in Colombian Spanish. In R. Gess, & E.J. Rubin (Eds.), Selected papers from the 34th Linguistic Symposium on Romance Languages (LSRL) (pp. 329 349). Amsterdam: John Benjamins. Travis, C. (2007). Genre effects on subject expression in Spanish: Priming in narrative and conversation. Language Variation and Change, 19, 101 135. Tsimpli, I.M., & Dimitrakopoulou, M. (2007). The Interpretability Hypothesis: Evidence from wh interrogatives in second language acquisition. Second Language Research, 23 (2), 215 242. Tsimpli, I.M., & Sorace, A. (2006). Differentiating interfaces: L2 performance i n syntaxsemantics and syntaxdiscourse phenomena. In D. Bamman, T. Magnitskaia, & C. Zaller (Eds.), Proceedings of the 30th annual Boston University Conference on Language Development, BUCLD 30 (pp. 653 664). Somerville, MA: Cascadilla Press. Tsimpli, I., Sorace, A., Heycock, C., & Filiaci, F. (2004). First language attrition and syntactic subjects: a study of Greek and Italian near native speakers of English. International Journal of Bilingualism, 8 257 277. Valenzuela, E. (2005). L2 ultimate attainment and the syntax discourse interface: the acquisition of topic constructions in non native Spanish and English. Doctoral dissertation, McGill University. Valenzuela, E. (2006). L2 end state grammars and incomplete acquisition of the Spanish CLLD construction s. In R. Slabakova, S. Montrul, & P. Prvost (Eds.), Inquiries in linguistic development: in honor of Lydia White (pp. 283 304). Amsterdam: John Benjamins. VanPatten, B. Keating, G. & Leeser, M. (2012). Missing verbal inflections as a representational prob lem Evidence from self paced reading. Linguistic Approaches to Bilingualism, 2 (2), 109 140 Wartenburger, I., Heekeren, H.R., Abutalebi, J., Cappa, S.F., Villringer, A., & Perani, D. (2003). Early setting of grammatical processing in the bilingual brain. Neuron, 37 1 20.

PAGE 310

310 Weber Fox, C., & Neville, H. (1996). Maturational Constraints on Functional Specializations for Language Processing: ERP and Behavioral Evidence in Bilingual Speakers. Journal of Cognitive Neuroscience, 8 (3), 231 256. Weir, C. (2005). La nguage Teaching and Validation. Houndmills: Palgrave Macmillan. Wexler, K. (1993). The subset principle is an intentional principle. In E. Reuland, & W. Abraham (Eds.), Knowledge and Language: Orwell's Problem and Plato's problem. White, L. (1989). Univers al Grammar and Second Language Acquisition Amsterdam: John Benjamins. White, L. (2003a). Second Language Acquisition and Universal Grammar. Cambridge: Cambridge University Press. White, L. (2003b). On the nature of interlanguage representation: Universal Grammar in second language. In C. Doughty, & M. Long (Eds.), The handbook of second language acquisition (pp. 19 42). Oxford: Blackwell. White, L. (2008). Different? Yes. Fundamentally? No. Definiteness Effects in the L2 English of Mandarin Speakers. In R. Slabakova, J. Rothman, P. Kempchinsky, & E. Gavruseva (Eds.), Proceedings of the 9th Generative Approaches to Second Language Acquisition Conference (pp. 251 261). Somerville, MA: Cascadilla Proceedings Project. White, L. (2009). Grammatical theory: inter faces and L2 knowledge. In W.C. Ritchie, & T.K. Bhatia (Eds.), The New Handbook of Second Language Acquisition (pp. 49 65). Bingley: Emerald Group Publishing, Ltd. White, L. (2011a). Second language acquisition at the interfaces. Lingua, 121 577 590. Whit e, L. (2011b). The Interface Hypothesis: How far does it extend? Linguistic Approaches to Bilingualism, 1, 108 110. White, L., & Genesee, F. (1996). How native is near native? The issue of ultimate attainment in adult second language acquisition. Second La nguage Research, 12 233 265. White, L., Valenzuela, E., Kozlowska MacGregor, M, & Leung, Y K.I. (2004). Gender and number agreement in nonnative Spanish. Applied Psycholinguistics, 25, 105 133. Wilson, F., Keller, F., & Sorace, A. (2009). The antecedent preferences of anaphoric demonstratives and personal pronouns in L2 German. Proceedings of the Boston University Conference on Language Development 33 (pp. 634 645). Somerville, MA: Cascadilla Press. Winkler, I., Kujala, T., Tiitinen, H., Sivonen, P., Alku P., Lehtokoski, A., et al. (1999). Brain responses reveal the learning of foreign language phonemes. Psychophysiology, 36 (5), 638 642.

PAGE 311

311 Witzel, N., Witzel, J., & Forster, K. (2012). Comparisons of online reading paradigms: Eye tracking, moving window, and maze. Journal of Psycholinguistic Research 41 105 128. Zapata, G., Snchez, L., & Toribio, A.J. (2005). Contact and contracting Spanish. International Journal of Bilingualism 3 4 377 395. Zubizarreta, M.L. (1998). Prosody, focus, and word order. Cambr idge, MA: MIT Press.

PAGE 312

312 BIOGRAPHICAL SKETCH Tiffany Judy received her PhD in Hispanic Linguistics from the University of Florida in 2013. She holds a BA in Spanish from the University of Northern Iowa and an MA in Hispanic Linguistics from the Universit y of Iowa. Her primary research interest concerns second language acquisition, spanning such bilingual populations as child bilinguals, adult second and third language speakers and heritage language acquisition. Previous and current research projects exami ne the acquisition of morphological, syntactic, semantic and discourse related properties in languages such as Spanish, Portuguese and English. Other current research projects include examining the acquisition of English as a second language by native spea kers of Farsi and an edited volume regarding the acquisition of Spanish by native speakers of less commonly studied languages such as Nahuatl, Romanian and Serbian. In addition to scholarly research, Tiffany is active in serving her university and the fiel d of second language acquisition as well as teaching. She has served as a peer reviewer for leading linguistics journals, aided in the organization of several linguistics conferences and mentored undergraduate linguistics students.